Crystal structure of human arginase I complexed with thiosemicarbazide reveals an unusual thiocarbonyl μ-sulfide ligand in the binuclear manganese cluster

Abstract

The crystal structure of the human arginase I-thiosemicarbazide complex reveals an unusual thiocarbonyl μ-sulfide ligand in the binuclear manganese cluster. The C=S moiety of thiosemicarbazide bridges Mn²⁺A and Mn²⁺B with coordination distances of 2.6 Å and 2.4 Å, respectively. Otherwise, the binding of thiosemicarbazide to human arginase I does not cause any significant structural changes in the active site. The crystal structure of the unliganded enzyme reveals a hydrogen bonded water molecule that could support proton transfer between a μ-water molecule and H141 to regenerate the nucleophilic μ-hydroxide ion in the final step of catalysis.

Arginase is a 105 kDa homotrimer containing a binuclear manganese(II) cluster in each subunit required for the hydrolysis of L-arginine to yield L-ornithine and urea.¹ Two isozymes, arginase I and arginase II, have been identified in humans and the amino acid sequences of these isozymes are related by 60% identity. In recent years, increasing attention has focused on arginase as a potential therapeutic target due to the overexpression of these isozymes in a variety of diseased tissues and organs, e.g., the airway of asthma patients², the spinal cord fluid in an animal model of multiple sclerosis³, and the corpus cavernosum of diabetic men suffering from erectile dysfunction.⁴

The first X-ray crystal structure of an unliganded mammalian arginase was that of rat arginase I, which revealed a Mn²⁺-Mn²⁺ cluster bridged by a nonprotein ligand interpreted as a μ-hydroxide ion that functions as a nucleophile in catalysis.⁵ The subsequently determined structure of rat arginase I complexed with the boronic acid substrate analogue 2(S)-amino-6-boronohexanoic acid (ABH)⁶ revealed the binding of the tetrahedral boronate anion form of the inhibitor, which mimics the tetrahedral transition state.⁷ Recently determined crystal structures of human arginases I and II complexed with ABH and/or the related boronic acid substrate analogue S-(2-boronoethyl)-L-cysteine (BEC)⁸ revealed similar inhibitor binding modes.^9,10

Despite the high affinity of ABH binding to human arginase I (K_d = 5 nM)¹⁰, consideration of ABH as a possible drug candidate for the treatment of human diseases linked to arginase hyperactivity is tempered by the relative scarcity of boron-containing drugs.¹¹ Thus, we have continued to explore new functional groups for manganese coordination in the design and development of new arginase inhibitors. We now report the X-ray crystal structure of human arginase I complexed with thiosemicarbazide determined at 1.95 Å resolution.

For structure determination, human arginase I was overexpressed in E. coli, purified, and crystallized as described^10,12 with the exception that the protein solution contained 1.4 mM thiosemicarbazide. The structure was refined to final R_twin and R_free/twin values of 0.169 and 0.219, respectively. The structure of unliganded human arginase I was also determined at 1.90 Å resolution and refined to final R_twin and R_twin/free values of 0.198 and 0.244, respectively. Complete experimental details are reported in the Supplementary Information.

The root-mean-square (r.m.s.) deviation of 314 Cσ atoms between unliganded rat arginase I and unliganded human arginase I is 0.64 Å, indicating that these enzymes are quite similar in structure (as expected by their amino acid sequence identity of 87%). However, interesting differences are evident in active site solvent structure and appear to result from an alternative conformation of T246: in human arginase I, the T246 hydroxyl group of this residue is oriented towards the manganese ions and forms a hydrogen bond with solvent molecule #76, which in turn forms a hydrogen bond with the metal-bridging hydroxide ion (solvent molecule #119); in rat arginase I, the T246 hydroxyl group is oriented away from the manganese ions (Figure 1).

Figure 1.

Omit electron density map of unliganded human arginase I calculated with Fourier coefficients |F_obs/A| −|F_calc/A| for twin domain A and phases from the refined enzyme model less the atoms of T246 (contoured at 2.7σ, green) and water molecules #76, #119 and #211 (contoured at 3.0σ, blue). The T246 conformation in rat arginase I (magenta) is superimposed.

Interestingly, the metal-bridging hydroxide ion is also within hydrogen bonding distance of solvent molecule #211, which also forms a hydrogen bond with H141. Solvent molecule #211 also interacts weakly with Mn²⁺ _A, but the Mn²⁺ _A-O separation of 2.8 Å is too long to be considered an inner-sphere coordination interaction. That solvent molecule #211 forms a hydrogen bonded bridge between the metal-bridging hydroxide ion and H141 is consistent with the proposed role of H141 as a proton shuttle in the regeneration of the nucleophilic metal-bridging hydroxide ion from a metal-bridging water molecule.⁵ In other words, solvent molecule #211 may serve as a “proton wire” to facilitate proton transfer in catalysis.¹³

The binding of thiosemicarbazide to human arginase I does not cause any significant structural changes in the active site, and the r.m.s. deviation is 0.39 Å for 313 Cσ atoms between the structures of the thiosemicarbazide-complexed and unliganded enzymes. However, a significant structural change is observed in the manganese coordination polyhedra: the C=S moiety of thiosemicarbazide bridges Mn²⁺_A and Mn²⁺_B with coordination distances of 2.6 Å and 2.4 Å, respectively (Figure 2). These metal coordination distances are consistent with other metalloprotein crystal structures. A search of the Protein Data Bank (PDB)¹⁴ retrieves 62 unique cysteine-Mn²⁺ interactions in 18 protein structures, 12 unique methionine-Mn²⁺ interactions in 6 protein structures, and 3 unique S-Mn²⁺ interactions involving non-protein groups, e.g., thiophosphate derivatives. The average S-Mn²⁺ coordination distance is 2.6 ± 0.2 Å for cysteine ligands and 2.8 ± 0.1 Å for methionine ligands. Additional geometric data are reported in Figure S1 of the Supporting Information.

Figure 2.

(a) Stereoview of simulated annealing gradient maps showing thiosemicarbazide (3.1σ contour, magenta) and its electron-rich sulfur atom (7.3σ contour, green) bound to human arginase I. Dashed lines indicate manganese coordination (red) and hydrogen bond (green) interactions. Atom color codes: carbon (yellow), oxygen (red), nitrogen (blue), manganese (pink), sulfur (green). (b) Summary of intermolecular interactions.

Analyses of the PDB and the Cambridge Structural Database (CSD)¹⁵ indicate that the human arginase I-thiosemicarbazide complex is only the second crystal structure ever determined of a thiosemicarbazide complexed with Mn²⁺, the first such complex being aqua-2,2′-bipyridyl)-(thiosemicarbazidediacetato-O,O′,S)-manganese(II) sesquihydrate (CSD accession code YARSEP) in which the manganese-sulfur coordination distance is 2.6 Å.¹⁶

Further analysis of the CSD (see Supporting Information) yields a total of 68 structures of thiosemicarbazide complexes with the following metal ions: Mn (1), Fe (3), Co (6), Ni (22), Cu (6), Zn (7), Rh (3), Ag (8), Cd (6), Pt (1), Hg (2), Pb (2), and Bi (1). The thiosemicarbazide S-C-N-N dihedral angle is occasionally distorted in these structures, and out-of-plane deviations of up to 21° are observed. At 64°, the distortion of the S-C-N-N dihedral angle of thiosemicarbazide bound to human arginase I is even more pronounced (Figure 2). This distortion appears to facilitate the formation of numerous direct and water-mediated hydrogen bonds. Comparable distortions of O-C-N-O dihedral angles are observed for N-hydroxyurea inhibitors of matrix metalloproteinases and may similarly be facilitated by intermolecular hydrogen bond interactions.¹⁷

Given that the N-OH groups of N-hydroxy-L-arginine and N-hydroxy-nor-L-arginine displace the metal-bridging hydroxide ion of unliganded rat arginase I,¹⁸ it is surprising that the N-NH₂ group of thiosemicarbazide does not do likewise. It is especially surprising that the electron-rich sulfur atom of thiosemicarbazide is preferred for manganese coordination given the fact that the electron-rich sulfur atom of thiosemicarbazide is a relatively soft ligand and Mn²⁺ is a relatively hard metal ion.¹⁹

Regardless, insofar that thiosemicarbazide is an analogue of urea, it is notable that this structure provides the first experimental evidence in support of a metal-bridging mode for the urea product as first proposed by Kanyo and collegues.⁵ The binding affinity of urea is weak (K_d ~ 1 M)²⁰, and isothermal titration calorimetry similarly indicates weak affinity for thiosemicarbazide with K_d > 0.1 mM (i.e., beyond the detection threshold), so the K_d value of thiosemicarbazide likely resides somewhere between these values. Nevertheless, we conclude that the unusual C=S---Mn²⁺ interactions shown in Figure 2 highlight the potential of thiosemicarbazide as a useful fragment²¹ for the design of amino acid thiosemicarbazide inhibitors that will be described in due course.²²

Supplementary Material

si20070412_032. Supporting Information Available.

Experimental procedures and PDB and CSD search parameters. This material is available free of charge via the Internet at http://pubs.acs.org.