The localization of enzymes in fixed sections of soybean root nodules by electron microscopy

Biochemical investigations have revealed that many of the enzymes associated with respiration and oxidative phosphorylation in mitochondria are to be found within Gram-negative bacteria (Gel’man, Lukoyanova & Ostrovskii, 1967; Harold, 1972). Some of these are associated with particulate membrane fractions of the bacteria, and this suggests a structural basis for respiration similar to that which occurs in other organisms. However, the exact location of the respiratory enzymes and cytochrome components of the respiratory chain is in some dispute. The pattern of respiratory pigments varies with the metabolic state of bacteria, and this is particularly true of Rhizobia, where free-living and symbiotic (bacteroid) stages have distinct complements of cytochromes (Appleby, 1969). The bacteroid stage exists in a very different environment from the free-living form, being enclosed by a host membrane of plasmalemma origin (Goodchild & Bergersen, 1966) and bathed in leghaemoglobin (Bergersen & Goodchild, 1973) (the principal features of this arrangement can be seen in Fig. 8). In the present study we have attempted to localize histochemically some components of the respiratory system in the bacteroids of active nitrogen-fixing soybean root nodules whilst still in the host cells.

It was established in preliminary work that some enzyme activity remained in soybean nodule tissue after short fixation in certain fixatives, and that better preservation for electron microscopy was obtained when fixed tissues were processed than when unfixed material was used. As a consequence, 1-mm slices of fresh nodules were cut into drops of 4% depolymerized paraformaldehyde in 50 mM Na/K phosphate buffer, pH 7·4, at room temperature, and allowed to fix for 15 min. The tissue was then washed 3 times in buffer and the following histochemical procedures were carried out. In all cases the tissue was incubated at 25 °C for 1 h.

1. Ferrocytochrome c: oxygen oxidoreductase (1·9·3·1). (a) After Seligman, Karnovsky, Wasserdrug & Hanker (1968). The incubation medium consisted of 50 mM Na/K phosphate buffer, pH 7·4, containing 0·5 rag ml⁻¹3,3’-diaminobenzidine tetra HC1 (DAB), 1 mg ml⁻¹ cytochrome c and 0·04 mg ml⁻¹catalase. Various media were made up with and without catalase, and cytochrome c. Media were prepared in subdued light, and incubations were performed only with freshly prepared media in darkness, to prevent auto-oxidation of the DAB (Hirai, 1971). DAB samples from different sources, and of differing grades of purity were compared; the histochemical results were found to be identical, (b) Methods modified after Novikoff & Goldfischer (1969). The medium consisted of 8 9 ml 50 mM sodium acetate/acetic acid buffer, pH 6, containing 20 mg DAB, 1 ml of 0·05 M MnCl₂, 0·1 ml of 0·01% H₂O₂. Freshly prepared solutions were used and incubations were performed in light-tight vials.

2. Other oxidases (Nir & Seligman, 1971). Material was incubated in 50 mM Na/K phosphate buffer, pH 74, containing 1 mg ml⁻¹ of N,N’ -bis(4-aminophenyl)-1,3-xylylene-diamine (BAXD). Solutions were also made up containing catalase and/or cytochrome c. Media were prepared in subdued light and used in darkness, although there is no evidence to suggest that solutions of BAXD undergo auto-oxidation in the light.

3. Succinate: (acceptor) oxidoreductase (1·33·99·1) (Seligman, Nir & Plapinger, 1971). The medium consisted of 0·5 M sodium succinate in 50 mM Na/K phosphate buffer pH 7· 4, containing 1 mg ml⁻¹ distyryl nitroblue (DS-NBT).

4. Reduced-NAD: (acceptor) oxidoreductase (1· 6· 99· 3) (Seligman et al. 1971). Material was incubated in 50 mM Na/K phosphate buffer, pH 7· 4, containing 1 mg ml⁻¹ DS-NBT and 0· 5 mg ml⁻¹ NADH₂.

5. Adenosinetriphosphate phosphorylase (3· 6· 1· 33) (Hall, 1969). The medium consisted of 50 mM sodium cacodylatc HC1, pH 6· 8, containing 2 mM substrate and 2 mM salts. The substrates used were adenosine tri-, di- and monophosphates (ATP, ADP, AMP) and the salts were either calcium or magnesium nitrates.

The effects of the following inhibitors upon enzyme activity were examined: 10⁻⁷ M rotenone, 10⁻² – 10⁻³ M amobarbitol, 10⁻³ M quinine, 10⁻⁴ M atebrin, and 10⁻⁴ M sodium arsenite. For oxidase enzymes the following inhibitors were used: 10⁻³ M potassium cyanide, 10⁻³ M aminotriazole, 5 mM sodium azide, 0·05% catalase, 2 mM sodium pyruvate. The sulphydryl inhibitor parachloromercuribenzoic acid (PCMB), 10⁻⁵ M, was used to inhibit salt-stimulated ATPase activity. When inhibitors were used, tissues were preincubated in the medium containing the inhibitor, without the dyestuff used for localization. Control reactions also involved various temperature regimes (4 °C during incubation, or preheating to 60 °C for 20 min) and the omission of the relevant substrates from the incubating media (e.g. succinate when testing for succinic dehydrogenase). Sometimes alternative substrates were employed (e.g. ADP, AMP in place of ATP). Reagents were obtained from British Drug Houses, Sigma Chemicals Inc. and Polysciences Inc. Penn. U.S.A.

After incubation, tissue samples were washed several times in appropriate buffer, then soaked for 6 h at 25 °C in 1 % OsO₄ made up in the same buffer. Excess osmium was removed by washing in buffer, after which samples were dehydrated in an ethanol series and introduced into the following resin mixture modified after Spurr (1969), using propylene oxide: 10 ml ERL, 6 ml DER, 26 ml NSA, 0-4 ml of the catalyst Si. Blocks were cut with a diamond knife using a Reichert OM U3 ultramicrotome: silver sections were collected on 300-mesh grids. In some cases sections were stained with 2% uranyl acetate, washed in distilled water, and stained with lead citrate (Reynolds, 1963). Grids were examined with an AEI 801 transmission electron microscope, using a 30μm objective aperture, at 60 kV.

This was found with both the Seligman et al. (1968) and Novikoff & Goldfischer (1969) methods. Considerable binding of DAB was also observed between cell walls and in lignificd tissues in controls and in test sections and was considered to be artifact. The results are illustrated in Figs. 1– 11. Dense, non-uniform products occurred in the bactcroid cytomembrane, with smaller amounts in the mesosomes and nuclear region (Fig. 1). All of these reactions were inhibited by cyanide (Fig. 2). Host cell mitochondria showed an intense reaction between the cristae membranes (Fig. 3) which was not completely inhibited by cyanide (Fig. 4). These two regions of activity also varied in their response to heat, which inhibited activity in the bacteroids but not in the mitochondria (Fig. 5). Aminotriazole only slightly inhibited the mitochondrial reaction (Fig. 10), but almost completely inhibited background cytomembrane activity (Fig. 6). Smaller non-infected cells lying between infected host cells contain numerous microbodies which reacted intensely with DAB (Fig. 7). The reaction was inhibited by cyanide but only partially inhibited by azide. It could occur when tissues were incubated with DAB in the absence of H₂O₂. Microbody reactions were probably due to peroxidase/catalase activity (Frederick & Newcomb, 1971; Longo, Dragonelli & Longo, 1972).

It has been reported that false binding of DAB oxide occurs if the DAB medium is allowed to autoxidize (Hirai, 1971). To test for this, freshly prepared medium was kept in the light in stoppered vials for 1 h and then used for a further incubation period of 1 h in the light. Autoxidized DAB should then show binding to the leghaemoglobin in the bacteroid compartments (Bergersen & Goodchild, 1973) and to the other haemoproteins. In such media very little binding occurred (Figs. 8, 9). On the other hand, tissue incubated in medium which had been left to age for 12 h and then used showed binding to the membranes. As a further control for DAB oxidase, tissues were fixed in 2·5% glutaraldchyde for 12 h. No reaction was observed (Fig. 12).

According to Nir & Seligman (1971) the reagent BAXD may reveal different localization of oxidase enzymes from those obtained with DAB and similar reagents. The cristae of mitochondria in corn root tips show reaction products with both DAB and BAXD. In root nodules, mitochondria within the host cell cytoplasm and bacteroids show products of BAXD oxidation (Fig. 14). The bacteroid cytomembranes showed only slight activity, whereas mesosomal regions were much more active. The mesosomal reaction was not inhibited by 1 mM KCN (Fig. 13), nor by catalase or pyruvate. It was sensitive to some fixatives (FAA, Ca formol) but not to others (4% paraformaldchyde, 15 min). BAXD oxidase activity was also present within the nuclear regions of the bacteroids (Fig. 14). This was insensitive to cyanide (Fig. 13) and was not overcome by catalase. Thus it was similar to the nuclear DAB oxidase (Fig. 1); the 2 reagents may have been oxidized by the same enzyme system.

Substrate-dependent, Ca-stimulated ATPase activity (Fig. 15), inhibited by PCMB (Fig. 11), was localized as deposits of lead phosphate in the cytomembrancs of most of the bacteroids. In this instance the bacteroid population of any particular host cell was heterogeneous with respect to ATPase activity. When either ADP or AMP was supplied in place of ATP, reaction products were not formed. In the absence of either Ca or Mg ions reaction products were not observed. A combination of ATP and Mg²⁺ gave reaction products in a few bacteroids (less than about 20% of the total in any one cell): reaction products with this medium were also found within the plasmalemma (Fig. 16) and plasmodesmata of the host cell.

The results are shown in Figs. 17 – 20. Activity of both enzymes was widely distributed. In Fig. 17 NADH oxidoreductase activity is seen in mitochondria, microbodies and bacteroids. This was not inhibited by rotenone or amobarbitol. Within bacteroids, activity of both enzymes can be detected in mesosomes and cytomembranes (Figs. 18, 19). One problem associated with this reaction is in the controls. Heat controls were effective, but an intense reaction with DS-NBT was obtained in bacteroids at 4 °C (Fig. 20).

Seligman et al. (1971) have discussed the use of DS-NBT for the localization of dehydrogenase enzymes in mitochondria. Localization is determined by the electron density of the osmicated distyryl formazan and is observed as black deposits on the cytomembranes and the mesosomes of bacteroids, treated with either DS-NBT plus NADH or succinate. The mesosomal reaction was far more intense than was the cyto-membrane staining, but the cytomembrane reaction was more intense with the NADH medium than with the succinate medium. The electron density of bacteroids treated with either medium was very similar. We conclude that NADH oxidoreductase and succinate oxidoreductase have similar distributions in the bacteroids and that this distribution is largely mesosomal.

The use of inhibitors did not allow separate localization of the various oxidoreductase systems within the host cells. Neither rotenone nor amobarbitol inhibited NADH oxidoreductase activity. In cell-free extracts of other bacteria amobarbitol, for example, was found to cause 50% inhibition of NADH dehydrogenase (Taniguchi & Kamen, 1965; Gel’man, Lukoyanova, Zhukova & Oparin, 1963) and rotenone 50% inhibition of NADH oxidase (White & Smith, 1964). If bacterial membranes are impermeable to ATP, ADP, NADH, and NADPH (Harold, 1972) then they may also be impermeable to both of the inhibitors employed.

Other reports of the mesosomal localization of tetrazolium salts in bacteria (Takagi et al. 1963) indicate the lack of formazan deposits within the cytomembranes. Takagi and his co-workers propose that the mesosomes contain a different set of dehydrogenase enzymes from the cytomembrane. Mudd et al. (1961) consider the cytomembrane and mesosomes to act as a single functional system, and this hypothesis is consistent with the present observations, and pertinent to the distribution of other oxidase systems.

Oxidorcductase activity within the microbodies was indicated by the reduction of DS-NBT. Donaldson, Tolbert & Schnarrenberger (1972) have indicated the likelihood of NADH activity within these organelles associated with cytochrome c oxidase. Some DAB oxidation may thus have been due to ferrocytochrome oxidoreduction, as well as peroxidase and catalasc.

Two BAXD oxidase systems were localized in the nodules, one within the host cells, and another within the bacteroid mesosomes. The mesosomal oxidase was not inhibited by KCN. Variable intensities of reaction within the bacteroid compartments due to oxidase reactions may have arisen because of leakage from the bacteroid compartments (Bergerscn & Goodchild, 1973). Leghaemoglobin within the compartments is normally in the reduced state (Bergersen & Goodchild, 1973), and would not cause the oxidation of DAB, BAXD, or arylamine substrates. Exposure to oxygen would bring about the formation of ferroleghaemoglobin and subsequent oxidation of the dyestuffs. From the reactions with inhibitors we may conclude that metalloprotein enzymes are involved which are sensitive to aldehyde fixation. Phenol oxidases (which contain copper, see Malstrom & Ryden, 1968) are associated with the ferroleghaemoglobin oxidation to ferrileghaemoglobin (Keilin & Wang, 1945) and could account for the fibrillar staining within the bacteroid compartments. BAXD activity has been implicated in phenoloxidase activity in other plants (Nir & Seligman, 1971). The peroxidative activity of haemoprotein is well known (Pearse, 1961) and has been investigated in nodules by Truchet & Buvat (1972).

The localization of BAXD oxidase within the mesosomes may have been due to a cyanide-sensitive NADH oxidase system isolated from Rhizobium japonicuni bacteroids (Dr I. R. Kennedy, personal communication). Particulate fractions of other Gram-negative bacteria also contain NADH oxidases (Gel’man et al. 1967). Cyanide sensitivity in bacteria is dependent upon concentration effects. For example, in aerobic cultures of Mycobacterium tuberculosis NADH oxidase is cyanide-resistant at 5 × 10 ⁻³ M but shows only 60% inhibition at 10⁻³M (Gel’man et al. 1967).

The cyanide-resistant oxidase within the nuclear region is more likely to have been a peroxidase enzyme, effectively oxidizing DAB and BAXD. Phenylenediamine is oxidized by fresh nodule slices (Marks and Sprent, unpublished) and has been associated with the peroxidative activity of chromatin in higher plant cells (Raa, 1973). Peroxidation of DAB is inhibited by aminotriazole (Fig. 6). This corresponds to the inhibition of microbody peroxidases in other tissues (Novikoff & Goldfischer, 1969; Hirai, 1971).

DAB binding also gave some mesosomal localization, although the bulk of the reaction product was localized within the cytomembrane. Reith & Schuler (1972) have demonstrated a relationship between this reaction, and the oxidation/utilization of DAB as a substrate for the in vitro cyanide-sensitive ferrocytochrome c: oxygen: oxido-reductase system of isolated mitochondria from rat liver. It is thus likely that DAB binding in some cases may represent the localization of this enzyme. DAB staining has been recorded within the mesosomes and cytomembrane of other strains of Rhizobium. (Gourrier, personal communication; Truchet & Buvat, 1972). However, since at least 3 oxidase systems occur within the bacteroids, stringent controls and careful interpretation are necessary.

Ferrocytochrome c oxidoreductase has 2 components, cytochrome a and a₃, both of which are cyanide-sensitive (Barman, 1969). Neither cytochrome a nor o are found within bacteroids (Appleby, 1969; Kretovich, Romanov & Korylov, 1973), which contain only the cytochromes c₅₅₀, c_552l b₅₅₉, and a hydroxylase haemoprotein P₄₅₀. What then may be the cause of DAB oxidation? One candidate might be ferrocytochrome: (nitrate) oxidoreductase (1·9·6·1), which is able to utilize cytochrome c by way of NADPH, methylene blue and other dyes as electron acceptors (Barman, 1969). Electron-transport inhibitors also inhibit this enzyme. However, nitrate reductase develops within the nodules only if the host plant is grown in a medium containing combined nitrogen. The soybean plants used here were nitrogen fixing, and grown in the absence of combined nitrogen. Nodules grown in this fashion would be expected to contain nitrogenase, the Mo component of which is similar to one component of nitrate reductase (Evans & Russell, 1971). Pyridine nucleotides will transfer electrons from dye substrates (benzylviologen) to nitrogenase. An alternative scheme for electron flow to nitrogenase may occur by way of a non-haem iron-protein and flavoprotein (see review by Evans & Russell, 1971). One of these schemes may bring about the oxidation of DAB within the bacteroid cytomembrane, in the absence of ferrocytochrome c: oxygen oxidoreductase.

The size and nature of the lead phosphate deposits are characteristic of those produced during ATP hydrolysis in bacterial membranes (Woelz, 1964; Vaituzis, 1973). Ca-ATPase was localized in a patchy fashion over the cytomembranes: these discontinuities were similar to the DAB (cytochrome oxidase) marker distributions. This suggests that both enzymes are localized in related but discrete regions of the cytomembranes. We do not know if there are any relationships between the distributions of these 2 enzymes. It is likely that the Ca-stimulated ATPase is concerned with that part of phosphorylation coupled to the respiratory chain (Harold, 1972) and so a close structural relationship’might be expected. Koenig & Vial (1973) have shown that, in animal cells, fixation in the presence of leadions will inhibit the mitochondrial ATPase system but will allow the localization of other membrane ATPase systems. It would therefore seem likely that the membrane ATPase enzyme of bacteria is similar in form to other membrane phosphorylase enzymes (i.e. it can be localized under similar histochemical conditions), but is different in that it is linked to the respiratory chain.

We should like to thank the Agricultural Research Council for generous financial assistance.