|Home | About | Journals | Submit | Contact Us | Français|
In Rhodospirillum rubrum, nitrogenase activity is regulated posttranslationally through the ADP-ribosylation of dinitrogenase reductase by dinitrogenase reductase ADP-ribosyltransferase (DRAT). Several DRAT variants that are altered both in the posttranslational regulation of DRAT activity and in the ability to recognize variants of dinitrogenase reductase have been found. This correlation suggests that these two properties are biochemically connected.
The posttranslational ADP-ribosylation of proteins is the result of the enzymatic transfer of ADP-ribose from NAD, typically resulting in alteration of the functional properties of the respective proteins. The interest in the field has concentrated on two types of phenomena: poly-ADP-ribosylation of nuclear proteins in eukaryotes in the process of DNA excision repair (26, 27) and mono-ADP-ribosylation by bacterial enzymes as the mechanism of diphtheria, cholera, and pertussis toxins in eukaryotic host cells (3, 20). However, the reversible regulation of metabolic functions by ADP-ribosylation in animal tissues (8, 19, 31) and bacterial cells (16, 17, 25) is also physiologically important, and the regulation of the dinitrogenase reductase from the nitrogen fixation system in photosynthetic bacteria provides an attractive model system. In Rhodospirillum rubrum, the posttranslational regulation of the nitrogenase by the ADP-ribosylation has been well characterized. Nitrogenase is a protein complex of two components: dinitrogenase (an α2β2 tetramer of the nifD and nifK gene products) contains the active site of dinitrogen reduction, and dinitrogenase reductase (an α2 dimer of the nifH gene product) transfers electrons to dinitrogenase. The posttranslational regulation of the complex involves reversible mono-ADP-ribosylation of dinitrogenase reductase at R101 (22).
Two enzymes have been found to perform this reversible regulation in R. rubrum. Under certain conditions, dinitrogenase reductase ADP-ribosyltransferase (DRAT, the gene product of draT) transfers an ADP-ribosyl group from NAD to one subunit of the dinitrogenase reductase dimer, and the ADP-ribosylated dinitrogenase reductase is no longer competent to transfer electrons to dinitrogenase (13, 14). The ADP-ribosyl group on the inactivated dinitrogenase reductase can be removed by the dinitrogenase reductase-activating glycohydrolase (DRAG, the gene product of draG), thus recovering dinitrogenase reductase activity (5, 15–17, 24).
Presumably to avoid the possibility of futile cycling of ADP-ribosylation, the activities of both DRAT and DRAG are themselves regulated posttranslationally (12, 29). Under conditions appropriate for nitrogen fixation (energy sufficiency and a deficiency in fixed nitrogen), DRAG is active, causing dinitrogenase reductase to be in the active form. However, following an environmental shift that makes nitrogenase activity undesirable, such as the depletion of the energy or introduction of a good source of fixed nitrogen, DRAG loses its activity and DRAT becomes active. This results in the ADP-ribosylation of a fraction of the dinitrogenase reductase. Surprisingly, it has been found that DRAT activation is only transient, so that some time after the environmental shift, neither DRAT nor DRAG is active (28). This typically results in a plateau of nitrogenase activity, reflecting the amount of dinitrogenase reductase that was not ADP-ribosylated by DRAT during its active stage. When conditions change again to those favorable to nitrogenase activity, DRAG recovers its activity and restores nitrogenase activity by removing the ADP-ribosyl group from dinitrogenase reductase. The central issue in understanding this regulatory system has therefore become the nature of the posttranslational regulation of DRAT and DRAG.
To better understand the behavior of this posttranslational regulatory system, a screen was developed for mutants whose nitrogenase activity was no longer regulated by DRAT. A strain (UR484) that lacks DRAG but has excess DRAT was created; the chromosomal copy of draG was mutated, and multiple copies of draT were provided on a replicating plasmid (Table (Table1).1). This strain displayed low nitrogenase activity under all conditions (see Table Table2),2), which resulted in poor growth on nitrogen-free medium and allowed us to seek mutants that grew better because they possessed higher nitrogenase activity. UR484 (excess DRAT) was then mutagenized with N-methyl-N′-nitro-N-nitrosoguanidine (NTG) by a modification of a published protocol (1). Thirty independent mutagenized samples were inoculated into 10 ml of MN− (nitrogen-free) medium (culture media are described in references 4, 10, and 11, cultured for 10 days under a regimen of 30-min dark–90-min light cycles to increase the population of the desired mutants, then diluted and plated on MN− medium and screened under the same light-dark regimen. After 7 days, 100 fast-growing candidates were chosen, individually inoculated in liquid medium, and analyzed for nitrogenase activity under different conditions.
Three independent mutants (UR592, 594, and 595) were chosen for direct examination of their ADP-ribosylation based on the criteria that they displayed relatively high nitrogenase activity under nitrogen-fixing conditions and little or no decrease in that activity in response to darkness or ammonium (Table (Table2).2). In vitro DRAT activity assays (14) were performed on extracts of the fast-growing mutants and controls, with the mutants showing 93 to 104% of the DRAT activity seen in the parent strain, UR484 (data not shown). The nifH regions from all the mutants were each PCR amplified and sequenced, and the results showed that all three mutants possessed the identical mutation in nifH, resulting in an E112K substitution. This mutation was moved into an otherwise wild-type background and caused the same phenotype, indicating that this mutation was indeed causative of the phenotype of good growth in the presence of elevated levels of DRAT. The mutation was given the allele number nifH1073, and the protein product is referred below as NifH-E112K.
Based on the structure of dinitrogenase reductase, Azotobacter vinelandii (21) E112 lies on the same face of dinitrogenase reductase as the R101 residue that is ADP-ribosylated by DRAT, so it is not surprising that an alteration at this site might perturb DRAT interaction with dinitrogenase reductase.
To explore the nature of interaction between DRAT and dinitrogenase reductase, mutants with altered DRAT that would be capable of regulating the activity of NifH-E112K were sought. A library of plasmids carrying a mutagenized draT (with random PCR mutagenesis) was introduced into strain UR662 (NifH-E112K and no DRAG), creating strain UR663. Mutagenesis of the BamHI-EcoRI fragment of draT was performed by a PCR procedure (30), and the resulting strains were screened for poor growth on nitrogen-free medium. Approximately 5,000 colonies were screened, and 100 slowly growing colonies were picked and retested in liquid medium. The nitrogenase activities in UR663 and in two of the slowly growing mutants are shown in Table Table3.3. UR663 has a fairly high nitrogenase activity under nitrogen-fixing conditions and no loss of activity after a shift to the dark. The mutants (UR666 and 668) showed rather lower nitrogenase activity initially, which is presumably the basis for the poor growth, and a further loss of activity upon a shift to the dark, consistent with the ability of the altered DRAT in these strains to modify NifH-E112K.
The draT regions from the plasmids in these mutants were sequenced and, where necessary, reconstructed. Plasmids pKT114 and pKT115 (strain UR666) had identical causative mutations, creating a K103E substitution, while plasmid pKT143 (UR668) had a causative mutation creating a Q81R substitution.
Concurrent with these experiments, a related project involved the screening of strains with PCR-mutagenized draT-containing plasmids for any conferring constitutive DRAT activity. This procedure had identified two substitutions, DRAT-K103E and DRAT-N248D, that caused significant ADP-ribosylation of wild-type dinitrogenase reductase under nitrogen-fixing conditions. A mutant with DRAT-K103E has just been described in the preceding section as having the capability of regulating NifH-E112K, so the isolation of the DRAT-K103E substitution from this very different screen was striking, and we wondered if the two properties of “constitutive DRAT activity” with the wild-type dinitrogenase reductase and ability to regulate the NifH-E112K were actually related. DRAT-N248D was therefore examined for altered substrate recognition, and DRAT-Q81R was tested for constitutive DRAT activity.
The data in Table Table33 show that DRAT-N248D was generally similar to DRAT-K103E and DRAT-Q81R in its ability to regulate NifH-E112K, indicating that the N248D substitution also confers altered substrate recognition on DRAT. The results from the activity analysis of DRAT-Q81R are shown in Fig. Fig.1.1. Based on both detectable ADP-ribosylation (Fig. (Fig.1A)1A) and activity (Fig. (Fig.1B),1B), DRAT-Q81R possesses constitutive DRAT activity, albeit at a lower level than that seen with DRAT-K103E. Because independent mutants, isolated for different phenotypes, show both altered substrate recognition and altered regulation, it seems likely that there is a common biochemical interaction shared by these processes, as rationalized at the end of this report.
The correlation between altered substrate recognition and altered regulation was interesting, but it did not resolve the paradox that we were detecting regulation of the activity of NifH-E112K by the constitutive DRAT variants, but without the detection of an ADP-ribosylated dinitrogenase reductase on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). To test the possibility that the constitutively active DRAT was actually modifying some other protein in the nitrogenase system (an electron carrier to dinitrogenase reductase, for example), the constitutively active DRAT variants were examined in vivo with NifH-R101Y; this substitution eliminates the site of ADP-ribosylation on dinitrogenase reductase and therefore also eliminates DRAT-DRAG regulation but allows low nitrogenase activity (18). The constitutive DRAT variants had no effect on the nitrogenase activity in this background (data not shown), showing that their ability to regulate NifH-E112K involved a specific recognition of this form of dinitrogenase reductase.
The results shown in Fig. Fig.22 resolve the paradox by demonstrating that different DRAT variants ADP-ribosylate NifH-E112K but that the ADP-ribosylated form of NifH-E112K fails to show a detectable shift in the SDS dimension. Figure Figure2A2A shows the behavior of the wild type, with a loss of nitrogenase activity and the appearance of an ADP-ribosylated spot (the “new” spot higher and to the right) upon a shift of the culture to darkness. Strain UR666 (NifH-E112K and excess DRAT-K103E) shows several differences (Fig. (Fig.2B)2B) from the wild type. Dinitrogenase reductase runs as two spots (ADP-ribosylated and nonribosylated) that are both shifted two charge positions to the left (basic side) because of the E112K substitution. There is ADP-ribosylated dinitrogenase reductase under all conditions due to the high level of DRAT and the absence of DRAG, but there is a significant loss of nitrogenase activity upon a shift to darkness. The shift in the proportions of unmodified and ADP-ribosylated spots is difficult to detect, as the dinitrogenase reductase is already substantially modified, as indicated by the nitrogenase activity assay. To make a more compelling case that DRAT-K103E actually becomes active after a shift to darkness, we created a new strain, UR673, with NifH-E112K, normal levels of DRAT-K103E, and no active DRAG. This strain shows a very clear increase in the proportion of dinitrogenase reductase after a shift to darkness and a substantial drop in detected nitrogenase activity in vivo (Fig. (Fig.2C).2C). In contrast to the strain in Fig. Fig.2B,2B, this strain shows little modified subunit in the light, consistent with low DRAT activity under these conditions. Taken together, the results in Fig. Fig.2B2B and C suggest that DRAT-K103E has a very low activity under nitrogen-fixing conditions when NifH-E112K is the substrate but is activated after a shift to the dark. This is in contrast to the behavior of DRAT-K103E with the wild-type dinitrogenase reductase as a substrate (Fig. (Fig.1),1), where it has a substantial level of constitutive activity.
Strain UR663 (NifH-E112K, elevated levels of wild-type DRAT, but no DRAG) displays a significant amount of ADP-ribosylation under both growth conditions (Fig. (Fig.2D)2D) but no apparent regulation in response to darkness. In designing our selections, we had assumed, based on the absence of an upper band on SDS, that NifH-E112K could not be modified by wild-type DRAT, but this is clearly incorrect. However, most of the dinitrogenase reductase is active in this strain, providing high nitrogenase activity, and that was the basis for the selective conditions that we used. A comparison of the nitrogenase activities in Fig. Fig.2B2B and D shows that DRAT-K103E, but not wild-type DRAT, ADP-ribosylates NifH-E112K after a shift to darkness. When NifH-E112K was examined with normal levels of wild-type DRAT, little ADP-ribosylation and no activity regulation was seen (compare Fig. Fig.2E2E to to22C).
As a final demonstration that the identification of the left and right spots in Fig. Fig.22 as unmodified and ADP-ribosylated dinitrogenase reductase was correct, strain UR671 (NifH-E112K, but without DRAT or DRAG) was analyzed (Fig. (Fig.2F).2F). Neither regulation of nitrogenase activity nor the right-hand (acidic) spot was detected.
These results suggest that while wild-type DRAT can modify NifH-E112K, it does so poorly under all conditions and DRAT activity is not stimulated upon a shift to darkness as it is in the presence of the wild-type dinitrogenase reductase. It is probably not correct to suggest that wild-type DRAT is “constitutively active” with NifH-E112K, as we see a rather similar low-level modification of wild-type dinitrogenase reductase under similar conditions when there is no active DRAG. As noted previously, there is always a low level of ADP-ribosylated dinitrogenase reductase detectable in strains that lack any active DRAG, as the posttranslational regulation of DRAT activity is not complete (12).
The results of the present work demonstrate that both protein partners in this complex are important for proper regulation and that the “inappropriate” complexes alter the regulation of DRAT activity. A summary of the regulation of the various combinations of DRAT and dinitrogenase reductase is shown in Table Table4.4.
Two results are particularly important for the model of a DRAT-dinitrogenase complex as the target for regulation. First, the nature of the substrate protein can clearly alter the regulation of DRAT activity. This is clear from the constitutive activity of DRAT-K103E in the presence of wild-type dinitrogenase reductase and from the failure of wild-type DRAT to become active in the presence of NifH-E112K. Second, the fact that DRAT variants capable of regulating NifH-E112K (DRAT-K103E, -N248D, and -Q81R) are uniformly perturbed in the regulation of their activity in the presence of wild-type dinitrogenase reductase demonstrates a striking correlation between substrate recognition and DRAT regulation. These results are reminiscent of the observation that the wild-type DRAT of R. rubrum has different requirements for optimal in vitro activity when a dinitrogenase reductase from another organism is used as a substrate, although the physiological significance of that observation had been unclear (13).
Taken together, these results strongly suggest that the posttranslational regulation of DRAT activity is a function of DRAT and its substrate. Cross-linking analysis has demonstrated that DRAT forms a complex with dinitrogenase reductase, even in the absence of ADP-ribosylation (7). This result has been interpreted to mean that the DRAT-dinitrogenase reductase complex might be the predominant form of DRAT in the cell and represents the form that is actually subject to posttranslational regulation (7). It is our working hypothesis that only this complex is competent to receive some presently unknown signal that modulates DRAT activity. Such a hypothesis has been presented previously (6, 29), based on the observation that overexpressed DRAT decreased nitrogenase activity in vivo without ADP-ribosylation, which suggested that a complex between the two proteins existed even under conditions where modification was not occurring. We are continuing to test this hypothesis through the biochemical analysis of the DRAT variants described in this report.
This work was supported by the College of Agricultural and Life Sciences at the University of Wisconsin—Madison and by USDA grant 96-35305-3696 and Hatch project 4024 to G.P.R.
We thank Paul Ludden and members of his laboratory for suggestions and assistance.