Substrate Interactions with Human Ferrochelatase

Home , Binding site, Ferrochelatase, Magnesium chelatase

Substrate interactions with human ferrochelatase

Amy Medlock, Larkin Swartz, Tamara A. Dailey, Harry A. Dailey, and William N. Lanzilotta*

Department of Biochemistry and Molecular Biology, University of Georgia, Athens, GA 30602

Edited by JoAnne Stubbe, Massachusetts Institute of Technology, Cambridge, MA, and approved December 20, 2006 (received for review July 21, 2006) Ferrochelatase, the terminal enzyme in heme biosynthesis, cata- antibody raised against N-methylmesoporphyrin (N-MeMP) IX lyzes the insertion of ferrous iron into protoporphyrin IX to form catalyzes the insertion of metal ions (14–17), the general hypothesis protoheme IX. Human ferrochelatase is a homodimeric, inner has been that N-alkyl porphyrins are transition-state analogs (18– mitochondrial membrane-associated enzyme that possesses an 20). The crystal structure of B. subtilis ferrochelatase with bound essential [2Fe-2S] cluster. In this work, we report the crystal N-MeMP has served as the basis of mechanistic models for ferro- structure of human ferrochelatase with the substrate protopor- chelatases (21). These models, as well as resonance Raman and phyrin IX bound as well as a higher resolution structure of the site-directed mutagenesis studies, assume that the N-MeMP ob- R115L variant without bound substrate. The data presented reveal served in the crystal structure is bound in the active site of that the porphyrin substrate is bound deep within an enclosed ferrochelatase in an orientation that is identical to the spatial pocket. When compared with the location of N-methylmesopor- orientation of the natural substrate/product (22–24). Additionally, phyrin in the Bacillus subtilis ferrochelatase, the porphyrin is it has been assumed that the 36° macrocycle distortion observed in rotated by Ϸ100° and is buried an additional 4.5 Å deeper within the N-MeMP-bound crystal structure represents a catalytic inter- the active site. The propionate groups of the substrate do not mediate that occurs during normal turnover. Data presented herein protrude into solvent and are bound in a manner similar to what provide evidence that these assumptions may not be valid. has been observed in uroporphyrinogen decarboxylase. Further- To provide insight into the interaction of the physiological more, in the substrate-bound form, the jaws of the active site substrate with ferrochelatase, we have determined the structure mouth are closed so that the porphyrin substrate is completely of human ferrochelatase with bound protoporphyrin IX. The engulfed in the pocket. These data provide insights that will aid in human ferrochelatase enzyme used in this investigation was an the determination of the mechanism for ferrochelatase. E343K variant that has a higher affinity for protoporphyrin IX in comparison to the wild-type enzyme. We have also collected heme biosynthesis ͉ protoporphyrin IX ͉ x-ray crystallography ͉ higher resolution data for the previously reported R115L variant metal insertion of human ferrochelatase (25) without bound substrate. The position of the substrate in the E343K variant is distinctly etallated tetrapyrroles are present in most organisms and different from the previously reported orientation of N-MeMP Mparticipate in essential biochemical processes that include in the B. subtilis ferrochelatase (21). In addition to the spatial photosynthesis, oxygen transport, drug metabolism, transcrip- orientation of substrate within the active site of human ferro- tional regulation, NO synthesis, and oxidative phosphorylation. chelatase, this work also shows that the substrate-bound form of Metallation of tetrapyrroles is catalyzed by a group of enzymes the enzyme possesses a ‘‘closed’’ active site conformation that is named chelatases. This group includes, but is not limited to, notably different from the structure of the inhibitor-bound magnesium chelatase, which is essential for chlorophyll produc- B. subtilis ferrochelatase or the structure of the human enzyme tion (1), and ferrochelatase, which is essential for heme produc- without substrate. In both of the latter cases, the active sites are tion (2). Because of the diverse functions of heme, the latter distinctly ‘‘open.’’ These observations provide insight into the enzyme plays a critical role in human health. Human genetic ferrochelatase mechanism of catalysis and inhibition. defects affecting this enzyme have been identified and result in Results the disease erythropoietic protoporphyria (3). Human ferrochelatase, with its [2Fe-2S] cluster (4, 5), represents the convergence General Description of the Overall Structure and Substrate Binding of tetrapyrrole synthesis with iron supply and must play a key role Sites. The crystals of the E343K variant of human ferrochelatase in overall body iron metabolism (6). belong to the triclinic space group P1 (see Table 1), whereas the Ferrochelatases from a variety of sources have been cloned, space group for the initial structure of human ferrochelatase expressed, and characterized to various extents (2, 4), but it is the (containing the amino acid substitution R115L) reported by Wu Bacillus subtilis and mammalian ferrochelatases that have been et al. (25) is orthorhombic (P212121). In the latter case, the studied most extensively. These two enzymes represent the broadest asymmetric unit contained a single biological dimer. The dif- diversity among ferrochelatases examined to date with Ͻ10% ferent crystal symmetry between the substrate-bound form and sequence identity. The B. subtilis protein is a water-soluble, mono- the free enzyme suggests that substantial conformational dif- meric protein with no cofactors (7), whereas the human enzyme is ferences exist between the two forms of the enzyme. Upon an inner mitochondrial membrane-associated homodimer with a phasing and refinement, it was revealed that the asymmetric unit [2Fe-2S] cluster in each subunit (8). Nevertheless, there is clear of the E343K variant of human ferrochelatase contains two structural similarity between these two enzymes. A comparison of the structures reveals a root-mean-square deviation of only 2.4 Å ␣ Author contributions: A.M. and W.N.L. designed research; A.M., L.S., T.A.D., and W.N.L.

for the C atoms. The majority of the conserved residues are located performed research; A.M., L.S., and W.N.L. analyzed data; and H.A.D. and W.N.L. wrote the BIOCHEMISTRY in the active site pocket. paper. Since the initial work by DeMatteis’ group that identified and The authors declare no conﬂict of interest. characterized the ‘‘green pigment’’ in livers of 3,5-dicarbethoxy- This article is a PNAS direct submission. 1,4-dihydrocollidine-treated mice as N-methylprotoporphyrin, Abbreviation: N-MeMP, N-methylmesoporphyrin. there has been considerable interest in N-alkylporphyrins as inhib- Data deposition: The atomic coordinates for the R115L and E343K human ferrochelatase itors of ferrochelatase (9–12). Because of the tight binding com- structures have been deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 2HRC petitive inhibition of ferrochelatase by N-methylprotoporphyrin and 2HRE, respectively). (13), the fact that nonenzymatic metal insertion into porphyrins is *To whom correspondence should be addressed. E-mail: [email protected]. facilitated by macrocycle distortion, and the observation that an © 2007 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0606144104 PNAS ͉ February 6, 2007 ͉ vol. 104 ͉ no. 6 ͉ 1789–1793 Downloaded by guest on September 23, 2021 Table 1. Data collection and reﬁnement statistics R115L E343K

Space group P212121 P1 Wavelength 0.98 1.54 Resolution range, Å 40.0–1.7 50.0–2.5 Unique observations 100,275 53,845 Completeness 99.6 (98.2)* 96.7 (93.5) † Rsym,% 0.08 (14.7) 0.08 (28.2) I/␴ 29.1 (5.8) 13.5 (3.4) Unit cell (a, b, c) 88.5, 92.9, 110.4 61.9, 88.3, 93.2 Protein atoms 5,782 11,564 Solvent atoms 601 338 Resolution limits 40.0–1.7 50.0–2.5 Rcryst,% 22.1 21.6 Fig. 2. Structural alignment of the substrate-bound (E343K) and substrate- Rfree,% 24.2 27.8 free (R115L) forms of human ferrochelatase. The substrate-free form of rmsd bonds, Å 0.006 0.008 human ferrochelatase is shown in green, and the substrate-bound form of rmsd angles, ° 1.21 1.61 human ferrochelatase is shown in magenta. Regions of significant movement average B factor, Å2 21.8 31.1 in the substrate-bound form have been highlighted in red for clarity and include residues 90–115, 302–313, and 349–361. The [2Fe-2S] clusters for the *Numbers in parentheses denote values for the outermost resolution shell. substrate-free and substrate-bound forms are shown in yellow and orange, † ϭ͚ ͚ Ϫ͗ ͘ ͚ ͗ ͘ Rsym hkl [ I(Ihkl,I Ihkl )]/ hkl,I Ihkl , where Ihkl is the intensity of an individual respectively. measurement of the reflection with indices hkl and ͗Ihkl͘ is the mean intensity of that reflection. variant possess [2Fe-2S] clusters, as was reported previously for human ferrochelatase. copies of the biological dimer (Fig. 1), for a total of four peptide Comparison of the R115L variant that lacks bound substrate monomers and six protoporphyrin IX molecules. This finding is with the substrate-bound (E343K variant) form reveals that the shown in Fig. 1, which highlights the relative positions of the four enzyme with porphyrin bound possesses a significantly more monomers, six porphyrin molecules, and two detergent mole- ‘‘closed’’ active site conformation. The overall difference be- cules, as well as the [2Fe-2S] clusters. The fact that substrate was tween these two conformations is illustrated in Fig. 2. There is observed in the electron density is of interest because substrate an average deviation of 3.5 Å between the R115L and E343K was not added during the expression, isolation, or crystallization variants for the backbone atoms of residues 90–115. Substantial of the E343K variant. One protoporphyrin molecule is found in differences in the torsion angles of the peptide backbone for the each of the four active sites, and two are located at a noncrys- substrate-bound structure in comparison with the substrate-free tallographic 2-fold axis. Interestingly, the two protoporphyrin model are also observed for residues 302–313 and 349–361 (Fig. molecules that are located at the noncrystallographic 2-fold axis 2). In addition, there is a reorientation of numerous amino acid are positioned at an interface just outside the entrance to the side chains in the active site pocket. The overall result is an active active sites of monomers B and D. This is the same region that site that completely engulfs the substrate. detergent molecules occupy in the original structure of the R115L human ferrochelatase variant (25). Consistent with pre- Protoporphyrin IX Binding and Active Site Structure. The binding of protoporphyrin IX is identical in all four of the ferrochelatase vious work, we also see electron density for two detergent Ϫ molecules bound at this interface (Fig. 1). All enzyme subunits monomers in the asymmetric unit. A difference (Fo Fc) composite omit map and the model for the active site are shown in both the E343K variant and the higher resolution R115L in Fig. 3. The omit map was generated by using the simulated annealing protocol with 7% of the model omitted during map generation. The planar and pseudosymmetrical nature of the protoporphyrin IX molecule makes it possible to model the substrate in either orientation with minimal impact on the R factor (Ͻ0.5%). However, if the porphyrin is rotated by 180°, then significant peaks appear in the difference map at the vinyl groups, suggesting that one orientation is preferred over the other. It is important to note that the preparations of the E343K ferrochelatase protein generally contain slightly fewer than 0.5 protoporphyrin molecules per ferrochelatase dimer (26). Thus, the E343K amino acid substitution does not cause ferrochelatase to irreversibly bind porphyrin but only increases the affinity of the enzyme for porphyrin. In contrast, the crystallographic data confirm that the substrate occupancy is 100% and the conformation is ‘‘closed’’ for all of the ferrochelatase monomers in the asymmetric unit. Because the substrate propionates are buried Fig. 1. Overall backbone trace of the asymmetric unit for the E343K human ␣ inside the active site and do not contribute to surface charge or ferrochelatase. The C trace for monomers A, B, C, and D are colored green, crystallographic contacts, it is reasonable to conclude that all yellow, magenta, and cyan, respectively. The two detergent molecules and six protoporphyrin IX molecules are also shown with the carbon, oxygen, and ferrochelatase molecules in the closed conformation, indepen- nitrogen atoms colored yellow, red, and blue, respectively. Only the detergent dent of the presence or absence of enclosed substrate, would molecules are labeled. The [2Fe-2S] clusters are represented as spheres and crystallize under identical conditions. Because there is colored bright red. porphyrin-free enzyme in the initial crystallization setup, the

1790 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0606144104 Medlock et al. Downloaded by guest on September 23, 2021 Fig. 3. Wall-eyed stereoview of the substrate and key residues in the active site of the human E343K ferrochelatase. (A) Top view of the substrate and residues R115, Y123, and S130. (B) Side view of the substrate and residues M76, R115, H263, and the point mutation E343K. Carbon, oxygen, nitrogen, and sulfur atoms are colored light brown, red, blue, and cyan, respectively. The light green cage represents a Fo Ϫ Fc composite omit map contoured at 3␴. The omit map was generated by using the simulated annealing protocol with 7% Fig. 4. Wall-eyed stereoview of an anion-binding site (A) and a bound of the model being omitted per cycle. imidazole molecule in the active site of the human R115L ferrochelatase (B). Carbon, nitrogen, oxygen, and chloride atoms are colored tan, blue, red, and yellow, respectively. The green cage represents a Fo Ϫ Fc composite omit map ␴ lack of porphyrin-free ferrochelatase in the crystal lattice sug- contoured at 3 . The omit map was generated by using the simulated annealing protocol with 7% of the model being omitted per cycle. gests that the substrate-free E343K variant is not in a closed conformation and that substrate binding alone results in the closed conformation. These conclusions are consistent with the side chains of Y123 and S130 (an 8␴ peak remained when longstanding use of crystallization techniques to isolate enan- modeled as a water molecule). When a chloride ion is placed in tiomerically pure compounds from racemic mixtures. this position, there was no additional density observed in the In the structural model for the E343K variant of human difference map. The hydrogen bonding in this area is also ferrochelatase, the protoporphyrin IX appears to be spatially consistent with the presence of a chloride ion (Fig. 4A). Spe- positioned in the active site by a few key interactions. One of the cifically, the imide proton from the peptide bond between protoporphyrin propionates forms a salt bridge with R115. This residues H341 and I342 points toward the chloride atom. The interaction is consistent with the published chemical modifica- proton on the side chain of S130 must also point toward the tion data (27). The other propionate participates in a hydrogen- chloride atom because the ␥O of S130 serves as a hydrogen bond bonding network with the side chains of residues S130 and Y123 acceptor for another backbone proton (the peptide bond be- (Fig. 3A). These types of salt bridge and hydrogen-bonding tween residues 131 and 132). These observations are consistent interactions have been observed before with the propionates of with the binding of chloride in other crystallographic models bound substrate in the crystallographic models reported for (30–32). Given the findings with the E343K variant, it is pro- S-adenosyl-L-methionine-dependent bismethylase dehydroge- posed that one of the substrate propionates displaces the chlo- nase/ferrochelatase (CysG) (28) as well as uroporphyrinogen ride ion upon substrate binding. decarboxylase (29). In addition to the salt bridge and hydrogen- The second feature revealed by these data is the presence of bonding interactions, there are other amino acids on either side an imidazole in the active site of the R115L variant (Fig. 4B). Of of the tetrapyrrole that are within 3.5 Å of the macrocycle center. specific interest is the observation that the plane of the imidazole These residues are shown in Fig. 3B and include residues M76 ring resides parallel to the imidazole ring from the side chain of and H263. Interestingly, W310 (W256 in mouse numbering), a residue H263. This position is strikingly similar to the position of residue that has been proposed to be involved in saddling of the the pyrrole ring of bound protoporphyrin substrate in the E343K porphyrin during catalysis (24), is farther from the porphyrin variant described above. The arrangement of water molecules in than predicted from the B. subtilis:N-MeMP structure, with a the active site further supports our conclusion that this density BIOCHEMISTRY closest approach of 3.7 Å. is due to a bound imidazole molecule (Fig. 4B). Given that our purification procedure employs a metal affinity matrix for which R115L Ferrochelatase. A higher resolution structure of the R115L imidazole is used to elute the protein, it is likely that the human ferrochelatase has also been solved (Table 1). This imidazole observed in the structure arises from the elution variant form of human ferrochelatase has essentially identical buffer. catalytic properties as the wild-type recombinant human ferrochelatase. Two previously unreported features are revealed by Discussion our data (Fig. 4). First, a molecule with substantially larger Although a variant form of the human enzyme was used in this electron density than water is bound in close proximity to the study, these data provide significant insights into the binding

Medlock et al. PNAS ͉ February 6, 2007 ͉ vol. 104 ͉ no. 6 ͉ 1791 Downloaded by guest on September 23, 2021 Fig. 5. Comparison of the protoporphyrin IX and N-MeMP binding modes for human and B. subtilis ferrochelatase, respectively. (A) Structural alignment of the E343K human ferrochelatase model containing protoporphyrin IX (blue cartoon) with the model reported for B. subtilis ferrochelatase containing N-MeMP (green cartoon; PDB ID code 1C1H). (B) Wall-eyed stereoview showing the relative positions of protoporphyrin IX, N-MeMP, and the side chains of strictly conserved amino acids within the active site of the two structures.

mode of the physiological substrate. The data reveal that when in the positions of atoms, several side chains are also reoriented protoporphyrin is bound in the active site of ferrochelatase, the within the active site pocket. In particular, the guanidino side enzyme surrounds the substrate in a snug pocket. The pyrrole chain of R164 is flipped 180° out of the pocket, the benzyl ring nitrogen of the protoporphyrin IX ‘‘A’’ ring is positioned directly of F337 is flipped 90° and rotated into the pocket, and the side under and 3.2 Å distant from the ring nitrogen of the conserved chains of H341 and M76 are rotated Ϸ90°. The overall effect is and centrally located H263 side chain. that the ‘‘mouth’’ of the active site pocket is closed in the The spatial position of the protoporphyrin substrate in the presence of porphyrin and forms a snug fit around the porphyrin active site of human ferrochelatase is significantly different from macrocycle. The two porphyrin propionates interact with active what has been reported for the position of the ferrochelatase site residues via hydrogen and ionic bonds to ensure a highly inhibitor, N-MeMP, in the structure of the B. subtilis enzyme. specific spatial orientation of the porphyrin in the active site. Although these two enzymes have little sequence identity, they This observation is consistent with the high degree of porphyrin do share similar monomeric structures, positioning of key active substrate specificity (20, 33) and spectroscopic studies (34) on site residues, and kinetic characteristics. In fact, a number of the enzyme. Similar observations have been made for another published spectroscopic and catalytic studies on mouse ferro- heme biosynthetic enzyme, uroporphyrinogen decarboxylase chelatase base their conclusions on the B. subtilis structures (28). In this case, the propionate groups are also held in place (22–24). Thus, although possible, it seems unlikely that these two through ionic interactions with arginine residues and hydrogen- proteins would bind substrate and inhibitors in significantly bonding interactions with tyrosine and serine residues. In com- different positions. The current data showing that the macro- parison, the published B. subtilis ferrochelatase structures with cycle is 4.5 Å ‘‘deeper’’ in the pocket and is rotated 100° in and without bound N-MeMP do not show any difference in comparison to what was observed in the N-MeMP-bound fer- active site amino acid side chain orientations or overall active site rochelatase structure (Fig. 5) suggest that the previous assign- pocket shape (21). ment of specific catalytic roles to various active site residues One somewhat surprising finding is that the enzyme-bound needs to be reevaluated. porphyrin macrocycle is only modestly distorted (Ϸ11.5° bend). The structure of human ferrochelatase with bound protopor- This distortion is significantly less than the 36° bend found with phyrin has significant differences in the position of several loop N-MeMP bound to the B. subtilis enzyme (21). The bound regions (residues 90–115, 302–313, and 349–361) when com- porphyrin in the E343K variant has a modest saddle conforma- pared with the structure of human ferrochelatase without bound tion that is consistent with theoretical calculations made by substrate. The positions of residues 90–115 are displaced an av- Sigfridsson and Ryde (35) for a B. subtilis ferrochelatase:por- erage of 3.5 Å, and the spatial positions of residues 302–313 and phyrin complex. Their theoretical model proposes a tilt of all 349–361 are also substantially different from what is seen in the four rings, with one pyrrole ring being 13–15° out of plane and structure without bound substrate. In addition to the difference the other three being 1–10° out of plane. However, because their

1792 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0606144104 Medlock et al. Downloaded by guest on September 23, 2021 model was based on the position of N-MeMP in the B. subtilis of the data presented above, it would seem appropriate to enzyme structure rather than what is reported here for the reevaluate these current models. physiological substrate, protoporphyrin IX, some differences may be expected between the theoretical model and the crys- Materials and Methods tallographic data for human ferrochelatase. Enzyme Expression, Isolation, and Crystallization. Mutagenesis, ex- Previous resonance Raman studies have examined conforma- pression, and purification of human ferrochelatase were per- tional changes that occur to porphyrins and metalloporphyrins formed as described (8, 26, 37). Concentration buffer for variants on binding to ferrochelatase (22–24). Although these data are consisted of 0.045 M Tris-Mops (pH 8.1), 0.09 M potassium consistent with some form of macrocycle distortion, the conclu- chloride, 0.9% sodium cholate, 0.225 M imidazole, and 10% sions drawn by the authors were necessarily influenced by use of glycerol. Crystallization was carried out by using the hanging drop method in the EasyXtal crystallization tool (Qiagen, Va- the B. subtilis:N-MeMP structure as the reference model. With lencia, CA). Mother liquor for the R115L crystals consisted of the current structural data, new assignments for the spectral 0.05 M calcium chloride, 0.1 M Bistris (pH 6.5), and 30% vol/vol shifts can be made that may provide further insight into the poly(ethylene glycol) methyl ether 550. Mother liquor for the ferrochelatase mechanism. Of additional interest is the crystal- E343K ferrochelatase crystals consisted of 0.1 M Bistris (pH lographic observation that human ferrochelatase not only binds 5.5), 0.2 M magnesium chloride, and 25% wt/vol PEG 3350. porphyrin within the active site but also has associated ‘‘free’’ Crystals typically took a minimum of 1 week at 18°C to grow. For porphyrin nearby, but outside, the active site. This observation both variants, the glycerol concentration was gradually increased may account for the published resonance Raman data that to 20% before flash freezing the crystals in liquid nitrogen. reported spectra characteristic of both planar and ruffled macrocycle conformations (23). These data were obtained for the Data Collection, Structure Determination, and Refinement. Data for mouse equivalent of the human E343Q variant enzyme. This the R115L variant were collected on Beamline 8.2.2. at the variant form of mouse ferrochelatase, and the human E343K and Advanced Light Source (ALS) in Berkeley, CA, by using a single E343H variants (26), have significant amounts of bound proto- crystal rotated a full 360°. Data for the E343K variant were porphyrin when purified. collected at the University of Georgia on a Rigaku (Tokyo, Before the current data, catalytic models for ferrochelatase Japan) RU-200 rotating anode equipped with Osmic focusing were largely influenced by the proposal that N-alkylporphyrins mirrors and an R-axisIIc image plate detector. A single crystal were transition-state analogs. Thus, when the structure of B. was used, and a full 360° of data were collected. In all cases, subtilis ferrochelatase with bound N-MeMP became available molecular replacement, model building, and refinement were performed by using the programs O (38) and CNS (39). All (21), the assumption was made that the N-MeMP was bound in graphic representations were rendered by using PyMOL (40), the active site in an orientation similar to what would occur with Molscript (41), Bobscript (42), and Raster3D (43). the natural substrate protoporphyrin IX. This assumption has resulted in a number of studies that proposed specific roles for This work was supported by American Heart Association Grant active site residues and has even served as the basis for a recently 0465228B (to W.N.L.) and National Institutes of Health Grant DK32303 proposed model for metal substrate specificity (36). On the basis (to H.A.D.).

1. Willows RD, Hansson M (2003) in The Porphyrin Handbook, eds Kadish KM, 22. Lu Y, Sousa A, Franco R, Mangravita A, Ferreira GC, Moura I, Shelnutt JA Smith KM, Guilard R (Academic, New York), pp 1–48. (2002) Biochemistry 41:8253–8262. 2. Dailey HA, Dailey TA (2003) in The Porphyrin Handbook, eds Kadish KM, 23. Franco R, Ma JG, Lu Y, Ferreira GC, Shelnutt JA (2000) Biochemistry Smith KM, Guilard R (Academic, New York), pp 93–121. 39:2517–2529. 3. Todd DJ (1994) Br J Dermatol 131:751–766. 24. Shi Z, Franco R, Haddad R, Shelnutt JA, Ferreira GC (2006) Biochemistry 4. Dailey HA, Dailey TA, Wu CK, Medlock AE, Wang KF, Rose JP, Wang BC 45:2904–2912. (2000) Cell Mol Life Sci 57:1909–1926. 25. Wu CK, Dailey HA, Rose JP, Burden A, Sellers VM, Wang BC (2001) Nat 5. Shepherd M, Dailey TA, Dailey HA (2006) Biochem J 397:47–52. Struct Biol 8:156–160. 6. Wingert RA, Galloway JL, Barut B, Foott H, Fraenkel P, Axe JL, Weber GJ, 26. Sellers VM, Wu CK, Dailey TA, Dailey HA (2001) Biochemistry 40:9821–9827. Dooley K, Davidson AJ, Schmid B, et al. (2005) Nature 436:1035–1039. 27. Dailey HA, Fleming JE (1986) J Biol Chem 261:7902–7905. 7. Al-Karadaghi S, Hansson M, Nikonov S, Jonsson B, Hederstedt L (1997) 28. Stroupe ME, Leech HK, Daniels DS, Warren MJ, Getzoff ED (2003) Nat Struct Structure (London) 5:1501–1510. Biol 10:1064–1073. 8. Burden AE, Wu C, Dailey TA, Busch JL, Dhawan IK, Rose JP, Wang B, Dailey 29. Phillips JD, Whitby FG, Kushner JP, Hill CP (2003) EMBO J 22:6225–6233. HA (1999) Biochim Biophys Acta 1435:191–197. 30. Park C, Schultz LW, Raines RT (2001) Biochemistry 40:4949–4956. 9. De Matteis F, Gibbs AH, Smith AG (1980) Biochem J 189:645–648. 31. Li SJ (2006) Biopolmers 81:74–80. 10. De Matteis F, Gibbs AH, Tephly TR (1980) Biochem J 188:145–152. 32. Natesh R, Schwager SL, Evans HR, Sturrock ED, Acharya KR (2004) 11. Tephly TR, Wagner G, Sedman R, Piper W (1978) Fed Proc 37:35–39. Biochemistry 43:8718–8724. 12. Tephly TR, Gibbs AH, Ingall G, De Matteis F (1980) Int J Biochem 12:993–998. 33. Honeybourne CL, Jackson JT, Jones OT (1979) FEBS Lett 98:207–210. 13. Dailey HA, Fleming JE (1983) J Biol Chem 258:11453–11459. 34. Dailey HA (1985) Biochemistry 24:1287–1291. 14. Cochran AG, Schultz PG (1990) Science 249:781–783. 35. Sigfridsson E, Ryde U (2003) J Biol Inorg Chem 8:273–282. 15. Venkateshrao S, Yin J, Jarzecki AA, Schultz PG, Spiro TG (2004) J Am Chem 36. Al-Karadaghi S, Franco R, Hansson M, Shelnutt JA, Isaya G, Ferreira GC Soc 126:16361–16367. (2006) Trends Biochem Sci 31:135–142. 16. Blackwood ME, Rush TS, Romesberg F, Schultz PG, Spiro TG (1998) 37. Mayer MR, Dailey TA, Baucom CM, Supernak JL, Grady MC, Hawk HE, Biochemistry 37:779–782. Dailey HA (2004) J Struct Funct Genomics 5:159–165. 17. Lavallee DK (1987) The Chemistry and Biochemistry of N-Substituted Porphyrins 38. Jones TA, Zou JY, Cowan SW, Kjeldgaard M (1991) Acta Crystallogr A (Wiley, New York). 47:110–119.

18. Lavallee DK (1988) in Mechanistic Principles of Enzyme Activity, eds Liebman 39. Brunger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve BIOCHEMISTRY JF, Greenberg A (VCH, New York), pp 279–314. RW, Jiang JS, Kuszewski J, Nilges M, Pannu NS, et al. (1998) Acta Crystallogr 19. Dailey HA (1990) in Biosynthesis of Heme and Chlorophylls, ed Dailey HA D Biol Crystallogr 54:905–921. (McGraw–Hill, New York), pp 123–162. 40. DeLano WL (2002) PyMOL v0.99 (DeLano Scientific, San Carlos, CA). 20. Dailey HA, Jones CS, Karr SW (1989) Biochim Biophys Acta 999:7–11. 41. Kraulis PJ (1991) J Appl Cryst 24:946–950. 21. Lecerof D, Fodje M, Hansson A, Hansson M, Al-Karadaghi S (2000) J Mol Biol 42. Gouet P, Robert X, Courcelle E (2003) Nucleic Acids Res 31:3320–3323. 297:221–232. 43. Merritt EA, Bacon DJ (1997) Methods Enzymol 277:505–524.

Medlock et al. PNAS ͉ February 6, 2007 ͉ vol. 104 ͉ no. 6 ͉ 1793 Downloaded by guest on September 23, 2021