Molecular Cloning, Functional Characterization, and Subcellular Localization of Soybean Nodule Dihydrolipoamide Reductase^,

Abstract

Nodule ferric leghemoglobin reductase (FLbR) and leaf dihydrolipoamide reductase (DLDH) belong to the same family of pyridine nucleotide-disulfide oxidoreductases. We report here the cloning, expression, and characterization of a second protein with FLbR activity, FLbR-2, from soybean (Glycine max) nodules. The cDNA is 1,779 bp in length and codes for a precursor protein comprising a 30-residue mitochondrial transit peptide and a 470-residue mature protein of 50 kD. The derived protein has considerable homology with soybean nodule FLbR-1 (93% identity) and pea (Pisum sativum) leaf mitochondria DLDH (89% identity). The cDNA encoding the mature protein was overexpressed in Escherichia coli. The recombinant enzyme showed K_m and k_cat values for ferric leghemoglobin that were very similar to those of DLDH. The transcripts of FLbR-2 were more abundant in stems and roots than in nodules and leaves. Immunoblots of nodule fractions revealed that an antibody raised against pea leaf DLDH cross-reacted with recombinant FLbR-2, native FLbR-2 of soybean nodule mitochondria, DLDH from bacteroids, and an unknown protein of approximately 70 kD localized in the nodule cytosol. Immunogold labeling was also observed in the mitochondria, cytosol, and bacteroids of soybean nodules. The similar biochemical, kinetic, and immunological properties, as well as the high amino acid sequence identity and mitochondrial localization, draw us to conclude that FLbR-2 is soybean DLDH.

Legume N₂ fixation requires the presence of abundant functional leghemoglobin (Lb) in the cytosol of nodule infected cells. The main function of this hemoprotein is to deliver O₂ to the bacteroids at a concentration compatible with nitrogenase activity and bacteroid respiration (for review, see Appleby, 1984). To carry O₂, Lb must be present in the ferrous form and, indeed, young, actively N₂-fixing nodules contain only ferrous or oxyferrous Lb (Appleby, 1984; King et al., 1988; Monroe et al., 1989). However, the chemical nature of Lb itself and various conditions existing in nodules are conducive for Lb oxidation (Becana and Klucas, 1992; Lee et al., 1995). Using diffuse reflectance and direct transmission spectroscopy, Lee and Klucas (1984) showed that ferric Lb generated in soybean (Glycine max) nodule slices by treatment with hydroxylamine is rapidly reduced to ferrous Lb and that ferric Lb is present in senescent sweet clover (Melilotus officinalis) and soybean nodules (Lee et al., 1995). These observations indicate that mechanisms exist in the nodules to maintain Lb in the functional, reduced state.

Saari and Klucas (1984) isolated from soybean nodules a protein capable of catalyzing the NADH-dependent reduction of ferric to ferrous Lb. This protein with ferric Lb reductase activity (FLbR), designated FLbR-1, was characterized and shown to be a homodimer of 110 kD with FAD and a redox-active disulfide group per subunit (Ji et al., 1991). FLbR-1 exhibits high affinity for ferric Lb (K_m = 9 μm) and NADH (K_m = 50 μm), which is consistent with its proposed role in maintaining Lb in its functional state (Saari and Klucas, 1984; Becana and Klucas, 1990; Ji et al., 1991). In soybean nodules, two additional proteins with FLbR activity are present (Ji et al., 1991), and at least two gene copies have been detected (Ji et al., 1994a). A cDNA encoding FLbR-1 was isolated, and the deduced protein sequence showed high homology with dihydrolipoamide dehydrogenase (DLDH) and, to a lesser extent, with glutathione reductase, mercuric reductase, and trypanothione reductase from various organisms (Ji et al., 1994a; Pullikuth and Gill, 1997). All of these enzymes belong to the NAD:disulfide oxidoreductase family and are homodimers containing FAD and a pair of redox-active Cys residues involved in the electron transfer from NAD(P)H and flavin to the substrates (Williams, 1991).

In pea (Pisum sativum) leaf mitochondria, the same DLDH isozyme is shared by the pyruvate dehydrogenase complex (E3 component) and the Gly decarboxylase complex (L protein; Bourguignon et al., 1996). Both enzymatic complexes catalyze oxidative decarboxylations in plant mitochondria, which are essential for the functioning of the tricarboxylic acid and the photorespiratory cycles, respectively (Rawsthorne et al., 1995). Pea chloroplasts also contain a DLDH isozyme, but this is distinctly different from its mitochondrial counterpart (Conner et al., 1996).

The high homology between FLbR and DLDH (Pullikuth and Gill, 1997), the physiological relevance that FLbR may have in legume N₂ fixation (Saari and Klucas, 1984), the critical role of DLDH in energy metabolism (Rawsthorne et al., 1995), and the observation that both FLbR and DLDH are present in plant tissues as various isozymes (Ji et al., 1991; Conner et al., 1996) prompted us to study other proteins with FLbR activity present in nodules and to establish their relationships with DLDHs. In this work, we have cloned and characterized a novel gene encoding a second protein exhibiting FLbR activity, designated FLbR-2, from soybean nodules. The protein was overexpressed in Escherichia coli, its biochemical properties were compared with those of FLbR-1 and DLDH, and its subcellular localization was determined by subcellular fractionation and immunogold labeling.

RESULTS

Isolation, Cloning, and Sequencing of FLbR-2 cDNA

We have isolated a cDNA (GenBank accession no. AF074940) encoding a novel protein with FLbR activity, FLbR-2, from a soybean nodule library (Ji et al., 1994a). Primers designed to conserved sequences of FLbR-1 were used to clone four independent PCR products. The full-length sequence of FLbR-2 was highly homologous at the nucleotide and amino acid levels with soybean nodule FLbR-1 (>93% identity; Ji et al., 1994a) and pea leaf DLDH (>86% identity; Bourguignon et al., 1996). The 5′- and 3′-untranslated regions (UTRs) were PCR amplified using internal primers for the FLbR-2 cDNA in combination with λ-primers (Arredondo-Peter et al., 1997). The cDNA was 1,779 bp long and contained an open reading frame of 1,503 bp, which codes for a 500-amino acid protein. Based on the N-terminal of native FLbR-1 (nFLbR-1) of soybean nodules (Ji et al., 1991), the precursor protein of nFLbR-2 was predicted to include a 30-residue signal peptide and a 470-residue mature polypeptide (Fig. 1), with an estimated molecular mass of 49,618 D. The existence of a signal peptide was further demonstrated in this work by automated N-terminal Edman degradation of nFLbR-1 and nFLbR-2, which confirmed that the signal peptide is processed, consistent with previous results of Ji et al. (1991). Both FLbR-1 and FLbR-2 appear to possess an identical leader sequence (Fig. 1).

Figure 1.

Comparison of deduced amino acid sequences of FLbRs and selected DLDHs from plants and other organisms. Residues identical in at least six of eight residues are indicated in white lettering on a black background. Twenty-eight N-terminal and nine C-terminal amino acids were sequenced automatically and are indicated with a continuous line. The cleavage site of the precursor proteins of FLbR-1 and FLbR-2 is indicated with an arrow. The conserved FAD- and NADH-binding domains are marked with a discontinuous line, and the presumptive active site Cys residues are marked with asterisks. Sequence alignment was performed using the PileUp program. GenBank accession numbers are as follows: soybean FLbR-1, S70187; soybean FLbR-2, AF074940; cowpea (Vigna unguiculata) FLbR, AF181096; pea DLDH, X63464; Arabidopsis E3, AF228640; human DLDH, P09622; yeast (Saccharomyces cerevisiae) DLDH, P09624; E. coli DLDH, P00391.

The transit peptide of FLbR-2 and FLbR-1 closely resembles those of other proteins that are targeted to the mitochondria (von Heijne et al., 1989). The peptide is enriched in Arg (17%), Ser (13%), Leu (13%), and Ala (13%) and lacks acidic residues (Fig. 1). The cleavage site fits into the RGF↓A motif, which can be clearly assigned to the RXY↓S/A or “R-3” mitochondrial motif described by Gavel and von Heijne (1990), assuming that the two aromatic residues, Tyr and Phe, are functionally interchangeable. Indeed, prediction programs of subcellular localization, including Mitoprot II, PSORT, and Target-P, indicated that both nFLbR-1 and nFLbR-2 are mitochondrial enzymes. It is interesting that a recently reported protein homolog of nFLbR-2 from cowpea nodules (GenBank accession no. AF181096; Luan et al., 2000) has an identical cleavage site motif and therefore also a putative 30-amino acid signal peptide for mitochondrial targeting (Fig. 1).

Expression and Overproduction of rFLbR-2

A 1,483-bp fragment coding for the mature part of the protein was cloned into the NdeI and XhoI sites of the expression vector pET-28a(+) and used to transform E. coli BL21(DE3). Analysis by SDS-PAGE of the cell extracts showed the overexpression of a protein of approximately 50 kD after induction of the cells with isopropyl-β-d-thiogalactopyranoside (IPTG) for a period of 1 to 6 h. Longer incubation times in the presence of IPTG led to the degradation of the recombinant protein, producing a fragment of approximately 16 kD, probably as a result of proteolytic activity (data not shown). The recombinant FLbR-2 (rFLbR-2) separated by SDS-PAGE and stained with Coomassie blue was, apparently, the major induced protein.

Enzyme Purification

rFLbR-2 was purified to near homogeneity as judged by SDS and isoelectric focusing (IEF) gels. The purification method involved affinity chromatography using a Ni-Probond column, which selectively bound the poly-His tag-fused FLbR-2, followed by anion-exchange chromatography (Table I). The metal-chelate affinity purification step was critical to separate rFLbR-2 from E. coli DLDH. The enzyme was purified 42-fold in terms of ferric Lb-reducing activity and 323-fold in terms of lipoamide reductase activity. The 6×-His tag was removed with a biotinylated thrombin system, as confirmed by N-terminal sequencing and SDS-gel analysis. The specific activities of rFLbR-2 were 958 and 83,055 units mg⁻¹ for FLbR and DLDH activities, respectively (Table I). The recombinant enzyme cross-reacted with antibodies raised against nFLbR (probably a mixture of three isozymes; see also Fig. 3B) from soybean nodules (Ji et al., 1994b) and against DLDH from pea leaf mitochondria (see below).

Table I.

Purification of soybean rFLbR-2 synthesized in E. coli BL21 (DE3)

Purification Step	Total Proteina	2,6-Dichloroindophenol Reductionbc	Ferric Lb Reductionb	FLbR Purification	Lipoamide Reduction	DLDH Purification
	mg	units mg−1	units mg−1	−fold	units mg−1	−fold
Crude cell extract	268	162	23	1	257	1
Ni-Probond	9.2	1,760	481	21	20,398	79
Poros HQ-X	1.9	3,153	958	42	83,055	323

^a Protein yield (mg) from 3 g of cell paste. 

^b One unit of enzyme activity catalyzes the reduction of 1 nmol substrate min⁻¹. 

Figure 3.

A, Native PAGE of soybean nFLbR (lane 1), rFLbR-2 (lane 2), and porcine heart DLDH (lane 3). Samples (approximately 6 μg of protein) were electrophoresed on a 7.5% (w/v) polyacrylamide gel and stained with Coomassie Brilliant Blue. B, IEF of purified nFLbR from soybean nodules (lane 2) and rFLbR-2 (lane 3). Samples (approximately 5 μg of protein) were electrophoresed on a 7.5% (w/v) polyacrylamide gel and stained with Coomassie Brilliant Blue/crocein scarlet. Values of pI are indicated for standards (lane 1) and FLbRs. The following IEF standards (Bio-Rad, Hercules, CA) were used: phycocianin (4.65), β-lactoglobulin B (5.10), bovine carbonic anhydrase (6.00), and human carbonic anhydrase (6.50).

Protein Sequence and Mass Spectrometry (MS) Analysis

N-Terminal sequencing of rFLbR-2 revealed that the sequence of the 32 first amino acids is GSHMASGSDENDVVVIGGGPGGYVAAIKAAQL, which comprises four amino acid residues of the pET-28 vector downstream of the excision site for thrombin plus the 28 amino acid residues of the deduced protein sequence. To unequivocally demonstrate the existence of the predicted C terminus in the FLbR-2 sequence, SDS-purified rFLbR-2 was subjected to digestion with cyanogen bromide. The HPLC-resolved peptides were analyzed by electrospray MS and/or N-terminal sequencing (Fig. 2). Peptide 6 was proved to be the rFLbR-2 specific N-terminal fragment by electrospray MS molecular mass determination and N-terminal sequencing. Peptide 21 corresponded to residues 449 to 491 as determined by N-terminal sequencing. Using this technique, we could determine the existence of peptide 6 in the IEF-separated upper isozyme of the nFLbR. This peptide had an identical mass of 1,057 D when analyzed by electrospray MS (Fig. 2).

Figure 2.

HPLC profile of the cyanogen bromide (CNBr) digestion of rFLbR-2. Peak 6 was further analyzed by N-terminal sequencing and electrospray MS (inset). Dotted lines, Blank; solid line, CNBr (3 mg).

Electrospray MS analysis showed that rFLbR-2 has a molecular mass of 50,027 D, which is virtually identical to the predicted mass, 50,030 D. The molecular mass determined by MS corresponded to only the polypeptide, because the flavin was released during the analysis, producing a peak at 786.1 D (data not shown). Collectively, these data indicated that the recombinant protein was correctly synthesized in E. coli, that FAD is the flavin coenzyme of FLbR-2, and that FAD is not covalently bound to the protein.

Biochemical Properties of rFLbR-2

Native PAGE suggested that FLbR-2 is a homodimer because the recombinant protein showed exactly the same mobility as the dimeric soybean nFLbR (Fig. 3A). rFLbR-2 and nFLbR were isoelectrofocused in an effective pI range of 6.5 to 4.0 and compared to pI standards (Fig. 3B). rFLbR-2 matched the upper isozyme of the three reported by Ji et al. (1994b), with an estimated pI of 5.8. The other nFLbRs isozymes had pI values of 5.7 and 5.6, respectively. The slight difference (0.1 unit) from the reported pI values of Ji et al. (1991) is probably due to a narrower range of ampholytes used in this study.

UV-visible spectra of rFLbR-2 revealed high similarities to those of pig and yeast DLDH and oxidized soybean FLbR-1 (Ji et al., 1994b). The rFLbR-2 showed a UV absorption peak at 278 nm and a visible absorption peak at 460 nm, with two shoulders at 435 and 485 nm, consistent with the presence of FAD as a prosthetic group.

Because FLbR-2 is able to use both ferric Lb and lipoamide as substrates, a comparison of the kinetic constants for the two reactions was considered essential to gain information concerning the putative function(s) of the isozyme. The calculated K_m values of rFLbR-2 were 29 μm for ferric Lb, 58 μm for NADH, and 3.38 mm for lipoamide (Table II). Consequently, the k_cat to K_m ratios for lipoamide and Lb were 116 and 55 mm s⁻¹, respectively. The latter two values are very similar to those of DLDHs from yeast, bovine, and pig mitochondria, although the k_cat/K_m ratio for NADH was considerably greater for rFLbR-2. Also, the K_m values (250–500 μm) for lipoamide of yeast, bovine, and pea leaf DLDHs were significantly lower than the corresponding value of rFLbR-2 (Kim, 1996; Neuburger et al., 2000). On the other hand, rFLbR-1 showed significantly higher affinity (5-fold), k_cat (4-fold), and k_cat/K_m ratio (18-fold) for ferric Lb than rFLbR-2 and also differs in kinetic behavior from yeast and bovine DLDHs (Table II).

Table II.

Kinetic properties of rFLbRs and DLDHs

Enzymes	Substrates	Km	Vmaxa	kcat	kcat/Km	Reference
		μm	units mg−1	s−1	mm−1 s−1
rFLbR-2	Lipoamideb	3,381	467,000	392	116	This work
	NADHb	58	163,000	137	2,362
	Soybean ferric Lb	29	1,840	1.6	55
rFLbR-1	Lipoamideb	716	NDc	ND	ND	Ji et al. (1994b)
	NADHb	46	16,000	31	674
	Soybean ferric Lb	6.3	450	6.2	984
DLDHd	Lipoamideb	245–279	ND	37–39	133–159	Kim (1996)
	NADHb	73	25,000	33	452	Ji et al. (1994b)
	Soybean ferric Lb	28	350	1.1	39

^a One unit of enzyme activity catalyzes the reduction of 1 nmol substrate min⁻¹. 

^b Determined as lipoamide-dependent NADH oxidation. 

^c ND, Not determined. 

^d Values for lipoamide reduction were calculated using yeast and bovine DLDH. Values for NADH and ferric Lb reduction were calculated using pig heart DLDH. 

Expression and Localization of FLbR-2 in Soybean

Reverse transcription (RT)-PCR analysis was used to study tissue-specific expression of flbr-2 in soybean (Fig. 4A). The transcript is more abundant in the stems and roots than in the leaves and nodules. As expected, Lb transcript was detectable only in nodules, whereas that of constitutively expressed Gln synthetase was found in all four tissues examined, especially in roots and nodules. A similar analysis was performed to determine the effect of nodule age on flbr-2 expression (Fig. 4B). The flbr-2 transcript was more abundant in mature nodules (4–6 weeks old) with a slight decrease in senescent nodules (10 weeks old). A similar pattern was followed by the Lb mRNA, although in this case there was no apparent decline in senescent nodules. As expected, there were no major changes in the level of the Gln synthetase transcript.

Figure 4.

A, Expression of flbr-2 in different soybean tissues (L, leaves; S, stems; R, roots; N, nodules). B, Effect of nodule age (in weeks) on the expression of flbr-2. Gln synthetase (GS) and Lb were used as markers of metabolic activity in plant tissues and nodules, respectively. Ubiquitin (Ubi) was used as an internal control of the RT-PCR analysis.

Immunoblots of whole extracts and various fractions from soybean nodules revealed that an antibody to pea leaf mitochondria DLDH recognized rFLbR-2 (Fig. 5, lane 1), as well as proteins of mitochondria and bacteroids (Fig. 5, lanes 4 and 5). The major (or single, in some cases) immunoreactive band was at approximately 50 kD and is therefore consistent with the expected molecular mass of FLbR-2, as well as with the expected mitochondrial localization of FLbR-2 and DLDH. Immunoblots also showed an immunoreactive band of approximately 70 kD in the soybean nodule cytosol (Fig. 5, lane 3). Extracts from soybean leaves or from pea nodules and leaves were also analyzed to verify results. Immunoreactive bands of 50 kD were observed in soybean and pea leaves and in pea nodules. The 70-kD cytosolic protein of soybean nodules was not found in soybean and pea leaves or in pea nodules (Fig. 5, lanes 6–8).

Figure 5.

Immunoblot analysis of FLbR-2 in soybean and pea nodules and leaves using anti-pea DLDH antibody. Lanes: 1, Pure rFLbR-2; 2, soybean nodule extract; 3, soybean nodule cytosol; 4, soybean nodule mitochondria; 5, soybean nodule bacteroids; 6, soybean leaf extract; 7, pea nodule extract; 8, pea leaf extract. Estimates of molecular mass of the relevant proteins (shown in kilodaltons) were obtained using prestained SDS M_r markers (Bio-Rad). Protein loaded was 10 μg for all lanes, except for lane 1 (50 ng) and lane 5 (20 μg).

The subcellular localization of FLbR-2 in soybean and pea nodules was also investigated by immunogold labeling using the same antibody (Fig. 6). The main areas examined were three-way cell junctions within the infected zone, because these cells are known to contain a high concentration of mitochondria adjacent to the intercellular space (Millar et al., 1995). There was significant labeling of mitochondria in both infected and uninfected cells (Fig. 6, A and C). However, the most intense labeling was seen on some of the bacteroids (Fig. 6, A and B), particularly on the electron-dense nuclear material (Fig. 6B). Labeling was also observed in the cytosol of infected cells, which, although sparse (Fig. 6, A and B), was significantly greater than that obtained with non-immune serum (Table III). Some minor labeling was only occasionally observed in peroxisomes of uninfected cells (Fig. 6C). The pattern of immunolabeling was similar in pea nodule sections with the exception that the gold labeling in the cytosol was not significantly above the background (non-immune serum; Table III).

Figure 6.

Immunogold localization of FLbR-2 in soybean nodules. Sections were probed with an antibody raised against pea leaf DLDH (A–C) or non-immune serum (D), followed by secondary antibodies conjugated to 15-nm gold particles. A, Three-way junction of infected cells in the active, N₂-fixing zone of a soybean nodule. Gold particles can be seen on mitochondria (m) and bacteroids (b), as well as in the cytosol (small arrows). B, Labeling of electron-dense nuclear material (large arrows) within the bacteroids (b), with scant labeling in the cytosol (small arrow). C, Uninfected cell with labeling on mitochondria (m) and peroxisomes (px). Note the adjacent infected cell (*), with labeling in mitochondria. D, Serial section to A and C that was incubated in non-immune serum. No gold particles are visible on mitochondria (m) or in the cytosol, but one can be seen in a bacteroid (b; arrow). s, Intercellular space. Bars = 500 nm.

Table III.

Number of gold particles on infected cell components from soybean and pea nodules

Legume	Treatment	Mitochondria	Bacteroids	Cytosol
		no. of gold particles per 25 μm2
Soybean	DLDH antibody	4.60 ± 0.54**	15.20 ± 3.21**	5.00 ± 1.89*
	Non-immune serum	0.30 ± 0.15	1.80 ± 0.36	1.60 ± 0.31
Pea	DLDH antibody	7.70 ± 0.52**	5.10 ± 0.71**	2.80 ± 0.51
	Non-immune serum	0.30 ± 0.15	0.60 ± 0.27	2.70 ± 0.87

Means ± se were taken from 10 micrographs, each with an area of 25 μm². Values significantly different from the corresponding non-immune serum according to the Student's t test are marked with ^* (P < 0.05) or ^** (P < 0.01).

Sections of soybean nodules that had been immunogold labeled with anti-DLDH antibody pre-absorbed with 40 μg mL⁻¹ of rFLbR-2 protein (Fig. 7A) showed similar labeling intensities to sections directly labeled with the antibody (Fig. 6, A–C). However, sections that had been labeled with the antibody pre-absorbed with 400 μg mL⁻¹ (data not shown) or 1,600 μg mL⁻¹ (Fig. 7B) had no labeling in mitochondria and very scant labeling over the cytosol and bacteroids. These results confirm that the FLbR-2 protein located in the mitochondria is immunoprecipitated by the anti-DLDH antibody and that there are additional proteins in the cytosol and bacteroids that are also recognized by the antibody, albeit with lower specificity.

Figure 7.

Immunogold localization of FLbR-2 in soybean nodules. A, Section probed with anti-pea DLDH antibody pre-absorbed with 40 μg mL⁻¹ rFLbR-2 protein. Arrows show labeling of mitochondria (m), bacteroids (b), and cytosol. B, Section probed with anti-pea DLDH antibody pre-absorbed with 1,600 μg mL⁻¹ rFLbR-2 protein. No gold particles are visible on mitochondria (m), but a few can still be seen within bacteroids and in the cytosol (arrows). s, Intercellular space. Bars = 500 nm.

DISCUSSION

A major goal of this work was to isolate proteins with FLbR activity from soybean nodules. The sequence of the FLbR-1 cDNA was used to design primers and amplify by PCR other FLbR-encoding cDNA sequences from a soybean nodule library. This strategy proved to be successful and enabled us to isolate a cDNA clone encoding a protein, FLbR-2, homologous to, but clearly different from, FLbR-1. Southern analysis of genomic DNA using soybean cDNA for FLbR-1 as a probe indicated that at least two copies of the flbr gene are present in the soybean genome (Ji et al., 1994a). One of these copies should be logically attributed to flbr-2, but additional flbr genes may be still present in soybean. Because the FLbR-1 and FLbR-2 cDNAs were obtained from an identical soybean library, both genes are expressed in the same growth stage of the plant and the differences at the nucleotide level cannot be accounted for by allelic variations between cultivars. Sequence analyses revealed high homology of FLbR-2 with soybean FLbR-1 and pea DLDH at the nucleotide and amino acid levels. These analyses also showed that the active disulfide center, FAD-binding domain, and NADH-binding domain are highly conserved in FLbRs and other oxidoreductases, namely, mercuric acid reductase, trypanothione reductase, and glutathione reductase (Fig. 1). Phylogenetic analyses showed that FLbR-1 clustered with the DLDH group, whereas the other oxidoreductases form a separate group (Pullikuth and Gill, 1997). As expected, soybean nodule FLbR-2 (this work) and its cowpea nodule homolog (Luan et al., 2000) are also closer to the DLDH proteins than to the other enzymes of the oxidoreductase family (data not shown).

There were at least two major differences between FLbR-1 and FLbR-2. The first is the existence of 23 extra amino acids at the C terminus of FLbR-1. The C terminus of FLbR-2 shares high homology with the family of eukaryotic DLDHs as opposed to soybean FLbR-1 (Fig. 1). A comparison of the 3′-coding region of the FLbR-1 and FLbR-2 cDNAs reveals that one base (g^1,512) is deleted in the latter, resulting in a frameshift and in the consequent formation of an early stop codon. This change in the FLbR-2 cDNA sequence was detected in several independent clones sequenced in both directions in this region of the cDNA. The function of the 23 additional residues remains unknown, but they could conceivably influence substrate specificities and/or subcellular localization. The second difference is that FLbR-1 and FLbR-2 differ in the gene expression pattern within the soybean plant (compare figure 6 in Ji et al., 1994a with Fig. 4A in this paper). The transcripts of flbr-1 are particularly abundant in nodules and leaves, whereas the transcripts of flbr-2 are more abundant in the shoot and root. Although this expression pattern does not preclude a role of the enzyme as FLbR, it indicates that its main function is not related to N₂ fixation. In fact, it is consistent with the conclusion that FLbR-2 is indeed the soybean DLDH.

Immunoblot analysis and immunogold labeling using an antibody to pea leaf DLDH (Figs. 5 and 6) allowed a number of important conclusions to be drawn. (a) The antibody recognizes rFLbR-2 purified from E. coli as well as nFLbR-2 from soybean and pea. In all cases, major immunoreactive bands were detected at approximately 50 kD. A second immunoreactive band at approximately 104 kD may correspond to the homodimer, which is the native state of FLbRs (Ji et al., 1991; Luan et al., 2000), assuming that this protein was particularly resistant to denaturation. (b) FLbR-2 is indeed localized in mitochondria, confirming that, as predicted, the enzyme is synthesized in the nodules as a preprotein bearing a mitochondrial transit peptide. (c) An immunoreactive protein is also present in the bacteroids, as expected for the cross-reaction between plant and bacterial DLDHs. (d) Immunoreactive bands at 50 kD are also present in soybean and pea leaves, as expected for mitochondrial DLDHs and in agreement with flbr-2 expression in leaf tissue (Fig. 4). (e) There is a 70-kD immunoreactive protein as well as gold labeling in the cytosol of soybean nodules but not of pea nodules. This provides correlative, but strong, evidence that the labeling observed in the soybean nodule cytosol is attributable to the 70-kD protein rather than to FLbR-2 or DLDH.

The location of nFLbR-2 in nodule mitochondria is at odds with the capacity of this enzyme to reduce ferric Lb, which is a cytosolic protein. We are forced to conclude that FLbR-2 is not involved in Lb reduction in vivo. However, we cannot exclude the possibility that other FLbR isozymes or DLDH-like proteins play such a role. Thus, the finding of a protein immunoreactive to DLDH antibody present in the soybean nodule cytosol deserves further investigation. The similar biochemical, kinetic, and immunological properties of soybean nodule FLbR-2 and pea leaf DLDH, their very high amino acid sequence identity, and their colocalization in mitochondria all draw us to the conclusion that FLbR-2 is soybean nodule DLDH. There is no previous information about any DLDH of legume nodules, despite its critical role in energy metabolism. In pea leaves, there are at least two DLDHs, one isozyme located in the mitochondria, which is shared by the pyruvate dehydrogenase and Gly decarboxylase complexes (Bourguignon et al., 1996), and another isozyme located in the chloroplasts (Conner et al., 1996). In the nodules, the role of DLDH is probably related to its participation in the pyruvate dehydrogenase complex, which catalyzes the key regulatory step of carbohydrate metabolism (the oxidative decarboxylation of pyruvate) in the mitochondria (Bourguignon et al., 1996; Conner et al., 1996). These organelles are very abundant and active in both the nodule parenchyma (inner cortex) and at the periphery of infected cells (Millar et al., 1995; Dalton et al., 1998). Therefore, we can predict that nodule FLbR-2 (or DLDH) will be essential to maintain respiratory activity in nodules and, hence, optimal N₂ fixation. However, as indicated by the immunolocalization data presented in this work, it will be important to identify other DLDH isozymes or closely related DLDH proteins in nodule compartments other than the mitochondria, especially in the cytosol.

MATERIALS AND METHODS

Amplification and Isolation of FLbR-2 cDNA

Two oligonucleotide primers (forward: 5′-CATATGGCGTCCGGATCTGAC-3′; reverse: 5′-AAGCTTCATCTCCAAGTCATTGTAA-3′) were designed to amplify a 1.5-kb fragment from a λgt11 cDNA library from soybean (Glycine max) nodules (Ji et al., 1994a). The oligonucleotides contained an NdeI site at the ATG starting codon and an HindIII site for the antisense primer, respectively (underlined). The nodule library was used as the template for PCR amplification. PCR components and concentrations were as follows: 0.5 μm for each sense and antisense primer, 200 μm for each deoxyribonucleoside triphosphate (dNTP), 1.5 mm MgCl₂, and 2.5 units of Taq DNA polymerase (Life Technologies, Rockville, MD) in a final volume of 25 μL of the PCR buffer. This consisted of 10 mm Tris-HCl (pH 8.3), 50 mm KCl, 1.5 mm MgCl₂, and 0.001% (w/v) gelatin. Before amplification, tubes were incubated at 95°C for 3 min to ensure that the template DNA was completely denatured. Amplification was carried out for 40 cycles at 55°C/1 min for annealing, 72°C/1.5 min for extension, and 95°C/1 min for denaturation. An additional annealing and extension step were performed at 55°C/1 min and 72°C/5 min, respectively. The total volume of the PCR samples was electrophoresed in a 1.0% (w/v) agarose gel. The PCR products were isolated from the melted agarose using the Geneclean kit (Bio 101, Vista, CA) and resuspended in 10 μL of sterile water.

Cloning and Sequencing of the cDNA

An aliquot (4 μL) of the resuspended DNA was used to clone each PCR product into the linearized vector pCR2.1 (Invitrogen, Carlsbad, CA) following the manufacturer's instructions. Several independent clones were sequenced in both directions by the dideoxy method (Sanger et al., 1977) at the DNA Sequencing Facility of the University of Nebraska-Lincoln. Database searches were performed with the BLAST program (Altschul et al., 1997). Sequence alignments and homology analyses were performed using the PileUp and Bestfit programs, respectively, of the Genetics Computer Group (Madison, WI). Signal peptide analyses and predictions of subcellular localization were performed using the programs MitoProtII (Claros, 1995), PSORT (Nakai and Kanehisa, 1992), and TransitP (Center for Biological Sequence Analysis, Department of Biotechnology, Technical University of Denmark, Copenhagen).

An flbr-2 internal reverse primer (5′-CGGCCCCATACTGAGCC-3′) and a λ-forward primer were used to obtain by PCR the 5′-UTR and the transit peptide sequence of the FLbR-2 cDNA. To obtain the 3′-UTR, combinations of flbr-2 internal forward oligonucleotide (5′-TAGTCATTGGGGCAGGCTAC-3′) with λ-forward and λ-reverse primers were used (Arredondo-Peter et al., 1997). The PCR components and concentrations were the same as before. Bands having the expected sizes were cloned into pCR2.1 and sequenced as described above.

Cloning and Heterologous Expression of the cDNA

The construct pCR2.1::FLbR-2 was digested with NdeI and XhoI restriction enzymes and separated on a 1% (w/v) agarose gel. The 1.5-kb fragment was extracted from the gel using the Geneclean kit and directionally subcloned in the pET-28a(+) vector (Novagen, Madison, WI) into the NdeI and XhoI sites. Cells of E. coli strains XL1 blue (CLONTECH, Palo Alto, CA) and BL21(DE3) (Invitrogen) were transformed with the resultant construct pET-28a::FLbR-2. Positive colonies were selected in Luria-Bertani plates containing kanamycin (100 μg mL⁻¹).

To determine the optimal incubation time, transformed E. coli BL21(DE3) cells were grown in small flasks until the A₆₀₀ reached 0.5, and IPTG was added at a final concentration of 1 mm. Aliquots were analyzed every 1 h for a total of 7 h and a final aliquot was taken after 18 h. Cells were dissolved in buffer containing SDS and electrophoresed in 7.5% (w/v) polyacrylamide denaturing gels.

Purification of Recombinant and Native Proteins

Host cells containing the pET-28a(+):FLbR-2 construct were grown in a 4-L flask using 1 L of Luria-Bertani broth containing kanamycin (100 μg mL⁻¹). Cultures were vigorously aerated at 37°C and grown until the A₆₀₀ reached 0.6; then, 1 mm IPTG was added. Cells were harvested after 6 h by centrifugation, and the cell paste was stored at −80°C until used. An aliquot of the cell paste (3 g) was thawed on ice and resuspended in 50 mL of buffer containing 20 mm potassium phosphate (pH 6.8), 1 mm phenylmethylsulfonyl fluoride, and 10 μg mL⁻¹ each of pepstatin, leupeptin, and chymostatin. Lysozyme (2 mg mL⁻¹) was added, and the suspension was incubated at 4°C with gentle stirring for 15 min and then sonicated with four 1-min pulses at 70% power (Sonifier 450, Branson Ultrasonics, Danbury, CT). After sonication, DNase (40 units mL⁻¹) and RNase (3 units mL⁻¹) were added, and the lysed cell suspension was incubated for another 15 min at 4°C. The suspension was then cleared by centrifugation at 48,000g for 15 min, and the resulting supernatant was chromatographed on a Probond Ni-chelating resin (25 mm i.d. × 50 mm) column (Invitrogen), equilibrated with 20 mm potassium phosphate (pH 7.8). rFLbR-2 was eluted with 20 mm potassium phosphate (pH 6.6) containing 500 mm imidazole and detected at 280 nm. The collected yellowish fraction was diluted to 0.5 mm EDTA and dialyzed three times against 50 mm Tris-HCl (pH 7.5) containing 0.5 mm EDTA. This fraction was centrifuged for 10 min at 48,000g, and the supernatant was concentrated using YM-30 filters (Amicon, Danvers, MA).

The concentrated solution was loaded onto a strong anion-exchange (4.6 × 100 mm; 10-μm particle size) column (Poros HQ-H; Applied Biosystems, Foster City, CA) equilibrated with 50 mm Tris-HCl (pH 7.5). Chromatography was performed on a BioCad workstation (Applied Biosystems) with detection at 280 and 400 nm. The column was washed with 4 column volumes of equilibrating buffer, and then a linear NaCl gradient (0–500 mm) with 20 column volumes was applied. rFLbR-2 eluted at approximately 370 mm NaCl. Collected fractions were pooled and made to 200 mm EDTA. This solution was salt washed and concentrated on Centricon C-30 devices (Amicon). The fused rFLbR-2 containing a His tag was digested for 2 h at 20°C with biotinylated thrombin, and the thrombin was subsequently removed using streptavidin-agarose (Novagen). Protein purification was monitored by SDS-PAGE (Laemmli, 1970). Protein was determined by a dye-binding assay (Bio-Rad) using bovine serum albumin as a standard.

Soybean nFLbR-1 and nFLbR-2 were purified from nodules as reported by Ji et al. (1994b), except that the ammonium sulfate precipitation step was omitted and an additional anion-exchange chromatography step was performed on the BioCad workstation as described above.

Amino Acid Sequence of rFLbR-2

N-Terminal Sequencing

The partial N-terminal sequences were determined after SDS-PAGE for rFLbR-2 or after IEF for nFLbR-2 (Jun et al., 1994). Proteins were transferred onto Immobilon-P^SQ (Millipore) membranes and subjected to automated Edman degradation (Procise 494; Applied Biosystems) using protocols recommended by the manufacturer at the Protein Core Facility of the University of Nebraska, Lincoln.

Protein Cleavage and Fragment Analysis

Approximately 50 μg of rFLbR separated by SDS-PAGE or nFLbR separated by IEF was stained with Coomassie blue or with Coomassie blue plus crocein scarlet, respectively. The destained gel was washed extensively with water. Gel slices containing the rFLbR band and nFLbR top band (pI 5.8) were excised and placed in 200 μL of 70% (v/v) formic acid; then, 3 mg of cyanogen bromide in 70% (v/v) formic acid was added. Blank pieces of the same gels were used as a control. Gel pieces were incubated with cyanogen bromide in the dark for 24 h under an Ar atmosphere. Gel slices were subsequently dried in a vacuum evaporator, washed with 200 μL of water, and redried. Peptides were eluted from the gel slices by adding 200 μL of 0.1% (v/v) trifluoroacetic acid/60% (v/v) acetonitrile, and the samples were agitated on a rocker at room temperature for 1 h. This step was repeated twice. Pooled solutions containing peptides were dried as above and resuspended in 40 μL of 0.05% (v/v) trifluoroacetic acid/25% (v/v) acetonitrile. This solution was made up to 200 μL with 0.1% (v/v) trifluoroacetic acid in water. Samples were filtered through a 0.22-μm filter and analyzed by reversed-phase HPLC (Waters, Milford, MA) using a C₁₈ (2.1 × 250 mm; 5-μm particle size) column (Vydac, Hesperia, CA). Peptides were eluted with a gradient of acetonitrile at 200 μL min⁻¹ and detected at 210 nm. Selected HPLC peaks were analyzed by N-terminal sequencing as described above and/or by electrospray MS at the Center for Mass Spectrometry, University of Nebraska, Lincoln.

Characterization of rFLbR-2

Purified rFLbR-2 was electrophoresed in a 7.5% (w/v) polyacrylamide SDS gel, blotted onto a nitrocellulose membrane, and probed with anti-soybean nFLbR antibodies at a dilution of 1:1,000 (Ji et al., 1991) as described by Sarath and Wagner (1989). Pig DLDH was included as a control. Native and SDS-PAGE (7.5%, w/v) were carried out using conventional protocols (Laemmli, 1970). IEF was performed in 7.5% (w/v) polyacrylamide gels, using ampholytes with pI values in the range of 4.45 to 9.60 (Bio-Rad) according to Jun et al. (1994).

The kinetics of the reactions and the characterization of enzymes were performed with a Cary 1-Bio spectrophotometer (Varian, Mulgrave, Australia) using a 200-μL microcuvette (1.0-cm path length). Final enzyme concentrations were between 0.2 and 2 μg of rFLbR protein per assay. Soybean ferric Lb reduction and lipoamide-mediated NADH oxidation were assayed following the decreases of A₅₇₄ (extinction coefficient of 10.2 mm⁻¹ cm⁻¹) and A₃₄₀ (extinction coefficient of 6.2 mm⁻¹ cm⁻¹), respectively, as indicated by Saari and Klucas (1984) and Ji et al. (1994b), respectively. All kinetics measurements were made in a final assay buffer containing 50 mm potassium phosphate (pH 6.5), 2 mm EDTA, and fixed (500 μm) or variable (40–500 μm) NADH concentrations. Averages of values from three experiments are reported. Original data were fitted to the Michaelis-Menten equation using Sigmaplot software (v3.0; SPSS, Chicago).

Expression of FLbR-2 in Soybean

Total RNA was extracted from nodules using the hot phenol method followed by LiCl precipitation (de Vries et al., 1982). For the RT-PCR analysis of soybean tissues, total RNA (5 μg) was treated with 2 units of DNase I at 37°C for 10 min to remove traces of contaminating DNA. After addition of 2.5 mm EDTA, samples were incubated at 65°C for 15 min to inactivate DNase. For RT, RNA samples were annealed to the primer 5′-CTCGAGGATCCGCGGCCGC-(T)₂₀-3′ at 70°C for 10 min, and then the cDNAs were synthesized using 200 units of reverse transcriptase (Superscript, Life Technologies) in a reaction mixture containing 10 mm dithiothreitol, 1.25 mm dNTPs, and buffer (20 mm Tris-HCl [pH 8.4], 50 mm KCl, 2.5 mm MgCl₂). The reaction proceeded at 42°C for 55 min and was stopped at 70°C for 15 min. The remaining RNA present in the samples was removed by incubation with 1 unit of RNase H at 37°C for 20 min. The reaction mixture was diluted to 120 μL, and 5 μL was used as template for PCR amplification.

For the PCR reactions, two gene-specific primers were designed based on the flbr-2 sequence. Primers were 5′-GGTATTGAAGGTCTATTCAAGAAAAAC-3′ (forward) and 5′-TATCTTGTCCAACCCAAGTTCGGC-3′ (reverse). The reaction mixture contained 5 μL of first-strand cDNA, 0.25 mm dNTPs, 1.5 mm MgCl₂, 0.2 μm of primers, and 1.25 units of Taq polymerase (Life Technologies) in a total volume of 25 μL. The PCR-cycling conditions comprised an initial denaturation step at 94°C for 2 min, 30 to 35 cycles at 94°C for 30 s, 60°C for 30 s, and 72°C for 45 s, and a final elongation step at 72°C for 10 min. As an internal control, PCR was performed simultaneously with ubiquitin primers (Horvath et al., 1993). As markers of nodule activity, fragments of Lb a (Hyldig-Nielsen et al., 1982) and of constitutive cytosolic Gln synthetase (Miao et al., 1991; Roche et al., 1993) from soybean were PCR amplified using specific primers. Primers for Lb were 5′-CACTGAGAAGCAAGATGCTTTGG-3′ (forward) and 5′-AGCTGCTGCCAATTCATCGTAG-3′ (reverse). Primers for Gln synthetase were 5′-AGCTGGTGATGAGATTTGGGCA-3′ (forward) and 5′-AACCACGTATGGGTCCATGTTG-3′ (reverse).

Immunolocalization of FLbR-2 in Soybean and Pea Nodules

Isolation of soybean nodule mitochondria, cytosol, and bacteroids was performed as previously described (Moran et al., 2000). Crude extracts were obtained by grinding leaves or nodules of soybean or pea in 30 mm 3-[N-morpholino]-propanesulfonic acid (pH 7.2), 2 mm EDTA, 2% (w/v) polyvinylpolypyrrolidone, 1 mm phenylmethylsulfonyl fluoride, and 2.5 μg each of leupeptin and pepstatin. Extracts were filtered through Miracloth (Calbiochem, San Diego) and cleared by centrifugation, and the supernatants were desalted using Centricon-30 (Amicon). Proteins were resolved in 7.5% (w/v) polyacrylamide SDS gels and electroblotted onto a polyvinylidene difluoride membrane. Immunoblots were carried out following standard protocols using a 1:2,000 dilution of rabbit polyclonal antibody raised against pea leaf mitochondria (Turner et al., 1992). The secondary antibody was goat anti-rabbit IgG alkaline phosphatase conjugate (Sigma, St. Louis) at a dilution of 1:30,000. Immunoreactive proteins were detected using 5-bromo-4-chloro-3-indolyl-phosphate/nitroblue tetrazolium as the substrate (Sigma).

For electron microscopy analysis, mature, N₂-fixing soybean and pea nodules were fixed in 4% (w/v) paraformaldehyde and 0.1% (v/v) glutaraldehyde in 50 mm potassium phosphate buffer (pH 7.0). Slices (200 μm) of four nodules from each species were obtained with a Vibratome 1000 (Agar Scientific, Stansted, UK), immersed overnight in 1.8 m Suc, and then frozen rapidly in liquid N₂. The frozen slices were then freeze substituted in methanol containing 0.5% (w/v) uranyl acetate using an EM AFS freeze substitution apparatus (Leica, Vienna) at −90, −65, and −45°C over a period of 68.5 h, before being embedded and polymerized in Lowicryl HM23 (Polysciences, Warrington, PA) at −45°C. Ultrathin sections (80 nm) were cut on an Ultracut E microtome (Leica) and collected on pioloform/carbon-coated Ni grids. The sections were then immediately immunogold labeled according to the procedures of James et al. (1996). They were first placed for 1 h on a blocking/diluting buffer containing 1% (v/v) Tween 20 and 1% (w/v) bovine serum albumin in Tris-buffered saline (10 mm Tris-HCl [pH 7.5], 150 mm NaCl, 0.5 g L⁻¹ polyethyleneglycol-20 K, 14 mm Na₃N) and then incubated for 2 h in a 1:100 dilution (in buffer) of the primary antibody (rabbit anti-pea leaf DLDH). After the grids were washed, they were incubated in a 1:50 dilution of goat anti-rabbit antibodies conjugated to 15-nm gold particles (Amersham Pharmacia Biotech, Piscataway, NJ) for 1 h. The sections were viewed and photographed using a 1200 EX transmission electron microscope (JEOL, Tokyo). As a control, serial sections of soybean nodules were immunogold labeled using a 1:100 dilution of anti-pea leaf DLDH that had been pre-absorbed for 1 h at room temperature with various concentrations (40, 400, and 1,600 μg mL⁻¹) of FLbR-2 protein.