ADP-Ribosylation of Variants of Azotobacter vinelandii Dinitrogenase Reductase by Rhodospirillum rubrum Dinitrogenase Reductase ADP-Ribosyltransferase

Abstract

In a number of nitrogen-fixing bacteria, nitrogenase is posttranslationally regulated by reversible ADP-ribosylation of dinitrogenase reductase. The structure of the dinitrogenase reductase from Azotobacter vinelandii is known. In this study, mutant forms of dinitrogenase reductase from A. vinelandii that are affected in various protein activities were tested for their ability to be ADP-ribosylated or to form a complex with dinitrogenase reductase ADP-ribosyltransferase (DRAT) from Rhodospirillum rubrum. R140Q dinitrogenase reductase could not be ADP-ribosylated by DRAT, although it still formed a cross-linkable complex with DRAT. Thus, the Arg 140 residue of dinitrogenase reductase plays a critical role in the ADP-ribosylation reaction. Conformational changes in dinitrogenase reductase induced by an F135Y substitution or by removal of the Fe₄S₄ cluster resulted in dinitrogenase reductase not being a substrate for ADP-ribosylation. Through cross-linking studies it was also shown that these changes decreased the ability of dinitrogenase reductase to form a cross-linkable complex with DRAT. Substitution of D129E or deletion of Leu 127, which result in altered nucleotide binding regions of these dinitrogenase reductases, did not significantly change the interaction between dinitrogenase reductase and DRAT. Previous results showed that changing Lys 143 to Gln decreased the binding between dinitrogenase reductase and dinitrogenase (L. C. Seefeldt, Protein Sci. 3:2073–2081, 1994); however, this change did not have a substantial effect on the interaction between dinitrogenase reductase and DRAT.

Nitrogenase activity in Rhodospirillum rubrum is regulated by reversible ADP-ribosylation of dinitrogenase reductase at Arg 100. Dinitrogenase reductase ADP-ribosyltransferase (DRAT) catalyzes the transfer of the ADP-ribose moiety from NAD to Arg 100, rendering dinitrogenase reductase inactive. The enzyme is reactivated upon removal of ADP-ribose by dinitrogenase reductase-activating glycohydrolase (DRAG). DRAT is very specific for native dinitrogenase reductase as a substrate. No other acceptor molecules have been found, although several have been tested; these include oxygen-denatured dinitrogenase reductase, arginine, dansylarginine, and a hexapeptide of dinitrogenase reductase containing Arg 100, the site of ADP-ribosylation (14, 16). DRAT can modify dinitrogenase reductases from Azotobacter vinelandii, Klebsiella pneumoniae, and Clostridium pasteurianum, although these organisms do not contain endogenous ADP-ribosylation systems (14).

Seefeldt and coworkers have created and characterized several site specifically altered A. vinelandii dinitrogenase reductases which have decreased ability relative to wild-type dinitrogenase reductase to support substrate reduction; these researchers thereby have identified regions of dinitrogenase reductase important in the transfer of electrons to dinitrogenase (11, 20, 21). Figure 1 shows the structure of dinitrogenase reductase (4) and the positions of the residues described in this paper. Characterization of the altered dinitrogenase reductase in which Asp 129 was replaced with Glu suggests that Asp 129 is involved in MgATP hydrolysis (11), which is coupled to the transfer of electrons from dinitrogenase reductase to dinitrogenase. Studies also have shown that both Arg 140 and Lys 143 are important for the docking of dinitrogenase reductase and dinitrogenase (22). Seefeldt and coworkers have also characterized an altered dinitrogenase reductase in which Leu 127 was deleted (Leu127Δ) (10, 21). Leu 127 is located between Asp 125, which is located in the nucleotide binding site and interacts with the Mg²⁺ ion associated with the nucleotide (23), and Cys 132, which is a ligand for the Fe₄S₄ cluster (7). The Fe₄S₄ cluster of the nucleotide-free Leu127Δ dinitrogenase reductase has chemical properties similar to those of the Fe₄S₄ cluster of the MgATP-bound wild-type dinitrogenase reductase. Leu127Δ dinitrogenase reductase also binds tightly to dinitrogenase in the presence and absence of MgATP.

FIG. 1.

Crystal structure of A. vinelandii dinitrogenase reductase as determined by Georgiadis et al. (4). The following residues are highlighted: Arg 100, Arg 140, Phe 135, Lys 143, Asp 129, Asp 43, and Leu 127 and the iron atoms of the Fe₄S₄ cluster. This structure is based on the form of A. vinelandii dinitrogenase reductase with one molecule of ADP bound.

In this work, we used several well-characterized altered forms of dinitrogenase reductase to further define the structural properties required for dinitrogenase reductase to interact with DRAT and also for dinitrogenase reductase to be a substrate for ADP-ribosylation.

MATERIALS AND METHODS

Purification of the altered dinitrogenase reductases.

Purification of the altered dinitrogenase reductases was performed by Lance Seefeldt and coworkers at Utah State University as previously described (11, 20–22). All of these proteins are “native” in that they are dimers with Fe₄S₄ clusters with a normal electron paramagnetic resonance signal.

Purification of DRAT.

DRAT was purified from R. rubrum strain UR356 as previously described (5).

ADP-ribosylation of the altered dinitrogenase reductases.

ADP-ribosylation reactions were performed with microcentrifuge tubes that had been placed inside 9-ml vials that were made anaerobic by repeated evacuation and flushing with nitrogen. A solution containing 100 mM dithionite was added to the vial on the outside of the microcentrifuge tube (16). This procedure creates an anaerobic environment that permits oxygen to be scavenged with a minimum amount of dithionite in the reaction mixture (dithionite reduces NAD, the ADP-ribose donor). Dinitrogenase reductase was incubated in a 40-μl reaction mixture containing 9.0 μg of purified protein, 15 mM HEPES (pH 7.6), 0.1 mM sodium dithionite, 2 mM NAD, 1.25 mM MgADP, and 1.5 μg of purified DRAT at 30°C for 3 min. (MgADP stimulates the ADP-ribosylation of A. vinelandii dinitrogenase reductase [15]). The reaction was stopped by the addition of 40 μl of sodium dodecyl sulfate (SDS) buffer containing 130 mM Tris (pH 6.8), 4.2% (wt/vol) SDS, 20% (vol/vol) glycerol, 0.003% (wt/vol) bromphenol blue, and 10% (vol/vol) 2-mercaptoethanol (added fresh), and the mixture was boiled for 1 min. After this step, the reaction mixture was diluted 10-fold, and 5.0 μl was loaded onto a polyacrylamide gel (10% [wt/vol] total acrylamide; ratio of acrylamide to bisacrylamide, 172:1) to resolve the modified and unmodified subunits of dinitrogenase reductase by electrophoresis (9). The separated proteins were transferred to a nitrocellulose membrane, which was then incubated with polyclonal antibodies against dinitrogenase reductase (1:5,000, 1 h). The position of dinitrogenase reductase on the blot was visualized by chemiluminescence (Amersham), and the proteins were quantitated by densitometry scanning. Some mutant forms of dinitrogenase reductase migrate anomolously on SDS gels.

Cross-linking of the altered dinitrogenase reductases and DRAT.

Cross-linking reactions were also performed with microcentrifuge tubes prepared as described above (6). Dinitrogenase reductase was incubated in a 40-μl reaction mixture containing 9.0 μg of purified protein, 15 mM HEPES (pH 7.6), 0.1 mM sodium dithionite, 2 mM NAD, 1.25 mM MgADP, 3.0 μg of purified DRAT, and 5 mM (1-ethyl-3-(3-dimethylaminopropyl)-carbodimide (EDC). The cross-linking reaction was incubated at 30°C for 3 min and then stopped by the addition of 40 μl of SDS buffer. After the mixture was boiled for 1 min, 5.0-μl reaction samples were loaded onto a gel as described above. The separated proteins were electrophoretically transferred to a nitrocellulose membrane, which was then incubated with polyclonal antibodies against DRAT (1:1,000, 1 h). The blot was then incubated with anti-rabbit immunoglobulin G–horseradish peroxidase conjugate (1:3,000, 40 min). The position of the DRAT-dinitrogenase reductase complex on the blot was visualized by chemiluminescence (Amersham), and quantitation was done by densitometry scanning.

Preparation of apo-dinitrogenase reductase.

Apo-dinitrogenase reductase for this study was prepared by Priya Rangaraj as previously described (13, 24). Briefly, 2.0 mg of purified dinitrogenase reductase was incubated with 20 μmol of α, α′-dipyridyl in the presence of 2.5 mM MgATP and 2 mM sodium dithionite. After incubation at 25°C for 30 min, apo-dinitrogenase reductase was passed over a Sephadex G-25 column (1 by 10 cm) equilibrated with 25 mM Tris-HCl (pH 7.4) and 2.0 mM sodium dithionite to remove α,α′-dipyridyl. To confirm the complete conversion of holo-Fe protein to apo-Fe protein, the apo-Fe protein preparation was coupled to MoFe protein in an in vitro acetylene reduction assay and shown to contain no substrate reduction ability. However, the prepared apo-dinitrogenase reductase was still active in its ability to support iron-molybdenum cofactor synthesis (19). Zheng et al. have demonstrated the ability of apo-dinitrogenase reductase prepared in this manner to be reconstituted with a normal Fe₄S₄ cluster by NifS (24).

RESULTS AND DISCUSSION

The following altered A. vinelandii dinitrogenase reductases, which have a decreased ability, relative to wild-type dinitrogenase reductase, to support substrate reduction, were characterized for their ability to be substrates for DRAT-catalyzed ADP-ribosylation: R140Q, K143Q, F135Y, D43N, D129E, and L127Δ. A decrease in dinitrogenase reductase ADP-ribosylation could be due to a decreased ability of the altered dinitrogenase reductase to form a complex with DRAT; therefore, the interaction between these altered dinitrogenase reductases and DRAT was studied via chemical cross-linking of the two proteins. Table 1 depicts the previously described properties of these altered dinitrogenase reductases and the data obtained here. Figure 1 shows the crystal structure of A. vinelandii dinitrogenase reductase, with the above-mentioned residues highlighted.

TABLE 1.

Characteristics of altered A. vinelandii dinitrogenase reductases

Dinitrogenase reductase	Activitya	Binding to dinitrogenaseb	% Dinitrogenase reductase dimer ADP-ribosylatedc	Amt of DRAT cross-linked (pg)d
Wild-type	100	Normal	78	145
R140Q	38	Weak	0	20
K143Q	25	Weak	30	90
F135Y	0	Weak	6	4
D129E	0	Normal	55	70
D43N	0	Weak	0	0
L127Δ	0	Very strong	35	110

^aAbility of dinitrogenase reductase to support substrate reduction by dinitrogenase. Values are expressed as percent wild-type dinitrogenase reductase activity. Data are from Seefeldt and coworkers (11, 20–22; Lanzilotta and Seefeldt, unpublished data). 

^bAbility of dinitrogenase reductase to bind to wild-type dinitrogenase. Data are from Seefeldt and coworkers (11, 20–22; Lanzilotta and Seefeldt, unpublished data). 

^cModified dinitrogenase reductase as a percentage of total dinitrogenase reductase was determined by densitometry scanning of the modified and unmodified protein bands on a dinitrogenase reductase immunoblot and visualization by the chemiluminescence method (see Materials and Methods). 

^dThe quantity of DRAT present in the reaction mixture as a DRAT-dinitrogenase reductase cross-linked complex was determined by densitometry scanning of the cross-linked complex band on a DRAT immunoblot and visualization by the chemiluminescence method (see Materials and Methods). Within the errors of this type of measurement, DRAT forms a 1:1 complex with dinitrogenase reductase. 

R140Q dinitrogenase reductase.

Replacement of R140 with a Q residue renders dinitrogenase unable to be ADP-ribosylated. R140Q dinitrogenase reductase was incubated in an ADP-ribosylation reaction as described in Materials and Methods. Analysis of this reaction on dinitrogenase reductase immunoblots showed that R140Q dinitrogenase reductase was not ADP-ribosylated (Fig. 2, lane 2b). Reaction conditions were such that 78% of wild-type dinitrogenase reductase was ADP-ribosylated (Fig. 2, lane 1b). (One hundred percent ADP-ribosylation is represented by equal quantities of modified and unmodified dinitrogenase reductase subunits.) ADP-ribosylation of R140Q dinitrogenase reductase was not observed even when the incubation period was increased to 20 min (data not shown). These data indicate that the R140Q substitution disrupts the ability of dinitrogenase reductase to be a substrate for ADP-ribosylation.

FIG. 2.

Dinitrogenase reductase immunoblot of an SDS-polyacrylamide gel showing the ADP-ribosylation of the altered dinitrogenase reductases. The ADP-ribosylation reactions were performed as described in Materials and Methods with 9.0 μg of the specific dinitrogenase reductase incubated with 2.0 mM NAD, 1.25 mM MgADP, and 1.5 μg DRAT. Lanes a represent control reactions in which NAD (the ADP-ribose donor) and MgADP were excluded from the reaction mixtures. Lanes b represent complete reaction mixtures. Numbered lanes represent reactions performed with the following dinitrogenase reductases: lane 1, wild type; lane 2, R140Q; lane 3, K143Q; lane 4, F135Y; lane 5, D43N; lane 6, D129E; and lane 7, L127Δ. The ADP-ribosylated subunit of dinitrogenase reductase migrates at the position labeled “modified,” and the non-ADP-ribosylated subunit migrates at the position labeled “unmodified.” In some cases, the mutation slightly affects the position of migration (e.g., compare lanes 1a and 2a).

The inability of R140Q dinitrogenase reductase to be a substrate for ADP-ribosylation may be caused by a decrease in the binding of DRAT to this altered dinitrogenase reductase. Therefore, a chemical cross-linking method was used to study the binding between these two proteins. Dinitrogenase reductase and DRAT were incubated in a reaction mixture with the cross-linking reagent, EDC. Formation of a cross-linked complex between DRAT and dinitrogenase reductase was monitored on immunoblots of SDS-polyacrylamide gels developed with DRAT antibodies. When wild-type dinitrogenase reductase was incubated with DRAT in this cross-linking reaction mixture, a protein complex was formed that cross-reacted with DRAT (Fig. 3, lane 1) and dinitrogenase reductase antibodies (data not shown) and migrated at approximately 65 kDa on SDS-polyacrylamide gels. This complex was identified as DRAT covalently cross-linked to only one subunit of the dinitrogenase reductase dimer (the two subunits of dinitrogenase reductase dissociate under SDS gel electrophoresis conditions). This complex was not formed in control reactions in which dinitrogenase reductase was excluded from the reaction mixture (Fig. 3, lane C).

FIG. 3.

DRAT immunoblot of an SDS-polyacrylamide gel showing the cross-linking of DRAT and the altered dinitrogenase reductases. The complete cross-linking reaction was performed as described in Materials and Methods with 9.0 μg of the specific dinitrogenase reductase incubated with 2 mM NAD, 1.25 mM MgADP, 3.0 μg of DRAT, and 5 mM EDC (final reaction concentrations). Numbered lanes represent reactions performed with the following dinitrogenase reductases: lane C (control reaction), none; lane 1, wild type; lane 2, R140Q; lane 3, K143Q; lane 4, F135Y; lane 5, D43N; lane 6, D129E; and lane 7, L127Δ. The cross-linked complex of DRAT and dinitrogenase reductase (DRAT/DR) and free DRAT migrate at the indicated positions.

When R140Q dinitrogenase reductase and DRAT were incubated together in this cross-linking reaction mixture, R140Q dinitrogenase reductase formed a cross-linkable complex with DRAT (Fig. 3, lane 2). Quantitation of the complex by gel scanning showed that approximately 20 pg of DRAT was cross-linked to R140Q dinitrogenase reductase under the same conditions in which 140 pg of DRAT was cross-linked to wild-type dinitrogenase reductase. Although there was less cross-linking between DRAT and R140Q dinitrogenase reductase than between DRAT and wild-type dinitrogenase reductase, these data show that R140Q dinitrogenase reductase and DRAT are capable of forming a complex, albeit one in which R140Q dinitrogenase reductase is not ADP-ribosylated. These results are consistent with the R140Q mutation on dinitrogenase reductase disrupting some other step in the ADP-ribosylation reaction besides DRAT-dinitrogenase reductase binding.

The Fe₄S₄ cluster region of dinitrogenase reductase is important for ADP-ribosylation.

We also wanted to test if the conformation of the Fe₄S₄ cluster region is important for maintaining the proper orientation of Arg 100 such that ADP-ribosylation of dinitrogenase reductase can occur. Crystallographic data show that Phe 135 is located at the interface between the dinitrogenase reductase subunits near the Fe₄S₄ cluster (4). The F135Y dinitrogenase reductase protein contains an intact Fe₄S₄ cluster, as determined by electron paramagnetic resonance studies; however, circular dichroism studies show that the environment of the Fe₄S₄ cluster in this protein is different from that in wild-type dinitrogenase reductase (20). Therefore, F135Y dinitrogenase reductase was analyzed for its ability to be ADP-ribosylated. Only 6% of F135Y dinitrogenase reductase was ADP-ribosylated when incubated with DRAT and NAD under the stated reaction conditions (Fig. 2, lane 4b), in which 78% of wild-type dinitrogenase reductase was ADP-ribosylated (Fig. 2, lane 1b). This result indicates that F135Y dinitrogenase reductase is not an effective substrate for ADP-ribosylation.

Cross-linking studies show that the extent of formation of a cross-linkable complex between DRAT and F135Y dinitrogenase reductase is approximately 40 times lower than that between DRAT and wild-type dinitrogenase reductase under these conditions (Fig. 3, lane 4). This result indicates that the F135Y mutation changes the conformation of dinitrogenase reductase in such a way as to decrease the ability of DRAT to cross-link with dinitrogenase reductase. There are two possible explanations for this decrease in cross-linking: (i) F135Y dinitrogenase reductase may not interact with DRAT, and thus no cross-linkable complex is formed between the two proteins, or (ii) F135Y dinitrogenase reductase interacts with DRAT but in a conformation such that cross-linking between the two proteins does not occur. If DRAT and F135Y dinitrogenase reductase do still interact, it is in such a way that an unproductive complex forms and ADP-ribosylation of F135Y dinitrogenase reductase does not occur.

The Fe₄S₄ cluster is required for ADP-ribosylation and cross-linking with DRAT.

Consistent with these results is the inability of wild-type apo-dinitrogenase reductase to be ADP-ribosylated in the DRAT-catalyzed ADP-ribosylation reaction (Fig. 4A, lane 2). The Fe₄S₄ cluster of wild-type A. vinelandii dinitrogenase reductase was removed as described in Materials and Methods by treatment of dinitrogenase reductase with the chelator α,α′-dipyridyl (14, 24). Apo-dinitrogenase reductase also did not form a cross-linkable complex with DRAT when added to the previously described cross-linking reaction mixture (Fig. 4B, lane 2). These results show that removal of the Fe₄S₄ cluster does change the conformation of dinitrogenase reductase such that apo-dinitrogenase reductase does not form a productive complex with DRAT and therefore is not a substrate for ADP-ribosylation. Apo-dinitrogenase reductase is effective in its role in iron-molybdenum cofactor synthesis (18).

FIG. 4.

SDS-polyacrylamide gel electrophoresis analysis of the ADP-ribosylation of apo-dinitrogenase reductase and the cross-linking of DRAT and apo-dinitrogenase reductase. Lanes 1 represent reactions with wild-type holo-dinitrogenase reductase. Lanes 2 represent reactions with wild-type apo-dinitrogenase reductase prepared as described in Materials and Methods (apo-dinitrogenase reductase activity, 0.0 nmol of ethylene formed/min/mg). (A) Dinitrogenase reductase immunoblot of an SDS-polyacrylamide gel showing ADP-ribosylation reactions in which 9.0 μg of apo- or holo-dinitrogenase reductase was incubated with NAD, MgADP, and DRAT as described in the legend to Fig. 2. The ADP-ribosylated subunit of dinitrogenase reductase migrates at the position labeled “modified” and the non-ADP-ribosylated subunit migrates at the position labeled “unmodified.” (B) DRAT immunoblot of an SDS-polyacrylamide gel showing cross-linking reactions in which 9.0 μg of holo- or apo-dinitrogenase reductase was incubated with DRAT, NAD, MgADP, and EDC as described in the legend to Fig. 3. The cross-linked complex of DRAT and dinitrogenase reductase (DRAT/DR) and free DRAT migrate at the indicated positions.

The side-chain rings of Phe 135 in dinitrogenase reductase seem to be involved in the formation of a hydrophobic pocket around the Fe₄S₄ cluster (4). Substitution of Phe 135 with tyrosine would disrupt the hydrophobic character of the side chains and therefore might disrupt the environment surrounding the Fe₄S₄ cluster. This conformational change might disrupt the environment of Arg 100, which is also located at the interface between the dinitrogenase reductase subunits. Similar conformation changes could occur if the Fe₄S₄ cluster were removed from dinitrogenase reductase. These conformational changes would be consistent with our results that F135Y and apo-dinitrogenase reductase are not able to be ADP-ribosylated.

ADP-ribosylation of dinitrogenase reductase altered in the nucleotide binding region.

The ADP-ribosylation of A. vinelandii dinitrogenase reductase is stimulated by the presence of MgADP. Previous data suggested that this effect is due to the binding of MgADP to dinitrogenase reductase (15), which contains two nucleotide binding sites per dimer, rather than to DRAT itself. These nucleotide binding sites have been extensively studied for their role in nitrogenase substrate reduction. The nucleotide binding sites are located in the cleft between the dinitrogenase reductase subunits and are oriented with the triphosphates toward the Fe₄S₄ cluster. The role of nucleotide hydrolysis in the function of dinitrogenase reductase has been compared to the role of GTP hydrolysis in G proteins (8). MgATP hydrolysis is coupled to the transfer of electrons from dinitrogenase reductase to dinitrogenase. Seefeldt and coworkers have characterized several altered forms of dinitrogenase reductase in which mutations have been made in the nucleotide binding region (10, 11, 21). Substitution of Glu for Asp 129 resulted in a dinitrogenase reductase that could still interact with dinitrogenase but that failed to hydrolyze MgATP, and thus no electron transfer occurred (11). Substitution of Gln for Asp 43 resulted in a dinitrogenase reductase that had a decreased affinity for dinitrogenase and that was unable to support substrate reduction (W. N. Lanzilotta and L. C. Seefeldt, unpublished data). Seefeldt and coworkers have also characterized an altered dinitrogenase reductase in which Leu 127 was deleted (Leu127Δ) (10, 21). Leu 127 is located between Asp 125, which is in the nucleotide binding site and which interacts with the Mg²⁺ ion associated with the nucleotide (23), and Cys 132, which is a ligand for the Fe₄S₄ cluster (7).

These three altered forms of dinitrogenase reductase were studied for their ability to be substrates for ADP-ribosylation. Previous studies showed that D129E, L127Δ, and D43N dinitrogenase reductases bound MgADP (11, 21; Lanzilotta and Seefeldt, unpublished data); since MgADP was added to the reactions, the dinitrogenase reductases studied were in their MgADP-bound forms. D129E dinitrogenase reductase was ADP-ribosylated when incubated with DRAT and NAD (Fig. 2, lane 6b). Cross-linking studies showed that D129E dinitrogenase reductase could also form a cross-linkable complex with DRAT (Fig. 3, lane 6). Furthermore, L127Δ dinitrogenase reductase could also be ADP-ribosylated by DRAT (Fig. 2, lane 7b) and could form a cross-linkable complex with DRAT (Fig. 3, lane 7). In contrast, D43N was not a substrate for ADP-ribosylation (Fig. 2, lane 5b) and also could not form a cross-linkable complex with DRAT (Fig. 3, lane 5). The major observable difference between these altered dinitrogenase reductases is that D43N dinitrogenase reductase has a much weaker affinity for dinitrogenase (Lanzilotta and Seefeldt, unpublished data) than does either D129E (11) or L127Δ (21) dinitrogenase reductase. It is believed that DRAT and dinitrogenase interact with similar regions of dinitrogenase reductase. Therefore, the D43N mutation may cause a much greater overall conformational change in dinitrogenase reductase, thereby affecting the ADP-ribosylation site.

Nucleotide dependence of the ADP-ribosylation of L127Δ dinitrogenase reductase.

We have shown that L127Δ dinitrogenase reductase with MgADP bound is a substrate for ADP-ribosylation and can form a cross-linkable complex with DRAT. Seefeldt and coworkers (21) have shown that the Fe₄S₄ cluster of L127Δ dinitrogenase reductase has properties very similar to the Fe₄S₄ cluster of wild-type dinitrogenase reductase with MgATP bound, whether or not MgATP is bound to the L127Δ protein. Their results suggest that deletion of Leu 127 brings dinitrogenase reductase into an “MgATP-bound state” in the absence of any bound nucleotides (21). However, the following studies on the ADP-ribosylation properties of L127Δ dinitrogenase reductase with various adenine nucleotides bound clearly showed that there is a distinct difference between the nucleotide-free and MgATP-bound forms of L127Δ dinitrogenase reductase as perceived by DRAT.

L127Δ dinitrogenase reductase was incubated with DRAT, NAD, and various nucleotides and analyzed for its ability to be ADP-ribosylated. L127Δ dinitrogenase reductase can be ADP-ribosylated in its nucleotide-free form (Fig. 5, lane 1), and in its MgADP-bound form (Fig. 5, lane 3). However, MgATP inhibits the ADP-ribosylation of L127Δ dinitrogenase reductase by DRAT (Fig. 5, lane 6). To ensure that in this reaction all the L127Δ dinitrogenase reductase was in its MgATP-bound form, an ATP-generating system (phosphocreatine and creatine phosphokinase) was added to this reaction to convert any residual ADP to ATP. Control experiments showed that this ATP-generating system had no effect on the ADP-ribosylation of nucleotide-free L127Δ dinitrogenase reductase (Fig. 5, lane 4). The following cross-linking studies between DRAT and L127Δ dinitrogenase reductase also showed that there is a difference in the conformation of L127Δ dinitrogenase reductase in its nucleotide-free form (Fig. 6B, lane 1) and in its MgATP-bound form (Fig. 6B, lane 3). The extent of formation of a cross-linkable complex between DRAT and L127Δ dinitrogenase reductase is decreased when L127Δ dinitrogenase reductase is in its MgATP-bound form versus when it is in its nucleotide-free form. This decrease in cross-linking is also observed with DRAT and nucleotide-free or MgATP-bound wild-type dinitrogenase reductase (Fig. 6A, lanes 1 to 3).

FIG. 5.

Adenine nucleotide dependence of the ADP-ribosylation of L127Δ dinitrogenase reductase. Shown is a dinitrogenase reductase immunoblot of an SDS-polyacrylamide gel of ADP-ribosylation reactions (as described in Fig. 2) in which 9.0 μg of L127Δ dinitrogenase reductase was incubated with DRAT, NAD, and the following components: lane 1, none; lane 2, 1 mM ADP; lane 3, 1 mM MgADP; lane 4, 1.5 mM phosphocreatine and 0.3 U of creatine phosphokinase per ml (ATP-generating system); lane 5, 1 mM ATP plus ATP-generating system; and lane 6, 1 mM MgATP plus ATP-generating system. DRAT (used for these reactions) was passed through a Sephadex G-25 column equilibrated with 100 mM morpholinepropanesulfonic acid (MOPS) (pH 7.0)–1 mM dithiothreitol–20% glycerol to remove the ADP that was present during DRAT purification. The ADP-ribosylated subunit of dinitrogenase reductase migrates at the position labeled “modified,” and the non-ADP-ribosylated subunit migrates at the position labeled “unmodified.”

FIG. 6.

Adenine nucleotide dependence of cross-linking between DRAT and wild-type or L127Δ dinitrogenase reductase. Shown is a DRAT immunoblot of an SDS-polyacrylamide gel of cross-linking reactions (as described in Fig. 3) in which 9.0 μg of wild-type dinitrogenase reductase (A) or 9.0 μg of L127Δ dinitrogenase reductase (B) was incubated with DRAT, NAD, EDC, and the following components: lane 1, none; lane 2, 1.5 mM phosphocreatine and 0.3 U of creatine phosphokinase per ml (ATP-generating system); and lane 3, 1 mM MgATP plus ATP-generating system. DRAT was prepared as described in the legend to Fig. 5 to remove the ADP present during purification. The cross-linked complex of DRAT and dinitrogenase reductase (DRAT/DR) and free DRAT migrate at the indicated positions.

These results show that although the L127Δ dinitrogenase reductase Fe₄S₄ cluster has very similar properties when the protein is in its nucleotide-free or MgATP-bound form (21), other regions of L127Δ dinitrogenase reductase, particularly the region that interacts with DRAT, are clearly in a different conformation depending on the nucleotide state of the protein.

Substitution of Gln for Lys 143 does not affect the binding of dinitrogenase reductase to DRAT.

Seefeldt and coworkers have shown that Lys 143 of dinitrogenase reductase is important for the docking of dinitrogenase reductase with dinitrogenase (22). Changing Lys 143 to Gln resulted in a decreased affinity of the altered dinitrogenase reductase for binding to dinitrogenase (22). The crystal structure of dinitrogenase reductase shows Lys 143 positioned near the Arg 100 residue (4). To determine if the Lys 143 residue is also important for the interaction of dinitrogenase reductase with DRAT, K143Q dinitrogenase reductase was tested for its ability to be ADP-ribosylated and to be cross-linked to DRAT.

When K143Q dinitrogenase reductase was added to DRAT and NAD in an ADP-ribosylation reaction, 30% of the K143Q dinitrogenase reductase population was ADP-ribosylated (Fig. 2, lane 3b), compared to 78% of the wild-type dinitrogenase reductase population (Fig. 2, lane 1b). Cross-linking studies showed that K143Q dinitrogenase reductase and DRAT can interact and form a cross-linkable complex (Fig. 3, lane 3). Therefore, the substitution of Gln for Lys 143 has only slight effects on the interaction between DRAT and dinitrogenase reductase; therefore, dinitrogenase reductase Lys 143 is not a critical residue for the ADP-ribosylation of dinitrogenase reductase. In contrast, Lys 143 of dinitrogenase reductase is important in the interaction of dinitrogenase with dinitrogenase reductase. These data suggest that DRAT and dinitrogenase may interact with similar, but not identical, sites of dinitrogenase reductase.

A number of NAD binding proteins contain in their active sites arginine residues which interact with NAD. The crystal structure of horse liver alcohol dehydrogenase shows two arginine residues in the NAD binding site (3). The side chain of Arg 47 is hydrogen bonded to the adenine-proximal phosphate group of NAD, and the side chain of Arg 369 is hydrogen bonded to the nicotinamide-proximal phosphate group of NAD. Li et al. have confirmed the role of these arginines in NAD binding by analyzing the crystal structure of alcohol dehydrogenase bound to an analog of NAD (thiazole-4-carboxamide adenine dinucleotide) (12). The crystal structure for Pseudomonas putida lipoamide dehydrogenase complexed with NAD⁺ shows the side chain of Arg 266 hydrogen bonded to a water molecule which is hydrogen bonded to the nicotinamide ribose hydroxyl group (17). Furthermore, diptheria toxin (DT) bound to NAD has also been crystallized, and the NAD binding site has been characterized (1). His 21 is hydrogen bonded to the adenosine ribose hydroxyl group of NAD. Bell and Eisenberg (1) structurally aligned the NAD binding sites of Escherichia coli heat-labile enterotoxin and pertussis toxin to the NAD binding site of DT. They showed that Arg 7 of heat-labile enterotoxin and Arg 9 of pertussis toxin are in the same position in the NAD binding pocket as is His 21 of DT. Therefore, these arginine residues may play a role similar to that of His 21 in binding to the adenine-proximal ribose ring or possibly to the adenine-proximal phosphate group of NAD. The sites of binding of NAD to ADP-ribosyltransferases appear to be structural motifs distinctly different from the well-characterized Rossman fold observed for the dehydrogenases (2).

The N-C bond between the nicotinamide and ribose groups of NAD must be oriented near Arg 100 of dinitrogenase reductase for ADP-ribosylation of Arg 100 to occur. With this as a base position for NAD, molecular modeling (performed by Wayne Schultz) of the NAD molecule in its DT-bound conformation (1) allows us to speculate about the interaction of NAD and A. vinelandii dinitrogenase reductase. This molecular modeling shows that the NAD molecule could interact with both Arg 100 and Arg 140. Two possible orientations for NAD on dinitrogenase reductase are that (i) the NAD molecule is positioned such that it bridges the two subunits of dinitrogenase reductase or (ii) it is positioned in the cleft of the two subunits of dinitrogenase reductase. The crystal structure of dinitrogenase reductase reported by Georgiadis et al. (4) shows that the distance between the guandino nitrogens of the Arg 100 and Arg 140 side chains located in the same subunit is approximately 11 Å and the distance between the guandino nitrogens of the Arg 100 and Arg 140 side chains located in different subunits is approximately 8 Å (4). For the NAD molecule bound to DT, the distance between the nitrogen atom of the nicotinamide ring and the adenine-proximal phosphate group is approximately 7 Å (1).

Arg 140 of dinitrogenase reductase is required for the ADP-ribosylation of dinitrogenase reductase. By analogy to the previously described NAD binding proteins, Arg 140 may be involved in hydrogen bonding to either the ribose hydroxyl groups or the phosphate groups of NAD during DRAT-catalyzed ADP-ribosylation.

Conclusions.

The following conclusions can be made about the region of dinitrogenase reductase that is important for the ADP-ribosylation of dinitrogenase reductase. (i) The amino acid Arg 140 is required for dinitrogenase reductase to be ADP-ribosylated. The side chain of Arg 140 may hydrogen bond to the phosphate or ribose groups of NAD and thus help position NAD in the proper conformation for ADP-ribosylation of dinitrogenase reductase to occur. (ii) A native conformation in the Fe₄S₄ region of dinitrogenase reductase is also required for ADP-ribosylation to occur, as evidenced by the inability of F135Y and apo-dinitrogenase reductase to be ADP-ribosylated. (iii) The conformation of the L127Δ dinitrogenase reductase is readily recognized by DRAT; complex formation with DRAT and ADP-ribosylation of L127Δ dinitrogenase reductase are inhibited by MgATP.