Ironsulfur Clusters Can Be Reassembled To Replace Dnic

To re-assemble the FeS clusters, recombinant cysteine desulfurase (IscS protein) was prepared from E. coli [56]. IscS is a member of the recently identified FeS cluster assembly machinery [57]. The enzyme catalyzes desulfurization of L-cysteine and provides sulfide for the synthesis of FeS clusters [58]. When the ferredoxin DNIC (Fig. 4, trace a) was incubated with L-cysteine and IscS, the observed UV-visible absorption spectrum was typical for the ferredoxin [2Fe—2S] clusters, with two absorption peaks at 415 and 459 nm (Fig. 4, trace d). The results clearly show that NO-generated ferredoxin DNIC can be repaired by the addition

415 nm

459 nm

350 400 450 500 550 Wavelength (nm)

Fig. 4. Repair of ferredoxin DNIC in vitro. Formation of the repaired [2Fe—2S] centers was monitored by UV-Visible absorption spectroscopy. NO-treated ferredoxin (trace a) was incubated with l-cysteine alone (trace b), IscS alone (trace c), or both together (trace d); trace e is from undamaged [2Fe—2S] ferredoxin. Reproduced from Yang et al. [56].

of L-cysteine and IscS in vitro. Of note, the FeS clusters of the protein are reconstructed without the need for additional iron [56].

A similar process has also been observed for tetranuclear clusters in a protein in which the FeS center appears to have a strictly structural role. Quite unexpectedly, the E. coli DNA repair enzyme endonuclease III was found to contain a [4Fe—4S] center [59], providing the first example of this type of metalloprotein among DNA repair enzymes. Homologous proteins have since been identified in mammalian species, including humans [60]. The [4Fe—4S] center in endonuclease III is resistant to oxidation and reduction [59], and structural studies indicate that the metal center instead helps position parts of the protein for substrate recognition [61]. Despite its lack of redox or apparent catalytic activity, the endonuclease III [4Fe—4S] center is susceptible to nitrosylation by NO [15]. FeS nitrosylation inactivates the enzyme, as it does with many other proteins. However, similarly to ferredoxin, intact [4Fe—4S] centers can be restored to the endonuclease III by incubation in vitro of the DNIC-containing form with ferrous iron, L-cysteine and IscS, resulting in full recovery of the enzymatic activity [15]. As is the case for SoxR, the endonuclease III DNICs are rapidly removed in vivo and replaced by functional FeS centers, without the need for new protein synthesis [15]. Thus, mechanisms to reconstruct FeS clusters from protein DNIC seem to be general.


The studies on the appearance and processing of DNICs upon NO exposure of SoxR and ferredoxin demonstrate that this modification has the key properties necessary to act in a specific stress-signaling role. First, the modification (DNIC) is formed directly by the activating agent (NO). Second, in SoxR the formation of DNIC is activating rather than inhibiting. Third, the modification is rapidly reversible, so that the activity of SoxR and the downstream genetic response can be finely modulated.

A new example of nitrosylation of an iron center in gene control was recently described. The NorR protein governs a defensive response in enterobacteria that induces the synthesis of flavorubredoxin and an associated flavoprotein that constitute an NO reductase activity, which converts toxic NO to the less toxic N2O [62]. NorR activation appears to be rather specific for NO, in contrast to SoxR (see below). A recent biochemical analysis demonstrated that NorR contains a mononuclear iron center, and that exposure of the protein to NO generates a nitrosylated form [63]. Most notably, the nitrosylation is reversible, which is a key feature of such a regulatory system, as noted earlier. It will be of interest to determine how the nitrosylation is reversed (passively or actively), and whether a similar mechanism is employed in other regulatory scenarios.


The observed activation of SoxR by diverse signals (oxidation, nitrosylation, iron removal) has prompted efforts to understand the structural basis for signal transduction by this sensor. Although a direct three-dimensional structure for SoxR has not been reported, related protein structures are available to aid with the interpretation of mutational and other analyses [64].

The mechanism by which activated SoxR stimulates transcription of the soxS gene is shared by the various members of the eponymous MerR regulatory family. The soxS promoter has -10 and -35 elements that are close to consensus sequences, but these elements are separated by a sub-optimal 19-bp spacing [65]. The majority of E. coli promoters dependent on a70 RNA polymerase have 17 ± 1 bp spacing. In the soxS promoter, shortening the —10/-35 spacing to 16-19bp gives strong transcription in the absence of SoxR [65], consistent with the spacing as the critical regulatory feature of this promoter. The MerR target promoter has similar properties, and an unusual characteristic is that the inactive and active forms of both MerR [66] and SoxR [48,50] bind their target promoters with approximately equal affinity. SoxR and MerR thus act allosterically, via structural transitions in the protein-DNA complex.

Topological experiments with MerR suggested that transcriptional induction is accomplished by localized distortions mediated by the protein [67]. Data consistent with such a mechanism for SoxR emerged from footprinting studies of the changes exerted by SoxR on the soxS promoter [51].

The foregoing focuses on changes at the DNA level, but these changes must be driven by structural transitions in the protein in response to signals such as [2Fe—2S] oxidation. It was noted earlier that deletions or insertions affecting the SoxR C-terminus led to a constitutively active protein [68]. Analysis of single amino acid substitutions in constitutively active SoxR proteins also showed a preference for alterations of the SoxR C-terminus [69]. This diversity of activating changes suggested that the C-terminus might act to restrain SoxR transcriptional activity.

More recent mutational analysis revealed a new class of single amino acid changes that disrupt the SoxR signal transduction circuit without preventing DNA binding [42]. Biochemical analysis of the mutant proteins against the backdrop of structural information from a few related proteins [70-72] provided some important insights [73]. This analysis suggested important roles for the center of the subunit interface, formed by helical coils in each of the subunits, and the likely intimate proximity of [2Fe—2S] domain, anchored near the C-terminus, with the DNA-binding domain formed by the N-terminus.

How various signals might result in analogous structural transitions in the SoxR-DNA complex? An additional important clue came from a seemingly discordant result: it was reported that the chelator 1,10-phenanthroline activates SoxR in vivo, even under anaerobic conditions [74]. Removal of the SoxR [2Fe—2S] seemed a likely effect of the chelator, and we later confirmed the formation of apo-SoxR in cells treated with an iron chelator that moderately activates SoxR (Ding and Demple, unpublished data). In contrast with these results, various studies showed an absolute requirement of the SoxR [2Fe—2S] centers for transcription in vitro [50,51,75,76]. However, SoxR proteins with cysteine-to-alanine substitutions that prevent formation of the [2Fe—2S] centers have modest transcriptional activity in vivo that gives up to 10-fold stimulation of soxS transcription [76]. Transcriptional stimulation by apo-SoxR in vitro was finally achieved when we found that it is strongly dependent on negative supercoiling of the DNA template (E. Hidalgo and Demple, unpublished data).

There are thus at least three active forms of SoxR: the protein with oxidized [2Fe—2S] centers; SoxR containing DNICs; and the apo-protein, dependent on supercoiling. Both apo-SoxR and DNIC-containing SoxR are expected to have disrupted metal-binding domains compared to SoxR with reduced [2Fe—2S] centers. This view suggests that [2Fe—2S] oxidation might also disrupt some aspect of the C-terminus structure. As suggested schematically here (Fig. 5), C-terminal disruption in all three cases would release the SoxR homodimer to take up the active conformation, with the resulting transcriptional activity. Clearly, detailed structural studies will be needed to test this hypothesis.

Was this article helpful?

0 0

Post a comment