The process of infection of lupin nodule cells by rhizobia was examined using thin-section and freeze-fracture electron-microscopic techniques to characterize the properties of different membranes and to establish relationships between them. The membranes of the Golgi bodies and the endoplasmic reticulum stained with zinc iodide-osmium tetroxide but not with phos-photungstic acid or silver. By contrast the infection thread membranes, peribacteroid membranes, plasma membranes and membranes of cytoplasmic vesicles did not stain with zinc iodide-osmium tetroxide but stained with phosphotungstic acid and silver. The peribacteroid membranes and plasma membranes are, however, different from each other since the particle density on the E face of freeze-fracture replicas of plasma membranes was twice that on the E face of the peribacteroid membranes.

An examination of the tips of the infection threads in the cytoplasm of the plant cells, showed that the rhizobia bud off from the infection threads enclosed in the infection thread membranes. The rhizobia continue to divide still surrounded by membranes of plant origin, namely the peri-bacteroid membranes. Cytoplasmic vesicles are observed in both thin-section and freeze-fracture preparations of nodule tissue closely associated with, and apparently produced by, Golgi bodies. Formation of the walls and membranes of the infection threads and of the peribacteroid membranes involves fusion of the cytoplasmic vesicles with these membranes. It is proposed that the process of infection of plant cells in lupin nodules involves a change in the function of the Golgi body system for the biogenesis of plant cell walls and plasma membranes to include the synthesis of the walls and membranes of the infection threads and also the peribacteroid membranes.

The development of nodules on the roots of legumes is initiated on the surface of root hairs by rhizobia, which induce the formation of infection threads (Nutman, 1965). The threads elongate, presumably as a result of continued division of rhizobia in the tips until entry is made into target cells within the root cortex (Newcomb, 1976). The rhizobia bud off from the tips of the infection threads into the plant cytoplasm surrounded by membranes which we prefer to call the peribacteroid membranes.*

Relationships between the peribacteroid membranes and other plant membranes have been the subject of considerable discussion (Dixon, 1969; Bergersen, 1973, 1974; Libbenga & Bogers, 1974). Reports have conflicted on whether the peribacteroid membranes are formed from the infection thread membranes (Bergersen & Briggs, 1958; Goodchild & Bergersen, 1966; Dixon, 1967; Gourret & Fernandez-Arias, 1974) or by de novo synthesis of new membranes after the rhizobia have burst through the infection thread membranes into the cytoplasm of the plant cells (Dart & Mercer, 1963; Jordan, Grinyer & Coulter, 1963; Generozova & Yagodin, 1972; Mackenzie, Vail & Jordan, 1973). That the peribacteroid membranes are similar to the plasma membranes is suggested by reports that both membranes have the same thickness (Dart & Mercer, 1963), the same staining characteristics with uranium and lead (Dixon, 1967) and also with phosphotungstic acid (Robertson et al. 1974). Particle densities on faces of freeze-fracture replicas of plasma membranes and peribacteroid membranes are similar, according to Tu (1975), although Mackenzie et al. (1973) reported them to be different.

Little is known about the biogenesis of the peribacteroid membrane within the cytoplasm of the plant cells, although it has been suggested that the Golgi bodies and endoplasmic reticulum might be involved in the production of vesicles which subsequently fuse with the peribacteroid membranes (Jordan et al. 1963; Generozova & Yagodin, 1972; Truchet & Coulomb, 1973). In this paper we present the results of a study using thin-section and freeze-fracture electron-microscopic techniques to examine the role of Golgi bodies in the biogenesis of infection threads and of peribacteroid membranes in lupin nodules.

Preparation of nodule tissue

Lupin seeds (Lupinus angustifolius L. cv. Uniwhite or Bitter blue) were surface-sterilized, germinated on the surface of sterile agar and planted in sterilized water-washed pumice. Before covering with pumice the germinated seeds were inoculated with Rhizobium lupini NZP2257 obtained by suspending organisms, grown for 6 days on yeast-mannitol agar slopes (Graham, 1964), in sterile distilled water. The troughs were placed in a controlled environment cabinet with a 12-h day-length, a day/night temperature regime of 25/21 °C and a light intensity of 130 W m−2 over a spectrum range of 400-750 nm. Plants were supplied with sterile, nitrogen-free nutrient solution (Hoagland & Amon, 1938) at one fifth strength for the first week following inoculation and thereafter with non-sterile nutrient solution of the same strength. Nodules, which occurred almost entirely on the main roots of the plants, were harvested between 9 and 20 days following inoculation of germinated seeds. This period corresponds to the period of development of N2-fixation in lupin nodules (Robertson, Farnden, Warburton & Banks, 1975).

Fixation and embedding

Nodules were fixed and embedded, after cutting into 1-mm3 blocks, using one of the following methods.

Method (1)

Blocks were fixed in 2% glutaraldehyde in 0·1 M phosphate buffer, pH 7-2, for 2 h at 0·4 °C; washed in phosphate buffer 3 times; fixed in 1% OsO4 in phosphate buffer for 2 h at 0·4 °C; dehydrated in an ethanol series, passed through propylene oxide at room temperature and embedded in epoxy resin (Durcupan ACM, Fluka AG, Switzerland).

Method (2)

Blocks were fixed in 2% glutaraldehyde in 0·1 M cacodylate buffer, pH 7·2, for 1 h at 0·4 °C; washed in cacodylate buffer 3 times; fixed in 1% OsO4 in cacodylate buffer for 1 h at 0·4 °C; washed in cacodylate buffer 3 times; dehydrated in an acetone series at 0·4 °C and embedded in Durcupan ACM.

Method (3)

Blocks were fixed in 2% glutaraldehyde in 0·1 M cacodylate buffer, pH 7·2, for 40 min at room temperature; washed in cacodylate buffer 3 times; fixed in 1% OsO4 in 0·04 M veronal-acetate buffer, pH 6·1, for 30 min at room temperature; dehydrated in an acetone series at 0·4 °C and embedded in Durcupan ACM.

Method (4)

Blocks were fixed in a mixture of 2% formaldehyde and 3% glutaraldehyde in 01 M phosphate buffer, pH 7·2, for 5 min under vacuum and then for 1 h at room temperature in fresh fixative; washed in phosphate buffer 3 times; fixed in 1% OsO4 in phosphate buffer for 1h at room temperature; washed in 1% NaCl 3 times; dehydrated in an acetone series at 0·4 °C; passed through propylene oxide at room temperature and embedded in Epon 812.

Method (5)

As for Method (4) except that following the wash in 1% NaCl the blocks were immersed in 0·5% magnesium uranyl acetate in 1% NaCl for 1 h at room temperature before dehydrating.

Method (6)

Blocks were fixed in a mixture of 2% formaldehyde and 3% glutaraldehyde in 0-06 M cacodylate, pH 7·2, containing 5 mM Ca2+, for 5 min under vacuum and then for 1 h at room temperature in fresh fixative; washed in the cacodylate buffer system 3 times; fixed in 1% OsO4 in the cacodylate buffer system; washed in the cacodylate buffer system 3 times; dehydrated in an acetone series at 0·4 °C and embedded in Epon 812.

Method (7)

Blocks were fixed in zinc oxide-osmium tetroxide (Niebauer, Krawczyk, Kidd & Wilgram, 1969) for 22 h at room temperature in the dark; washed in distilled water 3 times; dehydrated in an acetone series at room temperature and embedded in Durcupan ACM.

Staining of thin sections

Uranium and lead. Sections were mounted on unsupported copper grids and stained at room temperature with saturated uranyl acetate in 50% ethanol for 10–20 min followed by lead citrate (Venable & Coggeshall, 1965) for 5–10 min.

Phosphotungstic acid

Sections were stained with phosphotungstic acid in chromic acid for 20 min using the method of Roland, Lembi & Morré (1972) but omitting the preliminary treatment with periodic acid.

Silver

Sections, mounted on Formvar-supported or unsupported gold grids, were stained with silver using the method of Thiéry (1967) for localization of carbohydrates as described previously (Robertson, Lyttleton, Williamson & Batt, 1975). To summarize, grids were floated on 1% periodic acid for 30 min; 0 2% thiocarbohydrazide in 20% acetic acid for 24 h and finally on 1% silver proteinate for 30 min. Appropriate washing steps were included between each reagent.

Freeze-fracturing

Nodules were prepared for freeze-fracturing, after cutting into 1 -mm3 blocks, by one of the following methods.

Method (1)

Blocks were immersed in 30% glycerol for 1 h at 0·4 °C.

Method (2)

Blocks were fixed in 2% glutaraldehyde in 0·1 M phosphate, pH 7-2, for 5 min under vacuum, in fresh fixative for 1 h at 22 °C; held in phosphate buffer for 23 h at 0·4 °C and immersed in 30% glycerol for 1 h at 0·4 °C.

Method (3)

Blocks were fixed in 2% glutaraldehyde in phosphate buffer for 5 min under vacuum, in fresh fixative for 1 h at 22 °C and immersed in 30% glycerol for 1 h at 0·4 °C.

Method (4)

Blocks were fixed in 2% glutaraldehyde in phosphate buffer for 5 min under vacuum, in fresh fixative for 1 h at 22 °C and immersed in 30% glycerol for 5 days at 0·4 °C.

After treatment with glycerol the blocks were frozen in melting Freon 12 and stored under liquid nitrogen. Freeze-fracturing was carried out as described previously (Bullivant & Ames, 1966; Bullivant, 1973). Freeze-fracture replicas and thin sections were examined with a Philips EM 200 microscope at 60 kV and a Philips EM 301 at 80 kV. Magnification was calibrated using a grating replica of 2160 lines/mm. Thickness of membranes in thin sections was determined by measuring the distance between the mid-points of dark lines of trilaminar membranes viewed on micrographs. The mean value of at least 10 determinations and standard errors of means (S.E.M.) are presented. Particle densities on faces of freeze-fracture replicas were determined by counting particles on micrographs in a 6-mm square frame at × 60 000 (equivalent to 0·01 μm2). The frame was positioned at random on approximately horizontal surfaces of freeze-fracture replicas on which particles were clearly visible. The means of determinations and standard errors of means are presented. The nomenclature describing faces of freeze-fracture replicas is according to Branton et al. (1975).

Characterization of membranes in lupin nodules

A study of thin sections and freeze-fracture preparations of nodules, removed from plants between 9 and 20 days following inoculation of germinated seeds, was carried out to characterize the properties of the different membranes (Table 1). Both Uni-white and Bitter blue cultivars of L. angustifolius were used in this study. However, no differences in ultrastructure were detected between them or between individual membranes at different stages of nodule development.

Table 1.

Properties of membranes in thin sections and freeze-fracture replicas of lupin nodules

Properties of membranes in thin sections and freeze-fracture replicas of lupin nodules
Properties of membranes in thin sections and freeze-fracture replicas of lupin nodules

The plasma membranes, infection thread membranes, peribacteroid membranes and membranes of cytoplasmic vesicles all stain heavily with uranium and lead, silver, and phosphotungstic acid but not with zinc iodide-osmium tetroxide. The thickness of these membranes is also the same (Table 1). The plasma membranes and the peribacteroid membranes are different from each other, however, since the particle density on the E face of freeze-fracture replicas of plasma membranes is twice that on the E face of the peribacteroid membranes (Figs. 1,2; Table 1).

Figs.1–21

are electron micrographs of thin sections and freeze-fracture replicas of nodule tissue from 9-to 20-day plants. The direction of shadow on the freeze-fracture replicas is indicated by circled arrowheads Fig. 1. shows the E face of the plasma membrane with plasmodesmata and the P and E faces of peribacteroid membranes, × 80 000.

Figs.1–21

are electron micrographs of thin sections and freeze-fracture replicas of nodule tissue from 9-to 20-day plants. The direction of shadow on the freeze-fracture replicas is indicated by circled arrowheads Fig. 1. shows the E face of the plasma membrane with plasmodesmata and the P and E faces of peribacteroid membranes, × 80 000.

Fig. 2.

shows the P face of the plasma membrane with plasmodesmata and the P and E faces of peribacteroid membranes, × 80 000.

Fig. 2.

shows the P face of the plasma membrane with plasmodesmata and the P and E faces of peribacteroid membranes, × 80 000.

The membranes of the Golgi bodies and endoplasmic reticulum do not stain with silver or phosphotungstic acid but stain heavily with zinc iodide-osmium tetroxide, except for one or two cisternae on the distal face of the Golgi bodies. The membranes of the endoplasmic reticulum and Golgi bodies are thinner than the membranes mentioned above and no increase in membrane thickness was detected from the proximal to the distal face of the Golgi bodies. Cytoplasmic vesicles in close association with the Golgi bodies, however, do show a membrane thickness of 5-5 nm. The staining and freeze-fracture properties of the membranes of the bacteroid envelope are distinct from the plant membranes (Table 1).

It should be pointed out that phosphotungstic acid staining of membranes was not consistent either between different nodule preparations or between thin sections on the same grid. A study of the effect of changing the fixative (glutaraldehyde, formaldehyde-glutaraldehyde), the dehydrating agent (ethanol, acetone, propylene oxide), the embedding material (Epon 812, Araldite, Spurrs resin) or the conditions of staining, including the temperature or concentration of the various reagents as reported by several authors (Roland et al. 1972; Quintarelli, Bellocci & Geremia, 1973; Scott & Webb, 1975), showed that for every preparation of nodule tissue the optimum conditions for phosphotungstic acid staining had to be determined. Slight variations in temperature or concentration of the staining reagents altered the intensity and in some cases the specificity of the stain. Omission of periodic acid from the staining procedure improved the consistency of the stain slightly and had the added advantage that other membranous structures, which did not stain positively with phosphotungstic acid, were faintly distinguishable.

The process of infection of plant cells by rhizobia

Rhizobia occur in infection threads, as observed in cross-sections cut some distance behind the tips (Fig. 3), apparently tightly packed and surrounded by wall material. This wall material is continuous with the plant cell walls and has the same properties as the walls after staining with uranium and lead (Fig. 3) and also with silver (Fig. 4). The infection threads are bound by membranes which are continuous with, and have the same staining properties as, the plasma membranes (Figs. 35). A study of thin sections of infection threads, cut closer to the tips within the cytoplasm of the plant cells, shows that a thin layer of material of low electron density occurs in the immediate vicinity of the rhizobia (Fig. 6). This material may be rhizobial extracellular poly-saccharide. Spaces of low electron density are also observed between the infection thread membranes and the infection thread walls (Fig. 6). These spaces are unlikely to be artifacts of fixation since infection threads showing the presence and absence of large spaces are observed in the same thin section (Fig. 6). Golgi bodies and cytoplasmic vesicles, containing material with the same staining properties as the walls of the infection threads, are observed in the cytoplasm closely associated with the tips of the infection threads. These vesicles frequently give the appearance of fusing with the infection thread membranes (Fig. 6).

Fig. 3.

Thin section of a nodule, from a 12-day plant, prepared by Method 4 and stained with uranium and lead. The micrograph shows a cross-section of an infection thread cut some distance behind the tip and associated with the plant cell wall, × 14000.

Fig. 3.

Thin section of a nodule, from a 12-day plant, prepared by Method 4 and stained with uranium and lead. The micrograph shows a cross-section of an infection thread cut some distance behind the tip and associated with the plant cell wall, × 14000.

Fig. 4.

Thin section of a nodule, from an 11-day plant, prepared by Method 6 and stained with silver. The micrograph shows a cross-section of an infection thread associated with the cell wall, × 41 000.

Fig. 4.

Thin section of a nodule, from an 11-day plant, prepared by Method 6 and stained with silver. The micrograph shows a cross-section of an infection thread associated with the cell wall, × 41 000.

Fig. 5.

Thin section of a nodule, from an n-day plant, prepared by Method 2 and stained with phosphotungstic acid. The micrograph shows a cross-section of an infection thread associated with the cell wall, ×20000.

Fig. 5.

Thin section of a nodule, from an n-day plant, prepared by Method 2 and stained with phosphotungstic acid. The micrograph shows a cross-section of an infection thread associated with the cell wall, ×20000.

Figs. 6–12.

Thin sections of nodules, from 9-day plants, prepared by Method 1.

Fig. 6. Thin section, stained with uranium and lead, showing a cross-section of an infection thread cut close to the tip. The arrows indicate cytoplasmic vesicles containing material with the same staining properties as the walls of the infection threads. Two of these vesicles appear to be in the process of fusing with an infection thread membrane. × 21 000.

Figs. 6–12.

Thin sections of nodules, from 9-day plants, prepared by Method 1.

Fig. 6. Thin section, stained with uranium and lead, showing a cross-section of an infection thread cut close to the tip. The arrows indicate cytoplasmic vesicles containing material with the same staining properties as the walls of the infection threads. Two of these vesicles appear to be in the process of fusing with an infection thread membrane. × 21 000.

The tips of the infection threads (Fig. 7) contain little infection thread wall material. The rhizobia are, however, separated from the plant cytoplasm by the infection thread membranes which form pocket-like protrusions into the cytoplasm of the plant cells (Fig. 7). These protrusions are, in some cases, so voluminous and contain so few rhizobia (Fig. 8) that it is possible to confuse them with true vacuoles unless thin sections are stained with phosphotungstic acid or silver (Table 1). The protrusions of the infection thread membranes appear to be formed by fusion of cytoplasmic vesicles, containing variable amounts of cell wall material, with the infection thread membranes (Fig. 8). This wall material appears to be dispersed or possibly degraded after being released into the protrusions of the infection thread membranes (Fig. 8).

Fig. 7.

Thin section, stained with uranium and lead, showing the tip of an infection thread. The rhizobia occur in pocket-like protrusions of the infection thread membrane which contain little infection thread wall material. Some bacteroids, surrounded by peribacteroid membranes, occur in the plant cytoplasm, × 14000.

Fig. 7.

Thin section, stained with uranium and lead, showing the tip of an infection thread. The rhizobia occur in pocket-like protrusions of the infection thread membrane which contain little infection thread wall material. Some bacteroids, surrounded by peribacteroid membranes, occur in the plant cytoplasm, × 14000.

Fig. 8.

Thin section, stained with silver, showing a pocket-like protrusion of the infection thread membrane in the plant cytoplasm. Dispersion or possible degradation of wall material is visible (arrowheads). Cytoplasmic vesicles containing variable amounts of wall material occur in the plant cytoplasm (arrows), × 37000.

Fig. 8.

Thin section, stained with silver, showing a pocket-like protrusion of the infection thread membrane in the plant cytoplasm. Dispersion or possible degradation of wall material is visible (arrowheads). Cytoplasmic vesicles containing variable amounts of wall material occur in the plant cytoplasm (arrows), × 37000.

A study of thin sections of nodules, cut in the region of cell infection, shows some of the rhizobia separated from the tips of the infection threads in the cytoplasm of the plant cells but still surrounded by membranes, the peribacteroid membranes (Figs. 7, 9). That some of these organisms are definitely free from the tips and not in finger-like protrusions of the infection thread membranes was established from a study of serial sections. Golgi bodies, microtubules and cytoplasmic vesicles containing wall material are observed in the regions of cell infection (Fig. 9). In some cases the microtubules appear to form links between the cytoplasmic vesicles and the peribacteroid membranes (Fig. 10). The Golgi bodies, which do not stain or stain only very lightly with silver (Table 1), can be faintly recognized in Figs. 11, 12 and more clearly in Fig. 13 where the plane of the section has cut through the stacks of cisternae. The Golgi bodies can also be recognized in sections stained with phosphotungstic acid, although once again the cisternae do not stain specifically (Fig. 14). In both silver- and phosphotungstic-acid-stained preparations the Golgi bodies are observed closely associated with cytoplasmic vesicles which stain specifically and contain variable amounts of wall material (Figs.11 –14). These vesicles are often observed in the process of fusing with, or having fused with, the peribacteroid membranes (Figs. 12, 15). The Golgi bodies can be clearly seen associated with cytoplasmic vesicles in thin sections stained with uranium and lead (Fig. 16) or after fixation of nodules with zinc iodide-osmium tetroxide (Fig. 17).

Fig. 9.

Thin section, stained with uranium and lead, showing an early stage of infection in a nodule cell. The arrows indicate cytoplasmic vesicles containing wall material, one of which appears to have fused with a peribacteroid membrane. Note the Golgi bodies and the presence of microtubules, × 21 000.

Fig. 9.

Thin section, stained with uranium and lead, showing an early stage of infection in a nodule cell. The arrows indicate cytoplasmic vesicles containing wall material, one of which appears to have fused with a peribacteroid membrane. Note the Golgi bodies and the presence of microtubules, × 21 000.

Fig. 10.

Thin section, stained with uranium and lead, showing a possible association between a cytoplasmic vesicle (arrow), a microtubule and a peribacteroid membrane. × 47 000.

Fig. 10.

Thin section, stained with uranium and lead, showing a possible association between a cytoplasmic vesicle (arrow), a microtubule and a peribacteroid membrane. × 47 000.

Fig. 11.

Thin section, stained with silver, showing specific staining of the plant cell wall, plasma membranes, peribacteroid membrane and cytoplasmic vesicles containing wall material (arrows). Some vesicles occur in close association with Golgi bodies which are only just distinguishable, ×41 000.

Fig. 11.

Thin section, stained with silver, showing specific staining of the plant cell wall, plasma membranes, peribacteroid membrane and cytoplasmic vesicles containing wall material (arrows). Some vesicles occur in close association with Golgi bodies which are only just distinguishable, ×41 000.

Fig. 12.

Thin section, stained with silver, showing specific staining of the plant cell wall, plasma membranes, peribacteroid membranes and cytoplasmic vesicles (arrows) some of which contain wall material. One of the cytoplasmic vesicles appears to have fused with a peribacteroid membrane (double arrow). Some of the vesicles may be involved in cell wall synthesis (*). A Golgi body is faintly visible associated with small cytoplasmic vesicles which stain with silver (arrowheads), × 30000.

Fig. 12.

Thin section, stained with silver, showing specific staining of the plant cell wall, plasma membranes, peribacteroid membranes and cytoplasmic vesicles (arrows) some of which contain wall material. One of the cytoplasmic vesicles appears to have fused with a peribacteroid membrane (double arrow). Some of the vesicles may be involved in cell wall synthesis (*). A Golgi body is faintly visible associated with small cytoplasmic vesicles which stain with silver (arrowheads), × 30000.

Fig. 13.

Thin section of a nodule, from an 11-day plant, prepared by Method 6 and stained with silver. The micrograph shows silver staining of peribacteroid membranes and cytoplasmic vesicles (arrows) in close association with a Golgi body, × 41 000.

Fig. 13.

Thin section of a nodule, from an 11-day plant, prepared by Method 6 and stained with silver. The micrograph shows silver staining of peribacteroid membranes and cytoplasmic vesicles (arrows) in close association with a Golgi body, × 41 000.

Figs. 14,15.

Thin sections of nodules, from a 20-day plant, prepared by Method 3 and stained with phosphotungstic acid. Fig. 14. Phosphotungstic acid staining of the plasma membranes, the peribacteroid membranes and cytoplasmic vesicles (arrows) in the vicinity of a Golgi body which is only faintly distinguishable, × 39000.

Figs. 14,15.

Thin sections of nodules, from a 20-day plant, prepared by Method 3 and stained with phosphotungstic acid. Fig. 14. Phosphotungstic acid staining of the plasma membranes, the peribacteroid membranes and cytoplasmic vesicles (arrows) in the vicinity of a Golgi body which is only faintly distinguishable, × 39000.

Fig. 15.

Cytoplasmic vesicles apparently in the process of fusing with the peribacteroid membranes, × 64000.

Fig. 15.

Cytoplasmic vesicles apparently in the process of fusing with the peribacteroid membranes, × 64000.

Fig. 16.

Thin section of a nodule, from a 20-day plant, prepared by Method 2 and stained with uranium and lead showing a Golgi body and associated cytoplasmic vesicles, × 64 000.

Fig. 16.

Thin section of a nodule, from a 20-day plant, prepared by Method 2 and stained with uranium and lead showing a Golgi body and associated cytoplasmic vesicles, × 64 000.

Fig. 17.

Thin section of a nodule, from an 11 -day plant, fixed with zinc iodide-osmium tetroxide by Method 7. The section was examined without post-staining. Staining is observed on the inner surface of the endoplasmic reticulum and in the cisternae of the Golgi bodies except for one of the cisternae on the distal face which is only lightly stained. No specific staining occurs on a group of closely associated structures (arrows) including a vesicle attached to the distal face of a Golgi body, cytoplasmic vesicles and peribacteroid membranes. Vésiculation of the endoplasmic reticulum, observed in these preparations, is caused by fixation in zinc iodide-osmium tetroxide, × 36000.

Fig. 17.

Thin section of a nodule, from an 11 -day plant, fixed with zinc iodide-osmium tetroxide by Method 7. The section was examined without post-staining. Staining is observed on the inner surface of the endoplasmic reticulum and in the cisternae of the Golgi bodies except for one of the cisternae on the distal face which is only lightly stained. No specific staining occurs on a group of closely associated structures (arrows) including a vesicle attached to the distal face of a Golgi body, cytoplasmic vesicles and peribacteroid membranes. Vésiculation of the endoplasmic reticulum, observed in these preparations, is caused by fixation in zinc iodide-osmium tetroxide, × 36000.

The process of vesicle formation by the Golgi bodies was further examined using the freeze-fracture technique (Figs. 18, 19). The Golgi bodies appear to be producing vesicles which have particle densities on the P and E faces of freeze-fracture replicas similar to particle densities on the corresponding faces of replicas of peribacteroid membranes (Figs. 18, 19; Table 1). Although the particle densities on the P and E faces of the endoplasmic reticulum were not significantly different from the corresponding faces of the peribacteroid membranes (Fig. 20; Table 1), the endoplasmic reticulum did not appear to produce vesicles directly. Furthermore the membranes of the endoplasmic reticulum and the peribacteroid membranes appear to be different in that the endoplasmic reticulum tends to cross-fracture (Fig. 20 is an exceptional case), whereas the fracture plane tends to follow the contour of both the peribacteroid membranes and the cytoplasmic vesicles (Figs. 18, 19). Where the peribacteroid membranes do cross-fracture the P and E faces of the bacteroid envelope inner membranes are revealed (Fig. 21). Particles on the P and E faces of the peribacteroid membranes were aggregated when nodule tissue had been held in glycerol for 5 days at 0-4 °C after fixation with glutaraldehyde (Fig. 21), whereas particles on the faces of replicas of the endoplasmic reticulum were not aggregated by such treatment.

Figs. 18–20.

Freeze-fracture replicas of nodules, from 18-day plants, prepared by Method 2.

Fig. 18. A Golgi body, P and E faces of cytoplasmic vesicles and P and E faces of the peribacteroid membranes, × 80000.

Figs. 18–20.

Freeze-fracture replicas of nodules, from 18-day plants, prepared by Method 2.

Fig. 18. A Golgi body, P and E faces of cytoplasmic vesicles and P and E faces of the peribacteroid membranes, × 80000.

Fig. 19.

A Golgi body producing vesicles, of which both P and E faces are evident, × 80000.

Fig. 19.

A Golgi body producing vesicles, of which both P and E faces are evident, × 80000.

Fig. 20.

Both P and E faces of the endoplasmic reticulum can be seen, × 80000.

Fig. 20.

Both P and E faces of the endoplasmic reticulum can be seen, × 80000.

Fig. 21.

Freeze-fracture replica of a nodule from an 18-day plant, prepared by Method 4. The replica shows aggregation of particles on the P face of the peribacteroid membrane and no aggregation of particles on the P face of the bacteroid envelope inner membrane. × 80 000.

Fig. 21.

Freeze-fracture replica of a nodule from an 18-day plant, prepared by Method 4. The replica shows aggregation of particles on the P face of the peribacteroid membrane and no aggregation of particles on the P face of the bacteroid envelope inner membrane. × 80 000.

Golgi bodies are generally accepted as being involved in plant cell wall synthesis (see reviews by Northcote, 1968, 1972; Roland, 1973; Morré & Mollenhauer, 1974; Preston, 1974; Whaley, 1975). Characterization of the staining and freeze-fracture properties of membranes in developing lupin nodules has provided evidence which strongly suggests that Golgi bodies are also involved in the synthesis of infection thread walls, infection thread membranes and peribacteroid membranes. It is proposed that, as a consequence of rhizobial infection of plant cells, the function of the Golgi body system involved in the synthesis of cell walls and plasma membranes changes to include the synthesis of the infection thread walls and membranes and, subsequently, the synthesis of the peribacteroid membranes. According to this proposal the process of cell infection is one in which membrane synthesis is continued while cell wall synthesis is gradually discontinued.

A schematic diagram of the process of cell infection and peribacteroid membrane biogenesis is presented in Fig. 22. The rhizobia are represented, in a cross-section of an infection thread cut behind the tip, as being surrounded by a thin layer of what might be rhizobial polysaccharides. This layer is in turn surrounded by the infection thread wall material which is continuous with the plant cell wall. At the tip of the infection thread the rhizobia occur in pocket-like protrusions of the infection thread membrane which contain little or no plant wall material. The Golgi bodies are represented as having 3 functions, namely the synthesis of the plasma membranes and cell walls, the synthesis of the infection thread membranes and infection thread walls and the synthesis of the peribacteroid membranes. Synthesis of the tip of the infection thread involves fusion of cytoplasmic vesicles, derived from Golgi bodies, which contain variable amounts of cell wall material. The actively dividing rhizobia in the tip of the infection thread bud off, into the plant cytoplasm, surrounded by the infection thread membrane which we now call the peribacteroid membrane. Continued synthesis of the peribacteroid membranes involves fusion of cytoplasmic vesicles, derived from Golgi bodies, with the peribacteroid membranes. The evidence for these proposals is discussed below.

Fig. 22.

Diagrammatic representation of the process of infection of plant cells by rhizobia and biogenesis of peribacteroid membranes in lupin nodules. The details are described in the text.

Fig. 22.

Diagrammatic representation of the process of infection of plant cells by rhizobia and biogenesis of peribacteroid membranes in lupin nodules. The details are described in the text.

The formation of infection thread walls, infection thread membranes and peribacteroid membranes appears to be the result of fusion of cytoplasmic vesicles, containing variable amounts of wall material, with either the infection thread membranes or the peribacteroid membranes (Figs. 6, 12, 15). The silver-staining properties of the contents of some of the vesicles (Figs. 11, 12) are similar to those of vesicles reported to be involved in plant cell wall synthesis (Roland, 1973). Further evidence for a common biosynthetic origin for the plant cell walls and the infection threads arises from histochemical studies which suggest that the infection thread walls contain plant cell wall constituents (Dart, 1974).

The budding off from the infection threads of the rhizobia into the plant cytoplasm involves an endocytotic process as described for other legume systems (Goodchild & Bergersen, 1966; Dixon, 1967; Bergersen, 1974). Rhizobia were not observed free from enclosing membranes in the plant cytoplasm unless the plant cells were showing signs of senescence or damage due to poor fixation. The budding-off process, in lupin nodule cells, appears to depend on the occurrence of rhizobia in pocket-like protrusions of the infection thread membranes which contain very little wall material (Fig. 7). It is possible, however, that these pocket-like protrusions contain some rhizobial polysaccharides which are not stained by any of the staining techniques used.

The involvement of Golgi bodies in the production of cytoplasmic vesicles, which subsequently fuse with the infection thread membranes and peribacteroid membranes, is suggested by several observations. Golgi bodies are frequently observed in the cytoplasm of infected cells in the region of the tip of the infection thread and also throughout the cytoplasm of plant cells which contain large numbers of bacteroids. The Golgi bodies are often closely associated with cytoplasmic vesicles, some of which contain little or no electron-dense material, while others contain material with staining characteristics similar to those of the plant cell walls. Although the membranes of these vesicles stain with phosphotungstic acid and silver, the Golgi bodies themselves do not stain. This suggests either that the vesicles arise from some source other than the Golgi bodies or that the properties of the membranes of the vesicles change during budding off from the distal face of the Golgi bodies. Evidence supporting the conclusion that the Golgi bodies produce the vesicles is obtained from thin sections of nodule tissue fixed with zinc iodide-osmium tetroxide, which has been reported to stain the proximal face of Golgi bodies (Dauwalder & Whaley, 1972), and also from freeze-fracture preparations of nodule tissue. Changes in the properties of membranes on the distal faces of Golgi bodies, involved in the process of plant plasma membrane and cell wall synthesis have been reported by Roland (1973). Evidence for the involvement of Golgi bodies in the biogenesis of the encapsulation material surrounding the endophyte in Alnus crispa root nodules has been presented by Lalonde & Knowles (1975).

The process by which the cytoplasmic vesicles are directed to and recognize the target membranes, namely the peribacteroid membranes or infection thread membranes, is unknown. The study of thin sections suggests that microtubules might be involved in a manner similar to that described for plant cell wall synthesis (Packard & Stack, 1976; Northcote, 1969).

An obvious difference between the peribacteroid membranes and the plasma membrane is the absence of plant cell wall material from the surface of the peribacteroid membranes. Particles on fracture faces of plasma membranes have been suggested to be enzyme complexes involved in cell wall synthesis (Northcote, 1969) and in particular in the assembly of cellulose microfibrils (Mueller, Brown & Scott, 1976; Bailey & Northcote, 1977). Since the peribacteroid membranes arise initially from the plasma membranes it might be predicted that one difference between these membranes would be in the inability of the peribacteroid membranes to synthesize cell wall polymers. As a consequence of this the particle density on the E face of the peribacteroid membranes would be expected to be lower than that on the E face of the plasma membranes. This was in fact the case. The absence of particles on the E face of freeze-fracture replicas of invaginated plasma membranes in the haustoria of rust-infected flax has also been related to the lack of synthesis of host walls (Littlefield & Bracker, 1972).

The ability of rhizobia to induce the formation of infection threads and peribacteroid membranes in legumes suggests that these organisms are able to control, or at least influence, the systems of plant cell wall and membrane biosynthesis. Studies by Shore & MacLachlan (1975) indicate that wall-polymer synthesis in plants may be under hormonal control at the level of the Golgi bodies. Since rhizobia are known to produce auxins and cytokinins (Libbenga & Bogers, 1974; Syôno, Newcomb & Torrey, 1976) it seems possible that the process of infection of plant cells by rhizobia may involve hormonal control of the plant system of biosynthesis of cell walls and membranes.

The possible involvement of cellulolytic enzymes in the process of infection was raised by the apparent dispersion or degradation of cell wall material which appears to be released into the pocket-like protrusions of the infection thread membranes. The separation of the infection thread walls from the infection thread membranes could also suggest that degradation of wall material is occurring. As mentioned previously, rhizobia are known to produce auxins and Bal, Verma, Byrne & Mac-lachlan (1976) have detected cellulolytic enzymes in auxin-treated plant tissues. It is possible, therefore, that cellulolytic enzymes might be involved at some stage in the process of plant cell infection, although Hunter & Elkan (1975) were unable to detect any cellulolytic activity in soybean nodules.

Organelles described as lysosomes by Truchet & Coulomb (1973) and Tu (1976) could not be clearly identified in association with Golgi bodies or elsewhere in the cytoplasm of plant cells in lupin nodules in agreement with studies on Zea mays by Dauwalder, Whaley & Kephart (1969). It is possible, however, that cytoplasmic vesicles containing wall polymers (Fig. 9) might be classified as lysosomes after staining for phosphatase activity, since the outer surface of the plasma membrane which is derived from such vesicles shows heavy staining (Hall, 1969).

In conclusion we have presented evidence to suggest that the Golgi apparatus is involved in the biogenesis of the infection thread wall and membrane and also the peribacteroid membrane. It is proposed that the rhizobia infect plant cells by controlling the plant cell wall and plasma membrane systems of biogenesis.

We thank Wendy Ulyatt for growing the plants, Ivan Simpson and Douglas Hopcroft for photographic assistance, Keith Williamson for helpful advice, and Professor R. D. Batt for generous support.

Bailey
,
D. S.
&
Northcote
,
D. H.
(
1977
).
An ultrastructural study of the relationship between the plasma membrane and the cell wall of the coenocytic alga Hydrodictyon africanum
.
J.Cell Sci
.
23
,
141
149
.
Bal
,
A. K.
,
Verma
,
D. P. S.
,
Byrne
,
H.
&
Maclachlan
,
G. A.
(
1976
).
Subcellular localization of cellulases in auxin-treated pea
.
J. Cell Biol
.
69
,
97
105
.
Bergersen
,
F. J.
(
1973
).
Symbiotic nitrogen fixation by legumes
.
In Chemistry and Biochemistry of Herbage
, vol.
2
(ed.
G. W.
Butler
&
R. W.
Bailey
), pp.
189
226
.
London and New York
:
Academic Press
.
Bergersen
,
F. J.
(
1974
).
Formation and function of bacteroids
.
In The Biology of Nitrogen Fixation
(ed.
A.
Quispel
), pp.
473
498
.
Amsterdam & Oxford
:
North-Holland
.
Bergersen
,
F. J.
&
Briggs
,
M. J.
(
1958
).
Studies on the bacterial component of soybean root nodules: cytology and organisation in the host tissue
.
J, gen. Microbiol
.
19
,
482
490
.
Branton
,
D.
,
Bullivant
,
S.
,
Gilula
,
N. B.
,
Karnovsky
,
M. J.
,
Moor
,
H.
,
Mühlethaler
,
K.
,
Northcote
,
D. H.
,
Packer
,
L.
,
Satir
,
B.
,
Satir
,
P.
,
Speth
,
V.
,
Staehlin
,
L. A.
,
Steere
,
R. L.
&
Weinstein
,
R. S.
(
1975
).
Freeze-etching nomenclature
.
Science, N.Y
.
190
,
54
56
.
Bullivant
,
S.
(
1973
).
Freeze-etching and freeze-fracturing
.
In Advanced Techniques in Biological Electron Microscopy
(ed.
J. K.
Koehler
), pp.
67
112
.
New York
:
Springer-Verlag
.
Bullivant
,
S.
&
Ames
,
A.
(
1966
).
A simple freeze-fracture replication method for electron microscopy
.
J. Cell Biol
.
29
,
435
447
.
Costerton
,
J. W.
(
1970
).
The structure and function of the cell envelope of gram-negative bacteria
.
Rev. can. Biol
.
29
,
299
316
.
Dart
,
P. J.
(
1974
).
The infection process
.
In The Biology of Nitrogen Fixation
(ed.
A.
Quispel
), pp.
381
429
.
Amsterdam & Oxford
:
North-Holland
.
Dart
,
P. J.
&
Mercer
,
F. V.
(
1963
).
Development of the bacteroid in the root nodule of the barrel medic (Medicago tribuloides Desr.) and subterranean clover (Trifolium subterraneum L
.).
Arch. Mikrobiol
.
46
,
382
401
.
Dauwalder
,
M.
&
Whaley
,
W. G.
(
1972
).
The use of zinc iodide-osmium to demonstrate differences in the Golgi apparatus of cells of the Zea mays root tip
.
J. Cell Biol
.
55
,
55a
.
Dauwalder
,
M.
,
Whaley
,
W. G.
&
Kephart
,
J. E.
(
1969
).
Phosphatases and differentiation of the Golgi apparatus
.
J. Cell Sci
.
4
,
455
497
.
Dixon
,
R. O. D.
(
1967
).
The origin of the membrane envelope surrounding the bacteria and bacteroids and the presence of glycogen in clover root nodules
.
Arch. Mikrobiol
.
56
,
156
166
.
Dixon
,
R. O. D.
(
1969
).
Rhizobia (with particular reference to relationships with host plants)
.
A. Rev. Microbiol
.
23
,
137
158
.
Generozova
,
I. P.
&
Yagodin
,
B. A.
(
1972
).
Formation of the bacterial capsule membrane in the nodule of legumes
.
Fiziologiya Rast
.
19
,
348
353
.
Goodchild
,
D. J.
&
Bergersen
,
F. J.
(
1966
).
Electron microscopy of the infection and subsequent development of soybean nodule cells
,
J. Bad
.
92
,
204
213
.
Gourret
,
J.P.
&
Fernandez-Arias
,
H.
(
1974
).
Etude ultrastructurale et cytochimique de la différenciation des bactéroïdes de Rhizobium trifolii Dangeard dans les nodules de Trifolium repens L
.
Can. J. Microbiol
.
20
,
1169
1181
.
Graham
,
P. H.
(
1964
).
Studies on the utilisation of carbohydrates and Krebs cycle intermediates by rhizobia, using an agar plate method
.
Antonie van Leeuwhenhoek
30
,
68
72
.
Hall
,
J. L.
(
1969
).
Localization of cell surface adenosine triphosphatase activity in maize roots
.
Planta
85
,
105
107
.
Hoagland
,
D. R.
&
Arnon
,
D. T.
(
1938
).
The water-culture method for growing plants without soil
.
Univ. Calif, agrie, exp. Sta. Circular
,
110
.
347
.
Hunter
,
W. J.
&
Elkan
,
G. H.
(
1975
).
Role of pectic and cellulolytic enzymes in the invasion of the soybean by Rhizobium japonicum
.
Can. J. Microbiol
.
21
,
1254
1258
.
Jordan
,
D. C.
,
Grinyer
,
I.
&
Coulter
,
W. H.
(
1963
).
Electron microscopy of infection threads and bacteria in young root nodules of Medicago sativa
.
J. Bact
.
86
,
125
137
.
Lalonde
,
M.
, &
Knowles
,
R.
(
1975
).
Ultrastructure, composition, and biogenesis of the encapsulation material surrounding the endophyte in Alnus crispa var. mollis root nodules
.
Can. J. Bot
.
53
,
1951
1971
.
Libbenga
,
K. R.
&
Bogers
,
R. J.
(
1974
).
Root-nodule morphogenesis
.
In The Biology of Nitrogen Fixation
(ed.
A.
Quispel
), pp.
430
472
.
Amsterdam & Oxford
:
North-Holland
.
Littlefield
,
L. J.
&
Bracker
,
C. E.
(
1972
).
Ultrastructural specialization at the host-pathogen interface in rust-infected flax
.
Protoplasma
74
,
271
305
.
Mackenzie
,
C. R.
,
Vail
,
W. J.
&
Jordan
,
D. C.
(
1973
).
Ultrastructure of free-living and nitrogen-fixing forms of Rhizobium meliloti as revealed by freeze-etching
.
J. Bact
.
113
,
387
393.
Morré
,
D. J.
&
Mollenhauer
,
H. M.
(
1974
).
The endomembrane concept: a functional integration of endoplasmic reticulum and Golgi apparatus
.
In Dynamic Aspects of Plant Ultrastructure
(ed.
A. W.
Robbards
), pp.
84
137
.
London
:
McGraw-Hill
.
Mueller
,
S. C.
,
Brown
,
R. M.
&
Scott
,
T. K.
(
1976
).
Cellulosic microfibrils: Nascent stages of synthesis in a higher plant cell
.
Science, N. Y
.
194
,
949
951
.
Newcomb
,
W.
(
1976
).
A correlated light and electron microscope study of symbiotic growth and differentiation in Pisum sativum root nodules
.
Can. J. Bot
.
54
,
2163
2186
.
Niebauer
,
G.
,
Krawczyk
,
W. S.
,
Kidd
,
R. L.
&
Wilgram
,
G. F.
(
1969
).
Osmium zinc iodide reactive sites in the epidermal Langerhans cell
.
J. Cell Biol
.
43
,
80
89
.
Northcote
,
D. H.
(
1968
).
The organisation of the endoplasmic reticulum, the Golgi bodies and microtubules during cell division and subsequent growth
.
In Plant Cell Organelles
(ed.
J. B.
Pridham
), pp.
179
197
.
New York
:
Academic Press
.
Northcote
,
D. H.
(
1969
).
The synthesis and metabolic control of polysaccharides and lignin during differentiation of plant cells
.
In Essays in Biochemistry
, vol.
5
(ed.
P. N.
Campbell
&
G. D.
Greville
), pp.
89
137
.
London & New York
:
Academic Press
.
Northcote
,
D. H.
(
1972
).
Chemistry of the plant cell wall
.
A. Rev. Pl. Physiol
.
23
,
113
132
.
Nutman
,
P. S.
(
1965
).
Origin and developmental physiology of root nodules
.
In Encyclo-paedia of Plant Physiology
, vol.
xv/1
(ed.
W.
Ruhland
), pp.
1355
1379
.
New York
:
Springer-Verlag
.
Packard
,
M. J.
&
Stack
,
S. M.
(
1976
).
The preprophase band: possible involvement in the formation of the cell wall
.
J. Cell Sci
.
22
,
403
411
.
Preston
,
R. D.
(
1974
).
Plant cell walls
.
In Dynamic Aspects of Plant Ultrastructure
(ed.
A. W.
Robards
), pp.
256
309
.
London
:
McGraw-Hill
.
Quintarelli
,
G.
,
Bellocci
,
M.
&
Geremia
,
R.
(
1973
).
On phosphotungstic acid staining. IV. Selectivity of the staining reagent
.
J. Histochem. Cytochem
.
21
,
155
160
.
Robertson
,
J. G.
,
Farnden
,
K. J. F.
,
Warburton
,
M. P.
&
Banks
,
J. M.
(
1975
).
Induction of glutamine synthetase during nodule development in lupin
.
Aust. J. Pl. Physiol
.
2
,
265
272
.
Robertson
,
J. G.
,
Lyttleton
,
P.
,
Williamson
,
K. I.
&
Batt
,
R. D.
(
1975
).
The effect of fixation procedures on the electron density of polysaccharide granules in Nocardia corallina
.
J. Ultrastruct. Res
.
52
,
321
332
.
Robertson
,
J. G.
,
Taylor
,
M. P.
,
Craig
,
A. S.
&
Hopcroft
,
D. H.
(
1974
).
Development of membrane systems in legume root nodules
.
In Mechanisms of Regulation of Plant Growth
(ed.
R. L.
Bieleski
,
A. R.
Ferguson
&
M. M.
Cresswell
), pp.
31
36
.
Wellington
:
The Royal Society of New Zealand
.
Roland
,
J. C.
(
1973
).
The relationship between the plasmalemma and plant cell wall
.
Int. Rev. Cytol
.
36
,
45
92
.
Roland
,
J. C.
,
Lembi
,
C. A.
&
Morré
,
D. J.
(
1972
).
Phosphotungstic acid-chromic acid as a selective electron-dense stain for plasma membranes of plant cells
.
Stain Technol
.
47
,
195200
.
Saltón
,
M. R. J.
(
1967
).
Structure and function of bacterial cell membranes
.
A. Rev. Microbiol
.
21
,
4I7
442
.
Scott
,
J. E.
&
Webb
,
J.
(
1975
).
Phosphotungstate: phospholipid interactions. Direct staining and plasmal reactions
.
J. Histochem. Cytochem
.
23
,
306
308
.
Shore
,
G.
&
Maclachlan
,
G. A.
(
1975
).
The site of cellulose synthesis. Hormone treatment alters the intracellular location of alkali-insoluble β-1,4-glucan (cellulose) synthetase activities
.
J. Cell Biol
.
64
,
557
57I
.
SyŌno
,
K.
,
Newcomb
,
W.
&
Torrey
,
J. G.
(
1976
).
Cytokinin production in relation to the development of pea root nodules
.
Can. J. Bot
.
18
,
2155
2162
.
Thiéry
,
J. P.
(
1967
).
Mise en évidence des polysaccharides sur coupes fines en microscopie électronique
.
J. Microscopie
6
,
987
1018
.
Truchet
,
G.
&
Coulomb
,
Ph
. (
1973
).
Mise en évidence et évolution du système Phytolyso-somal dans les cellules des différentes zones de nodules radiculaires de Pois (Pisum sativum L.). Notion d’hétérophagie
.
J. Ultrastruct. Res
.
43
,
36
57
.
Tu
,
J. C.
(
1975
).
Structural similarity of the membrane envelopes of rhizobial bacteroids and the host plasma membrane as revealed by freeze-fracturing
.
J. Bact
.
122
,
691
694
.
Tu
,
J. C.
(
1976
).
Lysosomal distribution and acid phosphatase activity in white clover infected with clover yellow mosaic virus
.
Phytopathology
66
,
588
593
.
Venable
,
J. H.
&
Cocgeshall
,
R.
(
1965
).
A simplified lead citrate stain for use in electron microscopy
.
J. Cell Biol
.
25
,
407
408
.
Whaley
,
W. G.
(
1975
).
The Golgi Apparatus: Cell Biology Monograph
, vol.
2
(ed.
M.
Alfert
,
W.
Beermann
,
G.
Rudkin
,
W.
Sandritter
&
P.
Sitte
).
New York
:
Springer-Verlag
.
     
  • b

    bactcroid

  •  
  • bi

    bacteroid envelope inner membrane

  •  
  • cv

    cytoplasmic vesicle

  •  
  • cw

    cell wall

  •  
  • er

    endoplasmic reticulum

  •  
  • g

    Golgi body

  •  
  • itm

    infection thread membrane

  •  
  • itiu

    infection thread wall

  •  
  • pb

    peribacteroid membrane

  •  
  • pd

    plasmodesmata

  •  
  • Pi

    plasma membrane

  •  
  • r

    rhizobium

  •  
  • s

    starch granule

  •  
  • V vacuole
*

This nomenclature (Robertson, Taylor, Craig & Hopcroft, 1974) was adopted to avoid the use of the term ‘envelope’ (Dixon, 1967) which is also used to describe the cell wall of gram-negative bacteria of which the species Rhizobium is an example. The space between the peribacteroid membrane and the bacteroid envelope outer membrane is called the peribacteroid space, which seems appropriate since the space between the inner and outer membranes of the envelopes of gram-negative organisms is called the periplasmic space (Saltón, 1967; Costerton, 1970). The term ‘bacteroid’ (see review by Bergersen, 1973) is used to refer to all organisms which have budded off from the infection threads into the cytoplasm of the plant cells since we were unable to distinguish between N2-fixing and non-N2-fixing organisms in thin sections of lupin nodules.