- Sulfur is one of the six elements required during the early stages of the evolution of life, and enzymes involved in sulfur transfer and oxidation are increasingly being recognized as potential drug targets for antimicrobials as well as for therapies for cancer, neurodegenerative and inflammatory diseases. Bacteria are able to carry out a much broader range of sulfur chemistry than mammals and are able to use sulfate as a sole sulfur source, meaning that they can synthesize cysteine from sulfate, whereas mammals must take in either homocysteine or cysteine in their diet. Some bacteria, such as Escherichia coli, are able to obtain sulfate from alkanesulfonates through the expression a sulfur starvation utilization system, which includes a two-component system composed of an NADPH-dependent FMN-reductase, SsuE, and a monooxygenase, SsuD. Once cysteine (Cys) is available to a cell, the careful regulation of its concentration is important because even though Cys is required for life, at high levels it is toxic, especially for mammals. In bacteria, metazoa and fungi the Cys levels are primarily regulated by cysteine dioxygenase (CDO). In mammals, CDO is known to form a Cys-Tyr crosslink that greatly increases its enzymatic activity.
Here, I report structural studies and evolutionary considerations aimed at elucidating the mechanisms of SsuE and CDO. With regard to the work on SsuE, I report the first structures of any SsuE, and these crystal structures show that SsuE is structurally similar to closely related FMN-reductases. These structures revealed that SsuE forms a tetramer that is similar to related FMN-reductases, and an active site that is completed by the dimer interface. An evolutionarily conserved π-helix appears to link FMN binding with tetramer dissociation. Three different states of SsuE were captured at ~ 2.0 Å resolution: an apo enzyme, an FMN-bound enzyme and a FMNH2-bound enzyme. Based on these results, a reinterpretation of previous kinetics data on SsuE led to a novel proposal for the SsuE mechanism that is similar to those of its homologs. Furthermore, a general catalytic cycle was defined for NADPH-dependent FMN-reductases from the flavodoxin-like superfamily that provides a framework for understanding how the mechanism of these enzymes might change depending on cellular conditions and interactions with partner proteins.
With regard to the work on CDO, over 30 structures have been solved of wild type and mutant rat CDOs. In one set of studies, 14 crystal structures between 1.25 and 2.15 Å resolution, including a room temperature structure, are used to clearly define the pH dependence of Cys-persulfenate complex formation in the crystal, and fortuitously also provide the first high resolution view of an unreacted Cys bound in the active site. The main conclusions from this work are that persulfenate formation is consistently seen at pH values between 5.5 and 7, that is it not an artifact of freezing or synchrotron radiation, and at pH≥8 the unliganded active site iron shifts from 4- to 5-coordinate. We are able to identify that important active site pKa values lie between 5.0-5.5 and 7.0-8.0.
In a second study, 17 CDO crystal structures in the presence of Cys and inhibitors, ranging from 1.25 to 1.65 Å resolution and mostly of site-directed mutants, are used to define the role of an active site Cys-Tyr crosslink in catalysis as well as to determine the mechanisms of CDO inhibition by homocysteine and azide. Main conclusions are that a chloride ion binds to the active site iron in both the C93A and Y157F variants, even though the active site iron is still in the ferrous form. Upon exposure to Cys, the chloride is displaced but Cys does not bind in the same way as it does to wild-type CDO, with the Cys largely coordinating the iron only through its thiolate, and not through the α-amino group. Cys-persulfenate does not form in the C93A or Y157F active sites, indicating the crosslink is necessary for persulfenate formation in the crystalline enzyme. These results defines a key role for Tyr157 in Cys binding, through both positioning Cys in the active site and modulating the pKa of its α-amino group. The structures in the presence of homocysteine and azide revealed why the CDO-homocysteine complex is unproductive for catalysis, and and how azide binds to the wild-type CDO, associating with the iron and the hydroxyl of Tyr157 in the crosslinked enzyme.
Finally, in a third CDO study, I describe the crystal structures of two bacterial CDO homologs that were originally solved by structural genomics groups but had not been reported in the literature. These are of great interest as they represent Gln-type and Arg-type classes of bacterial CDO homologs defined by the residue aligning with Arg60 of rat CDO. For the Arg-type CDO from Bacillus subtilis, we reproduced the crystals and were able to obtain a Cys-bound complex at 2.30 Å resolution that shows that its mode of Cys binding is comparable to mammalian CDOs. Also, we obtained the original diffraction images and further refined the structure of the Gln-type CDO homolog from Ralstonia eutropha to 1.65 Å resolution, and discovered the active site contained an unexpected iron-bound dioxygen. From this structure, we identified a novel active site Arg that is evolutionarily conserved among Gln-type CDO homologs and is positioned in a way that allows us to conclude that these enzymes cannot bind Cys and thus are not authentic CDOs. This is consistent with the observation that the one Gln-type CDO homolog with characterized substrate specificity has been identified as a 3-mercaptopropionate dioxygenases.