In this work we have studied the function of the E. coli NAMNAT enzyme. Understanding the enzyme is necessary not only for its central role in metabolism but also for its possible role as a target for the development of new antibiotics. In an earlier study we initiated functional studies of two E. coli NAMNAT mutants [15]. One of the mutants, nadD72, is a frameshift mutation that changes the ten last amino acid residues of the protein and adds seventeen amino acids to the C-terminus [15]. When the amino acid sequences of bacterial NAMNATs are analyzed, two conserved residues, Tyr-205 and Ile-206 were found in the C-terminus. They are both located in the F helix of the E. coli enzyme, using the nomenclature given by Zhang et al. [10]. These two residues are changed in the NadD72 mutant. This lead us to ask whether it is the elongated C-terminus and/or the changed amino acids that cause the phenotype associated with the mutation. The other mutation, nadD74, changes Asp-13 to Val. The residue is next to the ATP-binding motif and is highly conserved [13].
To extend the study of NAMNAT we made three additional mutant alleles in the C-terminus and seven around the active site. For the in vivo studies we made a strain with a chromosomal deletion of the nadD gene, which makes it possible to study cloned, mutant enzymes in vivo. The gene on the chromosome was replaced by a cassette encoding chloramphenicol acetyltransferase [16] and the strain was called MS10. The gene nadD has been shown essential in Salmonella typhimurium [17]. However, some conflicting data on this matter was recently published [8]. The authors have isolated strains with transposon insertions affecting the nadD gene and the strains are viable on LB. The exact location of the insertion was not shown, moreover, attempts to delete the nadD gene from the chromosome failed. Our results are clear; we could only delete the nadD gene when the wild type nadD gene was present on a plasmid in the cell, thus, confirming the essentiality of the gene.
To study the mutant enzymes in vitro the different alleles were cloned in a vector under the control of the arabinose promoter. We first tried to fuse a His6-tag to the N-terminus of the proteins, but the expression level of the proteins was very low and some mutant proteins were undetectable when analyzed by Western blot (not shown). We changed to an IgG binding ZZ'-domain as a tag and the expression level increased considerably. A thrombin recognition sequence was engineered in the linker between the ZZ'-tag and the enzyme. However, removal of the ZZ'-tag from the purified proteins was not possible. Therefore, all investigated mutants and the wild type enzyme were assayed with the N-terminal ZZ-tag.
All enzymes were expressed as soluble proteins in strain TOP10. The final yield was typically 5–20 mg protein/l culture, similar to an earlier report [9]. The purity was about 90–95% as estimated by SDS-PAGE after Coomasie staining. The purified NadD72short protein gave two bands on the gel. One band had the correct enzyme size, the other, smaller, is probably a degradation product.
Investigation of the C-terminus
Based on the nadD72 mutation, we designed the mutant NadD72short with the same change of the ten C-terminal amino acid residues as nadD72, while the length of the protein is the same as in the wild type enzyme. We also constructed two mutants where either Tyr-205 or Ile-206 is changed to alanine.
We have earlier shown that the intracellular level of NAD+ correlates to growth ability on different media and at different temperatures [15]. Thus, we decided to investigate the effect on growth by the different mutant enzymes. Plasmids with either of the four mutant alleles, pZZNadD72, pZZNadD72short, pZZNadDY205A, pZZNadDI206A, or the wild type gene (pZZNadD) were transformed into strain MS10 with plasmid pKanNadD selecting for ApR. Transformants were restreaked on LB with ampicillin and arabinose to induce expression of the respective nadD alleles. The transformants were tested for loss of KmR to ensure plasmid exchange. Thereafter, MS10 with each respective plasmid were streaked on LB plates with or without arabinose and incubated at 30°C. The diameter of the colonies was measured. The results are shown in Table 4. Addition of 0.1 mM arabinose resulted in growth of all strains. In the absence of arabinose leakage expression of the wild type enzyme is enough to support normal growth while neither of the two mutant proteins NadD72 or NadD72short are active enough to do so. This indicates that it is the change in the last 10 amino acids and not the elongated C-terminus that impairs the enzyme. The two mutants with either of the two conserved residues changed, supported growth like the wild type enzyme. This shows that to affect enzyme activity more than one amino acid has to be changed in the C-terminus.
The same test was performed on minimal medium and it was found that NadD72 and NadD72short could not support growth in the presence of 0.1 mM arabinose. However, at 0.2 mM arabinose enough enzyme was produced to allow growth (not shown).
A test was also performed on strain MS10 containing plasmids carrying the nadD alleles without a ZZ'-tag. We could not detect any difference in growth behavior whether the nadD alleles were tagged or not (not shown). This makes us confident to use the ZZ'-tagged enzymes in vitro.
The enzyme activity of the mutants changed in the C-terminus were measured in vitro. The result can be seen in Figure 2. The activity of the wild type enzyme was set to 1. The enzymes NadD72 and NadD72short have almost no activity in good agreement with the in vivo phenotype. The other two mutants, NadDY205A and NadDI206A are less efficient than the wild type enzyme but not enough to show as a change in the growth phenotype.
The results obtained both in vivo and in vitro for NadD72short could indicate that it is not the extension of NadD72 that causes the deficiency in the protein but rather the change in the C-terminus. However, this conclusion is complicated by the finding during purification of NadD72short that two bands were visible on the protein gel, indicating instability of the protein. Thus, the results obtained for NadD72short are inconclusive.
Another possibility should also be considered. Since the nadD72 allele causes temperature sensitivity it is possible that the enzyme activity is lowered at 37°C the temperature at which the in vitro experiments were performed. To test this, we grew all mutants at 30°C, 37°C, and 42°C, respectively. We found that the Ts phenotype not only disappears in the presence of arabinose but that the cells grow better at the higher temperatures than at 30°C. Therefore, we do not think that the changes in enzyme activity are caused by changes in reaction temperature optimum.
To understand the role of the elongated C-terminus of the NadD72 mutant, we consider the role of the corresponding region of the human NMNAT. Human NMNAT has a 24 amino acid residues longer C-terminus than E. coli NAMNAT. It has been suggested that the C-terminus in human NMNAT plays a role in substrate recognition [18, 19]. The NadD72 enzyme has 17 extra amino acid residues and it is possible that the extension interferes with substrate binding, which would lead to low enzyme activity. All we can say with certainty is that the C-terminus is important for stability of the protein and that the two conserved amino acid residues do not have a great influence on activity.
Investigation of the active site
The nadD74 mutation leads to an amino acid change in position 13 (Asp to Val). The mutated residue is two amino acids away from the ATP-binding motif, T/HXGH (position 16 to 19). Crystal structure information was used to decide which amino acid positions to mutagenize to learn more about the active site. Residues within 6 Å distance from the oxygens of the two adjacent phosphate-groups of the bound NAAD molecule are shown in Figure 3. Based on their close contact and H-bonding abilities Thr-11, His-19, Asn-40, His-45, Arg-46, Asp-109 and Ser-110 were selected for mutagenesis. All these residues were changed to alanine by site-directed mutagenesis. The recombinant proteins were cloned and analyzed, as were the C-terminal mutants.
First, growth on LB plates with and without arabinose was tested. The results are shown in Table 4. All strains grew in the presence of arabinose as expected. In the absence of arabinose MS10/pZZNadDN40A and MS10/pZZNadDS110A behaved basically as MS10 with the wild type enzyme while the other mutations affected growth on LB to a varying degree. Second, the enzymatic activity for the active site mutants was determined in vitro. The results are summarized in Figure 2. As with the C-terminal mutants, the correlation between the two experiments is very good.
Asp-109 and Ser-110 is located in the region connecting strand d and helix D of NAMNAT [10]. This region is one of three regions observed to undergo large conformational changes upon substrate binding. Interestingly, mutations of these two amino acids affect enzymatic activity very differently. On the one hand, binding of the substrate brings Ser-110 closer to the substrate. It is possible that the side chain of Ser-110 is H-bonded to the 2'-OH of AMP-ribose. However, mutation of Ser-110 to alanine resulted in an enzyme with 80% activity as compared to that of the wild type NadD. Our results indicate that the interaction between Ser-110 and the ribose is dispensable for substrate coordination. On the other hand, the change of the highly conserved Asp-109 had a severe effect on activity. Asp-109 has been proposed to form an H-bond to the 2'-OH group of AMP [10]. It was also suggested by the same authors that the carboxylate oxygen of the residue might be involved in the coordination of a Mg2+ ion shown to be important for the enzyme function [10, 12]. The location of Asp-109 is ideal to position an Mg2+ ion which could act as a Lewis acid stabilizing the transition state in the transesterification reaction [10]. However, the functional importance of this conserved residue has not been investigated earlier. The serious decrease in enzymatic activity (0.038%) that we observe supports the above mechanism and serves as an experimental evidence for the involvement of Asp-109 in the catalysis.
The Thr-11 main chain nitrogen is H-bonded to the AMP phosphate in the crystal structure of the NadD-NAAD complex. This H-bond is expected to be independent of the side-chain character of the amino acid. However, the NadDT11A mutant leads to decreased enzyme activity (2%, Figure 2). The explanation might be that the hydroxyl group of Thr-11 forms an H-bridge to Asn-40, which has been shown to interact with 2'-hydroxyl group of NAMN-ribose [10]. Such an interaction could contribute to substrate binding and coordination. The change of Asn-40 to alanine leads to a decreased enzyme activity (23%, Figure 2). We conclude that the disruption of this H-bonding network might lead to inefficient substrate coordination. Therefore, the effect on enzyme activity by the T11A and/or N40A mutation is possibly indirect.
The mutation D13V (NadD74) was found to lead to decreased activity of the enzyme. A similar mutation has been described in CTP:glycerol-3-phosphate cytidyltranferase that is a member of the same enzyme super family [20]. The residue Asp-11 (which corresponds to Asp-13 in NAMNAT) was changed to alanine and enzyme activity was severely reduced. Both valine and alanine are hydrophobic amino acids and it is possible that the change disturbs ATP binding; resulting in lower enzyme activity. It is not clear whether the role of Asp in this position is catalytical or structural.
The role of the second histidine (His-19) in the conserved T/HXGH motif has been tested in several studies and been shown to play a role in ATP-binding and stabilization, but the role for the two histidine residues can vary between enzymes [14, 21]. Our results confirm the previous observations. The mutation H19A leads to a decrease in enzyme activity to 0.62% of the wild type activity.
Amino acid residues His-45 and Arg-46 are part of a flexible loop in the NadD enzyme, which moves upon substrate binding. The two residues are conserved within NMNAT from Bacteria and Eukarya but not in the known archaeal enzymes. His-45 is involved in a hydrophobic stacking interaction with the pyridine ring in NAMN and is also likely to form an H-bond with the NAMN phosphate group. Histidines are often involved in acid base catalysis, and prone to activate nucleophiles by abstracting a proton. The H45A change led to an enzyme with 4% residual activity, while the R46A mutant was the most affected of all tested active site mutants. If the role of Arg-46 in the substrate binding loop was simply to protect the bound substrates, higher rates of ATP hydrolysis is expected in the case of the R46A mutant. Since the R46A mutant did not produce AMP as by-product in the in vitro experiment, the role of this arginine side chain must be more than simply protecting the bound substrates from water molecules. The guanidinium group of Arg-46 lies in an ideal position to serve as a positively charged moiety that stabilizes the juxtaposed and negatively charged phosphates of the substrate molecules, as well as of the product NAAD. The archaeal orthologues lack the precise sequence homology in the sequence aligned to the H45-R46, but they have a conserved arginine (Arg-8 in Methanococcus jannaschii) that occupies the same position with its side chain in the ATP-enzyme complex as that of Arg-46 in the E. coli NadD enzyme [10, 12]. On the other hand another hypothesis should be considered as well. Mutational studies on NMNAT from Methanobacterium thermoautotrophicum indicate that the archaeal enzyme is involved in the reaction merely by placing the substrates in an ideal position [14]. Combining the facts above, we conclude that Arg-46 in E. coli NadD plays an important role in stabilizing the two adjacent, negatively charged, phosphate moieties during catalysis. A similar role might be attributed to Arg-8 in the archaeal counterparts but this has to be tested.