Structure of RNase Sa2 complexes with mononucleotides – new aspects of catalytic reaction and substrate recognition Vladena Bauerova ´ -Hlinkova ´ 1 , Radovan Dvorsky ´ 2 , Dus ˇ an Perec ˇ ko 1 , Frantis ˇ ek Povaz ˇ anec 3 and Jozef S ˇ evc ˇ ı ´ k 1 1 Institute of Molecular Biology, Slovak Academy of Sciences, Bratislava, Slovakia 2 Max Planck Institute for Molecular Physiology, Dortmund, Germany 3 Faculty of Chemistry and Agricultural Technology, STU, Bratislava, Slovakia Introduction The microbial RNase superfamily includes more than 150 enzymes, isolated from different fungi and bacte- ria. Most of them are small proteins that are involved in many aspects of cellular RNA metabolism, such as decay of mRNA, conversion of RNA precursors to their mature form, and turnover of certain RNases [1]. The function of RNases is the hydrolysis of the 3¢,5¢- phosphodiester bond of ssRNA. The process of RNA cleavage has been most thoroughly investigated for RNase T1 [2,3]. RNase T1 cleaves the O5¢-phosphodi- ester bond after guanosine in ssRNA by a two-step mechanism. In the first step, trans-esterification, Keywords binding subsite; complex structure; RNA hydrolysis; RNase; substrate recognition Correspondence V. Bauerova ´ -Hlinkova ´ , Institute of Molecular Biology, Slovak Academy of Sciences, Du ´ bravska ´ cesta 21, 84551 Bratislava, Slovakia Fax: +421 2 59307416 Tel: +421 2 59307410 E-mail: vladena.hlinkova@savba.sk (Received 24 June 2008, revised 23 May 2009, accepted 29 May 2009) doi:10.1111/j.1742-4658.2009.07125.x Although the mechanism of RNA cleavage by RNases has been studied for many years, there remain aspects that have not yet been fully clarified. We have solved the crystal structures of RNase Sa2 in the apo form and in complexes with mononucleotides. These structures provide more details about the mechanism of RNA cleavage by RNase Sa2. In addition to Glu56 and His86, which are the principal catalytic residues, an important role in the first reaction step of RNA cleavage also seems to be played by Arg67 and Arg71, which are located in the phosphate-binding site and form hydrogen bonds with the oxygens of the phosphate group of the mononucleotides. Their positive charge very likely causes polarization of the bonds between the oxygens and the phosphorus atom, leading to elec- tron deficiency on the phosphorus atom and facilitating nucleophilic attack by O2¢ of the ribose on the phosphorus atom, leading to cyclophosphate formation. The negatively charged Glu56 is in position to attract the pro- ton from O2¢ of the ribose. Extended molecular docking of mononucleo- tides, dinucleotides and trinucleotides into the active site of the enzyme allowed us to better understand the guanosine specificity of RNase Sa2 and to predict possible binding subsites for the downstream base and ribose of the second and third nucleotides. Structured digital abstract l MINT-7136092: RNase Sa2 (uniprotkb:Q53752) and RNase Sa2 (uniprotkb:Q53752) bind ( MI:0407)byx-ray crystallography (MI:0114) Abbreviations 2¢,3¢-GCPT, guanosine 2¢,3¢-cyclophosphorothioate; 2¢-GMP, guanosine 2¢-monophosphate; 3¢-AMP, adenosine 3¢-monophosphate; 3¢-CMP, cytidine 3¢-monophosphate; 3¢-GMP, guanosine 3¢-monophosphate; 3¢-UMP, uridine 3¢-monophosphate; exo-2¢,3¢-GCPT, exo-guanosine 2¢,3¢-cyclophosphorothioate. 4156 FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS 2¢,3¢-cyclophosphate is produced as an intermediate product. In the second step, hydrolysis, the cyclic inter- mediate is hydrolyzed in the presence of a water mole- cule, yielding an RNA strand that terminates with 3¢-guanylic acid. The most important catalytic residues in RNase T1 are Glu58 and His92, each of which functions as an acid and a base at different steps of the reaction [4,5]. An important role in catalysis was also ascribed to His40. Protonated His40 interacts with Glu58 through a hydrogen bond, enhancing the ability of Glu58 to activate the nucleophilic attack of the ribose O2¢ on the phosphorus atom of the phosphate ester, leading to cyclophosphate formation. Further- more, the positive charge of His40 helps to stabilize the negative charge on one of the cyclophosphate oxygen atoms [6,7]. This general acid–base mechanism was confirmed in a number of bacterial ribonucleases [8–12]. More recent measurements of k cat and K m of cleavage of the substrate analogs R p Gp(S)U and S p Gp(S)U by RNase T1, however, support a triester- like mechanism that depends on the protonation of a nonbridging phosphoryl oxygen [13]. All microbial RNases are either guanine-specific or show a marked preference for it. Guanine binds to the base recognition loop (residues 42–46; RNase T1 num- bering) and forms a hydrogen bond network with the enzyme [14]. Tyr42 (in RNase T1) or an arginine (in RNase Sa, barnase, and binase) has an important role in closing the guanine-binding site [10,15–17]. How- ever, although the interactions between guanine and the enzyme are highly specific, the molecular basis for guanine specificity or preference is still not completely understood [18,19]. Streptomyces aureofaciens strains BMK and R8 ⁄ 26 secrete two different guanyl-specific extracellular RNas- es, RNase Sa and RNase Sa2 [20,21]. They hydrolyze the phosphodiester bonds of RNA at the 3¢-side of guanosine nucleotides in a highly specific manner. The most thor- oughly studied is RNase Sa, which has been used as a model for the study of protein–protein [22] and protein– nucleotide recognition [10,23,24], protein folding and stability [25–28], protein dynamics [29], and cytotoxicity [30]. The mechanism of the catalytic reaction was studied by kinetic measurements [8,9] and supported by struc- tures of complexes of RNase Sa with guanosine 3¢-mono- phosphate (3¢-GMP), guanosine 2¢-monophosphate (2¢-GMP), and exo-guanosine 2¢,3¢-cyclophosphorothio- ate (exo-2¢,3¢-GCPT) [10,23,24]. Glu54 and His85 were identified as the catalytic residues acting as general acids ⁄ base. In contrast to the situation in RNase T1, there is no histidine analogous to His40. The importance of Gln38, Glu54, Arg65 and His85 in RNA catalysis has been shown by site-directed mutagenesis [31]. RNase Sa2 is homologous to RNase Sa. Their amino acid sequence identity is 53%, and the tertiary structure of RNase Sa2 is nearly identical to that of RNase Sa. The amino acids involved in the catalytic reaction are conserved in both enzymes [32]. In spite of this, the kinetic and enzymatic properties of the two enzymes differ [25,33,34]; for example, the catalytic constant k cat of RNase Sa2 at pH 7.0 is seven times lower than that of RNase Sa [34]. To better understand the mechanism of RNA cleavage and differences in the catalytic prop- erties of the two RNases, we have solved the structures of RNase Sa2 with a free active site, and in complexes with an analog of the reaction intermediate exo-2¢,3¢- GCPT, the catalytic cleavage product 3¢-GMP, and 2¢-GMP, which binds to the active site and functions as an RNase Sa2 inhibitor. Extended molecular docking of mononucleotides, dinucleotides and trinucleotides into the active site of RNase Sa2 contributed to a better understanding of enzyme–substrate recognition. Results Description of the structures Crystal structures of RNase Sa2 with a free active site (3D5G) and in complexes with 2¢-GMP (3DGY), exo- 2¢,3¢-GCPT (3D5I) and 3¢-GMP [crystal form I (3D4A) was prepared by diffusion of the mononucleotide, and crystal form II (3DH2) was obtained by cocrystalliza- tion] were solved by molecular replacement [35] and refined by refmac 5.0 [36] against 1.8–2.25 A ˚ data to final R-factors between 18% and 22% (Table 1). Struc- tures 3D5G, 3DGY, 3D5I and 3D4A have three enzyme molecules in the asymmetric unit, and structure 3DH2 has four. RNase Sa2 consists of one a-helix (residues 14–26) and five antiparallel b-strands (residues 7–9, 54–59, 70–75, 80–83, and 91–94) (Fig. 1). The antiparallel b-sheet, which contains three strands (residues 54–58, 71–74, and 79–83), forms the hydrophobic core of the protein. Mononucleotides binding into the active site of RNase Sa2 do not affect the overall fold of the protein. Superposition of 88 corresponding CA atoms of all 16 molecules (structures 3D5G, 3DGY, 3D5I, 3D4A, and 3DH2) yielded rmsd values in the range 0.17–0.56 A ˚ . Five N-terminal residues and loop 62–68 were removed from the superposition, owing to high flexibility. These segments were determined well only in molecules where they were stabilized by a neighboring molecule. The structure of RNase Sa2 was compared with the structures of other microbial RNases: RNase Sa (2SAR), barnase (1BRN), binase (1GOY), and RNase T1 (1RLS). As expected, the highest structural similarity was seen with RNase Sa (rmsd of 0.71 A ˚ ), V. Bauerova ´ -Hlinkova ´ et al. Structures of RNase Sa2–mononucleotide complexes FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS 4157 with the largest differences (up to 3.6 A ˚ ) at the N-terminus and in region 76–78, where there is one residue deletion in RNase Sa. Remarkably lower struc- tural similarities were observed between RNase Sa2 and barnase (rmsd of 1.2 A ˚ ), binase (rmsd of 1.2 A ˚ ), and RNase T1 (rmsd of 1.8 A ˚ ) (Table 2). Crystal packing The asymmetric units of RNase Sa2 with a free active site (3D5G) and in complex with 2¢-GMP (3DGY), exo-2¢,3¢-GCPT (3D5I) and 3¢-GMP crystal form I (3D4A) contain three enzyme molecules (A, B, and C) arranged in the same way. In the complex structures, only the active site of molecule B was accessible to the ligand. Molecules A and C form a crystallographic dimer by interacting through their active sites, so their active sites are occluded (Fig. 2A). The dimer interface is stabilized by six hydrogen bonds and a salt bridge. In the previously solved structure of RNase Sa2 [32], a similar dimer was formed in which Tyr87 from mole- cule C (Tyr87C) was flipped out of its usual position at the bottom of the active site and inserted into the active site of molecule A. The Tyr87 aromatic ring is positioned in the plane that is occupied by the guanine base in the RNase Sa–mononucleotide structures. A similar situation is also observed in 3D5G; however, Table 1. Refinement statistics of RNase Sa2 with free active site and complexed with 2¢-GMP, 2¢,3¢-GCPT, and 3¢-GMP (crystal forms I and II). AU, asymmetric unit; ESU, estimated standard uncertainties of atoms. Structure 1 2 3 45 Crystal form I Crystal form II Protein Data Bank ID 3D5G 3DGY 3D5I 3D4A 3DH2 Ligand – 2¢-GMP 2¢,3¢-GCPT 3¢-GMP 3¢-GMP Resolution (A ˚ ) 1.80 1.80 2.20 2.20 2.25 Molecules in AU Protein 3 3 3 3 4 Mononucleotide – 1 1 1 4 Waters 505 277 167 133 125 R (%) 17.7 21.5 20.8 22 19.8 R free (%) 24.3 24.6 26.2 26.5 26.1 ESU based on R free 0.13 0.12 0.23 0.24 0.28 Average B (A ˚ 2 ) 31.6 30.6 24.3 31.8 14.7 Protein atoms 28.6 29.2 23.7 31.5 14.5 Solvent molecules 40.5 38.7 31.7 37 20.2 Geometry statistics Bond lengths (A ˚ ) 0.022 0.013 0.014 0.014 0.013 Bond angles (°) 1.892 1.471 1.555 1.520 1.610 Chiral centers (A ˚ 3 ) 0.112 0.087 0.114 0.108 0.106 Planar groups (A ˚ ) 0.009 0.005 0.005 0.005 0.005 Fig. 1. Ribbon diagram of RNase Sa2. Table 2. Superposition of corresponding CA atoms of RNase Sa (2SAR, molecule A), barnase (1BRN, molecule L), binase (1GOY, molecule A) and RNase T1 (1RLS) on RNase Sa2 (3DG4A, mole- cule B). CA atoms that differ by more than 3 A ˚ were removed from the superposition. RNase No. of corresponding CA atoms rmsd (A ˚ ) RNase Sa2 ⁄ RNase Sa 90 0.76 RNase Sa2 ⁄ barnase 62 1.19 RNase Sa2 ⁄ binase 62 1.19 RNase Sa2 ⁄ RNase T1 47 1.80 Structures of RNase Sa2–mononucleotide complexes V. Bauerova ´ -Hlinkova ´ et al. 4158 FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS the electron density of the flipped-out Tyr87 side chain is weaker, suggesting a lower level of occupancy. In the structures 3DGY, 3D5I, and 3D4A, there is no electron density for Tyr87C in this alternative confor- mation, suggesting that the crystallographic dimer formation is independent of Tyr87C position. In the asymmetric unit of the RNase Sa2–3¢-GMP crystal form II (3DH2), there are four enzyme mole- cules (A, B, C, and D), each of which has 3¢-GMP molecules bound in its active site. In the crystal, mole- cules A and C, and B and D, interact through their active sites; however, this interaction differs from that mentioned above, as it is mediated by the 3¢-GMP molecules present in both active sites (Fig. 2B). Arg34C and Arg34D appear to play an important role in this interaction. Their d-guanido groups form hydrogen bonds with the phosphate group of the 3¢-GMP present in the active site of their own mole- cule while undergoing a stacking interaction with the guanine bases of 3¢-GMP from the neighboring A B Fig. 2. Stereoview of the A ⁄ C crystallo- graphic dimer (A, green; C, pink) in struc- tures 3D5G, 3GDY, 3D5I, and 3D4A (A), and in structure 3DH2 (B), in which molecules interact through their active sites. In the 3DH2 dimer, the interaction is mediated by the 3¢-GMP molecules present in RNase Sa2 active sites. Residues that form intermolecular hydrogen bonds are drawn as sticks and labeled. Intermolecular hydrogen bonds are shown as dashed lines. V. Bauerova ´ -Hlinkova ´ et al. Structures of RNase Sa2–mononucleotide complexes FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS 4159 molecule. The A–C and B–D interfaces are further stabilized by 10 intermolecular hydrogen bonds. 3¢-GMP in the active site of RNase Sa2 Because RNase Sa2 cleaves RNA specifically at the 3¢-side of guanosine, 3¢-GMP represents the product of the cleavage reaction. 3¢-GMP binds to the active site of RNase Sa2 in two modes. In the first one (Fig. 3B), seen in 3D4A and in molecules A and B of 3DH2, the mononucleotide binds in a similar way as in RNase Sa [10], binase [17], and barnase [37]. 3¢-GMP is in an anti-conformation, and the ribose adopts a C2¢-endo pucker. Guanine of 3¢-GMP forms five hydrogen bonds: three with the amide groups of Glu40, Asn41, and Arg42, and two with the carboxyl group of Glu43. The base is further stabilized by interactions with the aromatic rings of Phe39 and Tyr87, which form the bottom of the active site. Arg42 has an important role in guanine stabilization. In molecule B of the complex prepared by diffusion (3D4A), the planar d-guanido group of Arg42 undergoes a stacking interaction with the guanine base, forming a closed conformation of the active site [38]. The importance of this residue has been shown by kinetic measurements of the R59A mutation in barnase (Arg59 of barnase is structurally equivalent to Arg42 of RNase Sa2), which abolished 85% of the wild-type barnase activity [39]. In mole- cules A and B of 3DH2, the conformation of the ribose is stabilized by a hydrogen bond between O4¢ and Glu56 OE1. The phosphate group of 3¢-GMP forms several hydrogen bonds with the side chains of Glu56, Arg67, Arg71, His86, and Tyr87. The impor- tance of Glu56, Arg67, His86 and Tyr87 has been investigated in RNase Sa mutants by kinetic [31] and activity measurements (E. Heblakova, unpublished), suggesting a similar importance for these residues in RNase Sa2. In the second mode of 3¢-GMP binding, seen in mol- ecules C and D of 3DH2, the guanine base is shifted by 1.9 A ˚ towards Glu43 and Arg42, and the phosphate group by about 1.4 A ˚ . However, the weaker electron density for the mononucleotide and surrounding residues suggests that this manner of 3¢-GMP binding is less favorable and is probably not physiologically relevant. Exo-2¢,3¢-GCPT in the active site of RNase Sa2 Guanosine 2¢,3¢-cyclophosphorothioate (2¢,3¢-GCPT) is an analog of the cyclic reaction intermediate, with one of the two phosphate group oxygens replaced by sulfur. There are two isomers of 2¢,3¢-GCPT, endo-2¢,3¢-GCPT and exo-2¢,3¢-GCPT, which differ in the position of the sulfur atom. Streptomycete RNases cleave only the endo-isomer [24], whereas RNase T1 cleaves both the endo-isomer and the exo-isomer, although the hydroly- sis of the exo-isomer is much slower [40]. The guanine of exo -2¢,3¢-GCPT is bound to the active site in the same way as that of 3¢-GMP (3D4A, ABC Fig. 3. Electron density 2F o –F c (1r level), of mononucleotides exo-2¢,3¢-GCPT (3D5I) (A), 3¢-GMP (crystal form II, 3DH2) (B) and 2¢-GMP (3DGY) (C) in the active site of RNase Sa2. For clarity, side chains of Asn41 and Arg42 are not shown. Atoms of nitrogen, oxygen and phos- phorus are in blue, red, and cyan, respectively. In the enzyme, carbon atoms are yellow. For clarity, in the mononucleotide, carbon atoms are green. The sulfur atom, which replaces one of the phosphate oxygens in exo-2¢,3¢-GCPT, is dark green. Hydrogen bonds between the mononucleotide and RNase Sa2 are shown as dashed lines. Structures of RNase Sa2–mononucleotide complexes V. Bauerova ´ -Hlinkova ´ et al. 4160 FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS molecules A and B of 3DH2). Unlike the anti-confor- mation found in the complex with RNase Sa [24], exo- 2¢,3¢-GCPT in the active site of RNase Sa2 adopts a syn-conformation (Fig. 3A), causing the sulfur atom to point into the enzyme interior. The ribose O2¢ atom of exo-2¢,3¢-GCPT forms a hydrogen bond with the Glu56 OE1, and O3¢ forms a hydrogen bond with the side chain of His86. The only phosphate group oxygen forms two hydrogen bonds with Arg67 NH1 and NH2. The sulfur is within hydrogen bonding distance of Tyr87 OH, Arg71 NE, and His86 NE2. The side chain of Arg34 points towards the nucleotide. It is surprising that exo-2¢,3¢-GCPT adopts the syn- conformation, which is proposed to be catalytic, and is not cleaved by RNase Sa2. This is probably caused by the presence of the sulfur atom, which points into the active site and does not form contacts with the enzyme equivalent to those formed by oxygen. In endo-2¢,3¢- GCPT, the positions of the sulfur and oxygen atoms are exchanged, allowing this isomer to be cleaved. This has also been shown by a model of endo-2¢,3¢-GCPT built in the RNase Sa active site [24]. 2¢-GMP in the active site of RNase Sa2 To obtain a set of complexes of RNase Sa2 with the guanosine mononucleotides that were previously inves- tigated for RNase Sa [10,23,24], we also prepared an RNase Sa2–2¢-GMP complex. The guanine base of 2¢-GMP is bound in the same way as in RNase Sa2–3¢-GMP and RNase Sa2–exo-2¢,3¢-GCPT. The nucleotide is in the syn-conformation, whereas the ribose adopts the C3¢-endo pucker (Fig. 3C). The con- formation of the ribose is stabilized by four hydrogen bonds with Arg42, Arg34 and Glu56 side chains. The phosphate group of 2¢-GMP forms a hydrogen bond network with the side chains of Arg34, Glu56, Arg67, Arg71, His86, and Tyr87. The principal difference between the active sites of RNase Sa2–3¢-GMP and RNase Sa2–2¢-GMP seems to be in the conformation of the Arg34 side chain, which appears to depend on whether the mononucleotide is in the syn-conformation or anti-conformation. In RNase Sa2–3¢-GMP (anti-conformation), the side chain of Arg34 points outside of the active site and does not make any contact with the mononucleotide. In RNase Sa2–2¢-GMP (syn-conformation), the side chain of Arg34 forms hydrogen bonds with both ribose and phosphate. In RNase Sa, Arg34 is replaced by Gln32, which is oriented towards the mononucleotide only in the complex with 3¢-GMP (anti-conformation). Consequently, this substitution may account for some of the differences observed in substrate recognition and RNA cleavage between RNases Sa2 and Sa. Molecular docking of nucleotides After refinement, glucose, which had been used as a cryoprotectant, was found in several protein molecules in the vicinity of Tyr32, Asn33, and Arg34. The best electron density for glucose was found in molecule C of 3DH2 (Fig. 4A), where glucose forms two hydrogen ABC Fig. 4. (A) Electron density 2F o –F c (1r level) of glucose (GLC) in the vicinity of Tyr32, Asn33, and Arg34 (3DH2, molecule C). Glucose forms two hydrogen bonds with Asn33. Dinucleotides (B) and trinucleotides (C) with highest scoring rates docked into the active site of RNase Sa2. The trinucleotides are grouped into two clusters that differ in the position of the third nucleotide. One possible binding site is in the area of Asp66–Gly68. The other binding site is close to the region of Tyr32, Asn33, and Arg34, which corresponds to the glucose position. V. Bauerova ´ -Hlinkova ´ et al. Structures of RNase Sa2–mononucleotide complexes FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS 4161 bonds with Asn33. As glucose was bound close to the active site, we speculated that it might suggest a possi- ble location for the substrate-binding subsite. To sup- port this hypothesis, dinucleotides and trinucleotides were docked into the active site. Seven protein molecules that form complexes with mononucleotides in our structures were used for dock- ing. To verify the reliability of the docking procedure, several mononucleotides [3¢-GMP, 2¢-GMP, 2¢,3¢- GCPT, cytidine 3¢-monophosphate (3¢-CMP), uridine 3¢-monophosphate (3¢-UMP) and adenosine 3¢-mono- phosphate (3¢-AMP)] were docked into the free active site of the enzyme, and the resulting models were com- pared with those obtained from the crystal structures. With the standard-precision setup, guanosine mono- nucleotides were identified with the highest scores in five of the seven enzyme molecules (Table S1). This finding was even more pronounced when the high- precision setup was used, in which guanosine mono- nucleotides scored as the best in six cases. The position of the guanine base was very similar to that found in the crystal structures. The rmsd values of the super- posed base atoms between the docked and observed nucleotides were, in most cases, below 1 A ˚ . The phos- phate groups of most docked mononucleotides were situated in the region of the phosphate-binding site, although the rmsd values of the phosphorus atoms between docked and crystal nucleotides were higher, ranging between  1A ˚ and 2.3 A ˚ . No distinct binding mode was found for ribose. Comparing guanine with adenine, cytosine and ura- cil allowed us to better understand the guanosine spec- ificity of RNase Sa2. In all crystal structures and docked enzyme molecules, guanosine formed the high- est number of hydrogen bonds of all the bases, up to five, and had the best fit into the base-binding site. In addition, guanine underwent a stacking interaction with Phe39 and interacted with Arg42. Guanine forms the most efficient hydrogen-bonding network with the enzyme, and this seems to be very important for proper enzyme–base binding. Other bases form a lower number of hydrogen bonds, up to two, and have worse fits in the RNase Sa2 active site. For the pyrimidine bases, the base-binding site appears to be too large; for cytosine and uracil, we observed both horizontal shifts and rotation of the base with respect to the plane of the guanine, by up to  40°, disrupting the Phe39–base stacking interaction. To find possible binding subsites of RNase Sa2, four dinucleotides and 16 combinations of trinucleotides, all having a guanine as the leading base, were docked into the active site of the enzyme. In the five best-docked dinucleotides in each protein molecule, the position of the guanine base and most of the phosphate groups of the first nucleotide (Gp) corresponded well with the mononucleotides in the crystal structures. The same was true for the ribose, which ended in a syn-confor- mation or anti-conformation. Greater fluctuations were observed in the positions of the ribose and base of the second nucleotide. In all cases, the base of the second nucleotide interacted with the Asp66–Thr69 loop and with His86 (Fig. 4B). The majority of the five best conformations of docked trinucleotides formed two clusters (Fig. 4C). In one cluster, the position of the ribose and the base of the third nucleotide are located in the vicinity of Thr61 and Arg67–Thr69. In the second cluster, the ribose and the base of the third nucleotide are close to Tyr32, Asn33, and Arg34, which corresponds to the position of the bound glucose. The presence of the third nucle- otide appears to influence the position of the base of the second nucleotide, which is turned by 90° and sandwiched between His86 and Thr69 (Fig. 4C). The second phosphate group of the trinucleotide is posi- tioned between Asp66–Thr69 and Arg34 NH1 and NH2, which are  3.2 A ˚ from the phosphate group of the second nucleotide. This suggests that the Arg34 side chain may be important in binding the phosphate group of the second nucleotide. The putative binding subsites in RNase Sa2 were compared with those found in barnase and RNase T1. In barnase, the subsites were identified by kinetic mea- surements [41] and confirmed by crystallization with the tetranucleotide dCp 0 Gp 1 Ap 2 Cp 3 [37]. The most important barnase subsite, labeled p 2 , binds the phos- phate group of the third nucleotide. Occupation of the subsite for p2 gives rise to a 1000-fold increase in k cat ⁄ K m , composed of a 100-fold increase in k cat and a 10-fold decrease in K m [41]. Another important subsite is formed by His102, which binds the base of the third nucleotide. Comparison of the 16 RNase Sa2 docked trinucleotides with the barnase–dCGAC complex showed that the position of the second base of the tri- nucleotides in RNase Sa2 is close to the corresponding adenine in the barnase–dCGAC complex, which inter- acts with His102. This suggests that the role of His102 in barnase is taken over by His86 in RNase Sa2 (Fig. 4C). In RNase T1, two subsites were identified, formed by Asn36 and Asn98. The amide group of Asn36 inter- acts with the ribose of the leaving nucleoside, and Asn98 is partially responsible for the cytosine prefer- ence of the leaving nucleoside [42]. RNase Sa2 does not have a residue equivalent to Asn98 of RNase T1. However, Asn36 of RNase T1 correlates well with the positions of Asn33 and Arg34 in RNase Sa2, which, Structures of RNase Sa2–mononucleotide complexes V. Bauerova ´ -Hlinkova ´ et al. 4162 FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS according to modeling results, might form a subsite for the third nucleotide. Discussion The goal of the present work was to better understand the catalytic mechanism of RNase Sa2 and to account for the differences in catalytic activity between RNases Sa2 and Sa. On the basis of the crystal structures of RNase Sa2 with mononucleotides, we can confirm that the widely accepted reaction mechanism of guanyl-spe- cific RNases involving glutamic acid and histidine as important catalytic residues, as suggested by Takah- ashi and More [5], also applies to RNase Sa2. More- over, the structures provide more detailed information about the role of other residues during RNA cleavage, namely Arg67 and Arg71. Both arginines are found in the phosphate-binding site of RNase Sa2 and are con- served in all microbial RNases. The importance of Arg67 in RNA cleavage was suggested by a site-direc- ted mutagenesis study on RNase Sa [31]. An R65A mutation in RNase Sa caused k cat to decrease by three orders of magnitude. Because Arg65 in RNase Sa is structurally equivalent to Arg67 in RNase Sa2, and because, in all structures of both enzymes, these argi- nines have almost identical conformations and are in almost identical environments, we would expect that an R67A substitution in RNase Sa2 would have an effect on k cat that is very similar to that in RNase Sa. In RNase Sa2–exo-2¢,3¢-GCTP, Arg67 forms a hydrogen bond with the only oxygen in the phosphate group of the mononucleotide, and Arg71 is within hydrogen-bonding distance of sulfur, which replaces the other oxygen of the phosphate group. In the other RNase Sa2–mononucleotide structures, both arginines form hydrogen bonds with the oxygens of the phos- phate group of the mononucleotide (Fig. 3). At the optimum pH of RNA cleavage by RNase Sa2, pH 7.0–7.5, both arginines are protonated, allowing them to polarize the bonds between the oxygens of the phos- phate group and the phosphorus atom. This leads to an electron deficiency on the phosphorus atom, encouraging nucleophilic attack by the electron pair of O2¢ of the ribose (Fig. 5). The side chain of Glu56 is turned towards O2¢ of the ribose, with OE1 within hydrogen-bonding distance of O2¢. The favorable con- formation and distance allow Glu56 to interact with the hydrogen atom bonded to O2¢, weakening its attachment to the oxygen and facilitating O2¢ attack on the phosphorus atom. In both RNase Sa2–3¢-GMP structures (3D4A and 3DH2), His86 forms hydrogen bonds with two oxygens of the phosphate group (Fig. 3B), suggesting that it can be a proton donor for the leaving O5¢ RNA strand. Taking into consideration the conformation of both arginine side chains in the RNase Sa2–mononucleotide structures, Arg67 and Arg71 might also have addi- tional roles in RNA cleavage. In three of the four RNase Sa2–mononucleotide structures (3DGY, 3D5I, and 3DH2), the distance between NH1 and NH2 of Arg67 and the carboxyl group of Glu56 is below 4 A ˚ , and the charged groups of these two residues are fac- ing towards each other. Such a configuration might promote a conformation of Glu56 that is favorable for Fig. 5. The first step of RNA cleavage by RNase Sa2. At the pH optimum of RNA cleavage, 7.0–7.5, Arg67 and Arg71 are very probably pro- tonated, Glu56 is deprotonated, and its phosphate group is negatively charged. The positively charged Arg67 and Arg71 polarize the bonds between the oxygens of the phosphate group and phosphorus atom, causing electron deficiency on the phosphorus atom and, conse- quently, enhancing formation of the cyclophosphate intermediate. Negatively charged Glu56 can interact with the hydrogen atom bonded to O2¢, weakening its attachment to the oxygen and facilitating O2¢ attack on the phosphorus atom. The cyclophosphate intermediate is formed, and the 5¢-strand of RNA is leaving from the active site. The figure was drawn with ISIS ⁄ DRAW 2.5. V. Bauerova ´ -Hlinkova ´ et al. Structures of RNase Sa2–mononucleotide complexes FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS 4163 accepting a proton from O2 ¢ of the ribose. NH1 and NH2 of Arg71 form hydrogen bonds with the main chain oxygen of Gly68, and also, in some molecules, with the main chain oxygen of Arg67. This appears to help to maintain the functional conformation of the phosphate-binding site. As originally reported by Takahashi and More [5], in the next step, 2¢,3¢-cyclophosphate is hydrolyzed by a water molecule that enters the active site and inter- acts with catalytic histidine. Then, a free electron pair of the oxygen attacks the phosphorus atom, resulting in the opening of the cyclophosphate ring and leading to the formation of the final product – a strand of RNA ending with 3¢-GMP. In RNase Sa2–exo-2¢,3¢- GCPT, there is no water molecule close to His86, which may be attributable to the fact that exo-2¢,3¢- GCPT is not a functional substrate. However, in RNase Sa2–3¢-GMP (3DH2), there is a water molecule close to His86 NE2 that forms a hydrogen bond with O2¢ of the ribose. This water molecule, if present in the complex with real substrate, could perform the function of the catalytic water. In spite of the high similarity in amino acid sequences and tertiary structures of RNase Sa and RNase Sa2, their kinetic and physicochemical proper- ties differ (Table 3). To account for the differences in k cat between RNase Sa2 and RNase Sa, and to better understand the function of the amino acids involved in catalysis, we analyzed the active sites of RNase Sa2, RNase Sa (2SAR, 1RSN, and 1GMP), binase (1GOY) and barnase (1BRN) complexes. The conformations of the residues directly involved in binding of the guanine (residues 40–43; RNase Sa2 numbering) are almost identical in all bacterial RNases compared (Fig. 6). In the RNase Sa structure (2SAR), Arg40, which corre- sponds to Arg42 of RNase Sa2, is disordered, owing to the presence of a neighboring molecule. In the struc- tures with different crystal packing (e.g. 1GMP), Arg40 is ordered, forms a stacking interaction with a guanine, and adopts a closed conformation of the active site. Asn41 has an identical conformation in all structures that we compared. The main role of this res- idue is to stabilize the conformation of the loop form- ing the base-binding site, and its importance has been confirmed by site-directed mutagenesis studies with dif- ferent RNases [11,43]. The main difference is found in the position of Arg45, which is close to the base-bind- ing site. The structural counterparts of Arg45 are Val43 in RNase Sa and Arg61 in binase. The impor- tance of Arg61 in binase was shown by an R61V mutation, imitating RNase Sa, which increased the k cat of mutated binase seven-fold in comparison with the wild type [18]. The structural and conformational iden- tity of Arg45 (RNase Sa2) and Arg61 (binase) allows us to consider that an R45V mutation might have a similar effect on the k cat of RNase Sa2. Summary In this article, we have presented five structures of RNase Sa2, one with a free active site (3D5G), and others in complex with an analog of the reaction Table 3. Differences in physicochemical properties of RNase Sa2 and RNase Sa. No. of amino acids Sequence identity (%) pI a Catalytic activity at pH 7 (%) b T m (°C) RNase Sa2 97 53 5.3 14 41.1 c RNase Sa 96 3.5 100 47.1 d a From [33]. b From [34]. c From [56]. d From [25]. Fig. 6. Stereoview of the active sites of RNase Sa2 (blue, 3D4A), RNase Sa (purple, 2SAR), barnase (green, 1BRN), and binase (brown, 1GOY). The main changes in the active sites, which are in the Arg45 and Arg34 positions in RNase Sa2, correspond to Val43 and Gln32 in RNase Sa, Arg61 and Lys26 in binase, and Ala60 and Lys27 in barnase. Structures of RNase Sa2–mononucleotide complexes V. Bauerova ´ -Hlinkova ´ et al. 4164 FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS intermediate, exo-2¢,3¢-GCPT (3D5I), a product of the reaction, 3¢-GMP (3D4A and 3DH2), and the inhibitor 2¢-GMP (3DGY). In all complex structures, the guan- ine base of the mononucleotides forms a hydrogen bond network with the main chain nitrogens of Glu40, Asn41, and Arg42, and OE1 or OE2 of Glu43, and the phosphate-binding site contains Glu56, Arg67, Arg71, His86, and Tyr87. In the exo-2¢,3¢-GCPT complex, O2¢ and O3¢ form hydrogen bonds with OE1 of Glu56 and NE2 of His86, respectively. Arg67 and Arg71 interact with the oxygens of the phosphate group, and site- directed mutagenesis studies performed on their equiv- alents in RNase Sa have shown that they are necessary for the catalytic reaction. At the pH optimum for the reaction, both arginines are protonated, facilitating polarization of the bonds between the oxygens of the phosphate group and phosphorus atom, leading to electron deficiency on the phosphorus atom and, consequently, enhancing formation of the the cyclo- phosphate intermediate. We also propose that the seven-fold higher efficiency of RNA cleavage by RNase Sa than by RNase Sa2 can be at least partly explained by the Val43 (RNase Sa) to Arg45 (RNase Sa2) substitution. On the basis of molecular modeling studies, we propose two possible subsites for the third downstream nucleoside, formed by Thr61 and Arg67– Thr69 and Tyr32, Asn33, and Arg34, respectively. Experimental procedures Purification, crystallization, and data collection RNase Sa2 was purified by a procedure described by Hebert et al. [33], with yields of 10–50 mg from 1 L of cul- ture medium. The crystallization of RNase Sa2 with a free active site was performed as described previously [32]. Complexes of RNase Sa2 with 2¢-GMP (3DGY), exo-2¢,3¢- GCPT (3D5I) and crystal form I of RNase Sa2–3¢-GMP (3D4A) were prepared by diffusion of mononucleotides into the RNase Sa2 crystals with free active sites. The procedure involved adding small amounts of solid mononucleotide to crystallization drops containing crystals of RNase Sa2 until the concentration of the mononucleotide was close to satu- ration. Crystal form II of RNase Sa2–3¢-GMP (3DH2) was prepared by adding approximately twice the amount of 3¢-GMP into the crystallization drops as used for crystal form I. This caused original crystals to dissolve; however, new RNase Sa2–3¢-GMP crystals appeared within 1 day [44]. Diffraction data from all crystals were collected to 1.8– 2.25 A ˚ resolution at the EMBL X31 beamline at DESY (Hamburg), using radiation at a wavelength of 1.1 A ˚ at 100 K. The cryoprotectant solution was prepared by enrich- ing the mother liquor to 25% glucose (w ⁄ v). The crystals were monoclinic and belonged to the C2 space group. Opti- mal conditions for data collection were found using the program best [45]. denzo and scalepack were used for processing of all datasets [46]. Data collection and process- ing statistics for all five structures are summarized in Table S2. Structure determination and refinement Structures were solved by molecular replacement using molrep [35], with RNase Sa2 (1PY3) as a search model. Refinement was performed against 95% of the data using refmac5 [36]. The remaining 5% of the data were ran- domly excluded for the calculation of the R free factor [47]. The solvent molecules were modeled using warp [48]. All models were checked against (2F o –F c ; a c ) and (F o –F c ; a c ) maps and rebuilt using o [49] or xtalview [50]. Mono- nucleotides, sulfate anions and glucose molecules were built into clear 3r peaks in the difference electron density map after several cycles of refinement, and their presence was confirmed by a decrease in R and R free . In the final stages, the complex structures were refined using TLS. Tempera- ture factors, bond lengths and bond angles were restrained according to the standard criteria employed in refmac5. The geometry of all structures was verified with the pro- gram procheck [51]. Analysis of the Ramachandran plot indicated that the torsion angles for more than 90% of the amino acids in all structures are in the most favored regions, and that the rest lie in additionally allowed regions. The final refinement statistics for all five structures are given in Table 1. To evaluate the similarity of the struc- tures, CA atoms of all molecules were superposed with molecule A from 3D5G with the program multiprot [52]. Five N-terminal CA atoms and seven CA atoms in loop 61–67 were excluded from superposition because they were not modeled in most of the molecules, owing to poor elec- tron density. All figures were drawn by pymol [53]. The numbering of amino acids is according to RNase Sa2 unless indicated otherwise. Molecular docking of the nucleotides into the active site of RNase Sa2 Module glide [54] from maestro [55] was used for mole- cular docking. Structures of RNase Sa2 in which the ligand was found in the active site (B molecules of the complex structures 3DGY, 3D5I, and 3D4A, and all four molecules of 3DH2) were selected for docking. All nonprotein mole- cules (nucleotides, sulfates, glucoses, and waters) were removed, and input files containing protein molecules were preprocessed using the Protein Preparation command of glide. Interactions of probe atoms with proteins were calculated with the Receptor Grid Generation command of V. Bauerova ´ -Hlinkova ´ et al. Structures of RNase Sa2–mononucleotide complexes FEBS Journal 276 (2009) 4156–4168 ª 2009 The Authors Journal compilation ª 2009 FEBS 4165 [...]... Bauerova-Hlinkova et al Structures of RNase Sa2 mononucleotide complexes glide at the points of a regular three-dimensional grid ˚ around the active site, spaced by 1 A in all directions ˚ box, and centered at the geometwithin a rectangular 40 A rical center of the bound nucleotides Molecules of mononucleotides 2¢-GMP, 2¢,3¢-GCPT, 3¢-GMP, 3¢-CMP, 3¢-UMP, and 3¢-AMP, dinucleotides GG, GA, GC, and GU, and all possible... (1997) Additivity of protein–guanine interactions in ribonuclease T1 J Biol Chem 272, 963 5–9 639 16 Meiering EM, Bycroft M, Lubienski MJ & Fersht AR (1993) Structure and dynamics of barnase complexed with 3¢-GMP studied by NMR spectroscopy Biochemistry 72, 1097 5–1 0987 17 Polyakov KM, Lebedev AA, Okorokov AL, Panov KI, Schulga AA, Pavlovsky AG, Karpeisky MY & Dodson GG (2002) The structure of substrate- free... Garcia S, Both V, Sevcik J & Pace CN (1997) Purification of ribonucleases Sa, Sa2, and Sa3 after expression in Escherichia coli Protein Express Purif 11, 16 2–1 68 34 Ilinskaya ON, Koschinski A, Mitkevich VA, Repp H, Dreyer F, Pace CN & Makarov AA (2004) Cytotoxicity of RNases is increased by cationization and counter- Structures of RNase Sa2 mononucleotide complexes 35 36 37 38 39 40 41 42 43 44 45 46 47 48... positional constraint of 4.5 A between the geometrical center of the main chain nitrogen of Arg42 and O6 of the leading guanine Accuracy of the docking was assessed on the basis of scoring values calculated by glide Analysis of the positions and conformations of docked molecules was performed using pymol Acknowledgements The authors are very grateful to Dr Jacob Bauer for help with text editing and Dr Lubica... Hollanderova Z, Kormanec J & Sevcik J ¨ (1992) Cloning and sequencing of the gene encoding a ribonuclease from Streptomyces aureofaciens CCM3239 Gene 119, 14 7–1 48 22 Sevcik J, Urbanikova L, Dauter Z & Wilson KS (1998) Recognition of RNase Sa by the inhibitor barstar: structure of the complex at 1.7 A resolution Acta Crystallogr 54, 95 4–9 63 23 Sevcik J, Hill CP, Dauter Z & Wilson KS (1993) Complex of. .. 2, 26 1–2 77 9 Kery V, Both V, Sevcik J & Zelinka J (1986) The number and role of histidine residues in the active site of guanyloribonuclease Sa Gen Physiol Biophys 5, 40 5–4 14 10 Sevcik J, Dodson EJ & Dodson GG (1991) Determination and restrained least-squares refinement of the structures of ribonuclease Sa and its complex with 3¢guanylic acid at 1.8 A resolution Acta Crystallogr B 47 (Pt 2, 24 0–2 53 11... net charge on the solubility, activity and stability of ribonuclease Sa Protein Sci 10, 120 6–1 215 4168 Supporting information The following supplementary material is available: Table S1 Molecular docking of mononucleotides 3¢-GMP, 2¢-GMP, exo-2¢,3¢-GCPT, 3¢-CMP, 3¢-UMP and 3¢-AMP into the active site of RNase Sa2, using standard-precision and high-precision setups of module glide [54] from maestro [55]... 30, 866 6–8 670 Hebert EJ, Giletto A, Sevcik J, Urbanikova L, Wilson KS, Dauter Z & Pace CN (1998) Contribution of a conserved asparagine to the conformational stability of ribonucleases Sa, Ba, and T1 Biochemistry 37, 1619 2–1 6200 ˇ ˇ ´ ´ ´ ˇ Hlinkova V, Urbanikova L & Sevcı´ k J (2002) Crystallization and preliminary X-ray analysis of complex RNase Sa¢ with 3¢-GMP and 2¢,3¢-GCPT Biologia 57, 82 3–8 26 Popov... improvement and extension of crystallographic phases by weighted averaging of multiplerefined dummy atomic models Acta Crystallogr 53, 44 8–4 55 Jones TA, Bergdoll M & Kjeldgaard M (1990) O: a macromolecular modeling environment In Crystallo- FEBS Journal 276 (2009) 415 6–4 168 ª 2009 The Authors Journal compilation ª 2009 FEBS 4167 ´ ´ V Bauerova-Hlinkova et al Structures of RNase Sa2 mononucleotide complexes. .. 164 4–1 653 Sevcik J, Lamzin VS, Dauter Z & Wilson KS (2002) Atomic resolution data reveal flexibility in the structure of RNase Sa Acta Crystallogr 58, 130 7–1 313 Meiering EM, Serrano L & Fersht AR (1992) Effect of active site residues in barnase on activity and stability J Mol Biol 225, 58 5–5 89 Zegers I, Loris R, Dehollander G, Fattah Haikal A, Poortmans F, Steyaert J & Wyns L (1998) Hydrolysis of a . Structure of RNase Sa2 complexes with mononucleotides – new aspects of catalytic reaction and substrate recognition Vladena Bauerova ´ -Hlinkova ´ 1 ,. (A ˚ ) RNase Sa2 ⁄ RNase Sa 90 0.76 RNase Sa2 ⁄ barnase 62 1.19 RNase Sa2 ⁄ binase 62 1.19 RNase Sa2 ⁄ RNase T1 47 1.80 Structures of RNase Sa2 mononucleotide