Chlorinated natural products include vancomycin and cryptophycin A. Their biosyntheses involves regioselective chlorination by flavin-dependent halogenases. We report the structural characterization of tryptophan 7-halogenase (PrnA), which regioselectively chlorinates tryptophan. Tryptophan and FAD are separated by a 10Å-long tunnel and bound by distinct enzyme modules. The FAD module is conserved in halogenases and is related to flavin-dependent monooxygenases. Based on biochemical studies, crystal structures and by analogy with monooxygenases, we predict FADH 2 reacts with O 2 making peroxy-flavin which is decomposed by Cl −. The resulting HOCl is guided through the tunnel, to tryptophan, where it is activated to participate in electrophilic aromatic substitution. In addition to man-made chemicals, there are nearly four thousand chlorinated and brominated natural products (1), including drugs such as vancomycin (2), rebeccamycin (3) and cryptophycin A (4). The de-novo chemical synthesis of complex natural products is often too expensive or too difficult to be practical. Their production relies on fermentation and introducing diversity in such molecules requires protein engineering. This has been hampered by a lack of understanding of the molecular basis of the biological regioselective halogenation mechanism. Metal-dependent haloperoxidases were once thought to catalyze all halogenation reactions in biology and fall into two classes. Both, heme-iron-dependent enzymes (5) and vanadium-dependent enzymes (6), have been structurally characterized. Although different in structure, both form a metal-bound hydrogen peroxide, which reacts with halide ions producing a metal-bound hypohalite ion. This ion dissociates from the metal as hypohalous acid (5,6) where, in solution, it reacts with substrate. Such halogenation lacks regioselectivity and substrate specificity (7). Perhydrolases are now recognized not to be involved in halometabolite biosynthesis (8). A new halogenase was reported by Dairi et al. identifying the gene for the chlorinating enzyme in chlorotetracycline biosynthesis (9). The gene product showed no similarity to haloperoxidases. Studies of the antifungal compound pyrrolnitrin from Pseudomonas fluorescens identified two related genes (prnA and prnC), coding for two halogenating enzymes (10) (fig. S1). Both contain a flavin binding site (9-13) and exhibit weak sequence homology to flavin-dependent monooxygenase enzymes (14). PrnA catalyzes the regioselective chlorination of the 7-position of tryptophan (15). Turnover requires that FAD is reduced to FADH 2 (by flavin reductase) and that O 2 is present (11). One sentence: The crystal structure of tryptophan 7-halogenase indicates that the halogenation of important metabolites in bacteria proceeds by the novel use of hypohalous acid to achieve regioselective substitution.

Flavin-dependent halogenases have been shown to play a major role in biological halogenation reactions. For halogenating activity, flavin-dependent halogenases require reduced FAD, which is formed from FAD and NADH by a second enzyme, a flavin reductase. Although in a number of cases, a flavin reductase gene is present in the biosynthetic gene cluster of the halometabolites, it is unclear whether the corresponding flavin reductases interact directly with the halogenases. At least in a number of cases, flavin reductases from different bacterial strains can be used in combination with halogenases.1, 2-5 For the tryptophan 7-halogenase PrnA from Pseudomonas fluorescens BL915 which catalyzes the first step in pyrrolnitrin biosynthesis6 it could be shown that even chemically reduced FAD is used by the halogenase in the halogenation reaction.7 Based on the threedimensional structure of PrnA, it was postulated that flavin hydroperoxide is formed by the reaction of halogenase-bound reduced flavin with oxygen. This flavin hydroperoxide then reacts with chloride ion leading to the formation of hypochlorous acid, which is then guided along a tunnel about 10 Å long towards the substrate tryptophan (Figure 1). A lysine residue (K79) was suggested to hydrogen bond with hypochlorous acid and thus position it to react with tryptophan.8 Yeh et al. demonstrated chloramine formation by the reaction of HOCl with the ε-amino group of lysine,9 suggesting that chloramine rather than HOCl is the active agent. The importance of K79 is undisputed; exchange of K79 against an alanine residue leads to total loss of halogenating activity as demonstrated for PrnA8 and for the tryptophan 7-halogenase RebH from rebeccamycin biosynthesis.9 However, other factors must also be at work, since chlorination of tryptophan cannot be accomplished by chloramine (or HOCl) in solution.10, 11 Chloramine is a weaker halogenating agent than HOCl,12 and according to quantum mechanical calculations, N-chloramine formation reduces the electrophilicity ofthe chlorine species; in other words, the charge Q(Cl) is reduced to −0.07 compared to Q(Cl)=+0.017 in free HOCl.In the active site, glutamate 346 (E346) is positioned across the tunnel from K79, and the positioning of the substrate tryptophan is supported by a hydrogen bond between the NH group of the indole ring and the peptide bond oxygen between E346 and serine 347 (S347) (Figure 1). Evidence that E346 could be involved in the catalytic cycle was the observation that E346Q is two orders of magnitude less active.8 Whereas K79 is absolutely conserved in all the flavin-dependent halogenases known so far, E346 and S347 are conserved only in flavin-dependent tryptophan halogenases and they are not present in halogenases acting on substrates with a phenol or pyrrole ring. Dong et al. suggested that E346 is required for the ‡ Equal contribution

No abstract

The proinflammatory enzyme caspase-1 plays an important role in the innate immune system and is involved in a variety of inflammatory conditions. Rare naturally occurring human variants of the caspase-1 gene (CASP1) lead to different protein expression and structure and to decreased or absent enzymatic activity. Paradoxically, a significant number of patients with such variants suffer from febrile episodes despite decreased IL-1β production and secretion. In this study, we investigate how variant (pro)caspase-1 can possibly contribute to inflammation. In a transfection model, such variant procaspase-1 binds receptor interacting protein kinase 2 (RIP2) via Caspase activation and recruitment domain (CARD)/CARD interaction and thereby activates NF-κB, whereas wild-type procaspase-1 reduces intracellular RIP2 levels by enzymatic cleavage and release into the supernatant. We approach the protein interactions by coimmunoprecipitation and confocal microscopy and show that NF-κB activation is inhibited by anti–RIP2-short hairpin RNA and by the expression of a RIP2 CARD-only protein. In conclusion, variant procaspase-1 binds RIP2 and thereby activates NF-κB. This pathway could possibly contribute to proinflammatory signaling.