sign in sign up
<<
Home , Alclofenac, Cyclooxygenase

Molecular basis of (COXs) selective inhibition

Vittorio Limongellia,b,1, Massimiliano Bonomia, Luciana Marinellib, Francesco Luigi Gervasioc, Andrea Cavallid,e, Ettore Novellinob, and Michele Parrinelloa,1

aComputational Science, Department of Chemistry and Applied Biosciences, Eidgenössiche Technische Hochschule (ETH) Zürich, Università della Svizzera Italiana Campus, Via Giuseppe Buffi 13, CH-6900 Lugano, Switzerland; bDipartimento di Chimica Farmaceutica e Tossicologica, Università degli Studi di Napoli “Federico II”, Via D. Montesano, 49, I-80131 Naples, Italy; cComputational Biophysics Group, Spanish National Cancer Research Center (CNIO), calle Melchor Fernández Almagro, 3, E-28029 Madrid, Spain; dDepartment of Pharmaceutical Science, University of Bologna, Via Belmerolo 6, I-40126, Bologna, Italy; and eDepartment of Drug Discovery and Development, Italian Institute of Technology, Via Morego 30, I-16163, Genoa, Italy

Edited by Michael L. Klein, University of Pennsylvania, Philadelphia, PA, and approved January 25, 2010 (received for review November 20, 2009)

The widely used nonsteroidal anti-inflammatory drugs block the and in recent years, much effort has been paid in elucidating the cyclooxygenase enzymes (COXs) and are clinically used for the dynamic binding mechanism of COX inhibitors (7–9). Lanzo et treatment of , , and cancers. A selective inhibition al. (8) have found, by means of the fluorescence quenching tech- of the different isoforms, particularly COX-2, is desirable, and nique, that while the association of the COX-2 selective inhibitor consequently a deeper understanding of the molecular basis of SC-299 with COX-1 and COX-2 occurs at similar rates, the dis- selective inhibition is of great demand. Using an advanced compu- sociation of SC-299 from COX-2 takes hours, which is thousands- tational technique we have simulated the full dissociation process fold slower than in COX-1 (≈30 s). This finding clearly shows that of a highly potent and selective inhibitor, SC-558, in both COX-1 there is a correlation between the relative rates of dissociation and COX-2. We have found a previously unreported alternative and the selectivity of the isoenzyme inhibition, and this correla- binding mode in COX-2 explaining the time-dependent inhibition tion is confirmed also in additional experiments on other COX-2 exhibited by this class of inhibitors and consequently their long selective inhibitors (9). These phenomena have been so far as- residence time inside this isoform. Our metadynamics-based cribed to a more stable binding mode of selective inhibitors into approach allows us to illuminate the highly dynamical character the COX-2 isoform due to the presence of diverse residues such of the / recognition process, thus explaining a wealth as Val434 in COX-2 instead of in COX-1. However, of experimental data and paving the way to an innovative strategy this represents a very simplistic way to address the different rates for designing new COX inhibitors with tuned selectivity. of dissociation of an inhibitor from an , and very likely other molecular mechanisms might be involved (8–11). nonsteroidal anti-inflammatory drugs ∣ COX selectivity ∣ coxibs ∣ On the other hand, it has to be pointed out that processes that path collective variables ∣ metadynamics take from microseconds to hundreds of seconds, such as binding and unbinding of ligands into protein, are impossible to simu- onsteroidal anti-inflammatory drugs (NSAIDs) are widely late with standard computational techniques such as molecular Nused as therapeutic agents for the treatment of pain and in- dynamics (MD), whose typical timescale is hundreds of nanose- flammation, and in addition several evidences have been very re- conds. Thus, computer simulation of the binding and especially of cently reported on their chemopreventive effect in colorectal the unbinding, which for COX-2 selective inhibitors is responsible cancer (1). Their is based on the blockage of the differences in the rates of dissociation, is an ongoing of the cyclooxygenase enzymes (COXs) (2) by sterically hindering challenge we decided to deal with. the entrance of the physiological binder . The Thus, to reveal at an atomic level what might happen during classical NSAIDs such as , , or the complex formation and dissociation and with the aim to over- BIOPHYSICS AND are nonselective and inhibit indifferently all the COXs isoforms. come the large free-energy barriers toward the inhibitor-enzyme In the last few decades, the interest of scientists has been mostly undocking process in an affordable computational time, we used COMPUTATIONAL BIOLOGY focused on a selective inhibition between COX-1/COX-2. In fact, the recently developed well-tempered metadynamics (12), which the inhibition of COX-1, particularly in the gastrointestinal sys- is an evolution of standard metadynamics (13). For this purpose, tem, may lead to dangerous side effects such as ulcers. As a one of the most potent and selective COX-2 inhibitors, SC-558 consequence, a selective inhibition of COX-2 has been sought (IC50 ¼ 9.3 nM) (6), was selected. for decades, and recently a new generation of NSAIDs, namely We have found a previously unreported alternative binding coxibs, was found. However, some of them, such as mode of SC-558 into COX-2, which is very similar to that experi- (Vioxx®), have been withdrawn from the market due to their car- mentally found for some nonselective COX inhibitors (14, 15). diotoxicity (3–5). This was not the case of other COX-2 selective This has important consequences on the classical way of perform- drugs such as and , with the latter one largely ing rational , which is mainly based on understanding used as an anti-inflammatory agent in many diseases. As a con- only the main binding mode. Here we show that in so doing one sequence, today, in the rational design of new COXs binders, might miss crucial bits of information and one cannot compare medicinal chemists have to pay attention to the selectivity proper- the results of modeling with the experiments. In fact, our model ties of the designed drugs, and a fine-tuning of the COX selec- of ligand binding allows one to rationalize many previously tivity profile might be necessary to generate novel effective drugs with reduced side effects. In the rational modulation of the bin- Author contributions: V.L., L.M., F.L.G., A.C., E.N., and M.P. designed research; V.L. ders selectivity precious help comes from the x-ray crystallogra- performed research; M.B. contributed new reagents/analytic tools; V.L., M.B., and M.P. phy. In fact, the crystallographic structures show that selective analyzed data; and V.L. and M.P. wrote the paper. and nonselective inhibitors generally bind in two different pat- The authors declare no conflict of interest. terns. In COX-1, the space of the selectivity pocket is reduced This article is a PNAS Direct Submission. due to the presence of Ile523, while in COX-2 the presence of 1To whom correspondence may be addressed. E-mail: [email protected]. Val523 augments the available space providing a more stable ethz.ch or [email protected]. binding possibility for selective inhibitors (6). The kinetics of This article contains supporting information online at www.pnas.org/cgi/content/full/ selective and nonselective inhibitors are also different in general, 0913377107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0913377107 PNAS ∣ March 23, 2010 ∣ vol. 107 ∣ no. 12 ∣ 5411–5416 Downloaded by guest on September 28, 2021 unexplained experimental results, for instance, the key role hypothesis that the group of helices A–D forms the door through played by some residues in the binding path of the ligand as well which the arachidonic acid and other ligands pass (14). as the kinetic models for the binding mechanism of closely related This is a slow mode that needs to be sampled in a biased MD by analogues of SC-558 (8, 9, 16). Additionally, our study highlights the introduction of an appropriate CV. In order to address this the importance of the three alpha-helices’ flexibility at the effect we use a path collective variable (21) constructed with entrance of the enzyme, in line with what has been discussed the contact map (22) between the residues that determine the in literature (15). Finally, we contrast the behavior of COX-2 with gate flexibility (see Methods and SI Text). In addition we use a that of COX-1, in which only one binding mode for SC-558 has distance and a dihedral angle CV to identify the position and been found. orientation of the ligand relative to the enzyme (SI Text). This last set of CVs has been already successfully used in several me- Results tadynamics simulations of undocking processes of small ligands Biased MD Simulations. The well-tempered metadynamics is an from their binding sites (23, 24). The final choice of CV is the evolution of standard metadynamics (13), able of enhancing the result of a lengthy investigation reported in SI Text. sampling and reconstructing the free-energy profile of the pro- cess of interest by adding an adaptive bias on a selected number SC-558 Dissociation Process in COX-2. Under the action of metady- of collective variables (CVs) (12). The user defined CVs must be namics, the ligand leaves the starting position, which corresponds able to discriminate the initial and final states of the system and to the x-ray structure of the SC-558/COX-2 complex (PDB ID take into account all the slow modes of the process (17–19). One code 1cx2) (6), and explores the whole and finally of the very first problems encountered in our undocking study, takes its way out from the enzyme through the helices gate. which has been also one of the major topics in COX inhibition, We describe in detail the relevant minima found along the exiting was the identification of the part of the protein that allows COX path in the following paragraphs. binders to exit from the inner cyclooxygenase site. On the basis of apo and ligated x-ray structures of COX-2, it was stated that the The crystallographic pose. The first energy minimum, basin A in the enzyme can assume a relaxed and a tightened state (SI Text) free-energy surface (FES) depicted in Fig. 1, is the deepest and where Arg120 is free to move or makes a salt bridge with the corresponds to the x-ray conformation. This means that basin A Glu524, respectively (16). In the latter case the substrate is locked represents the most energetically stable pose where the ligand has into a catalytically competent conformation. Moreover, the evi- the highest probability to be found once docked in agreement dence that, when complexed with certain inhibitors, the COX-2 with the crystallographic structure. The bromophenyl ring is x-ray structures show some changes in the Cα position at the base placed in a hydrophobic cavity surrounded by Phe381, Leu384, of one of the helices A–D (20) led experimentalists to make the Tyr385, Trp387, Phe518, and Ser530. Also the Gly526 and

Fig. 1. The FES of the dissociation process as a function of the distance and dihedral CVs is shown at the bottom using isosurfaces of 2 kcal∕mol. The four main energy basins A–D found during the metadynamics simulations are highlighted in the FES graph. The four snapshots of the complex SC-558/COX-2 displayed in the surrounding boxes represent the following binding poses: basin A, crystallographic pose; basin B, alternative pose; basin C, poses at gate site; basin D, external pose. The ligand and the main interacting residues are displayed as licorice, while the protein is represented as a green cartoon with the α-helices forming the gate colored in orange. The interacting waters are shown as spheres, while hydrogens are not displayed for clarity. On the basis of the information derived from the FES, many aspects of the complex inhibition kinetics of COX-2 selective inhibitors can be elucidated. In fact, the presence of the second deep minimum (B) can explain the time-dependent behavior of many COX-2 selective inhibitors and, consequently, their long residence time inside the protein. In this respect we have been encouraged by the interpretation of flourescence quenching experiments of Lanzo et al., who have suggested for a SC-558 analogue an in and out movement from the selectivity pocket (8). They proposed that while in COX-1 the binding is a two-step process (see Fig. 4), in COX-2 a further not yet identified step takes place. Our calculations reveal the nature of this additional step, which considerably slows the off rate.

5412 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0913377107 Limongelli et al. Downloaded by guest on September 28, 2021 Ala527 backbone takes part in these hydrophobic contacts. The ligand/protein interactions conserved and with comparable trifluoromethyl moiety resides in a close pocket surrounded by average rmsd values calculated for the ligand heavy atoms Met113, Val116, Tyr355, Leu359, and Leu531 (Fig. 1). This is [1.07 (parmff99SB) vs. 1.06 (parm99)]. referred to as the common pocket in Fig. 2 since it is the same In view of these data, it is natural to suggest that this unique that hosts the aromatic ring bearing the carboxylate function of pose can provide an alternative way for SC-558 to bind to COX-2. many COX nonselective compounds such as ibuprofen. Finally, Several interactions conspire to make this pose stable. The the phenylsulphonamide moiety inside the selectivity pocket as- bromophenyl moiety is in the highly hydrophobic cage defined sumes a conformation in which one of the oxygen atoms H-bonds by Ile345, Val349, Leu359, Leu531, and Met535, while the tri- with Arg513 and is close enough to interact with His90 while the fluoromethylpyrazole occupies approximately the same cavity other oxygen forms a H-bond with a water molecule. The amide as the crystallized pose but is rotated by 180°. In such a way, it hydrogens of the sulphonamide group interact with the backbone improves its interactions with neighboring Leu352, Phe518, of Phe518 via two water bridges. Unfortunately, x-rays have not Val523, Gly526, and Ala527. Finally, the sulphonamide group en- been able to resolve the conformation of the sulphonamide group gages a bifurcated H-bond with Tyr355 and Arg120 side chains, in the selectivity pocket, and conformations dissimilar to the and additional interaction energy can be gained from the relative x-rays’ one have been already reported in previous theoretical closeness of the Val116 carbonyl group (Fig. 1). studies (25, 26). The involvement of residues such as Arg120 and Tyr355 in the ligand binding assumes an important value in the light of experi- The alternative pose. Under the action of metadynamics, the ligand ments (15, 16). In line with our results that demonstrate the role leaves the crystallographic pose and while exploring the catalytic of the polar interactions between SC-558 and the Tyr355 hydroxyl site it finds another minimum (SI Text). The presence of this sec- group, the mutation Tyr355Phe disfavors the binding of many ond minimum (basin B in Fig. 1) was a surprise and reveals the ligands to COX (16). Moreover, Kurumbail et al. (6) have ob- presence of a pose of great interest for several reasons. Such a served in many x-ray data that Arg120 is displaced from its usual pose was also found in metadynamics where different CV settings position in the ligated enzyme. This is due to a weakening of the have been used (SI Text). First of all, the depth of this second Arg120-Glu524 bond relative to the free form of the enzyme as basin suggests a good thermodynamic stability of the ligand at also reflected by the large B factor. This has led these authors to the site. We confirmed the stability of this alternative pose by suggest a temporary interaction of this residue with the ligand carrying out a 5-ns standard MD run. During this simulation, during its mechanism of biding to the enzyme. the complex was very stable with an average rmsd of the ligand An even stronger support to our suggestion that pose B is heavy atoms of only 1.06 Å. We also measured an rmsd of 2.24 Å highly relevant comes from the comparison to the crystal struc- for the heavy atoms of the residues that interact with the ligand, tures of COX complexed with several nonselective inhibitors. For with respect to their x-ray coordinates (SI Text). This large value instance, comparing our unique pose of SC-558 into COX-2 to reflects the fact that the ligand induces local conformational the binding mode of ibuprofen to COX-1 (PDB ID code 1eqg) changes in the protein. It is important to note that standard (15), it clearly emerges that the main interactions with the protein docking programs such as AutoDock (27, 28), which assume are well conserved. In fact, the carboxylate group of ibuprofen the protein to be rigid, fail to predict this second binding pose, takes part in a network of polar interactions involving Tyr355 as we have explicitly tested. In order to check if basin B is an and Arg120 similarly to what happens to the sulphonamide moi- artifact of the potential applied we used a different force ety (Fig. 3). In either case, the common pocket is occupied by field (parmff99SB) to validate this second minimum through a groups that are similar in size, the phenyl in the case of ibuprofen standard MD run of over 10 ns. The results are very close to those and the pyrazole in SC-558. The similarity is even greater if in- obtained using the parm99 force field with all the principal hibitors such as flurbiprofen or are considered, where a halogen atom is substituted in the phenyl ring, thus enforcing the hydrophobic interactions with Leu352, Phe518, and Val523. It can be seen that a similar role is played by the trifluoromethyl BIOPHYSICS AND group of SC-558 in pose B. The existence of two possible binding modes is fully compati- COMPUTATIONAL BIOLOGY ble with the presence of an additional binding pose for many diarylheterocycles, which has been invoked by experimentalists to explain the time-dependent inhibition exhibited by this class of inhibitors (7–9). In fact, the slow tight-binding inhibition of compounds chemically similar to SC-558 is interpreted by these authors as due to the presence of an additional binding step, and

Fig. 2. Schematic representation of the main binding sites found during our metadynamics simulations. The common and the selectivity pocket represent the SC-558 binding site in the crystallographic pose (basin A in Fig. 1), while Fig. 3. Comparison between (A) the alternative binding pose of SC-558 in the same common pocket with the side pocket represents the site for the COX-2 found during metadynamics simulations and (B) the x-ray binding alternative pose (basin B in Fig. 1). The gate site (basin C in Fig. 1) is when conformation of ibuprofen in complex with COX-1 (PDB ID code 1eqg). the ligand is in proximity of the protein gate assuming several similar The ligands and the interacting residues are represented as licorice, while conformations. Finally, the lobby-like site represents the binding site of the protein is represented as green cartoon with the α-helices forming the SC-558 in its external pose (basin D in Fig. 1). gate colored in orange. The hydrogens are not displayed for clarity.

Limongelli et al. PNAS ∣ March 23, 2010 ∣ vol. 107 ∣ no. 12 ∣ 5413 Downloaded by guest on September 28, 2021 in this sense a crucial role might be played by the rearrangement Interestingly, the FES exhibits only one energy minimum, and of the hydrogen bonding network formed by residues such as consequently a unique binding mode is found in the inner part Arg120, Tyr355, and Glu524, which are critical for the transition of the cyclooxygenase site (Fig. 4). This corresponds to the most from the relaxed to the tightened state of the enzyme (SI Text). stable pose predicted by AutoDock (27, 28) and also to the start- The involvement of these key residues in the newly binding mode ing point of our simulations. Interestingly, this conformation is of SC-558 suggests that the time-dependent inhibition kinetic of very similar to the crystallographic pose of SC-558 in COX-2, SC-558 results from the ability of the ligand to bind in two distinct although here the ligand is more weakly bound due to an only but equally strong ways. One can thus understand why COX-1 is partial insertion of the sulphonamide moiety in the selectivity resistant to the time-dependent inhibition of diarylheterocycles pocket, at variance with what happens to basin A of COX-2. This (6). In fact, the mutation in this region of Val523 in COX-2 to can be explained by the presence in COX-1 of the bulkier Ile523 Ile523 in COX-1 reduces the space availability in COX-1 and in place of Val523 in COX-2 (29). The similarity of the inner in this isoform the transfer of SC-558 from the original crystal- poses found in COX-1 and COX-2 is supported by the experimen- lographic pose to the new one becomes more difficult. tal fluorescence quenching data indicating that SC-299, a close analogue of SC-558, occupies a very similar deep position in Poses at the gate site. Once all the minima inside the are COX-1 and COX-2 (8). Moreover, the presence of the bulkier filled, the ligand points toward the B–D helices facing the internal Ile523 in COX-1 might also be responsible for the absence in this part of the helices that form the exit door of the enzyme (SI Text). isoform of the alternative binding pose found in COX-2, since the Here, it can assume a variety of closely related conformations sta- Ile side chain sensitively reduces the space available in that region bilized by several hydrophobic interactions in the site upper part to host a group such as the trifluoromethyl of SC-558 (Fig. 3). with residues such as Ile91, Val98, Trp99, Val102, Ile112, Tyr115, Comparing COX-1 and COX-2 one sees that not only the Val116, Tyr355, and Phe357 (basin C in Fig. 1). Most conforma- alternative binding site is uniquely present in the latter but that tions belonging to this site are characterized by the presence of a the accessible free-energy surface inside the binding pocket is salt bridge between Arg120 and Glu524. Moreover, while the sul- much larger. This explains the COX-2 selectivity of this com- phonamide group of SC-558 engages stabilizing interactions with pound and the higher residence time in COX-2 found for com- the surrounding water molecules, the phenyl ring of the sulpho- pounds similar to ours (8). In a study on a series of COX-2 namidephenyl moiety forms a π-π interaction with Tyr355. This selective inhibitors (9), the authors have advocated the possible corresponds to a local minimum whose stability was checked existence of an additional binding step to explain their time- in 5 ns of nonbiased MD. During this time all these interactions dependent inhibition kinetics. Our study validates their hypoth- were conserved as well as the salt bridge between Arg120 and esis and offers a microscopic description of this site. Glu524, which keeps the protein in the tightened conformation. Discussion The clinical use of COX inhibitors necessitates that their binding The external pose. Once the ligand has crossed the helices forming the gate of the enzyme, it reaches the outer part of the protein, mode at COX and, above all, their kinetics of inhibition are where a further local minimum is found (SI Text). It is worth not- established, not only to fully understand their mechanism of ing that this crossing event happens only when the following three inhibition, but also to provide precious insight for the rational phenomena occur concurrently: (i)theB–D helices breath, improvements of specificity and inhibitory potency. (ii) the gate formed by Tyr115 opens, and (iii)SC-558formsa The full unbinding process of the potent and highly selective double H-bond with Tyr355 and Arg120. This can be clearly seen COX-2 inhibitor, SC-558, simulated by using metadynamics, re- in our simulations approximately at 50 ns when the contact map vealed the presence of a binding mode alternative to that experi- CV has its maximum value for a relative large time interval mentally found in COX-2. The interest for the newly identified (SI Text). At the external site, although the ligand is mostly sol- pose increases when a comparison with the x-ray binding confor- vated by water molecules, it continues interacting with a small mation of some COX nonselective inhibitors is made. In fact, number of surfacing hydrophobic residues such as Val98, comparing this pose with the biding mode of ibuprofen or flur- Ile101, Val102, Ile105, Phe107, and Leu108 and with the biprofen, it emerges clearly that the overall interactions with the His94 side chain through a H-bond (basin D in Fig. 1). Once surrounding residues are conserved and, moreover, SC-558 again we checked the stability of this conformation via a 5-ns stan- engages additional hydrophobic contacts in a side pocket that dard MD run (SI Text). This preliminary binding site just outside further stabilize the newly found minimum. Many experimental theenzyme(lobby-likesiteinFig.2),intheupperpartof the door formed by the B–D helices, plays a pivotal role in the inhibitor/enzyme recognition process. Our hypothesis is complementary to the flourescence quenching experiments of Lanzoetal.,whosuggestedforaSC-558analogueaninand out movement from the selectivity pocket. They proposed that while in COX-1 the binding is a two-step process, in COX-2 a further not yet identified step takes place (8). Our calculations reveal the nature of this additional step, which is the reason of the slower dissociation rate of SC-558 from COX-2. The part of the FES with the ligand inside the cavity is quan- titatively well characterized and the free-energy differences are converged, while the determination of the relative energy of pose D with respect to the others is less accurate. In fact, while we have observed many recross events among A, B, and C, once out the Fig. 4. The FES of the undocking process of SC-558 in COX-1 as a function of 2 ∕ ligand was not able to return inside. Thus, the free-energy differ- the distance and dihedral CVs using isosurfaces of kcal mol. Only an inter- ence between the bound (basin A) and the unbound state (basin nal and an external energy minimum have been found. Once this inner mini- ≈−8 7 ∕ mum has been filled, SC-558 moves in the direction of the helices that form D) of . kcal mol has a semiquantitative value. the exit door. Close to the minimum a shallow wide basin is to be found whose conformation differs very slightly from that of the lowest minimum. SC-558 Dissociation Process in COX- 1. Similar CVs have been When the ligand is in between the two helices and the gate is in the open used to study the undocking of SC-558 from COX-1 (SI Text). conformation, it leaves the binding site moving to the external pose.

5414 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0913377107 Limongelli et al. Downloaded by guest on September 28, 2021 data have suggested that some of the residues involved in the The Path Collective Variables. In order to determine the appropriate collective alternative binding pose of SC-558 play a key role in regulating variables to describe the opening and closing of the three helices, we per- the inhibition kinetics of many COX inhibitors (15, 16). As stated formed a preliminary set of calculations using the path CVs of Branduardi et al. (21), represented in the space of the contact map variable (CMAP) in- in the Introduction, we think that the inhibition kinetics and the troduced by Bonomi et al. (22). The path method is extremely powerful selective behavior of COX inhibitors cannot be explained only whenever one wants to study a transition between states A and B. As state in terms of the presence of different crucial residues such as Va- A we chose the x-ray conformation of the enzymes in which all the contacts l434Ile, Arg513His, and Val523Ile, in the different isoforms. On contribute to have the enzymes in the closed state. State B represents the the contrary, our results clearly show the ability of SC-558 to bind open form of the enzymes, and, due to the lack of experimental information, COX-2 in two different ways, supporting the theory of the pre- it was derived by first preliminary metadynamics trials. In these runs, carried sence of an additional binding pose for many diarylheterocycles, out without path CVs, the ligand, pushed by the added bias, induced the opening of the helices and of the gates. Certainly, this is not the optimal and consequently provide an explanation for the increased time way to choose the open conformation of the enzymes, but the path method of permanence and the slow binding rate in COX-2 for ligands itself, when used together with metadynamics, is able to find transition paths structurally similar to SC-558. rather different from the initial one. Our results will be very helpful in the drug design using either Let SðRÞ be a reduced representation of a generic configuration R.Ifthe automated techniques such as virtual screening, targeting both choice of S is appropriate, we would expect the reactive trajectories to be the x-ray and the alternative binding site, or using the classical bundled in a narrow tube around the path. To trace this path, we follow sð Þ rational drug design. In fact, modifications of the chemical struc- the procedure of Branduardi et al. (21) introducing the two variables R and zðRÞ: ture of SC-558 can either improve or reduce the contacts with the residues present in either of the two possible binding sites. This p −λ‖SðRÞ−SðlÞ‖2 1 ðl − 1Þe would tune the time of permanence in the protein and conse- sðRÞ¼ ∑l¼1 ; quently the kinetic profile and selectivity of the newly designed P − 1 p e−λ‖SðRÞ−SðlÞ‖2 ∑l¼1 compounds. For instance, the addition of a polar substituent such as at position 3 of the pyrazole ring in SC-558 0 1 would improve the selectivity and the affinity for the alternative 1 P binding site in COX-2, engaging H-bond interactions with the zðRÞ¼− @ e−λ‖SðRÞ−SðlÞ‖2 A; λ ln ∑ close Arg513 side chain, present univocally in this isoform. Basing l¼1 the development of the newly designed compounds on the dyna- mical features of the ligand binding mechanism represents an in- where P is the number of frames and sðRÞ and zðRÞ measure the intercept and novative way of performing drug design that we are successfully distance of any microscopic configuration R from the path SðlÞ. We described experiencing designing new inhibitors that have shown promising the transition between the closed and open conformations of the enzyme COX-2 selective behavior in the preliminary results. This ap- using five frames Sð1Þ…Sð5Þ, with Sð1Þ and Sð5Þ being the closed (state A) proach assumes an even greater value if specific properties of and the open conformation (state B), respectively. Care must be taken that all the frames are equally spaced relative to the metric used. The λ value was these compounds, such as their high selectivity, are responsible set to 8.511 and 6.902 for COX-2 and COX-1, respectively. for their cardiotoxicity as some researchers have suggested Following ref. 22, we have used CMAP as reduced representation SðRÞ, (30). Certainly, much remains to be done to determine the bind- and we have measured the square distance ‖…‖2 of a generic state from ing mechanism of selective inhibitors in COXs, but if this relation a point belonging to the reference path using between cardiotoxicity and selectivity is confirmed, by providing ‖SðRÞ − SðlÞ‖2 ¼ ½CðRÞ − CðlÞ 2; the molecular basis for the rational quest of novel COX-2 tar- ∑ i;j i;j geted therapeutic agents with lower toxicity, our study represents j>i an innovative way of doing rational drug design since information ð Þ ðlÞ captured by static experiments such as x-rays or mutagenesis data where C R i;j and C i;j are the elements of the CMAP. A contact between atom i and j is defined as

can now be assisted by information derived from the transient BIOPHYSICS AND ri;j p ligand/protein interactions engaged along the path that leads 1 − ð Þ COMPUTATIONAL BIOLOGY to the main binding site. CðRÞ ¼ r0 ; i;j ri;j q 1 − ðr Þ Methods 0 Metadynamics Simulations. Before doing metadynamics simulations the bin- where ri;j is the distance between the two atoms and r0 is the typical distance ary complexes between COXs and SC-558 were equilibrated with a 5-ns MD at which the contact is formed (SI Text). under NPT conditions at 1 atm and 320 K using the parm99 version (31) of the We performed metadynamics only in the space of sðRÞ while zðRÞ was all-atom Amber force field (32), as implemented in version 2.6 of the NAMD constrained to zðRÞ < 0.15. This gives the possibility for the system to explore molecular dynamics simulation code, which was used throughout (33). All conformations different from the original path, while maintaining at the the simulations were carried out in explicit solvent and in cubic periodic same time the system reasonably close to the chosen intermediate frames. boundary conditions. In such a way, the loss of the alpha-helices folding is avoided. The estimation Fðs; tÞ at time t of the free-energy surfaces FðsÞ as a In addition to sðRÞ, we have also used a distance and a torsion CV to de- function of the CV s was determined by metadynamics (13) in its recently scribe the different conformations of the ligand during the simulations. The developed well-tempered variant (12), using the following formula: former is defined as the distance between the center of mass of the ligand and of a group of atoms of the . The latter is the dihedral angle de- T þ ΔT fined by four atoms, two on the major inertia axes of the ligand and the Fðs; tÞ¼− Vðs; tÞ; other two chosen on a conserved secondary structure of the proteins (SI Text). ΔT The VMD program (34) was used for visualization and data analysis while the figures were made using the PyMOL software (35). where Vðs; tÞ is the bias potential added to the system and T is the tempera- ture of the simulation. ΔT is the difference between the temperature of the Note. During the review process, a 2.75 Å resolved x-ray structure of COX-1 CV and the temperature of the simulation. The bias potential is made up by complexed with celecoxib, an analogue of SC-558, was recently reported (36). the sum of the Gaussians deposited along the trajectories of the CVs. This structure agrees remarkably well with our prediction of binding for SC- 558 in COX-1. In fact, the binding conformation of SC-558 in COX-1 calculated Thanks to this new formalism, one can increase barrier crossing and facil- through metadynamics is very similar to the crystallographic pose found ΔT itate the exploration in the CVs space by tuning . A Gaussian deposition for Celebrex in COX-1 with a low rmsd of 1.46 Å for the ligand heavy atoms rate of 1.1 kcal∕mol per picosecond was initially used and gradually de- (SI Text). This provides a further convincing evidence of the reliability of our creased on the basis of the adaptive bias with a ΔT of 2880 K. model, which predicts a binding conformation of SC-558 in COX-1 similar to

Limongelli et al. PNAS ∣ March 23, 2010 ∣ vol. 107 ∣ no. 12 ∣ 5415 Downloaded by guest on September 28, 2021 the crystallographic pose found in COX-2, although in COX-1 the ligand is ACKNOWLEDGMENTS. The authors thank Matteo Masetti, Davide Branduardi, more weakly bound due to an only partial insertion of the sulphonamide and Giovanni Bussi for useful discussions. This work was supported by a grant moiety in the selectivity pocket. This prediction has been now validated from the Swiss National Supercomputing Centre—CSCS under project by the crystallographic structure (PDB ID code 3kk6). ID s233.

1. Cuzick J, et al. (2009) Aspirin and non-steroidal anti-inflammatory drugs for cancer 18. Hummer G, Kevrekidis I-G (2003) Coarse molecular dynamics of a peptide fragment: prevention: An international consensus statement. Lancet Oncol 10:501–507. Free energy, kinetics, and long-time dynamics computations. J Chem Phys 2. Smith W-L, Borgeat P, Fitzpatrick F-A (1991) Biochemistry of Lipids, Lipoproteins 118:10762–10773. and Membranes, eds D-E Vance and J Vance (Elsevier Science, Amsterdam; London), 19. Parrinello M (2008) Physical Biology from Atoms to , ed AH Zewail (Imperial pp 297–3253. College Press, London), pp 247–265. 3. Scheen A-J (2004) [Withdrawal of rofecoxib (Vioxx): What about cardiovascular safety 20. Luong C, et al. (1996) Flexibility of the NSAID binding site in the structure of human of COX-2 selective non-steroidal anti-inflammatory drugs?]. Revue médicale de Liège cyclooxygenase-2. Nat Struct Biol 3:927–933. 59:565–569. 21. Branduardi D, Gervasio F-L, Parrinello M (2007) From A to B in free energy space. 4. Kearney P-M, et al. (2006) Do selective cyclo--2 inhibitors and traditional J Chem Phys 126:054103. non-steroidal anti-inflammatory drugs increase the risk of atherothrombosis? 22. Bonomi M, Branduardi D, Gervasio F-L, Parrinello M (2008) The unfolded ensemble Meta-analysis of randomised trials. BMJ 332:1302–1308. and folding mechanism of the C-terminal GB1 beta-hairpin. J Am Chem Soc 5. Chen L-C, Ashcroft D-M (2007) Risk of associated with selective 130:13938–13944. COX-2 inhibitors: Meta-analysis of randomised controlled trials. Pharmacoepidemiol 23. Gervasio F-L, Laio A, Parrinello M (2005) Flexible docking in solution using Drug Saf 16:762–772. metadynamics. J Am Chem Soc 127:2600–2607. 6. Kurumbail R-G, et al. (1996) Structural basis for selective inhibition of cyclooxygenase- 24. Masetti M, Cavalli A, Recanatini M, Gervasio F-L (2009) Exploring complex 2 by anti-inflammatory agents. Nature 384:644–648. protein-ligand recognition mechanisms with coarse metadynamics. J Phys Chem B 7. Copeland R-A, et al. (1994) Mechanism of selective inhibition of the inducible isoform 113:4807–4816. of G/H synthase. Proc Natl Acad Sci USA 91:11202–11206. 25. Soliva R, Almansa C, Kalko S-G, Luque F-J, Orozco M (2003) Theoretical studies on the 8. Lanzo C-A, Sutin J, Rowlinson S, Talley J, Marnett L-J (2000) Fluorescence quenching inhibition mechanism of cyclooxygenase-2. Is there a unique recognition site?. J Med analysis of the association and dissociation of a diarylheterocycle to cyclooxygenase-1 Chem 46:1372–1382. and cyclooxygenase-2: Dynamic basis of cyclooxygenase-2 selectivity. Biochemistry-US 26. Price M-L-P, Jorgensen W-L (2000) Analysis of binding affinities for celecoxib analogues 39:6228–6234. with COX-1 and COX-2 from combined docking and Monte Carlo simulations and 9. Walker M-C, et al. (2001) A three-step kinetic mechanism for selective inhibition of insight into the COX-2/COX-1 selectivity. J Am Chem Soc 122:9455–9466. cyclo-oxygenase-2 by diarylheterocyclic inhibitors. Biochem J 357:709–718. 27. Morris G-M, et al. (1998) Automated docking using a Lamarckian genetic algorithm 10. Yuan C, Rieke C-J, Rimon G, Wingerd B-A, Smith W-L (2006) Partnering between mono- and an empirical binding free energy function. J Comput Chem 19:1639–1662. mers of cyclooxygenase-2 homodimers. Proc Natl Acad Sci USA 103:6142–6147. 28. Huey R, Morris G-M, Olson A-J, Goodsell D-S (2007) A semiempirical free energy force 11. Prusakiewicz J-J, Duggan K-C, Rouzer C-A, Marnett L-J (2009) Differential sensitivity field with charge-based desolvation. J Comput Chem 28:1145–1152. and mechanism of inhibition of COX-2 oxygenation of arachidonic acid and 2-arachi- 29. Gierse J-K, et al. (1996) A single amino acid difference between cyclooxygenase-1 donoylglycerol by ibuprofen and . Biochemistry-US 48:7353–7355. (COX-1) and -2 (COX-2) reverses the selectivity of COX-2 specific inhibitors. J Biol Chem 12. Barducci A, Bussi G, Parrinello M (2008) Well-tempered metadynamics: A smoothly 271:15810–15814. converging and tunable free-energy method. Phys Rev Lett 100:020603. 30. Dogne J-M, Hanson J, Supuran C, Pratico D (2006) Coxibs and cardiovascular 13. Laio A, Parrinello M (2002) Escaping free-energy minima. Proc Natl Acad Sci USA side-effects: From light to shadow. Curr Pharm Des 12:971–975. 99:12562–12566. 31. Wang J, Cieplak P, Kollman P-A (2000) How well does a restrained electrostatic 14. Picot D, Loll P-J, Garavito R-M (1994) The x-ray crystal structure of the membrane potential (RESP) model perform in calculating conformational energies of organic protein synthase-1. Nature 367:243–249. and biological molecules?. J Comput Chem 21:1049–1074. 15. Selinsky B-S, Gupta K, Sharkey C-T, Loll P-J (2001) Structural analysis of NSAID binding 32. Cornell W-D, et al. (1995) A second generation force field for the simulation of by prostaglandin H-2 synthase: Time-dependent and time-independent inhibitors proteins, nucleic acids, and organic molecules. J Am Chem Soc 117:5179–5197. elicit identical enzyme conformations. Biochemistry-US 40:5172–5180. 33. Phillips J-C, et al. (2005) Scalable molecular dynamics with NAMD. J Comput Chem 16. So O-Y, Scarafia L-E, Mak A-Y, Callan O-H, Swinney D-C (1998) The dynamics of 26:1781–1802. prostaglandin H synthases—Studies with prostaglandin H synthase 2 Y355F unmask 34. Humphrey W, Dalke A, Schulten K (1996) VMD: Visual molecular dynamics. J Mol mechanisms of time-dependent inhibition and allosteric activation. J Biol Chem Graphics 14:33–38. 273:5801–5807. 35. DeLano W-L (2002) The PyMOL Molecular Graphics System (DeLano Scientific, 17. Gear C-W, Kevrekidis I-G, Theodoropoulos C (2002) ‘Coarse’ integration/bifurcation San Carlos, CA). analysis via microscopic simulators: Micro-Galerkin methods. Comput Chem Eng 36. Rimon G, et al. (2010) Coxibs interfere with the action of aspirin by binding tightly to 26:941–963. one monomer of cyclooxygenase-1. Proc Natl Acad Sci USA 107:28–33.

5416 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0913377107 Limongelli et al. Downloaded by guest on September 28, 2021