Inhibitor design for ribonuclease A: the binding of two 5′-phosphate uridine analogues

Description
Inhibitor design for ribonuclease A: the binding of two 5′-phosphate uridine analogues

Please download to get full document.

View again

of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

Medicine, Science & Technology

Publish on:

Views: 0 | Pages: 7

Extension: PDF | Download: 0

Share
Tags
Transcript
  structural communications Acta Cryst.  (2009). F 65 , 671–677 doi:10.1107/S1744309109021423  671 Acta Crystallographica Section F Structural Biologyand CrystallizationCommunications ISSN 1744-3091 Inhibitor design for ribonuclease A: the binding oftwo 5 000 -phosphate uridine analogues Vicky G. Tsirkone, a KyriakiDossi, a Christina Drakou, a Spyros E. Zographos, a MariaKontou b and Demetres D.Leonidas a * a Institute of Organic and PharmaceuticalChemistry, National Hellenic ResearchFoundation, 48 Vas. Constantinou Avenue,11635 Athens, Greece, and  b Department of Biochemistry and Biotechnology, University of Thessaly, 26 Ploutonos St., 41221 Larissa,GreeceCorrespondence e-mail: ddl@eie.grReceived 4 May 2009Accepted 5 June 2009 PDB References:  ribonuclease A, U5P complex,3dxg, r3dxgsf; UDP complex, 3dxh, r3dxhsf. In the quest for the rational design of selective and potent inhibitors formembers of the pancreatic ribonuclease A (RNase A) family of biomedicalinterest, the binding of uridine 5 0 -phosphate (U5P) and uridine 5 0 -diphosphate(UDP) to RNase A have been investigated using kinetic studies and X-raycrystallography. Both nucleotides are competitive inhibitors of the enzyme, with K  i  values of 4.0 and 0.65 m M  , respectively. They bind to the active site of theenzyme by anchoring two molecules connected to each other by hydrogen bondsand van der Waals interactions. While the first of the inhibitor molecules bindswith its nucleobase in the pyrimidinyl-binding subsite, the second is bound at thepurine-preferring subsite. The unexpected binding of a pyrimidine at the purine-binding subsite has added new important elements to the rational designapproach for the discovery of new potent inhibitors of the RNase A superfamily. 1. Introduction Ribonucleases (RNases) are enzymes that catalyze the degradationof RNA. The most well studied RNase is the mammalian pancreaticribonuclease A (RNase A; Raines, 1998). The RNA-binding site of RNase A is a deep groove in the molecular surface (Fig. 1) lined withpositively charged residues that bind the phosphate groups of thesubstrate RNA, mainly by electrostatic interactions. Along this deepcleft, several subsites have been identified that accommodate thephosphate groups, the riboses and the bases of RNA. These subsitesare denoted P 0  . . . P n , R 0  . . . R n  and B 0  . . . B n , respectively, where  n indicates the position of the group with respect to the position atwhich phosphodiester-bond cleavage occurs ( n  = 1; Raines, 1998).RNase A has a preference for purines at subsite B 2  and binds pyri-midines at subsite B 1  (Raines, 1998).In recent years, the members of the RNase A superfamily haveattracted considerable biomedical interest as targets for the discoveryof new pharmaceuticals for the treatment of inflammatory disordersand cancer. The fact that the ribonucleolytic activity of these enzymesis a prerequisite for the pathological activities related to the proteinsof this family has triggered a structure-assisted approach to the designof inhibitors, mainly for three human RNases: angiogenin (RNase 5;Ang), a potent inducer of neovascularization that manifests patho-logically during tumour growth and metastasis, and two eosinophilRNases that have been implicated in inflammation and viral repli-cation, eosinophil-derived neurotoxin (EDN) and eosinophil cationicprotein (ECP) (Russo  et al. , 2001). Initial efforts have mostly targetedthe parental protein RNase A, since it is more amenable to bindingstudies and its active site is conserved in all members of this super-family. Several structures of RNase A in complex with phospho-purine nucleotide derivatives have been reported to date by X-raycrystallography or NMR [d(Ap) 4  (McPherson  et al. , 1986); d(CpA)(Zegers  et al. , 1994; Toiron  et al. , 1996); UpcA (Richards & Wyckoff,1973; Gilliland  et al. , 1994); 2 0 ,5 0 -CpA (Wodak  et al. , 1977; Toiron  et al. , 1996); d(ApTpApA) (Fontecilla-Camps  et al. , 1994); ppA-3 0 -p,ppA-2 0 -p (Leonidas  et al. , 1997); 3 0 ,5 0 -ADP, 2 0 ,5 0 -ADP, 5 0 -ADP(Leonidas  et al. , 2003); dUppA-3 0 -p (Jardine  et al. , 2001); pdUppA-3 0 -p (Leonidas  et al. , 1999); AMP, IMP (Hatzopoulos  et al. , 2005) and # 2009 International Union of CrystallographyAll rights reserved  d(GMP) (Larson  et al. , 2007)]. Most of these studies have focused onthe purine-preferring subsite B 2  and structure-based inhibitor-designefforts have been directed towards ligand molecules that containvarious adenine derivatives, with the aim of exploiting the potentialinteractions offered by protein residues at this subsite. This approachhas generated pdUppA-3 0 -p, the most potent low-molecular-weightinhibitor of RNase A reported to date ( K  i  = 27 n M  ), which is alsoeffective against two major nonpancreatic RNases, Ang and EDN(Russo  et al. , 2001; Leonidas  et al. , 1999).In recent years, inhibitor-discovery efforts have shifted towardspyrimidine analogues that mostly explore the interactions with resi-dues in subsite B 1 , generating several compounds which also inhibitangiogenin as well as RNase A (Maiti  et al. , 2006; Leonidas  et al. ,2006; Ghosh  et al. , 2008). However, the molecular recognition of pyrimidine derivatives by subsite B 1  has only been studied in thecomplexes of RNase Awith U-2 0 -p and U-3 0 -p (Leonidas  et al. , 2003);the structural mode of binding of 5 0 -phosphopyrimidines has not yetbeen analyzed. With the aim of studying the molecular recognition of 5 0 -phosphouridines by RNase A, we have determined the crystalstructures of the U5P–RNase A and UDP–RNase A complexes andhave studied the inhibitory potency of these ligands towards RNaseA in solution. 2. Materials and methods Bovine pancreatic RNase A (type XII-A), U5P, UDP, cytidine 2 0 ,3 0 -cyclic phosphate (C>p) and other chemicals were obtained fromSigma–Aldrich (Athens, Greece). The enzymatic activity of RNase Awas measured using a spectrophotometric method at 303 K in 0.1  M  MES–NaOH buffer pH 6.0, 0.1  M   NaCl with an enzyme concentra-tion of 1  m M   (Hatzopoulos  et al. , 2005). The inhibition constants ( K  i )were determined by the method of Dixon (1953) using nonlinearregression analysis with the program  GraFit   (Leatherbarrow, 2007).Crystals of RNase A were grown at 289 K using the hanging-dropvapour-diffusion technique as described previously (Leonidas  et al. ,1997). Briefly, drops formed by mixing equal volumes of an RNase Asolution (18 mg ml  1 ) in water and reservoir solution [20 m M   sodiumcitrate buffer 5.5 and 25%( w / v ) PEG 4000] were equilibrated againstreservoirs containing 25%( w / v ) PEG 4000 and 20 m M   sodium citratebuffer pH 5.5. Single crystals (800    400    50  m m) appeared after7–10 d at 289 K. Crystals of the inhibitor complexes were obtained bysoaking RNase A crystals (Leonidas  et al. , 1997) in a solution of thecrystallization medium [20 m M  sodium citrate pH 5.5, 25%( w / v ) PEG4000] containing either 50 m M   U5P for 45 h or 50 m M   UDP for 2.5 hprior to data collection. Diffraction data to 1.4 A˚resolution werecollected on station PX10.1 (   = 1.0448 A˚), SRS Daresbury, Englandat 100 K [using a solution of 20 m M   sodium citrate buffer pH 5.5,25%( w / v ) PEG 4000 and20%( w / v ) MPDas a cryoprotecting medium]on a MAR 225 CCD detector using the MAR CCD diffraction data-collection protocol. The exposure time was 10 s per image, theoscillation range was 0.8  and a total of 162 and 180 images werecollected for the U5P and UDP complexes, respectively. Data wereprocessed using the  HKL  package (Otwinowski & Minor, 1997) andthe program  TRUNCATE   (French & Wilson, 1978). Phases wereobtained using the structure of free RNase A (Leonidas  et al. , 2006)as a starting model. Alternate cycles of manual building with theprogram  Coot   (Emsley & Cowtan, 2004) and refinement using themaximum-likelihood target function and anisotropic temperature-factor refinement of all non-H atoms with the program  REFMAC  (Murshudov  et al. , 1997) improved the model. Inhibitor moleculeswere included in during the final stages of the refinement procedureusing models from the  REFMAC   library. Details of data-processingand refinement statistics are provided in Table 1. structural communications 672  Tsirkone  et al.   Ribonuclease A  Acta Cryst.  (2009). F 65 , 671–677 Table 1 Crystallographic statistics. Values in parentheses are for the outermost shell.RNase A–U5P RNase A–UDPSpace group  C  2  C  2Unit-cell parameters (A˚,   )  a  = 100.035,  b  = 32.299, c  = 72.475,    = 90.00,   = 90.91,     = 90.00 a  = 100.003,  b  = 32.337, c  = 72.299,    = 90.00,   = 90.72,     = 90.00Matthews coefficient (A˚  3 Da  1 ) 2.10 2.09Resolution (A˚) 30.0–1.40 (1.42–1.40) 30.0–1.40 (1.42–1.40)Reflections measured 414856 297437Unique reflections 44340 (2291) 45026 (2290) R merge † 0.106 (0.263) 0.044 (0.111)Completeness (%) 95.3 (99.8) 97.8 (100.0) h  I  /   (  I  ) i  34.5 (4.1) 29.0 (8.2) R cryst ‡ 0.208 (0.217) 0.188 (0.186) R free § 0.254 (0.273) 0.225 (0.270)No. of solvent molecules 358 367R.m.s. deviation from idealityIn bond lengths (A˚) 0.009 0.009In angles (  ) 1.4 1.4Average  B  factor (A˚  2 )Protein atoms (mol  A /mol  B ) 19.9/19.9 17.6/16.3Solvent molecules 32.4 31.7Ligand atoms (mol  A /mol  B /mol  C  /mol  D )24.2/20.9/24.7 26.5/22.7/24.1/28.2†  R merge  = P hkl  P i  j  I  i ð hkl  Þ  h  I  ð hkl  Þij = P hkl  P i  I  i ð hkl  Þ , where  I  i ( hkl  ) and  h  I  ( hkl  i  are the i th and the mean measurements of the intensity of reflection  hkl  . ‡  R cryst  = P hkl   j F  o    F  c j = P hkl   F  o , where  F  o  and  F  c  are the observed and calculated structure-factor amplitudes of reflection  hkl  , respectively. §  R free  is the same as  R cryst  but for arandomly selected 5% subset of reflections not used in the refinement (Bru ¨ nger, 1992). Figure 1 A schematic diagram of the RNase A molecule with U5P (yellow) and UDP (cyan)molecules superimposed bound at the active site. The molecular surface andsecondary structure of the enzyme are also shown. Subsites P 0  (Lys66), B 2  (Asn67,Gln69, Asn71, Glu111, His119), P 1  (Gln11, His12, Lys41, His119), B 1  (Val43,Asn44, Thr45, Phe120, Ser123) and P 2  (Lys7, Arg10) are labelled and marked onthe molecular surface with different colours.  structural communications Acta Cryst.  (2009). F 65 , 671–677 Tsirkone  et al.   Ribonuclease A  673 Figure 2 The    A  2| F  o |    | F  c | electron-density map calculated from the RNase A model before incorporating the coordinates of U5P ( a ) or UDP ( b ) is contoured at the 1.0    level andthe refined structure of both the inhibitor molecules in the active site is shown. The numbering scheme used for each inhibitor molecule is also shown. Diagrams of theinteractions between RNase A and U5P in mol  A  ( c ) and mol  B  ( d ) and UDP in mol  A  ( e ) and mol  B  (  f  ) in the active site are shown. A standard colouring scheme is used(yellow for carbon, blue for nitrogen, red for oxygen and cyan for water molecules) and hydrogen-bond interactions are represented as dashed lines. C atoms in the inhibitormolecules are shown in grey.  The program  PROCHECK   (Laskowski  et al. , 1993) was used toassess the quality of the final structures. Analysis of the Ramachan-dran ( ’ –   ) plot showed that all residues lay in the allowed regions.Solvent-accessible areas were calculated by the program  NACCESS (Hubbard & Thornton, 1993). The atomic coordinates and structurefactors of the two complexes have been deposited in the Protein DataBank (http://www.pdb.org) with accession numbers 3dxg and 3dxh.Figures were prepared with the program  PyMOL  (DeLano, 2002). 3. Results and discussion Both ligands are competitive inhibitors of the enzyme with respect toC>p. U5P is a moderate inhibitor ( K  i  = 4.00    0.41 m M  ), while UDPis more potent ( K  i  = 0.65    0.06 m M  ). The complex structures arevery similar to that of the free RNase A (Leonidas  et al. , 2006) andthe binding of the inhibitors did not cause any significant confor-mational change.In the monoclinic crystals of RNase A there are two proteinmolecules in the crystallographic asymmetric unit (Leonidas  et al. ,2006). Two ligand molecules were found bound in the active site of the first protein molecule of the asymmetric unit, but only one wasfound in the second. This can be attributed, as previously (Leonidas et al. , 2006; Hatzopoulos  et al. , 2005), to the impediments imposed bythe crystal lattice that limit access to the active site of the secondprotein molecule in this crystal form. Our structural analysis wasbased on the ligand complex with the first RNase A molecule. All theatoms of U5P and UDP are well defined in the electron-density mapof the protein complexes (Figs. 2 a  and 2 b ).Upon binding to RNase A, each inhibitor molecule adopts adifferent conformation. However, the glycosyl torsion angle   0 inU5P and UDP adopts the frequently observed (Moodie & Thornton,1993)  anti  conformation. The ribose of U5P adopts the two mostpreferred conformations for free and protein-bound nucleotides(Moodie & Thornton, 1993): C3 0 - endo  and C2 0 - endo . The rest of thebackbone torsion angles are in the common range for protein-boundpyrimidines (Moodie & Thornton, 1993). In UDP, the ribose of thethree inhibitor molecules adopts the C1 0 - exo , C2 0 - endo  and C3 0 - endo puckering, respectively, which together with the rest of the backboneand phosphate torsion angles are also in the preferred range forprotein-bound pyrimidines (Moodie & Thornton, 1993).The two inhibitors bind at the active site with one molecule insubsite B 1  and the other in subsite B 2  (referred to hereafter as mol  A and mol  B , respectively). In the U5P complex the uracil engages inhydrogen-bond interactions with Thr45 at subsite B 1  (Table 2), theresidue that is responsible for the pyrimidine specificity of this site(Raines, 1998), while the rest of the molecule is involved in van derWaals interactions, mainly with His12 and Phe120 (Fig. 2 c ). The5 0 -phosphate group moves away from subsite P 1  towards P 0  and theclosest distance between the phosphate and the side chain of Lys66(the sole component of subsite P 0 ; Raines, 1998) is 4.9 A˚. U5P mol  B is bound with the uracil ring almost parallel to the side chain of His119 (Fig. 2 d ) and is involved in hydrogen bonding and van derWaals interactions with all residues of subsite B 2  (Table 2). The5 0 -phosphate group binds at P 1  and is hydrogen bonded to His119(Fig. 2 d ). In addition to the interactions between the ligands and theprotein, the two ligands also interact with each other. Thus, the5 0 -phosphate group of U5P mol  B  forms a hydrogen bond to the3 0 -hydroxyl group of the ribose of U5P mol  A , while two watermolecules mediate polar interactions between the 5 0 -phosphate of U5P mol  B  and the 2 0 - and 3 0 -hydroxyl groups of the ribose of U5Pmol  A . Upon binding to RNase A, the two U5P molecules displace 13water molecules from the active site of the unliganded enzyme(Leonidas  et al. , 2006). The two ligand molecules have a total solvent-accessible surface of 829 A˚  2 , which shrinks to 326 A˚  2 upon binding toRNase A. Polar and nonpolar groups contribute almost equally to theburied surface (233 and 270 A˚  2 , respectively).On binding, the two UDP molecules displace 15 water moleculesfrom the active site of the free enzyme (Leonidas  et al. , 2006) and, likeU5P, anchor one uracil to subsite B 2  (Fig. 2 e ) and another to B 1 (Fig. 2  f  ). The   -phosphate rather than the   -phosphate group of the5 0 -pyrophosphate group of UDP mol  B  binds to subsite P 1 , forminghydrogen-bond interactions with the side chain of His119 (Fig. 2 e ).This phosphate group also engages in hydrogen bonds to the twohydroxyl groups of the ribose of UDP mol  A , similarly to U5P(Table 2). The 5 0 -pyrophosphate of inhibitor mol  A  projects towards structural communications 674  Tsirkone  et al.   Ribonuclease A  Acta Cryst.  (2009). F 65 , 671–677 Table 2 Potential hydrogen bonds of U5P and UDP with RNase A in the crystal. Values in parentheses are distances in A˚. Asterisks indicate residues from a symmetry-related protein molecule.RNase A–U5P complex RNase A–UDP complexU5P/UDP atom Mol  A  Mol  B  Mol  A  Mol  B  Mol  C  O2 Thr45 N (2.9) Asn71 N  2 (2.9) Thr45 N (2.9) Asn71 N  2 (2.8) Asn34 N  2 (2.7)O2 Glu111 O " 2 (3.2) Gln69 N " 2 (3.3)O4 Water7 (3.2) Asn67 O  1 (3.2) Water35 (2.9) Asn67 O  1 (3.1) Water108 (2.7)O4 Water43 (2.8) Asn67 N  2 (3.4) Water49 (3.3) Asn67 N  2 (3.1)O4 Water15 (2.8) Water10 (2.7)N3 Thr45 O   1 (2.8) Gln69 O " 1 (3.4) Thr45 O   1 (2.7) Tyr76 O  * (2.7)O2 0 His12 N " 2 (3.3) His12 N " 2 (3.2) Arg10 N  1 (3.2)O2 0 Water20 (3.1) Lys41 N   (3.0)O2 0 Water207 (3.0) Water371 (3.0)O2 0 Water215 (2.8)O3 0 Water207 (2.7) Glu111 O " 1 (3.1)O3 0 Water286 (2.7) Water230 (3.0)O3 0 Water364 (3.2)O4 0 Water38 (3.2) Water20 (3.2) Glu2 O " 1 (3.2)O5 0 Water146 (3.3)O1A Water146 (2.6) Water263 (2.7) Water355 (2.8)O1P/O1B Water175 (2.8) Water286 (3.0) Water221 (2.9)O2P/O2B Water236 (2.9) Water207 (3.0) His119 N  1 (2.9)O2P/O2B Water244 (2.8) Water371 (2.7)O3P/O3B His119 N  1 (2.5) Water221 (2.9)O3P/O3B Water120 (3.4) Water20 (2.7)O3P/O3B Water236 (3.4) Water207 (3.2)  the solvent (the closest RNase A residue is Lys66, which is 5.6 A˚away). Upon binding to RNase A, the UDP molecules becomeburied. Thus, the combined solvent-accessible surface of the two freeligand molecules is 899 A˚  2 . When bound, this molecular surfaceshrinks to 361 A˚  2 , indicating that 60% of the UDP surface becomesburied. The greatest contribution comes from the nonpolar groups,which contribute 381 A˚  2 (71%) of the surface that becomes in-accessible. The shape-correlation statistic Sc, which is used to quan-tify the shape complementarity of interfaces and gives an idea of the‘goodness of fit’ between two surfaces (Lawrence & Colman, 1993), is0.67 and 0.70 for the combined molecular surface of the two U5Pmolecules and the two UDP molecules, respectively.A third UDP molecule was found to be bound in a location closeto the N-terminus of RNase A (Fig. 3). There, it participates in ahydrogen-bond network of interactions with residues from bothRNase A molecules of the asymmetric unit as well as with residuesfrom a symmetry-related protein molecule (Table 2). X-ray diffrac-tion data collected from RNase A crystals soaked with 20 m M   UDPdid not show any ligand binding at this location, indicating that thisbinding is nonspecific and can be attributed to the high concentrationof UDP (50 m M  ) used for soaking the RNase A crystals.Crystallographic data from RNase A crystals soaked with 5 m M   of either U5P or UDP for 1 h showed only one inhibitor molecule (mol  A ) bound at the active site. This indicated that the inhibitory effect of U5P and UDP in solution probably arises from the binding of mol  A at the active site. Both uridylyl compounds bind similarly at the activesite (Fig. 4 a ). The differences in the potency of UDP and U5P mightderive from a combination of positive factors associated with theaddition of the extra phosphate group (an enthalpic gain arising fromnew van der Waals interactions between the   -phosphate and resi-dues of the P 1  subsite and the entropic advantage of the increasedstructural constraints on the   -phosphate) partially counterbalancedby enthalpic losses arising from this same conformational restrictionof the   -phosphate, which would prevent this group from optimizingits interactions as in the U5P complex.U-2 0 -p and U-3 0 -p are potent inhibitors of RNase A, with  K  i  valuesof 7 and 82  m M  , respectively (Anderson  et al. , 1968). Superposition of the U5P and UDP complexes onto the U-2 0 -p or U-3 0 -p complexes(Leonidas  etal. , 2003) reveals that the uracil moieties bind similarly inall four complexes at subsite B 1  (Fig. 4 b  and 4 c ). However, while inthe U-2 0 -p and U-3 0 -p complexes the phosphate group binds at P 1 , inthe U5P and UDP complexes it binds away from this site (Figs. 4 b and4 c ). Instead, it is the phosphate group of the U5P or UDP moleculethat binds at subsite B 2  that binds close to P 1 . Interestingly, althoughthe structures of the RNase A complexes with either U-2 0 -p or U-3 0 -pwere determined from protein crystals soaked in a solution con-taining 50 m M   of each inhibitor for several hours (Leonidas  et al. ,2003), only one inhibitor molecule was found to be bound at theactive site. Therefore, the question raised from the present structuralstudy is why U-2 0 -p and U-3 0 -p are more potent inhibitors than U5Pand UDP when two molecules of the latter instead of one moleculebind at the active site. The answer may lie in the number of inter-actions of the phosphate group at subsite P 1 . In the U-2 0 -p and U-3 0 -pcomplexes this group forms many more interactions than in the U5Pand UDP complexes. Thus, while in the U-2 0 -p and U-3 0 -p complexesit forms hydrogen-bond interactions with Gln11, His12, Lys41, His119and Phe120 (Leonidas  et al. , 2003), in the U5P and UDP complexes itonly forms a hydrogen bond to His119. Therefore, it seems that theseadditional hydrogen-bond interactions in the U-2 0 -p and U-3 0 -pcomplexes counterbalance the interactions of the second uracil atsubsite B 2  in the U5P and UDP complexes. Differences in thepotency between U-2 0 -p and U-3 0 -p or 2 0 -CMP and 3 0 -CMP (U-2 0 -pand 2 0 -CMP have a tenfold smaller  K  i  for RNase A than U-3 0 -p and3 0 -CMP, respectively; Anderson  et al. , 1968) have similarly beenattributed to differences in the phosphate binding at subsite P 1 (Leonidas  et al. , 2003; Howlin  et al. , 1987; Zegers  et al. , 1994).A structural comparison of the UDP binding to that of pdUppA-3 0 -p (the most potent ribonucleolytic inhibitor found to date) revealsthat the two ligands follow a similar binding pattern and the uracilring of UDP at B 2  superimposes onto the adenine ring of pdUppA-3 0 -p (Fig. 4 d ). The pyrophosphate group of UDP does not bind at thesame location as the analogous group of pdUppA-3 0 -p, but this isprobably a consequence of steric impediments since it is not cova-lently bound to the second inhibitor molecule at the active site of theenzyme as is the case for pdUppA-3 0 -p. However, it is still the  -phosphate rather than the   -phosphate that binds at P 1 . Kinetic structural communications Acta Cryst.  (2009). F 65 , 671–677 Tsirkone  et al.   Ribonuclease A  675 Figure 3 Stereo diagram of the interactions of UDP bound at the interface of the two RNase A molecules of the crystallographic asymmetric unit. Residues labelled with asterisks arefrom a symmetry-related molecule.
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks