Novel infection process in the indeterminate root nodule symbiosis between Chamaecytisus proliferus (tagasaste) and Bradyrhizobium sp.

Authors

  • M. C. Vega-Hernández,

    1. Departamento de Microbiología y Biología Celular. Facultad de Farmacia. Universidad de La Laguna. 38071 La Laguna. Tenerife. Canary Islands. Spain;
    Search for more papers by this author
  • R. Pérez-Galdona,

    1. Departamento de Microbiología y Biología Celular. Facultad de Farmacia. Universidad de La Laguna. 38071 La Laguna. Tenerife. Canary Islands. Spain;
    Search for more papers by this author
  • F. B. Dazzo,

    1. Department of Microbiology and Center for Microbial Ecology. Michigan State University, East Lansing. MI 48824, USA
    Search for more papers by this author
  • A. Jarabo-Lorenzo,

    1. Departamento de Microbiología y Biología Celular. Facultad de Farmacia. Universidad de La Laguna. 38071 La Laguna. Tenerife. Canary Islands. Spain;
    Search for more papers by this author
  • M. C. Alfayate,

    1. Departamento de Microbiología y Biología Celular. Facultad de Farmacia. Universidad de La Laguna. 38071 La Laguna. Tenerife. Canary Islands. Spain;
    Search for more papers by this author
  • M. León-Barrios

    Corresponding author
    1. Departamento de Microbiología y Biología Celular. Facultad de Farmacia. Universidad de La Laguna. 38071 La Laguna. Tenerife. Canary Islands. Spain;
    Search for more papers by this author

Author for correspondence: M. León-Barrios Tel: +34 922318481 Fax: +34 922630095 Email:mileonba@ull.es

Summary

  •  The main characteristics of the symbiosis of tagasaste (Chamaecytisus proliferus ssp. proliferus var. palmensis), a woody legume forming N2-fixing indeterminate nodules in response to infection by strains of Bradyrhizobium sp. (Chamaecytisus), are reported here.
  •  The infection process in this legume was examined by bright field, phase contrast and transmission electron microscopy, and was found to be unlike any other previously described.
  •  First steps in the infection process involve initiation of infection threads within deformed root hairs and induction of foci of host-cell divisions in the inner root cortex. However, infection of root hairs aborts early, and instead, the bacteria use the crack-entry mode of host infection, whereby they penetrate the periphery of the emerging nodule through an intercellular route, eventually infecting host nodule cells directly through altered cell walls. No successful infection threads were detected at any stage of primary-host infection or nodule invasion. Indeterminate nodules were mainly formed on unbranched areas of lateral roots.
  •  This is the first description of such a combination of events in an infection process in the Rhizobium–legume root-nodule symbiosis.

Introduction

The rhizobia (Rhizobium, Bradyrhizobium, Azorhizobium, Sinorhizobium,Mesorhizobium and Allorhizobium) are a group of Gram-negative soil bacteria that can infect legumes and establish a symbiotic relationship. As a result of this association, a new organ, the root-derived nodule, is formed. Within the nodule, rhizobia differentiate into endosymbiotic bacteroids that can fix atmospheric N2.

Two main modes of infection have been described in the root nodule symbiosis. The mode most thoroughly studied involves primary entry of bacteria into deformed root hairs through infection threads, which progress to the root cortex and eventually reach the emerging nodule primordia (Napoli & Hubbell, 1976; Callaham & Torrey, 1981; Newcomb, 1981; Hirsch, 1992). The other infection mode is called crack entry, in which the rhizobia invade the root interior through natural wounds on the epidermis (Napoli et al., 1975; Boogerd & van Rossum, 1997). In addition, a mode of direct infection through undamaged epidermis has been described for the hairless tree legume, Mimosa scabrella (de Faria et al., 1988). This has been proposed as a third mode of infection for woody legumes that neither produce root hairs regularly nor produce nodules associated with lateral roots (Sprent, 1989).

The best-studied mode of primary host infection takes place through infection thread formation (Inf), where the rhizobia induce a marked curling of the root hair tip (Hac), forming the typical ‘shepherd crook’ deformation. The bacteria infect the root through the formation of a tubular infection thread that grows to the base of the root hair cell and penetrates into the underlying root cortex, releasing the rhizobia into newly divided cortical cells. Collectively referred to as the Inf phenotype, this mode of infection can be subdivided into Iti (Infection Thread Initiation), Itr (Infection Thread development in Root hairs), and Itn (Infection Thread development in Nodules) phenotypes. This commonly yields two different nodule morphotypes depending on the host. These are the indeterminate nodules as in clover and alfalfa, and determinate nodules as in soybean and cowpea (Hirsch, 1992). This is a well-documented infection process.

The alternative mode of crack entry infection occurs by intercellular penetration between root cells. After colonizing the root epidermis, the bacteria invade the root cortex through natural wounds caused by splitting of the epidermis where young lateral roots or nodule primordia have been stimulated to develop and emerge. The crack entry mode of primary host infection has been described in various (sub)tropical legumes such as Aeschynomene (Napoli et al., 1975; Alazard & Duhoux, 1990), Arachis (Chandler, 1978; Boogerd & van Rossum, 1997), Stylosanthes (Chandler et al., 1982), Sesbaniarostrata (Dreyfus et al., 1984; Duhox, 1984; Ndoye et al., 1994; Rana & Krishnan, 1995), and Neptunia (Subba-Rao et al., 1995). Two types of crack entry can be distinguished depending on the mode of bacterial dissemination within the nodule. In the case of Aeschynomene, Arachis, and Stylosanthes (Napoli et al., 1975; Chandler, 1978; Chandler et al., 1982) the intercellular rhizobia directly invade some root cortical cells and their dissemination within the nodule takes place by division of the infected cells without involving infection threads. In the case of Sesbania and Neptunia, dissemination of the microsymbiont involves an initial intercellular spread, followed later by intracellular infection involving formation of true tubular infection threads that penetrate nodule cells and release the endosymbiotic bacteria at infection droplets that have protruded from localized eroded areas of the infection thread wall (Ndoye et al., 1994; Subba-Rao et al., 1995). Crack entry has also been described in the infection of the nonlegume Parasponia by Bradyrhizobium in which both mechanisms of dissemination appear to take place (Trinick, 1979; Lancelle & Torrey, 1984; Bender et al., 1987).

Tagasaste, Chamaecytisusproliferus (L. fil.) Link ssp. proliferus var. palmensis (Christ) Hansen & Sunding, is a temperate woody legume in the tribe Genisteae of the Papilionoideae subfamily endemic to the Canary Islands. This evergreen shrub can grow to a height of 5 m and is of great agronomic and ecological value (Pérez de Paz et al., 1986; Francisco-Ortega et al., 1991). Its high nutritive value and palatability makes this plant a suitable forage for grazing animals, pigs and poultry (Borens & Poppi, 1986; Dann & Trimmer, 1986; Pérez de Paz et al., 1986; Borens & Poppi, 1990; Oldham et al., 1991; Snook, 1996). For centuries, tagasaste has been used as fodder in the Canaries (Pérez de Paz et al., 1986). It was introduced in the last century in other parts of the world, especially in Australia and New Zealand (Snook, 1996) where it was misnamed tree-lucerne (European name for alfalfa) alluding to its comparative forage value with alfalfa. In Australia, tagasaste has been reported to be nodulated by strains of rhizobia and bradyrhizobia (Gault et al., 1994), but in the Canaries only the slow-growing bradyrhizobia isolated from endemic woody legumes have been found to nodulate tagasaste effectively (León-Barrios et al., 1991; Santamaría et al., 1997), typically forming cylindrical, indeterminate nodules. The legume host range of these Canarian bradyrhizobia also includes Macroptilium atropurpureum (siratro), where the mature nodules are spherical rather than elongated. Genomic studies indicate that some of the Canarian bradyrhizobia isolated from tagasaste and other endemic woody legumes form a distinctive group of strains that could constitute a new species of the genus Bradyrhizobium (Jarabo-Lorenzo et al., 2000).

The aim of this work was to document the developmental morphology of the infection and nodulation processes in tagasaste by two strains of Bradyrhizobium sp. (Chamaecytisus) indigenous to the Canary Islands. Portions of this work were presented at the 12th International Congress on Nitrogen Fixation (Vega-Hernández et al., 2000).

Materials and Methods

Bacterial and plant growth conditions

Bradyrhizobium sp. (Chamaecytisus) strain BTA-1 was isolated from root nodules of tagasaste, Chamaecytisusproliferus (L. fil.) Link ssp. proliferus var. palmensis (Christ) Hansen & Sunding (Acebes-Ginovés et al., 1991), and strain BGA-1 from nodules of Teline stenopetala (Webb & Berth) Webb & Berth. var. stenopetala (León-Barrios et al., 1991). Inocula were prepared by growing bacterial cultures to late exponential phase in shaken flasks containing yeast mannitol broth at 28°C.

In order to facilitate germination, tagasaste seeds were scarified with concentrated sulphuric acid for at least 30 min, and then sterilized for 15 min in 5% (v/v) aqueous sodium hypochlorite. Macroptilium atropurpureum seeds were similarly sterilized for 15 min in 5% (v/v) sodium hypochlorite. Surface-sterilized seeds were rinsed six times in sterile distilled water, and germinated on 1% (w/v) agar plates in darkness at room temperature. Young seedlings with roots approx. 1 cm long were removed from the germination plates and immersed for 2 h in a suspension of strain BTA-1 or BGA-1. Uninoculated seedlings were used as control.

The first steps of the infection process were examined using the Fåhraeus slide culture technique (Fåhraeus, 1957), and nodule development was examined on inoculated plants grown in Leonard jars containing a 1 : 1 mixture of sand and vermiculite. In both cases, nitrogen-free Fåhraeus solution was added as nutrient (Fåhraeus, 1957). Plants were incubated in a growth chamber programmed with a 16-h photoperiod, 70% rh, and a day/night temperature of 25°C and 17°C, respectively.

Microscopy analyses

Slide cultures were observed daily by brightfield and phase contrast microscopy to examine the early steps of the infection process. Foci of cortical meristems were visualized by the hypochlorite clearing/methylene blue staining procedure of Truchet et al. (1989). Segments of inoculated plants were cleared by vacuum-infiltration with commercial sodium hypochlorite solution diluted to 70% of its original volume with water, rinsed in water, stained briefly with an aqueous solution of methylene blue (0.2 mg ml−1) and examined by brightfield microscopy.

Mature nodules from 6–8-wk-old plants were fixed in ethanol-acetic acid-formaldehyde (90 : 5 : 5, by volume), infiltrated in Paraplast, sectioned to 10 µm, stained with safranine/fast green (Johansen, 1940) and examined by brightfield and phase contrast microscopy.

Mature nodules on 2–4-wk-old plants were fixed by vacuum infiltration for 2 h with 3% (v/v) glutaraldehyde in 0.1 M phosphate buffer (PB; pH 7.2) at room temperature. Submerged nodules were washed in PB, postfixed in 1% (w/v) osmium tetroxide in the same buffer for 2 h at 4°C, washed again in PB, dehydrated in a graded ethanol series and embedded in Spurr’s resin (Spurr, 1969). Semi-thin-sections obtained using glass knives were stained with toluidine blue and examined by brightfield microscopy. Ultra-thin sections obtained using a diamond knife were stained with aqueous uranyl acetate and lead citrate, and examined by TEM using Jeol 1010 (Tokyo, Japan), and Philips CM-10 electron microscopes (Hillsboro, Oregon, USA).

Results

Appearance of nodulated tagasaste plants

The typical morphology of an inoculated tagasaste plant after 1 month of growth under the conditions for this study and a closer view of its nodulated roots are shown in Fig. 1(a),(b), respectively. Nodules developed frequently on lateral roots but not on the main root or at the lateral root axils. Nodules were typically cylindrical when mature.

Figure 1.

Example of a typical nodulated tagasaste plant (Chamaecytisus proliferus) used in this study. (a) The entire plant was place in a Petri dish and photographed at low magnification. The arrows indicate nodulated lateral roots. Bar, 2 cm (main picture). Bar, 1.2 mm (inset). (b) Higher magnification stereomicrograph of the same lateral roots. Note the cylindrical shape of the mature indeterminate nodule (arrow) on the dense hairy root. Bar, 1 mm.

Light microscopy of aborted root hair infections

Both bacterial strains showed the same pattern of symbiotic infection with tagasaste. Bacteria polarly attached to the root hair tips (figure not shown), and induced various types of root hair deformation within 24–48 h after inoculation (Fig. 2a,b). Root hairs with typical markedly curled shepherd’s crooks were found on inoculated seedlings 48 h and older (Fig. 2a). At this point, the initial formation of refractile infection threads within deformed root hairs (Iti+) could also be distinguished, but these initial infections always aborted at or near the site of inception (Fig. 2b). No infection threads were found that had grown to the base of the root hair and/or penetrated to the subepidermal cortex (Itr). Further examination of 2 and 3 wk-old seedlings revealed no successful infection threads in the vicinity of emerged root nodules, or elsewhere within the entire areas of the bulging epidermis overlying nodule primordia on lateral roots at this time. Possibly related to the unusual Ini+ Inr phenotypes of this symbiosis was the frequent development of localized areas of wall erosion at the tips of root hairs on inoculated plants only (Fig. 2c). Occasionally, root hair cytoplasm leaked out from these localized erosions (Fig. 2d). These results are a further indication of unsuccessful root hair infections and suggest that these bradyrhizobia probably use an alternate mode of primary host infection in order to nodulate tagasaste effectively.

Figure 2.

Phase contrast micrographs illustrating various typical alterations in root hairs of tagasaste (Chamaecytisus proliferus) inoculated with bradyrhizobia. (a) Markedly curled (‘shepherd crook’) deformation at the root hair tip. (b) An aborted infection of a deformed root hair. Arrows point to the infection thread that arrested shortly after formation. (c) Refractile, circular eroded areas (arrow) at a root hair tip. (d) A lysed root hair with its cytoplasm expressed through a hole at the tip. Residual portions of the root hair protoplast have separated from the inner face of the cell wall. Bar, 12 µm.

Light microscopy of nodule development and histology

Brightfield microscopy indicated that the initial foci of host cell divisions induced by the symbiotic bacteria occurred in the inner cortex, as is typical of indeterminate nodule formation (Fig. 3a). These foci of cortical cell divisions developed more frequently in lateral roots than in the main root. The mitotic activity of these cortical cells gave rise to nodule primordia (Fig. 3b) that emerged as nodules (Figs 1a, 1b). Longitudinal sections of mature tagasaste nodules exhibited the typical histological organization of elongated, indeterminate nodules with apical meristem, infection and fixation zones, and a proximal senescence zone (Fig. 4).

Figure 3.

Brightfield micrographs of cleared/stained roots exhibiting development of foci of cortical cell divisions initiated in the inner root cortex (a, arrows) and a nodule primordium prior to emergence on a lateral root (b). Bar, 250 µm.

Figure 4.

Brightfield micrograph of a semi-thin longitudinal section of an 8-wk-old indeterminant tagasaste nodule. The apical meristem (m), the central zone of N2 fixation (fz) and the proximal senescence zone (sz) are labelled. Bar, 250 µm.

Ultrastructural morphology of the tagasaste nodule

Transmission electron microscopy confirmed that the nodule invasion stage of the infection process did not involve any infection threads. Instead, primary invasion of young nodules by the bacteria involved crack entry through intercellular spaces of the outer apical cell layers (Figs 5a,b, 6a). Deeper within the nodule, the rhizobia proliferated in enlarged intercellular spaces that served as foci to facilitate further dissemination of intercellular invasion. Bacteria entered other cell layers through the middle lamellae within the nodule (Fig. 6b) resulting in a network of intercellular infection (Fig. 6c). Accompanying the intercellular dissemination of the bacteria was the successive collapse of host cells in the apical layers of the nodule, causing the deformation and convolution of their cell walls (Fig. 7a). This collapse phenomenon was seen only in uninfected host cells neighbouring intercellular bacteria in the outer cortical or invasion zone of the nodule. This was clearly observed in nodules aged 4 wk or older, probably due to more extensive rhizobial proliferation (Fig. 8c), whereas 2- or 3-wk-old nodules had flatter, nonconvoluted cells in this region. Furthermore, plant cells infected with endosymbiotic bacteria showed a normal cell shape (Fig. 7b,c).

Figure 5.

Brightfield micrographs of semi-thin sections at the periphery of an 18-d-old tagasaste nodule. Arrows indicate opened areas between epidermal host cells providing crack portals of entry for the symbiotic bradyrhizobia. Bar, 12 µm.

Figure 6 Transmission electron microscopy (.

Figure 6 Transmission electron microscopy (.

TEM) of intercellular invasion of tagasaste nodules by bradyrhizobia. (a) Bacteria (arrow) that have entered a peripheral intercellular space of the outer nodule cortex. (b) Bradyrhizobia colonizing an intercellular space deeper within the nodule. Note the packing of bacteria in the middle lamella between two separated host cell walls. (c) Low magnification view of the intercellular infection network of disseminating bradyrhizobia within the tagasaste nodule. Note the convoluted cell walls of the host cells neighbouring the intercellular bacteria. Bar, 2 µm.

Figure 7.

Cytology of tagasaste nodule cells. (a) Collapse of host cells in the nodule cortex, where intercellular dissemination of symbiotic bradyrhizobia is occurring. (b) Brightfield micrograph showing the intercellular dissemination of the bacteria and sparsely infected host cells containing released vegetative bacteria. (c) Transmission electron microscopy (TEM) of the fixation zone showing infected host cells completely filled with singly enclosed bacteroids. Bar, 5 µm in (a) and (b), and 1 µm in (c).

Figure 8 Transmission electron microscopy (.

Figure 8 Transmission electron microscopy (.

TEM) of nodule host cell invasion. (a,b,c) Rhizobia surrounded by an electron-transparent zone and electron-dense fibrillar material in the intercellular space. (b) Accumulation of loose fibrillar material (arrowhead) suggestive of wall loosening associated with the invading bacteria, and the presence of rhizobia near pockets of invaginated host cell wall (double arrowheads). (c) Extensive rhizobial colonization of an intercellular space. Double arrowhead points to an invaginated host cell wall. Line-bordered insert is an enlargement of the corresponding area. (d) Bacterium embedded within the wall forming a host cell wall loop (hcw) associated with electron-dense material (arrowhead). (e) Bacterium included in the host cell wall (hcw), surrounded by loosened cell-wall material (arrowheads) near a symbiosome containing a bacteroid with condensed nucleoid. The interface of the symbiosome and host cell wall is indicated by a double arrowhead. (f) Bacterium (asterisk) in the process of entering the plant cell. Note the proximity of the intercellular space, an altered host cell wall, an invaginated host cell plasmamembrane, and the presence in the bacterial cytoplasm of poly-β-hydroxybutyrate (PHB) granules (characteristics of vegetative state). Bar, 0.5 µm.

Closer examination of the invaded intercellular spaces within nodules revealed bacteria surrounded by a matrix of electron-transparent material that approximately followed their contour (perhaps capsule) and more electron-dense fibrillar material loosely associated with, and possibly derived from, the host cell wall (Fig. 8a). The extent of alterations in the nodule host cell walls was such that, sometimes, they appeared to be pliable and became invaginated, forming pockets that were slightly larger than individual bacteria (Fig. 8b,c). However, instead of inducing infection threads, the bacteria presumably entered nodule host cells directly from intercellular spaces by continued, localized erosion of their structurally altered cell walls followed by invagination of their plasmamembrane. Several lines of TEM evidence strongly suggest that the bacteria transverse the host cell wall. First, intercellular bacteria were frequently found to occupy the cell wall pockets that were invaginated into the adjacent host cell (Fig. 8c). Similarly, Fig. 8(d),(e) show bacteria embedded within altered structures of the host cell wall in contact with the host cytoplasmic membrane, and at the same time, adjacent to other bacteria that had differentiated into bacteroids with well-defined nucleoids. Sometimes extracellular bacteria with internal poly-β-hydroxybutyrate (PHB) granules (characteristic of bacteria in the intercellular spaces) were observed in the process of invaginating the plasmamembrane during invasion of the plant cell (Fig. 8f). Moreover, localized areas of the host cell walls that were thinner and/or altered in electron-density were often found near the symbiosomes (Fig. 9a–d). These observations are very suggestive of the mode of invasion of the host nodule cells. We interpret these changes in ultrastructure as reflecting localized areas of structurally altered host cell walls that constitute regions more easily penetrated by the bacteria during their invasion of the nodule cell. Division of recently infected meristematic nodule cells appears to expand the infected area without the involvement of infection threads (Fig. 10a,b).

Figure 9 Transmission electron microscopy (.

Figure 9 Transmission electron microscopy (.

TEM) of infected nodule cells. (a) Infected host cell with the cell wall thinned at several sites (arrowheads). (b–d) Symbiosomes containing bacteroids near the inner face of the host cellular wall (hcw) where a distinct electron-dense material has accumulated (arrowhead). Peribacteroid membrane (pbm), peribacteroid space (pbs). Bar, 0.5 µm.

Figure 10.

(a,b) Phase contrast micrographs of semi-thin cross-sections at the interface of the meristem-infection zone of a tagasaste nodule invaded by bradyrhizobia. Bacteria are present within recently divided host cells. Bar, 10 µm.

The central fixation zone had the typical organization of an indeterminate nodule characterized by the presence of densely infected plant cells (Figs 4, 7c). By contrast to intracellular vegetative bacteria, bacteroids commonly appeared as enlarged, elongated (and occasionally pleomorphic) rods that were singly enclosed within peribacteroid membranes (Figs 7c, 11). The centrally localized nucleoid region of the endosymbiotic bacteroids appeared densely compacted and associated with electron-dense (possibly polyphosphate) inclusions (Fig. 11). Interestingly, although the bacteria in the intercellular spaces frequently contained large intracellular inclusions of poly β-hydroxybutyric acid (Figs 6a,b 8a–c,f), these electron-transparent granules were rarely, if ever, found in the endosymbiotic bacteroid stage (Figs 7c, 8d,e, 11).

Figure 11.

Ultrastructure of an endosymbiotic bacteroid of Bradyrhizobium sp. strain BTA-1 within a tagasaste nodule cell. Shown are the peribacteroid membrane (pbm), peribacteroid space (pbs), and a bacteroid cell with a centrally localized nucleoid associated with electron-opaque polyphosphate inclusions (pp). Note the absence of electron-transparent intracellular granules of poly β-hydroxybutyrate. Bar, 0.15 µm.

The proximal senescence zone was characterized by the presence of degenerated bacteroids with associated membranes in a disorganized host cytoplasm (not shown). In no case were any infection threads found within the senescing nodule cells.

In striking contrast to the above description of symbiotic development with tagasaste, these same strains of Canarian bradyrhizobia infected siratro roots via formation of infection threads within deformed root hairs, induction of the initial foci of cell divisions in the root outer cortex and development of mature nodules of the spherical determinate type with a central infected tissue formed by groups of host cells containing true infection threads and symbiosomes surrounded by uninfected interstitial cells (not shown).

Discussion

Tagasaste formed indeterminate nodules with bradyrhizobia indigenous to the Canary Islands. The first stages of the infection process included rhizobial attachment to and deformation of root hairs, followed by marked curling, localized formation of bright spots, and initiation of intracellular infection threads. These infection threads always aborted before they grew to the base of the root hair. This phenotype has been described in other symbioses with rhizobial mutants defective in the genes that specify the synthesis of superficial polysaccharides (Finan et al., 1985; Carlson et al., 1987; de Maagd et al., 1988; Long, 1989; Hirsch, 1992; Rolfe et al., 1996; van Workum et al., 1998; Pellock et al., 2000). However, the two wild-type strains used in this study produce normal surface polysaccharides (León-Barrios et al., 1992a,b; Santamaría et al., 1997), and are fully capable of infecting deformed root hairs of Macroptilium through true infection threads that reached the root cortex (data not shown).

A thorough search, using combined light and transmission electron microscopy, confirmed the absence of infection threads in tagasaste nodules. Instead, both strains of bradyrhizobia invaded the periphery of nodules by the crack entry mode, followed by deeper dissemination though intercellular spaces, infection of host cells without infection thread formation and extension of infection by division of the recently infected meristematic cells.

The unique character in development of the tagasaste symbiosis lies in the fact that it shows a hybrid mix of combined events representing several different types of infection collectively unlike any other previously described. The preinfection stages of interaction with root hairs are triggered on tagasaste by strains BTA-1 and BGA-1 in the same way as the Rhizobium-clover or Bradyrhizobium-siratro symbioses. Although deformation of axillary hairs of Aeschynomene, Arachis and Stylosanthes occurs when inoculated with homologous rhizobia, these other crack-entry types of symbioses do not progress further in root hair infection (Napoli et al., 1975; Chandler, 1978; Chandler et al., 1982). By contrast, in the Bradyrhizobium-tagasaste symbiosis, these early interactions progress further to initiate infection threads within deformed root hairs, but these intracellular infection structures always abort before growing to the base of the root hair. The formation of infection threads within root hairs but not within root nodules, the initiation of meristematic foci in the inner (rather than the outer) cortex and the nodule morphology, all represent characteristics that distinguish the Bradyrhizobium-tagasaste symbiosis from the Rhizobium-lupin symbiosis (Tang et al., 1993; James et al., 1997).

Another important difference in symbiotic development between tagasaste and other legumes infected by crack entry is that the nodules emerge along lateral roots rather than being restricted to lateral root axils as is usually the case for Aeschynomene, Arachis, Stylosanthes,Sesbania, and Neptunia (Napoli et al., 1975; Chandler, 1978; Chandler et al., 1982; Ndoye et al., 1994; Rana & Krishnan, 1995; Subba-Rao et al., 1995). This fact, together with the early development of nodule primordia in the inner cortex of the tagasaste root, suggests that the strategy used by these strains to produce wounds to penetrate the epidermis is similar to the one used by bradyrhizobia to infect the nonlegume Parasponia. In this symbiosis, the stimulation of cell divisions in epidermal root hairs or divisions in the outermost cortical cells by the rhizobia leads eventually to the separation of epidermal cells and enlargement of the intercellular spaces, thereby developing an access point of entry (Lancelle & Torrey, 1984; Bender et al., 1987). Presumably, strains BTA-1 and BGA-1 stimulate the development of an expanding nodule primordium that splits the epidermis as it emerges, thus creating the portal of crack entry through which these bacteria can access the nodule surface. Young nodules contain wide enough open spaces in the epidermis through which the bacteria can enter their interior.

The Bradyrhizobium-tagasaste symbiosis shares additional characteristics with other symbioses using crack entry as the mode of infection. For instance, the bacterial dissemination through the nodule cortex and into the infection zone takes place by separating cortical cells at the middle lamellae as occurs in Arachis (Dongre et al., 1985). Like tagasaste, collapse of cells in the outer layers of the nodule also occurs in Aeschynomene and Stylosanthes (Chandler et al., 1982; Alazard & Duhoux, 1990).

It must be highlighted that the bacteria in intercellular spaces were often associated with structurally altered and weakened host cell walls, and accumulation of broken host cell wall fragments. This suggests the presence of highly active enzymes with lytic activity against walls. The holes that developed on the tips of tagasaste root hairs during the first stages of the infection process are also suggestive of the activity of wall-degrading enzymes. Similar holes have been detected on root hair tips of axenic white clover seedlings incubated with purified cellulases from Rhizobium leguminosarum bv. trifolii (Mateos et al., 1992; Mateos et al., 1996; Mateos et al., 2000). Although low cellulase activities have been detected in pure cultures of BTA-1 and BGA-1 (data not shown), the in situ concentrations of these and other wall-degrading enzymes made by the tagasaste bradyrhizobia in planta is not known. Other events that could possibly influence this process of localized wall modification in planta include the induction of host plant polygalacturonase by rhizobial components (Muñoz et al., 1998), the inhibition of the expression of peroxidase genes encoding the enzymes involved in lignification of the plant cell wall (Klotz & Lagrimini, 1996), the transient suppression of plant wall-bound peroxidase activity by homologous rhizobial components (Salzwedel & Dazzo, 1993) and localized disruption in crystalline wall architecture in host cells grown with chitolipooligosaccharide Nod factors (Dazzo et al., 1996). Some of these activities, documented in other plant systems, could contribute to the magnitude of the injuries found in the host cell wall during the infection process in tagasaste. The direct invasion of tagasaste nodule cells by bradyrhizobia without infection thread formation is similar to rhizobial invasion in Arachis nodules (Chandler, 1978), but distinctly different from rhizobial invasion via infection thread formation as occurs in nodules of the tropical legumes, Mimosa (de Faria et al., 1988), Sesbania (Ndoye et al., 1994) and Neptunia (Subba-Rao et al., 1995).

Another unique feature of the bradyrhizobia-tagasaste symbiosis is that, in spite of infection by crack entry, it nevertheless forms the developmental gradient of indeterminate nodules characterized in the mature stage by their cylindrical shape, within which the apical meristem, host cell infection, bacteroid N2 fixation and senescence structures are separated into discrete, distal-to-proximal zones. Indeed, this is the first reported case of a wild-type rhizobia-legume symbiosis in which crack entry leads to the development of an indeterminate N2-fixing nodule. The necessity of studying other root-nodule symbioses established between rhizobia and (sub)tropical legumes (especially woody legumes) must be highlighted to determine if this crack entry mode of infection evolved uniquely in the Bradyrhizobium-tagasaste symbiosis, or alternatively, is more common than is assumed among N2-fixing rhizobia-legume symbioses.

Acknowledgements

We thank Pedro Mateos for measuring the hydrolytic enzymatic activities. This work was supported by a grant PI1998/016 from the Canary Autonomous Government, and by the MSU Center for Microbial Ecology supported by National Science Foundation grant DEB 91–20006.

Ancillary