In this study we have demonstrated, through a series of genetics-based experiments, that the S. aureus sbnA and sbnB genes are necessary for staphyloferrin B production. We have also shown that S. aureus mutations in sbnA and sbnB are fully complementable in trans by both wild-type copies of each gene as well as through feeding of the molecule L-Dap itself, leading to the renewed production of the staphyloferrin B molecule. The data support the contention that the enzymes SbnA and SbnB function synergistically as a L-Dap synthase, catalyzing the first committed biosynthetic step towards staphyloferrin B synthesis in S. aureus. Overall, this is the first study that simultaneously investigates the roles of both genes encoding a cohesive L-Dap synthase.
The L-Dap molecule is a very unusual and rare amino acid. It is non-proteinogenic but it is often found structurally associated with secondary metabolites such as antibiotics (Table 4). To our knowledge, staphyloferrin B represents the only characterized siderophore that contains L-Dap as part of its structure (Figure 1A). The experiment shown in Figure 2A also reinforces the fact that only L-Dap, and not D-Dap, is incorporated into staphyloferrin B. This is in agreement with initial structural elucidation studies , the high resolution crystal structure of the siderophore , as well as enzymatic recognition of L-Dap as a substrate by staphyloferrin B NIS synthetases . The only siderophore with a component similar to L-Dap in its structure is achromobactin from Pseudomonas syringae , which has an overall structure and biosynthetic pathway that is very similar to that of staphyloferrin B. In place of L-Dap, achromobactin contains L-2,4-diaminobutyric acid which is condensed onto a unit of citrate and α-KG at both amino groups. L-2,4-diaminobutyric acid may be synthesized by a putative aminotransferase (AcsF) that is also encoded within the achromobactin biosynthetic gene cluster. In the case of achromobactin, synthesis of this diamino acid substrate requires only one enzyme as opposed to the two enzymes required for synthesis of L-Dap. Biochemical characterization of AcsF, along with its substrate specificity, awaits further investigation. Why some siderophore biosynthetic systems have evolved to select one diamino acid over another is an intriguing biological question.
Based on bioinformatics and the emerging diversity of members of the OCD enzyme family, the S. aureus SbnB enzyme likely does not contribute to proline production, and hence would not recognize L-ornithine as a substrate. In agreement with this hypothesis, under the experimental conditions of Li et al.  in testing S. aureus for proline synthesis, they could not attribute the production of L-proline to SbnB, but rather L-proline production was solely dependent on the pyrroline-5-carboxylate reductase pathway; it should be noted, however, that these experiments did not test whether sbnB was responsible for proline synthesis under low-iron growth conditions and thus do not completely eliminate the possibility that SbnB has OCD activity. The exact biochemical reactions catalyzed by SbnA and SbnB (and homologs) await detailed investigation.
SbnA and SbnB are likely functioning together as an L-Dap synthase and perhaps the mechanism is that originally proposed by Thomas and colleagues  for VioB and VioK with regards to viomycin biosynthesis in Streptomyces (Figure 3, scheme A). In this scheme for L-Dap synthesis, VioK (or SbnB) acts as an L-ornithine cyclodeaminase (based on sequence similarity to an OCD [1X7D]) that will convert L-Orn to L-Pro with the concomitant release of ammonia. The released ammonia is picked up by VioB (or SbnA) to be used as a nucleophile for the β-replacement reaction on (O-acetyl-) L-serine, thus generating L-Dap. The reaction catalyzed by VioB (or SbnA) is modeled after homologous cysteine synthases which use a sulfide group for β-replacement reactions to generate cysteine . Therefore, the action of VioB, or SbnA, would appear to be an amidotransferase in this reaction scheme.
However, more recent bioinformatic and phylogenetic analyses of these enzymes suggest that the mechanism of L-Dap synthesis may be quite different from that just described. This is especially true for SbnB, which is more closely related to NAD+-dependent amino acid dehydrogenases rather than characterized ornithine cyclodeaminases. Therefore, this prompted us to propose several new mechanisms of L-Dap synthesis (Figure 3, Schemes B-D), emphasizing the role of SbnB as an amino acid dehydrogenase, while SbnA would continue to serve the function of a β-replacement enzyme or aminotransferase. As illustrated in Figure 3, scheme B, SbnB acts as an NAD+-dependent L-Glu dehydrogenase that converts L-Glu to 2-oxoglutarate (or α-KG). This reaction will release an ammonia molecule to be used by SbnA in an identical manner to the second half of the reaction proposed in scheme A. The reaction depicted in scheme B is attractive since all products of this mechanism can be funneled towards staphyloferrin B biosynthesis (i.e. α-KG is a substrate for SbnC, while L-Dap is a substrate for SbnE and SbnF), as opposed to scheme A where the generation of L-Pro serves no purpose in staphyloferrin B biosynthesis. In scheme C, SbnA would act as the first enzyme in the pathway by condensing L-Ser with L-Glu to form a larger intermediate consisting of an L-Ser-L-Glu conjugate. In effect, SbnA would perform a β-replacement reaction on L-Ser by displacing the hydroxyl group on L-Ser with L-Glu. Dehydrogenase activity provided by SbnB would resolve and split the intermediate compound to give rise to L-Dap and 2-oxoglutarate. As in scheme B, all products from this reaction are used in the biosynthesis of staphyloferrin B. In scheme D, SbnB would serve as a 2-Ser dehydrogenase, converting L-Ser to 2-amino-3-oxopropanoic acid, an intermediate that would be primed for nucleophilic attack at the β-carbon by an ammonia molecule derived from the aminotransferase activity of SbnA. The source of this ammonia might be L-Glu, L-Asp, or L-Ala, which are all common amino group donors. In this scheme, if L-Glu is used as the amino donor, 2-oxoglutarate is produced and would be a substrate for the SbnC synthetase.
Unlike the substrate uncertainty exhibited by SbnB, the substrate for SbnA homologs have been defined through precursor labeling studies [37, 38]. Since, an SbnA homologue is involved in L-Dap production for viomycin (Table 4), then it is very likely that L-serine (or the O-acetylated derivative) is also the substrate for SbnA. Moreover, a recent study by Zhao et al.  characterized the gene zwa5A, an SbnA homologue (Table 4), and through genetic knockout of this gene in Bacillus thuringiensis, confirmed that it is involved in synthesizing L-Dap for the antibiotic zwittermicin A. Similar to our experiments, these researchers were able to restore the production of zwittermicin A in the zwa5A mutant by providing exogenous L-Dap to the culture media .
It is important to note that β-replacement reactions involving ammonia as the nucleophile are rare. Only recently was an L-2,3-diaminobutyric acid (L-Dab) synthase studied that is involved in mureidomycin A production . This enzyme, which catalyzes a similar reaction to the ones proposed for SbnA (Figure 3), will use L-Thr as the substrate (instead of L-Ser) and will displace the β-hydroxyl group with an ammonia molecule to form L-Dab. However, the source of the ammonia was not described and thus it is assumed that this enzyme may depend on cellular concentrations of free ammonia rather then receiving the ammonia from a dedicated dehydrogenase. The idea that an enzyme acquires free ammonia within a cell is intriguing. Certainly, the rate of diffusion of ammonia inside a cell can be a limiting factor and this is perhaps why both halves of an L-Dap synthase appear to be consistently co-expressed, and potentially are intimately associated with one another such that liberated ammonia by the dehydrogenase unit can be properly channeled to the aminotransferase unit. This would ensure catalytic efficiency and also assumes that extensive protein-protein interactions would occur between the two enzymes. Certainly, this idea is supported by the existence of single-polypeptide encoding genes found within the P. syringae and Acidobacterium capsulatum genome, in which half of the polypeptide shares significant similarity with SbnA and the other half shares significant similarity with SbnB (Table 4).
It is interesting that supplementation of the S. aureus culture medium with L-Dap enhanced staphyloferrin B output in wildtype cells (Figure 2B cf. 2C), a phenomenon that has previously been observed . It is tempting to speculate that L-Dap may be a critical molecule in terms of regulating staphyloferrin B production or that the presence of L-Dap is a signal for the organism to commit to staphyloferrin B synthesis. At this time, how such a signal may be transduced is unknown but perhaps L-Dap can induce allosteric effects on certain Sbn proteins that may, in turn, regulate staphyloferrin B production. Interestingly, enhancement of end product formation by L-Dap feeding has also been observed for zwittermicin A production in B. thuringiensis .
The biochemical schemes for L-Dap synthesis, as depicted in Figure 3, await experimentation with purified enzymes as well as screening with potential substrates, and these experiments are under investigation in our laboratory. Certainly, the actual mechanism of L-Dap synthesis may not be restricted to those mechanisms outlined here, but at least these provide a starting point towards the biochemical investigation of L-Dap synthase enzymes in different bacteria. No matter the mechanism, it is most surely to be novel. Regardless, the studies here have demonstrated the essentiality of SbnA and SbnB towards L-Dap synthesis in S. aureus, a nonproteinogenic amino acid component of staphyloferrin B that is critical to the iron coordinating function of the siderophore, as well as providing implications for the role that L-Dap may play in regulating production of the molecule.