Peribacteroid membranes and bacteroid envelope inner membranes have been isolated from developing lupin nodules. Isolation of the peribacteroid membranes was achieved by first preparing membrane-enclosed bacteroids free from other plant organelles or membranes. The peribacteroid membranes were then released by osmotic shock and purified by centrifugation to equilibrium on sucrose gradients. The bacteroids were broken in a pressure cell and the bacteroid envelope inner membranes were isolated using sucrose gradient fractionation of the bacteroid total envelope preparation. The density of the peribacteroid membranes decreased during the period of development of N2-fixation in lupin nodules from 1·148 g/ml for nodules from 12-day plants to 1·137 g/ml for nodules from 18-day plants. The density of the bacteroid envelope inner membranes from nodules from 18-day plants was 1·153 g/ml. The identity and homogeneity of the isolated membranes was established, by comparison with membranes in intact nodules, using phosphotungstic acid and silver staining of thin sections and particle densities on faces of freeze-fracture replicas of the membranes.
Analyses for NADH oxidase and succinate dehydrogenase, spectral analyses and gel-electro-phoretic analysis of proteins were also used to characterize the membrane and soluble protein fractions from the nodules. The ratio of lipid to protein was 6·1 for the peribacteroid membranes and 2·5 for the bacteroid envelope inner membranes. Leghaemoglobin was localized in the plant cytoplasm in lupin nodules and not in the peribacteroid space.
Movement of metabolites such as oxygen (Appleby, Turner & Macnicol, 1975), ammonia (Scott, Farnden & Robertson, 1976) and possibly organic acids (Bergersen & Turner, 1967) between the plant cytoplasm and the bacteroid-form of rhizobia in legume root nodules must involve transport through the peribacteroid membranes (Robertson, Lyttleton, Bullivant & Grayston, 1977) and the bacteroid envelope outer and inner membranes (Van Brussel, 1973). Although the outer membranes of the cell envelopes of gram-negative organisms, such as rhizobia, function as a barrier to macromolecules, the inner membranes constitute the major barrier to substances of low molecular weight (reviewed by Costerton, 1970). It seems likely that the peribacteroid membranes also have properties which influence the rate of movement of low molecular weight substances, since these membranes have originated from the plasma membranes (Robertson et al. 1977) which are considered to be important sites of active transport in plant cells (Bowling, 1976). The peribacteroid membranes are also considered to restrict leghaemoglobin to the spaces surrounding the bacteroids (Bergersen & Goodchild, 1973; reviewed by Appleby, 1974) although this has recently been disputed (Verma & Bal, 1976). Little is known about the functions of the bacteroid envelope outer and inner membranes except that, in comparison with free-living rhizobia, the envelopes are modified in some way which makes bacteroids more susceptible to osmotic shock after treatment with lysozyme (Van Brussel, 1973; Mackenzie, Vail & Jordan, 1973). Bacteroid envelope preparations have been shown to contain an electron-transfer system which is probably involved in N2-fixation (Appleby, 1969).
Questions regarding the role of the peribacteroid membranes and the bacteroid envelope inner membranes in the N2-fixing system in legume nodules cannot be satisfactorily answered until these membranes have been isolated and purified. In a preceding paper (Robertson et al. 1977) we characterized the peribacteroid membranes and the bacteroid envelope inner membranes in intact lupin nodules using thin section and freeze-fracture electron-microscopic techniques. In this paper we report for the first time the isolation of the peribacteroid membranes from nodules from 12- and 18-day lupin plants which represent the early and late stages of development of symbiotic N2-fixation (Robertson, Farnden, Warburton & Banks, 1975a). The preparation of the bacteroid envelope inner membranes from nodules from 18-day plants is also described.
MATERIALS AND METHODS
Chemicals and biochemicals
All chemicals were purchased from commercial sources and were either analytical grade or laboratory reagent grade.
Bovine serum albumin (Sigma Fraction V), dialysed against distilled water and freeze-dried, DNase (Worthington) and ferritin (Type 1, Sigma) were used. Leghaemoglobin, used for gel electrophoresis, was prepared from the plant soluble-protein fraction from nodules from 25-day lupin plants (Lupinus angustifolius L. cv. Bitter blue), using ammonium sulphate precipitation, oxidation to ferric leghaemoglobin using potassium ferricyanide and finally chromatography on G75 Sephadex (Pharmacia) as reported by Appleby (1974).
Preparation of nodule tissue for fractionation
Germinated lupin seeds (Lupinus angustifolius L. cv. Uniwhite) were inoculated at sowing with Rhizobium lupini strain NZP2257 from an agar slope culture and grown as described previously (Robertson et al. 1977). Nodules were first visible 9 days after inoculation of germinated seeds and the onset of N2-fixation occurred 11 days after inoculation. Full development of N2-fixation in nodules was reached approximately 20 days after inoculation (Robertson et al. 1975).
The procedure for preparing the peribacteroid membranes, the bacteroid total envelope fraction and the bacteroid envelope inner membranes is summarized in Figs. 1 and 2. For each 7 g wet weight of nodule tissue one centrifuge tube was used at each stage of the fractionation. Throughout the procedures sucrose concentrations, expressed as g/100g of solution, were monitored at 20 °C using an Abbé refractometer. The buffer was tris (hydroxymethyl) aminomethane-HCl (Tris-HCl), pH 8·0, and the preparations were carried out at 0·4 °C. Centrifugal forces are given for Rmax. All stages of the fractionation procedures were assessed for purity using electron microscopy.
Preparation of the peribacteroid membranes
Nodules were picked into 16% sucrose, 50 mM Tris, 10 mM dithiothreitol (Robertson, Warburton & Farnden, 1975b) and weighed. The wet weight of nodules consistently gave a mean value of 65 mg/plant from 12-day plants and 165 mg/plant from 18-day plants. Nodules were crushed in 2 vol. (w/v) of 16% sucrose, 50 mM Tris, 10 mM dithiothreitol, using a mortar and pestle (Fig. 1). The brei was filtered through 2 layers of Miracloth (Chicopee Mills Inc., New York) and centrifuged in a 50-ml tube at 250 g for 5 min using a Sorvall HB-4 rotor. The pellet (P1) contained starch granules and plant cell debris. The supernatant (SN1) was recentrifuged at 8000 g for 5 min using the Sorvall HB-4 rotor to yield a pellet (P2), consisting of a crude preparation of bacteroids, and a supernatant (SN2). Further centrifugation of the supernatant (SN2) at 100000 g for 30 min using a Beckman 30 rotor, yielded a pellet of microsomes and mitochondria (P3) and a supernatant (SN3) containing the plant soluble-protein fraction. The pellet (P2) was resuspended in 10 ml 16% sucrose, 50 HIM Tris, 2 mM dithiothreitol and this suspension was layered over 10 ml 55% sucrose, 50 mM Tris. Centrifugation at 16000 g for 15 min using a Sorvall HB-4 rotor removed starch-containing plastids (P5). The bacteroids which formed a firm pellicle (P4) at the 16/55% sucrose interface, were removed using a syringe equipped with a 16-cm 18-gauge needle and the volume and sucrose concentration were adjusted to 7 ml 30% sucrose, 50 mM Tris. This suspension was layered over 10 ml 35% sucrose, 50 mM Tris and centrifuged at 16 000 g for 15 min using a Sorvall HB-4 rotor. Under these conditions the bacteroids pelleted (P 6) and residual mitochondria remained in the 30% sucrose layer. Both the 30 and 35% sucrose layers were discarded taking care not to disturb the pellet of bacteroids (P8), the majority of which were still surrounded by peribacteroid membranes. The membrane-enclosed bacteroids (P6) were osmotically shocked by resuspending with rapid stirring in 7 ml 10 mM Tris. This suspension was layered over 10 ml 35% sucrose, 10 mM Tris and centrifuged at 16 000 g for 15 min using a Sorvall HB-4- rotor to give a pellet of osmotically shocked bacteroids (P 7). The peribacteroid membranes collected on top of the 35% sucrose layer and 10 ml of supernatant (SN4) were removed. The supernatant (SN4) was adjusted to 8% sucrose, 10 mM Tris, layered over 12 ml 12% sucrose, 10 mM Tris and centrifuged at 60000 g for 30 min using a Beckman SW 25 · 1 rotor. The pellet (P8) consisted of peribacteroid membranes and the supernatant (SN5) contained soluble proteins released during osmotic shock of the bacteroids (P6). The pellet (P8) was resuspended in 10 ml 47% sucrose, 10 mM Tris, overlaid with 3 ml 45% sucrose, 10 mM Tris followed by a linear gradient of 22 ml from 45 to 20% sucrose, 10 mM Tris. Centrifugation to equilibrium at 90000 g for 17 h at 2 ± 1 °C using a Beckman SW27 rotor gave a single band at 31 · 5% sucrose, which was isolated, diluted with 10 mM Tris and centrifuged at 73000 g for 25 min using a Beckman 30 rotor. The pellet of peribacteroid membranes was finally resuspended at approximately 0·5 mg protein/ml in 10 mM Tris and analysed immediately or stored under liquid nitrogen. The time between picking nodules and completing the isolation of the membranes was approximately 24 h. The yield of membranes was approximately 3 μg protein/g nodule (wet weight) from 12-day plants and from 18-day plants approximately 20 μg protein/g nodule.
Preparation of the bacteroid total envelope fraction and the bacteroid envelope inner membranes
Preparation of the bacteroid total envelope fraction and the bacteroid envelope inner membranes (Fig. 2) from nodules from 18-day plants was carried out using osmotically shocked bacteroids (P7) which were resuspended in 3·5 ml 12% sucrose, 10 mM Tris, 0·1 mM MgCl2, 20 μg/ml DNase, and then passed through a pressure cell (Aminco) at 400 kg cm−2 (3·9 ×104 kN m−2). The pressate was held at 0–4 °C for 30 min and then centrifuged at 10000 g for 5 min using a Sorvall HB-4 rotor to give a supernatant (SN6) and a pellet (P9) of unbroken bacteroids. The supernatant (SN 6) contained bacteroid soluble proteins and the bacteroid total envelope fraction, which consisted of bacteroid envelopes and bacteroid envelope inner and outer membranes detached from each other. The bacteroid total envelope fraction was prepared (Fig. 2) by layering the supernatant (SN 6) over a step-gradient of 8 ml 20% sucrose, 10 mM Tris over 5 ml 55% sucrose, 10 mM Tris, centrifuging at 50000 g for 30 min using a Beckman 30 rotor and removing the membranes from the 20/55% sucrose interface using a syringe.
The bacteroid envelope inner membranes were prepared (Fig. 2) by layering the supernatant (SN8) over a step-gradient of 10 ml 17% sucrose, 10 mM Tris over 10 ml 40% sucrose, 10 mM Tris and centrifuging at 73000 g for 75 min using a Beckman SW 25.1 rotor to give a bacteroid envelope inner membrane fraction (P10) at the 17/40% sucrose interface and a pellet (Pu) consisting of bacteroid envelopes and bacteroid envelope outer membranes. The bacteroid envelope inner membrane fraction (P10) was collected using a syringe, adjusted to 10 ml 47% sucrose, 10 mM Tris and overlaid with 3 ml 45% sucrose, 10 mM Tris followed by a linear gradient of 22 ml from 45–20% sucrose, 10 mM Tris. After centrifuging at 90000 g at 2 °C for 17 h using a Beckman SW 27 rotor, the bacteroid envelope inner membranes formed a broad band from 30 to 42% sucrose. The fraction banding between 30 and 38% sucrose was removed, diluted with 10 mM Tris and centrifuged at 73 000 g for 25 min using a Beckman 30 rotor. The pellet consisting of bacteroid envelope inner membranes was finally resuspended at approximately 0-5 mg protein/ml in 10 mM Tris and analysed immediately or stored under liquid nitrogen. The yield of bacteroid envelope inner membranes was approximately 15 μg protein/g nodule (wet weight) from 18-day plants.
Thin sections. All fixation, washing and dehydration steps were carried out at 0·4 °C. Suspensions of bacteroids or membranes in 1·5-ml conical propylene centrifuge tubes were fixed by the addition of an equal vol. of suspension medium containing 4% glutaraldehyde. After 1 h the suspension was diluted, if necessary, to just under 16% sucrose by slow addition of 0·1 M phosphate buffer, pH 7·2, and centrifuged at 10000g for 10 min using a Sorvall HB-4 rotor. Under these conditions the fixed material formed a firm pellet. A major portion of the supernatant was removed and carefully replaced (twice) with phosphate buffer. The pellet was then fixed in 1% osmium tctroxide in phosphate buffer for 1 h, washed in 1% sodium chloride (3 times), postfixed in 0-5% uranyl magnesium acetate, dehydrated through an acetone series and embedded in Epon 812 (TAAB). At the 50% resin stage the tip of the centrifuge tube was cut off, just above the pellet, and cut in half. The pellets which became detached from the bisected tip were oriented in the final resin so that thin sections from top-to-bottom of the pellets could be obtained for electron microscopy.
Membranes were also collected under vacuum on a Millipore filter (50-nm pore size) and then covered with a second filter which was clamped to the first using stainless steel rings. The whole unit was taken through the fixation and dehydration steps and the thin layer of fixed membranes was embedded in Epon 812.
Thin sections were cut on an LKB microtome fitted with a diamond or glass knife and stained using one of the following methods. Uranium and lead staining was carried out at room temperature by immersing sections mounted on copper grids, without support films, in saturated uranyl acetate in 50% ethanol for 2 min followed by lead citrate (Venable & Coggeshall, 1965) for 2 min. Phosphotungstic acid staining was carried out by floating sections mounted on gold grids on 1% periodic acid at room temperature for 30 min followed by immersion in 1% phosphotungstic acid in 1 M HC1 at 30 °C for 10 min. Silver staining was carried out on sections mounted on gold grids using the method of Thiéry (1967). Grids were floated successively at room temperature on 1% periodic acid for 30 min, 0·2% thiocarbohydrazide in 20% acetic acid for 30 min and 1% silver proteinate for 30 min with appropriate washing steps between each reagent.
Bacteroid and membrane fractions were prepared for freeze-fracturing by mixing pellets of the appropriate fractions with 50% glycerol/water. Sufficient of the suspension (approx. 10 /d) was transferred to a small brass cup to give a drop protruding from the top of the cup. After 15 min from the time of addition of the glycerol to the pellets, the suspensions were frozen by dropping the cups into melting Freon 12. Alternatively bacteroids or membranes were fixed in suspension using glutaraldehyde as described for thin-section preparations and centrifuged. The pellets were then mixed with 50% glycerol/water and the suspensions stored for 48 h at 0·4 °C before freezing. Freeze fracturing was carried out using the method of Bullivant & Ames (1966) and Bullivant (1973).
Freeze-fracture replicas and stained sections were examined with a Philips EM 200 electron microscope at 60 kV or a Philips EM 301 at 80 kV. Particle densities on freeze-fracture replicas were determined from electron micrographs as described previously (Robertson et al. 1977). Particles were counted on 20–30 faces of freeze-fracture replicas of membranes selected randomly in each preparation and means and standard errors of means are presented.
Fractions were either analysed immediately or, in the case of soluble-protein fractions containing sucrose, after passing through Sephadex G25 (Pharmacia) using the method of Neal & Florini (1973) and concentrating in dialysis sacs over dry Sephadex G200 at 0·4 °C. Analyses were carried out in duplicate and means are presented.
Protein estimation (Lowry, Rosebrough, Farr & Randall, 1951) was carried out following precipitation of samples with trichloroacetic acid, using bovine serum albumin as standard. Total nitrogen (Jaenicke, 1974) was determined after dialysing membrane preparations against 10 mM sodium 3,3-dimethyl glutarate buffer, pH 7·6, for 48 h at 0·4 °C. Total lipid was extracted from membranes, suspended in 10 mM Tris, using chloroform and methanol (Folch, Lees & Sloane Stanley, 1957). The extract was taken to dryness, the residue was redissolved in chloroform-methanol (2:1, v/v), passed through a small column of silicic acid (Unisil, Clarkson Chemical Co.) and finally dried to constant weight under vacuum and over silica gel. Leghae-moglobin was determined using the pyridine haemochromogen method (Bergersen, Turner & Appleby, 1973). Absorbance spectra were determined using a Varian Techtron 635 spectro-photometer at 20 °C with a band width of 0·2 nm and 10-mm light path cuvettes.
Gel electrophoresis of proteins was carried out after adjusting the final concentration of the fractions to 2% sodium dodecyl sulphate, 0·1% dithiothreitol, boiling for 2 min and, in the case of membrane preparations, centrifuging at 100000 g for 30 min using a Beckman 30 rotor at 20 °C. Centrifugation removed a small amount of material which otherwise collected at the origins of the gels, causing the protein bands to smear. After addition of small amounts of glycerol and bromophenol blue, aliquots containing 50–100 μg of protein were applied to tube gels, 10 cm in length, consisting of 15% acrylamide, 0·08% N,N′-methylenebisacrylamide, 0·1% sodium dodecyl sulphate, 0·1 M Tris-0·1 M glycine, pH 8·9 (Cashmore, 1976). Proteins were subjected to electrophoresis at 3 mA/gel and stained with Coomassie blue.
NADH oxidase was assayed at 25 °C using the method of Crane (1957). Succinate dehydro-genase (E.C. i.3.99.1) was assayed at 25 °C using modifications of the methods of Bonner (1955) and Singer, Oestreicher, Hogue, Contreiras & Brandao (1973). The reaction mixture contained: 20-50μl membrane preparation; 200μl 0·5 M KH2PO4 buffer, pH 7·2; 100μl 0·14 M Na succinate; 100 μl 0·05 M Na malonate (where required); 100 μl0·01 M K3Fe(CN) 6; 100μl 0·1 M KCN (neutralized); and water to a final volume of 1 ml. The membrane preparation was incubated for 20 min with the phosphate buffer, succinate and malonate (where required) before addition of KCN and K3Fe(CN)6. The decline in absorbance at 420 nm was recorded and the activity calculated from the difference between activities in the presence and absence of malonate.
The object of the sucrose step-gradient fractionation of nodule tissue (Fig. 1) was to isolate membrane-enclosed bacteroids free from all other plant membranes and organelles. An electron-microscopic study of thin sections of the fraction pelleted through 35% sucrose (P6, Fig. 1) showed that the preparation consisted entirely of bacteroids of which 53, 62 and 73%, in 3 separate experiments, were still surrounded by peribacteroid membranes (Figs. 3, 4). Similar results were obtained from freeze-fracture studies of this preparation (Fig. 12). The particle density on the freeze-fracture faces of the peribacteroid membranes in a preparation of membrane-enclosed bacteroids from nodules from 18-day plants was 2500±190 particles/μm2 for the P face and 240 ±42 particles/μm2 for the E face (Fig. 12). In general the fracture plane followed the peribacteroid membranes rather than cross-fracturing through the membrane-enclosed bacteroids. The bacteroids appeared shrunken (Figs. 3, 4) in comparison with bacteroids in intact tissue (Robertson et al. This shrinkage was observed to occur when nodules were crushed and bacteroids were isolated in sucrose concentrations above 0·4 M. The bacteroid envelope inner and outer membranes were clearly visible (Fig. 5), although the inner membranes were sometimes difficult to distinguish from the bacteroid cytoplasm if postfixation in uranyl magnesium acetate was omitted from the fixation schedule (Fig. 6). Where the post-fixation step had been included the outer membranes stained asymmetrically with uranium and lead (Fig. 5). In some bacteroids spaces occurred between the envelope inner and outer membranes (Fig. 5) in the region normally occupied by murein in the envelopes of free-living rhizobia (Fig. 7).
The peribacteroid membranes were released from the bacteroids (P6, Fig. 1) by osmotic shock and purified by centrifugation to equilibrium on sucrose gradients (Fig. 8). Peribacteroid membranes from nodules from 12-day plants banded at 34·3 ± 0·1% sucrose (equivalent to a density of 1·148 ± 0·004 g/ml), while membranes from nodules from 18-day plants banded at 32 ·0 ± 0 ·2% sucrose (1 ·137 ± 0 ·005 g/ml). These results are the means and standard errors of means for 3 separate preparations of membranes from nodules at both stages of development.
An electron-microscopic study of thin sections cut from top to bottom of a pellet of peribacteroid membranes isolated from a linear sucrose gradient (Fig. 9) showed a homogeneous preparation of smooth vesicles 0 ·2-1 ·2 μm in diameter (Fig. 10). A study of thin sections from a pellet obtained by Millipore filtration gave identical results. The peribacteroid membranes stained positively with phosphotungstic acid, although the intensity of staining of vesicles varied over a thin section. Staining of vesicles with silver (Fig. 11) was highly uniform from top to bottom of the pellet.
A freeze-fracture study of the peribacteroid membranes isolated from nodules from both 12- and 18-day plants also showed the presence of vesicles (Fig. 13) of which over 98% were recognizable as peribacteroid membranes on the basis of the particle densities on the freeze-fracture faces. The density of particles on freeze-fracture faces of vesicles in a preparation of peribacteroid membranes from nodules from 12-day plants was 2320 ± 105 particles/μm2 on the P face and 490 ± 38 particles/μm2 on the E face and, in a preparation from nodules from 18-day plants, 2250 ± 140 particles/μm2 on the P face and 210 ± 54 particles/μm2 on the E face. Freeze-fracture preparations of peribacteroid membranes (Figs. 13 –15) showed P and E faces in both concave and convex forms in approximately equal numbers, demonstrating that the preparation was a mixture of right-side-out and inside-out vesicles. Vesicles enclosing smaller vesicles, observed in freeze-fracture (Fig. 13) and also in thin-section preparations (Fig. 9), of peribacteroid membranes were found to have particle densities on both P and E faces which were the same as those of the larger vesicles.
Aggregation of particles on freeze-fracture faces of the peribacteroid membranes was observed under certain conditions of preparation for freeze-fracturing. Little or no aggregation occurred when membranes, freshly collected from the final linear sucrose gradient (Fig. 1), were mixed with glycerol for 15 min before freezing (Figs. 14, 15). If membrane samples were fixed with glutaraldehyde and then held in glycerol for 48 h before freezing, particle aggregation occurred (Fig. 16).
Osmotically shocked bacteroids
Two distinct types of bacteroids were observed in these preparations; a plasmolysed form with cytoplasm of high electron density and a non-plasmolysed form with cytoplasm of low electron density (Figs. 17, 18). A count of more than 1000 bacteroids from nodules from 12-day plants showed that 65% occurred in the plasmolysed form whereas 85% of a similar number of bacteroids from nodules from 18-day plants were plasmolysed. Plasmolysis was not observed in free-living rhizobia harvested from an agar slope and subjected to the same sucrose gradient and osmotic shock protocol used to isolate the bacteroids (P7, Fig. 1).
A freeze-fracture study of a preparation of shocked bacteroids (Fig. 22) also showed that a majority of bacteroids were plasmolysed. The particle density on freeze-fracture faces of bacteroid envelope inner membranes in a preparation of osmotically shocked bacteroids was 5000 ± 133 particles/μ m2 for the P face and 1530 ± 79 particles/μ m2 for the E face. In general the fracture plane followed the bacteroid envelope inner membranes. The convex and concave faces of the outer membranes were rarely observed (Fig. 22).
Bacteroid envelope membranes
The bacteroid total envelope fraction, prepared as described in Fig. 2 from nodules from 18-day plants, was found to contain bacteroid envelopes as well as envelope inner and outer membranes detached from each other. The bacteroid envelope outer membranes tended to roll up on themselves while the inner membranes formed vesicles (Figs. 19, 23). In rare cases the inner membranes fused with the outer membranes or formed vesicles at the free ends by rolling back on themselves (Fig. 19). Particles, 5 – 10 nm in diameter, were often attached to the cytoplasmic surface of the inner membranes (Fig. 19).
A sucrose gradient fractionation of a preparation of bacteroid total envelopes (Fig. 8) followed by an electron-microscopic study of thin sections of membranes isolated from the gradient showed that the membranes banded as follows: 34 · 5 – 40% sucrose, bacteroid envelope inner membranes (Fig. 20); 41 – 43 · 5% sucrose, inner membranes containing cytoplasm with a band of unknown composition; 44 · 7 – 48% sucrose, bacteroid envelope inner and outer membranes still attached to each other; 49-53% sucrose, bacteroid envelope outer membranes (Fig. 21) and also inner and outer membranes still attached to each other and containing cytoplasm.
The bacteroid envelope inner membranes were routinely prepared (Fig. 2) by sucrose gradient fractionation of the bacteroid total envelopes (SNG, Fig. 2). The mem branes banding between 30 and 38% sucrose on the final linear sucrose gradient (Fig. 8) consisted of a homogeneous preparation of vesicles (identical to Fig. 20) 0 · 2 – 1 · 0 μ m in diameter. On recentrifugation to equilibrium on a second linear sucrose gradient these membranes banded at 35 · 4% sucrose, equivalent to a density of 1 · 153 g/m 1 (Fig. 8). The 5 to 10-nm particles observed on the cytoplasmic surfaces of the bacteroid envelope inner membranes (Fig. 19) were also observed attached to either the inside or the outside, but not both, surfaces of the vesicles, although they were not always so clearly defined as in Fig. 19. The membranes did not stain with phosphotungstic acid and only very lightly with silver, in contrast to the heavy staining of the peribacteroid membranes.
A freeze-fracture study of a preparation of bacteroid envelope inner membranes (Fig. 24) showed that the particle density on the P freeze-fracture face was 4410 ±149 particles/μ m2 and on the E face 1390 ± 48 particles/μ m2. The fraction consisted of an equal mixture of right-side-out and inside-out vesicles. Contamination of the bacteroid envelope inner membranes with membranes which could not be identified as inner membranes was less than 5%. This included slight contamination from bacteroid envelope outer membranes and peribacteroid membranes.
Composition of the peribacteroid membranes and bacteroid envelope inner membranes
From the absorption spectra of membranes from nodules from 18-day plants (Fig. 25) the absorbance at 280 nm for a 1 mg protein/ml solution was calculated to be 71 for the peribacteroid membranes and 7 · 4 for the bacteroid envelope inner membranes. The oxidized form of the inner membranes showed a Soret band at 411 nm. At higher protein concentration (Fig. 25), in the presence of deoxycholate, the oxidized form of the inner membranes showed shoulders at 525 and 550 nm. On reduction with dithionite the Soret band showed a slight loss in intensity and shifted to 418 nm. At the same time peaks appeared at 560 and 530 nm, corresponding to the a and ft bands of cytochrome c. The failure to detect absorption bands near 600 nm indicated that cytochrome a was not present. The activity of NADH oxidase in the bacteroid envelope inner membrane preparations was 0-029 μ mol NADH/min/mg protein and the activity of succinate dehydrogenase was 0 · 27 μ mol succinate/min/mg protein.
The absorption spectrum of the peribacteroid membranes showed only a slight shoulder at 415 nm in the oxidized form, which shifted to 425 nm on addition of dithionite (Fig. 25). No absorption bands were observed in the 500 – 600 nm region in either the presence or absence of dithionite. No NADH oxidase or succinate dehy-drogenase activity was detected in peribacteroid membrane preparations and the level of leghaemoglobin, as determined using the alkaline haemochromogen method, was below the measurable level of 0 · 5 nmol haemoglobin/mg protein.
The lipid-to-protein ratio for the peribacteroid membranes, using the Lowry method for protein analysis, was 6-i and for the bacteroid envelope inner membranes, 2 · 5. On the basis of total nitrogen the lipid-to-protein ratio for the peribacteroid membranes was 5 · 0 and for the inner membranes, 1 · 8.
Gel electrophoretic analysis of membrane proteins showed that the protein pattern for the bacteroid envelope inner membranes differed from that for the peribacteroid membranes and no cross-contamination could be detected (Fig. 26). Leghaemoglobin, which consisted of two species of mobility 0 · 80 and 0 · 82 relative to the dye front in this electrophoretic system (Fig. 27), was not detected in peribacteroid membrane preparations.
Localization of leghaemoglobin in lupin nodules
Isolation of membrane-enclosed bacteroids raised the possibility that these bodies might contain leghaemoglobin. The level of this protein, however, in the soluble protein fraction (SN5, Fig. 1), obtained following osmotic shock of the membrane- enclosed bacteroids from 18-day plants, was below the measurable level of 0 · 5 nmol haemoglobin/mg protein. By comparison, the level of leghaemoglobin in the plant soluble-protein fraction (SN3, Fig. 1) was 12 nmol/mg protein. Electrophoresis of soluble proteins in the plant fraction (SN3, Fig. 1), the shock fraction (SN5, Fig. 1) and the bacteroid fraction (SN7, Fig. 2) confirmed that leghaemoglobin, which consisted of 2 species of slightly differing electrophoretic properties, occurred in only trace amounts, if at all, in the shock fraction (Fig. 27). The electrophoretic analysis also demonstrated that the 3 soluble protein fractions differed in composition. To establish whether leghaemoglobin could have leaked out through the peribacteroid membranes during isolation of the membrane-enclosed bacteroids, ferritin was included at all stages of nodule fractionation including crushing of the nodules and glutaraldehyde fixation of the pellet (P6, Fig. 1). Ferritin, which was clearly observed outside the peribacteroid membranes (Fig. 6) was detected in the peribacteroid spaces in fewer than 1% of the membrane-enclosed bacteroids.
The homogeneity of the isolated peribacteroid membranes and the bacteroid envelope inner membranes was established by comparing the properties of these membranes with the properties of the membranes in intact nodules (Robertson et al. 1977). The peribacteroid membranes stained positively with phosphotungstic acid but staining was so variable as to render this method unsatisfactory for determining the homogeneity of the membrane samples. This observation is in agreement with recent studies by Flowers & Hall (1976). Phosphotungstic acid staining was therefore used to indicate the presence or absence of the peribacteroid membranes in membrane preparations and not to determine the degree of purity of the preparations.
Silver staining of peribacteroid membrane vesicles was highly uniform, and non-staining contaminants could be clearly recognized in thin sections. Staining with silver was particularly useful once the membrane-enclosed bacteroids had been isolated free from other plant organelles and membranes, since the bacteroid envelope membranes stained only very lightly compared with the heavy staining of the peri-bacteroid membranes. The freeze-fracture method, which does not appear to have been widely used to characterize membranes isolated from plant tissues, was a particularly rapid and specific method for determining the purity of membrane preparations. Using this technique approximately 98% of the vesicles in peribacteroid membrane preparations were specified as peribacteroid membranes (Robertson et al. 1977) and approximately 95% of vesicles in the bacteroid envelope inner membrane preparations were specified as inner membranes.
The high lipid content of the peribacteroid membranes, although unexpected, since most metabolically active membranes contain less (Korn, 1969), provides a possible explanation for several properties of this membrane. The high absorbance at 280 nm, relative to that of the bacteroid envelope inner membranes, could be caused by conjugated bonds in plant lipids and both the phosphotungstic acid and silver staining of the peribacteroid membranes in thin sections could be attributed to the presence of specific lipids (Clegg, Morré & Lunstra, 1975; Thiéry, 1967). The failure of glutaraldehyde fixation to block aggregation of particles on freeze-fracture faces of peribacteroid membranes might also reflect the high lipid content of these membranes. Pinto da Silva & Miller (1975) have reported that glutaraldehyde does not block particle aggregation in myelin, which also contains a high percentage of lipid.
The significance of the high lipid content of the peribacteroid membranes to nodule function is an important question. It may indicate that only a small complement of enzymes, which includes ATPase activity (unpublished results) but not NADH oxidase or succinate dehydrogenase activities, is required to carry out the metabolic functions of this membrane. Alternatively the membranes may have physical properties such as an affinity for lipophilic molecules which might be of importance to bacteroid function. For example, both oxygen, which is required by the bacteroids to support N2-fixation (Bergersen, 1974), and ammonia, which appears to be excreted by the bacteroids (Scott et al. 1976), have lipophilic properties (Fischkoff & Vander- kooi, 1975; Mitchell & Moyle, 1969). The decrease in density of the peribacteroid membranes during the period of development of N2-fixation in lupin nodules could reflect a change in lipid-to-protein ratio in these membranes. Such a change might be related to changes in the metabolism of the bacteroids during development of N2-fixation (Robertson et al. 1975a) or possibly to the loss, from the peribacteroid membranes, of enzymes associated with plant cell wall synthesis (Robertson et al. 1977).
The successful isolation of the bacteroid envelope inner membranes was almost certainly dependent upon the fact that, in the cell envelopes of a majority of bacteroids, the layer of murein appeared to have been lost, leaving the inner and outer membranes almost completely detached from each other. Whether this detachment was in some way related to the development of N2-fixation in bacteroids is not known. Changes in cell envelope structure and inhibition of cell wall synthesis in bacteroids have been reported (Jordan & Coulter, 1965; Van Brussel, 1973; MacKenzie et al. 1973). Jordan & Grinyer (1965) have observed the presence of wide subwall spaces between the cell walls and the plasma membranes of bacteroids in nodules from Lupinus luteus. We have not examined the effect of lysozyme on preparations of bacteroids (Van Brussel, 1973) but it is possible that the yield of inner membranes could be improved by such treatment.
The density of the bacteroid envelope inner membranes of 1 · 153 g/ml was in agreement with the density of inner membranes from envelopes of Salmonella typhimurium (Osborn, Gander, Parisi & Carson, 1972). Preparations contained both NADH oxidase and succinate dehydrogenase activities demonstrating that the terminal oxidase system in bacteroids is associated with the inner membranes as is the case with other gram-negative organisms (Costerton, 1970). Only cytochromes b and c were detected in the bacteroid envelope inner membrane preparations, as previously reported for total envelope preparations from bacteroids from soybean (Tuzimura & Watanabe, 1964; Appleby, 1969). The 10-nm particles observed on the cytoplasmic surface of the inner membranes appear likely to be F1 ATPase particles (Oppenheim & Saltón, 1973) which have also been observed in thin-section preparations of mito-chondria postfixed with uranyl acetate (Telford & Racker, 1973). We have detected ATPase activity in preparations of bacteroid envelope inner membranes showing characteristic properties of Fr ATPase (unpublished results).
Isolation of membrane-enclosed bacteroids raised the important question of the site of leghaemoglobin in lupin nodules. Osmotic shock of membrane-enclosed bacteroids released no more than trace amounts of leghaemoglobin, as determined by haem analysis on the soluble proteins in the shock fraction and also by gel electro-phoretic analysis of these proteins. The soluble protein composition of the shock fraction was clearly different from that of the bacteroid soluble-protein fraction and also from that of the plant soluble-protein fraction which contained leghaemoglobin. Gel-electrophoretic analysis of proteins in the peri-bacteroid membranes, spectral analysis of these membranes and analyses for leghaemoglobin showed that leghaemoglobin was not trapped in the vesicles formed during osmotic shock of the membrane-enclosed bacteroids. It seems unlikely that the peri-bacteroid membranes could have been opening and closing, and therefore releasing leghaemoglobin, during isolation of membrane-enclosed bacteroids, since no ferritin was detected in the peri-bacteroid spaces when nodules were fractionated in the presence of ferritin. Thus it appears that leghaemoglobin occurs in the plant cytoplasm and not in the peri-bacteroid spaces in lupin nodules, in agreement with suggestions made recently by Verma & Bal (1976) for soybean nodules. It is therefore suggested that this protein plays a role in transporting oxygen, not from the peri-bacteroid membranes to the bacteroids (see review by Appleby, 1974), but from the plasma membranes through the plant cytoplasm to the peri-bacteroid membranes.
We wish to thank Wendy Ulyatt for growing the plants, Glenn Grayston for help with electron microscopy, Ivan Simpson and Douglas Hopcroft for photographic assistance, Alan Craig and Dr A. Petkau for helpful advice and Professor R. D. Batt for generous support.