STRUCTURAL AND BIOCHEMICAL ANALYSIS REVEAL PIRINS TO POSSESS QUERCETINASE ACTIVITY

Pirin is a recently identified eukaryotic protein implicated in transcriptional activation Homologues of Pirin are highly in both prokaryotes and eukaryotes, their function remains poorly understood. We present here the crystal structure of the YhhW gene product, a putative Pirin homologue, Escherichia coli and confirm its structural similarity to Pirin. The YhhW protein displays a bicupin fold with a single N-terminal metal coordination site. Molecular surface comparisons of YhhW and Pirin with structurally similar proteins quercetin as a potential ligand. We the of the We also demonstrate the release of carbon monoxide as a reaction

Pirin is a recently identified protein that is localized to sub-nuclear dot-like structures in a wide variety of human tissues (1). Pirins are highly conserved among mammals, plants, fungi and prokaryotes and have been classified as a subfamily of the cupin superfamily on the basis of both sequence and structural similarity (1,2). The cupin superfamily is one of the most functionally diverse protein classes and includes both enzymatic and nonenzymatic members, ranging from isomerases and epimerases that are involved in the modification of cell wall carbohydrates to non-enzymatic storage proteins in plant seeds and transcriptional cofactors in humans (2). Cupin proteins typically maintain a highly conserved G(x) 5 HxH(x) 3,4 E(x) 6 G metalbinding motif, along with a G(x) 5 PxG(x) 2 H(x) 3 N motif, and a conserved β-barrel fold. The structure of human Pirin (hPirin), the only structure of a Pirin determined to date, also displays two cupin domains, consistent with other bicupin family members (3).
The function of Pirin is not known and no enzymatic activity has been described for any of the homologues. Early studies demonstrated the ability of hPirin to interact with the nuclear factor I/CCAAAT box transcription factor (NF-I) (1) and the oncoprotein B cell lymphoma 3-encoded (Bcl- 3) in vivo (4), suggesting that hPirin may be involved in the modulation of NFI/CTF1 transcriptional activation, or possibly in other nuclear processes, such as DNA replication or repair (1). As observed with hPirin and its homologues, proteins involved with such fundamental cellular processes are highly conserved throughout evolution. Recent studies of tumour metastasis in advanced colorectal cancer have also implicated Pirin in cellular responses to polysaccharide-K, a chemoimmunotherapeutic agent from mushrooms (5).
Studies of hPirin homologues in plants have suggested a possible involvement of these proteins in programmed cell death, seed germination and seedling development. The Le-pirin gene from tomato exhibited dramatically increased mRNA levels in response to both camptothecin-and fusmonisin-B1-induced programmed cell death, which when taken together with the hPirin interaction studies implies a possible relationship between the Le-pirin encoded protein and NF-κBassociated pathways in plant defence mechanisms (6). The insertion mutants of the hPirin homologue from Arabidopsis display phenotypes of reduced germination levels in the absence of stratification and an abscisic acid-imposed delay in germination and early seedling development (7). These studies highlight the apparent importance of Pirin proteins in fundamental cellular processes of numerous species.
Until very recently, there had been no reports on the characterization of hPirin homologues in prokaryotes. Investigations of the cyanobacterium Synechocystis sp. PCC6803 indicated that the putative Pirin-encoding gene was highly induced under severe salt stress and other stress conditions (8). This work also demonstrated the gene to be negatively regulated by a LysR-type transcriptional regulator located immediately upstream of the gene, but in the divergent direction (8).
To date, hPirin is the only structure representative of this protein subfamily.
As investigations of bacterial homologues are underrepresented in the literature, we undertook structurebased functional analysis of YhhW for comparative purposes. While the overall sequence identity is not high (28% identity) and the bacterial variant is 60 residues shorter in length than that of human counterpart, the N-terminal region is very well conserved, particularly for residues forming the metal binding site. Here we present the crystal structure of YhhW refined to 2.0 Å. The structure aligns well with that of hPirin. Active site analysis has allowed us to identify the antioxidant quercetin as an enzymatic substrate and we demonstrate that both YhhW and hPirin have activity similar to that of quercetin 2,3-dioxygenase, a non-Pirin member of the bicupin family. Additionally, we have been able to confirm the release of carbon monoxide during the reactions of both proteins and have shown that activity can be precluded by inhibitors of quercetin 2,3-dioxygenase. This is the first report of quercetinase activity in either humans or E. coli and the first report of enzyme activity for any member of the Pirin family.

MATERIALS AND METHODS
Protein preparation, crystallization and x-ray diffraction data collection -Recombinant YhhW was expressed as a selenomethionine derivative in DL41 (DE3) E. coli cells under control by the T7 promoter in LeMaster medium (9). The expression construct contained a C-terminal hexa-histidine tag that permitted facile purification using batch elution over nickel-nitrilotriacetic acid resin (Qiagen). The final yield was 200 mg of pure protein per litre E. coli culture with no additional purification steps. The protein was concentrated to 10 mg mL -1 in 50 mM NaH 2 PO 4 , pH 8.0, and 300 mM NaCl for crystallization. All reagents for protein expression and purification were purchased from BIOSHOP Canada. All crystallization reagents were purchased from Hampton Research. Protein crystals of YhhW were generated through hanging-drop vapour diffusion by mixing 2 µL of protein solution with 2 µL of well solution. The final crystallization condition contained 26% polyethylene glycol 400, 0.1 M sodium acetate pH 4.6, and 0.1 M cadmium chloride.
The hPirin expression construct was generously provided by Dr. Zihe Rao from the Laboratory of Structural Biology, Tsinghua University. The protein was expressed in DL41 (DE3) E. coli cells under kanamycin selection, as per the previous reports (10). The cells were then lysed and the protein purified in a similar manner to YhhW. hPirin was purified to homogeneity via a single nickel-affinity step.
Data were collected at cryogenic temperature (100 K) using 35% polyethylene glycol 400 as a cryo-protectant at the 19-BM beamline of the Structural Biology Center, Advanced Photon Source, Argonne National Laboratory (Chicago, IL, USA) using a 3 x 3 mosaic CCD detector. Reflections were processed using DENZO and SCALEPACK (11). The space group was P3 2 21 with unit cell dimensions a = b = 62.769Å, c = 98.053Å.
Structure determination and refinement -The YhhW crystal structure was determined by single anomalous dispersion using a selenomethionine derivative. The heavy atom positions for 6 of 7 selenomethionine positions, along with 5 cadmium positions, were identified using SOLVE (12)(13)(14). Density modification was performed by RESOLVE (12)(13)(14). The figure of merit of phasing following density modification was 61% in RESOLVE. RESOLVE traced 153 residues, or 66%, of the residues in the asymmetric unit. This starting model and the phase from RESOLVE were used for unrestrained refinement with iterative water refinements using REFMAC5 (15). The new model, consisting of 218 of the 242 residues in the construct, was built using the refined phases by ARP/wARP 6.0 (16). The remaining residues were added manually and the complete model refined using REFMAC5 (15) and CNS (17). Though density was observed for three histidines of the purification tag none was apparent for the side chains of residues Val138 and Gln139, which have been modeled as glycine. A total of 222 water molecules and 6 cadmium ions have been added to the final model. The structure factors and coordinates have been deposited under the accession code 1TQ5 in the Protein Data Bank (PDB). Fold recognition and comparison were carried out using the Protein Structure Comparison Service, incorporating secondary structure matching, at European Bioinformatics Institute (http://www.ebi.ac.uk/msd-srv/ssm) to search the Secondary Structure Classification of Proteins (SCOP) database (18,19).
Activity Assays -The quercetin 2,3-dioxygenase activities of YhhW and hPirin were assayed at 295K for 5 minutes in a reaction mixture containing 50 mM NaH 2 PO 4 pH 8.0, 300 mM NaCl, 60 µM quercetin in dimethyl sulfoxide (DMSO), and 18.5 nM of enzyme. Activity was observed by following the decrease in the absorbance maximum for quercetin, which occurs at 384nm at pH 8.0. The inhibitors kojic acid, sodium diethyldithiocarbamate, and 1,10-phenanthroline monohydrochloride were dissolved in DMSO and added to the reactions at a final concentration of 50 nM (20)(21)(22).
Atomic absorption spectrophotometry -The iron and copper content of the YhhW protein sample was analyzed using atomic absorption spectrophotometry on a SpectrAA-20Plus instrument (Varian) employing an air-acetylene flame through the Analytical Services Unit at Queen's University. The detection wavelengths were 324.8nm (0.5nm slit width) and 248.3nm (0.2nm slit width) for copper and iron, respectively.
Inductively coupled plasma-mass spectrometry -YhhW samples were treated with 5mM EDTA and 5mM diethyldithiocarbamate and concentrated to 1.8 mg mL -1 in water, equivalent to a greater than 70fold molar excess of each chelating reagent relative to the amount of protein present.
Samples not subjected to metal chelation were concentrated to 1.3 mg mL -1 in water. For both samples, the concentrate flow-through was collected and the metal content measured as a control. All samples were processed and the metal content determined at the Queen's Facility for Isotope Research, Department of Geology, Queen's University. The amounts of magnesium, manganese, iron, cobalt, copper, zinc and cadmium were quantified for each sample.
Detection of CO liberation -The release of carbon monoxide (CO) during the reaction was monitored in amber glass vials with a total reaction volume 250µL following displacement of the headspace gas for 10s with 1% O 2 . A solution of quercetin in reaction buffer was used as a standard. Increasing CO levels in the headspace were monitored using a RGA3 gas chromatograph (Trace Analytical, Menlo Park, CA) equipped with a 13x molecular sieve and a chemicalspectrophotometric detector that measures, at 254 nm, elemental Hg from the reaction of CO with HgO (23).

RESULTS
The Crystal Structure of YhhW -The structure of YhhW, a putative member of the Pirin family, was determined using the single anomalous dispersion (SAD) method from a single selenomethionine derivative crystal ( Figure 1). The structure was refined to 2.0 Å with R and R-free values of 20.2% and 27.2%, respectively (Table 1). For the final structure, 90.2% of residues were in the most favoured regions of the Ramachandran plot, while 8.2% and 1.6% were in the additionally allowed and generously allowed regions of the plot, respectively. We have attributed the high R-free value to inherent problems with this crystal form, the data from which presents a slightly skewed intensity distribution as calculated in Truncate (24) (See Supplementary Material). The space group was P3 2 21 with one monomer in the asymmetric unit. The final model contained all of the 231 residues encoded by the native protein, however no side chain density was observed for residues Val138 and Gln139, thus these residues were modeled as glycine. This model also contained 222 water molecules and 6 cadmium ions, one of which was clearly located at the conserved metal coordination site. All cadmium ions exhibited full occupancy in the refined structure. Interestingly, the histidine purification tag was partially visible in the refined structure and was involved in the bridging of YhhW with a symmetry related molecule through interactions mediated by cadmium atoms. Two purification-tag histidine residues coordinate a cadmium ion in a small cleft between the N-and Cterminal domains. The histidine lying between these residues is rotated to face a symmetry-related molecule where it is involved in the coordination of a symmetry-related cadmium ion. The remaining cadmium sites were located on the molecular surface where they were involved in less elaborate intermolecular bridging (See Supplementary Material). Structural alignment with the previously determined human Pirin structure was excellent ( Figure 2), displaying only a root mean squared (r.m.s.) deviation of 1.81 Å despite an overall sequence identity of 28% (18,19).
YhhW is composed of two structurally similar domains arranged face to face. Each domain is comprised of two antiparallel β-sheets, with seven β-strands forming a β-sandwich. The N-and Cterminal domains are cross-linked with strand β1 forming part of one sheet of the C-terminal domain. A single-turn helix connects β1 in the C-terminal domain to β2 of the N-terminal domain. A second domain cross-link occurs between residues Asp136 to Lys141 which join strands β13 to β14. The more apparent difference between YhhW and hPirin occurs at the C-terminus. For hPirin, the C-terminus extends back toward the N-terminal domain in close proximity to the metal binding site and contains a long α-helix (3). YhhW does not contain an extended C-terminus.
The metal binding site of YhhW is located within the N-terminal domain, and is accordingly positioned in hPirin. Density for a cadmium ion, added as part of the crystallization condition, is clearly visible between residues His57, His59 and His101 ( Figure 1). Specifically, the metal is ligated by His59 and His101 with respective bond lengths of 2.28 and 2.54 Å between the ε-nitrogens and the metal. Two water molecules are also involved in the coordination of cadmium and are located 2.47 and 2.67Å from the cation. A break between strands β5 and β6 appears to provide sufficient flexibility in this region to permit the formation of the metal binding site. Iron is observed in the hPirin structure and the coordination involves Glu103 in addition to the three histidine residues (3). While Glu103 is conserved in YhhW, it is not involved in the coordination of the cadmium ion. Rather, Glu103 of YhhW is hydrogenbonded to the side chain of Gln91, which positions Glu103 away from the coordinated metal. A slight rotation of His57 that precludes its involvement of cadmium ligation is also observed in YhhW but is not seen in hPirin.
Attempts of identification of the coordinated divalent cation -Atomic absorption spectroscopy was initially employed to quantify the levels of copper and iron in the protein samples. The calculated molar ratio of copper to iron was 1.43. The concentrations of these metal ions were significantly higher than those of the buffer alone.
To complement this result, the metal content of YhhW protein samples was determined by inductively coupled plasma-mass spectrometry. No significant amounts of magnesium, manganese, cobalt or cadmium were found in any of the samples tested. The blank-subtracted molar ratios of Zn:Cu:Fe were 4.05:1.18:1 for the sample receiving no treatment with EDTA or diethyldithiocarbamate and 7.79:1.62:1 for the sample treated with chelating reagents.
Pirin and YhhW are structurally similar to quercetin 2,3-dioxygenase -The molecular surface of YhhW was compared with that of hPirin (PDB code 1J1L) (3), as well as with members of the bicupin class that displayed good structural alignments with YhhW ( Figure 2 (26). Of particular interest was the localization of a deep charged pocket in the N-terminal domain of YhhW and hPirin immediately next to the metal-coordination site. This region was most similar in charge, shape and size to that of quercetin 2,3-dioxygenase (PDB code 1JUH) ( Figure  2) (22). Additionally, manual docking of the quercetin substrate into the pockets of YhhW and hPirin suggested that this compound might be appropriate for catalysis by these proteins (data not shown). The sequence conservation between YhhW and quercetin 2,3-dioxygenase is poor, despite the conservation of the metal coordination residues. Figure 3 presents the structure-based sequence alignment of YhhW with quercetin 2,3-dioxygenase from A. japonicus. Cupin motif 1, G(x) 5 HxH(x) 3,4 E(x) 6 G, demonstrates good sequence similarity but the structural alignment suggests that a slight conformational difference occurs at the metal binding site, as His57 of YhhW is not well aligned with the corresponding histidine residue from quercetin 2,3-dioxygenase ( Figure 3). This is consistent with the rotation of His57 away from the coordinated cadmium in the YhhW crystal structure (Figure 1).
In contrast, cupin motif 2, G(x) 5 PxG(x) 2 H(x) 3 N, displays poor sequence conservation between these proteins, though both generally conform to the consensus sequence. Of particular note is the lack of sequence conservation outside of the cupin motifs in the N-terminal domain.
The global folds of YhhW and hPirin aligned reasonably well with reported structures of quercetin 2,3-dioxygenase (2.8 Å and 2.9 Å r.m.s.d., respectively), a copper-containing enzyme with an N-terminal domain that is structurally conserved with the Pirin family (3,22). The structural conservation of hPirin with other bicupin proteins, including quercetin 2,3-dioxygenase, had been previously reported, however the quercetinase activity of hPirin was not investigated (3). The copper binding site of quercetin 2,3-dioxygenase is formed by three histidines (His66, His68 and His112), a glutamate residue, and a single water molecule. These residues are thought to be functionally equivalent to His55, His57, His101, and Glu103 of YhhW. The N-terminal domain of YhhW aligns well with that of quercetin 2,3-dioxygenase with only 1.05 Å r.m.s.d. for 66 equivalent residues, while the N-terminal domain alignment of hPirin with quercetin 2,3-dioxygenase displays an r.m.s. deviation of only 1.5 Å for 84 equivalent residues, when all atoms, including the metal co-ordination residues, are considered as part of the alignment (27).
Quercetin is a substrate for YhhW and hPirin -hPirin and YhhW were assessed for their ability to utilize quercetin as a substrate. Quercetin 2,3dioxygenase converts quercetin to 2protocatechuoylphloroglucinol carboxylic acid and carbon monoxide (Figure 4a). Reactions were carried out similarly to those of quercetin 2,3dioxygenase (28). Striking modifications in the absorbance spectrum of quercetin were noted immediately upon addition of enzyme. The typical absorbance maximum for quercetin has been reported to occur at 367 nm, however this is shifted to approximately 384 nm in the reaction buffer (50 mM NaH 2 PO 4 pH 8.0, 300 mM NaCl) (29). In other solvent systems, such as water, methanol and aqueous buffers with neutral or acidic pH, the peak is consistent with the literature value of 367 nm (data not shown). Despite the pH-dependent shift in maximal absorbance, a significant loss of this maximum was noted on addition of YhhW or hPirin ( Figure 4b). The reactions for both hPirin and YhhW were completely inhibited by kojic acid, sodium diethyldithiocarbamate, and 1,10phenanthroline monohydrochloride, which are all quercetin 2,3-dioxygenase inhibitors (Figure 4b) (20)(21)(22). Additionally, we have been able to confirm that carbon monoxide is generated as a product of the YhhW and hPirin reactions (Figure 4c).

DISCUSSION
Primary sequence analysis predicted both hPirin and YhhW to belong to the cupin superfamily (1). Both proteins contain the aforementioned characteristic motifs of the cupin superfamily. While members of the bicupin protein family vary in length, most differences occur away from the conserved motifs at the loop regions and the Cterminus. The core of the fold is well maintained in all known bicupin structures, including those of YhhW and hPirin, and comprises a motif of six to eight antiparallel β-strands located within a conserved β-barrel structure (30). The broad range of enzymatic activities encompassed by the cupin superfamily is reflected in the low sequence homology found in members of this group. Pirin homologues, however, display significant sequence conservation in the N-terminal domain, particularly for cupin motifs I and II, which include the four conserved metal coordinating residues (1).
Structural similarity searches with YhhW revealed the fold to be most similar to that of hPirin, though significant fold homology was also observed with the general bicupin protein class. However, due to the broad range of functions represented amongst the bicupin proteins, the low sequence identity of YhhW and other putative Pirin homologues with non-Pirin bicupin proteins had prevented sequencebased enzymatic annotation for these proteins. Comparisons of the electrostatic surface potentials of Pirins with other bicupins suggested quercetin to be a Pirin ligand. While the structural similarity of hPirin with quercetin 2,3-dioxygenase had been reported, no attempts to measure the quercetinase activity of hPirin have been described (3). As previously described, the size, shape, charge and localization of the binding pockets are consistent between the Pirins and quercetin 2,3-dioxygenase. Indeed, we have been able to demonstrate the characteristic decrease in absorbance for quercetin upon addition of YhhW or hPirin and have confirmed the release of carbon monoxide during reactions with both of these proteins (Figure 4). Both YhhW and hPirin are also susceptible to inhibition by compounds typically employed as inhibitors of the quercetin 2,3-dioxygenase reaction. Surprisingly, outside of the metal coordination residues, there are no residues that line the binding pocket which are structurally conserved between the Pirins and quercetin 2,3-dioxygenase (Figure 3).
The extracellular enzyme quercetinase, now known more specifically as quercetin 2,3dioxygenase, was discovered more than thirty years ago in strains of Aspergillus (28). In this species, it is a secreted enzyme that is unique in its production of CO as a by-product of the degradation of quercetin to 2-proto-catechuoylphloroglucinol carboxylic acid ( Figure 4A) (31). The crystal structure of quercetin 2,3-dioxygenase from A. japonicus was the first non-iron dioxygenase to be reported (22). This result clearly showed that the enzyme contained only one copper center per monomer, which agreed with previous results from metal elementary analysis (22). More recently, extensive studies involving the structure solution of quercetin 2,3-dioxygenase in complex with its substrate and a variety of inhibitors have suggested a mechanism for reaction involving the mononuclear type 2 center, as well as clarified the mechanisms of inhibition of kojic acid and diethyldithiocarbamate (20,21,32). The structures of quercetin 2,3dioxygenase anaerobically complexed with kaempferol and quercetin, both known substrates for the enzyme, were virtually identical to each other but displayed significant differences relative to the unbound enzyme (32). A number of residues located in front of the active site entrance exhibited substantial alterations on substrate binding. Of these, residues 159-174 became ordered, showing clear electron density on substrate binding and formed a lid over the active site. While there are no additional residues in YhhW that would be capable of forming a lid over the active site entrance, such a mechanism may be possible for hPirin, given its extended C-terminus that lies in close proximity to the active site (3). Kaempferol and quercetin were demonstrated to bind the copper ion in a monodentate fashion through their O3 atom, displacing the water molecule present in the native structure. The substrates' B-ring (Figure 4a) binds in the active site through van der Waals interactions with residues Tyr35, Met51, Thr53, Glu73, Phe75, Phe114, Met123, Ile127, and the backbone atoms of Gly125. The A and C rings are also stabilized by van der Waals interactions. The substrate oxygen atoms O7 and O4' hydrogen bond with two well defined solvent molecules. Most interestingly, the enzyme bound substrates displayed a pyramidalization of the C2 flavonol atom, which caused the B-ring to bend out of the plane defined by the A and C rings. A detailed reaction mechanism, similar to that of the intradiol-type, iron-containing catechol dioxygenases (33), has been described on the basis of this work (21). As evidenced in Figure  3, no significant sequence conservation with quercetin 2,3-dioxygenase is apparent in the Nterminus of YhhW, apart from the metal coordination residues. Since the proposed mechanism for the quercetin 2,3-dioxygenases primarily involves van der Waals interactions of the substrate with the amino acids of active site pocket, rather than specific residue interactions, it is possible that the general chemical character of the active site is more important for activity than the identities of the individual residues.
Structural investigations into the enzymeinhibitor complexes of quercetin 2,3-dioxygenase bound to kojic acid and diethyldithiocarbamate revealed that these compounds bind the enzyme in similar manners, despite their different chemical nature. Both asymmetrically chelate the metal with their molecular plane facing the solvent, a geometry that is perpendicular to that of bound quercetin (20). These structures also highlight the flexibility of the highly conserved Glu73 residues with regard to metal coordination.
The large sulphur of diethyldithiocarbamate precludes the ligation of the Glu73 sidechain to the metal. In contrast, kojic acid complexes demonstrate full Glu73 metal-ligation. It is important to note that in the unbound structure of quercetin 2,3-dioxygenase Glu73 is involved in metal coordination with only 30% occupancy (20) whereas hPirin appears to exhibit full coordination of the metal by Glu103 in the absence of substrate or inhibitor (3). When these observations are taken together with the conservation of Glu103 in YhhW, the lack of coordination of Glu103 with cadmium in this crystal structure does not preclude the possibility of a role for Glu103 in the coordination of divalent cations in the enzyme-substrate complex.
Structural comparisons between the Pirins and quercetin 2,3-dioxygenase revealed that the metal coordination sites of these proteins are dissimilar in their hydrogen bonding patterns. Traditional tri-histidine coordination sites exhibit hydrogen bond formation between the histidine side chains and the protein backbone that confers rigidity, and likely metal specificity, to the region. The histidines of quercetin 2,3-diosygenase are also stabilized by a hydrogen bond between the Nε1 atom of the imidazole rings and carbonyl oxygens of the main chain (22). Neither hPirin nor YhhW presents hydrogen bonds involving any of the histidine side chains responsible for metal coordination. The coordination of cadmium by YhhW highlights the malleable nature of this region. In addition to the shift of Glu103 away from the cadmium ion, the side chain of His57 is rotated relative to the metal with its Nε2 atom not involved in cadmium coordination.
Recently, the protein YxaG from Bacillus subtilus was shown to possess quercetinase activity and was also the first described intracellular quercetinase enzyme, providing precedence for the existence of intracellular quercetinases. Unlike YhhW and hPirin, YxaG shows substantial sequence identity to quercetin 2,3-dioxygenase in the Nterminal domain (34). The ability of YxaG to exchange its divalent cation while still retaining some quercetinase activity is of substantial interest (35). Of particular note was the observation that replacement of iron with copper enhanced quercetinase activity in a statistically significant manner, though the protein purifies with iron as the most abundant divalent cation. With regard to the structure tri-histidine coordination site, one molecule of YxaG demonstrated hydrogen bonding between all three histidines and the protein backbone while in the second molecule in the asymmetric unit only two of the histidines display similar hydrogen bonding pattern (PDB code 1Y3T). Histidine B234 of YxaG, while involved in metal coordination, is not hydrogen bonded to the protein backbone. As for YxaG, the atomic absorbance spectroscopic analysis of hPirin determined there to be a predominance of iron in the sample and refinement of the hPirin structure also demonstrated full occupancy for the iron cation (3). It has been suggested that members of the Pirin family are capable of exchanging their divalent cation (3), a possibility supported by the analysis of YxaG.
The variation in hydrogen bonding patterns for the metal coordination sites between these proteins may provide a mechanism for the predicted exchange of metal cofactors. As YhhW crystallized with cadmium clearly bound in the metal-coordination site, further metal analysis of this protein was necessary to suggest the possible catalytically relevant divalent cations. The presence of iron and copper were confirmed by both atomic absorption spectrometry and inductively coupled plasma-mass spectrometry.
Surprisingly, the concentration of zinc in the samples was substantially greater than either that of copper or iron. While phosphomannose isomerase (36) and auxin-binding protein (37) provide evidence for the binding of zinc by members of the cupin superfamily, there have been no reports of zincdependent dioxygenases.
The differences between previously reported quercetin 2,3-dioxygenases and the Pirins may reflect cellular adaptations that allow for spatialtemporal modulation of quercetin levels. Flavonoids, such as quercetin, are a class of widespread naturally occurring compounds synthesized by a variety of plants, including foods for human consumption.
Of the flavonoids, quercetin is among the most extensively studied. The numerous available reports suggest that quercetin, and related compounds, have varied and significant cellular effects. Firstly, due to the chemical nature of flavonoids, quercetin is capable of stabilizing free electrons obtained from free radicals, such as the reactive oxygen species (ROS) in biological systems (38,39). Some flavonoids are also capable of inhibiting ROS formation by chelating metal ions that would otherwise be involved in Fenton reactions (39). Quercetin and other flavonoid compounds are known to inhibit the production of superoxide anions by xanthine oxidase (38). Secondly, quercetin acts as a protein kinase inhibitor at the ATP binding site through its ability to compete with ATP for binding to the enzyme (40). Increased concentrations of quercetin have been shown to reduce phosphorylation of Akt and ERK kinases, leading to the activation of caspase 3 and death in human primary neuronal cells (41,42). Finally, quercetin is able to inhibit topoisomerase II in humans (43), as well as serve as a regulator of transcription for NF-κB and AP-1 influenced genes (44). Taken together, these studies reflect the importance of quercetin regulation within both prokaryotic and eukaryotic cells and in particular the requirement of quercetin degradation to alleviate its inhibitory effects on various cellular pathways.
The involvement of quercetin in transcriptional regulation may be an important In conclusion, we have solved the crystal structure of YhhW, a Pirin homologue from E. coli O157:H7. The structure of YhhW is highly similar to human Pirin, particularly in the region of the metal binding site. Analysis of the cavity containing the metal-coordination site demonstrated YhhW and hPirin to be similar to quercetin 2,3-dioxygenase from Aspergillus japonicus. In a test of activity, YhhW and hPirin were both shown to perpetuate the loss of the characteristic absorption maximum for quercetin. This spectral change could be completely inhibited by the addition of inhibitors specific to the reaction. Additionally, we were able to detect the release of carbon monoxide for both YhhW and human Pirin, a known product of the quercetin 2,3dioxygenase reaction. Such activity may provide a mechanism that would avoid inhibition of key cellular proteins by quercetin.
connection to the function of members of the Pirin family, which are believed to function as transcriptional co-regulators. As previously stated, hPirin interacts with Bcl-3, an inhibitor of the IκB family of NF-κB inhibitors (3). It has the properties of a transcriptional co-activator and acts as a bridging factor between NF-κB/Rel and nuclear coregulators. hPirin is known to be one of several binding partners that can associate with Bcl-3 to form part of a larger quaternary complex on NF-κB DNA binding sites (3). As both quercetin and Bcl-3 are inhibitors of NF-κB proteins, it is possible that hPirin, as a quercetin 2,3-dioxygenase, acts to modulate transcriptional responses to quercetin. In a similar fashion, prokaryotic systems are also exposed to quercetin and studies have demonstrated that DNA gyrase is inhibited by quercetin (45). Given the ubiquitous distribution of quercetin, it would be of tremendous importance for all cells to rapidly degrade this compound in order to properly modulate cellular processes, with transcriptional activation being of utmost importance.  (46). The conserved metal coordination residues are shown in the N-terminal domain. A cadmium atom located in this site is highlighted in magenta. The 2Fo-Fc density for the metal coordination site is depicted with a 2σ contour level for the whole area with the 5σ contour level also shown for the cadmium atom (47).  (48). While there is little sequence similarity between the Pirins and quercetin 2,3-dioxygenase, the charge, localization, shape and size of the active sites are most similar amongst the bicupins (49). As an example, although glycinin g1 maintains the bicupin topology, it lacks a comparable ligand binding site in the corresponding domain. Figure 3. Structure-based sequence alignment of YhhW with quercetin 2,3-dioxygenase. The native protein sequences for YhhW and quercetin 2,3-dioxygenase have been aligned using SSM searches of the SCOP database (18,19). Conserved residues are indicated with an asterisk. Sequence segments showing good structural similarity have been capitalized. For clarity, the conserved structural elements, color-coded to match Figures 1 and 2, are indicated above the protein sequences. Furthermore, the sequence motifs I and II that are strictly conserved in the cupin family are highlighted in red. shown. Similar results were obtained for hPirin. Inhibition of the reaction is observed on addition of kojic acid, diethyl dithiocarbamate and 1,10-phenanthroline. Five measurements have been averaged for each run and the average absorbance at the wavelength of maximal absorbance (384nm) for each reaction is shown in the inset graph. The error bars represent one standard deviation for the average of each reaction condition. No statistically significant difference is observed for the inhibited reactions when compared to the standard quercetin solution, indicating a lack of quercetin turnover (c) CO release as monitored by the production of elemental Hg from the reaction of CO with HgO. Increasing amounts of CO are observed over the time course of the reaction. CO measurements were made immediately following the system purge (i) and at 2 minutes (ii), 5 minutes (iii) and 8 minutes (iv) post-purge.      F o is the observed structure factor, F c is the calculated structure factor based on the model. SAD data were collected in inverse beam mode.
No sigma cutoff was applied to the data. 5% of reflections were excluded from refinement for calculation of

R-free
Data given in parentheses are for the high resolution shell (2.07 -2.00 Å) r.m.s.d., root mean square deviation.