Potent Fibrinolysis Inhibitor Discovered by Shape and Electrostatic

Article pubs.acs.org/jmc

Potent Fibrinolysis Inhibitor Discovered by Shape and Electrostatic Complementarity to the Drug Tranexamic Acid† Jonas Boström,*,‡ J. Andrew Grant,§,∥ Ola Fjellström,‡ Anders Thelin,‡ and David Gustafsson‡ ‡

AstraZeneca Mölndal, CVGI iMed, Mölndal, Sweden Discovery Sciences, Alderley Park, U.K.

§

S Supporting Information *

ABSTRACT: Protein−protein interfaces provide an important class of drug targets currently receiving increased attention. The typical design strategy to inhibit protein−protein interactions usually involves large molecules such as peptides and macrocycles. One exception is tranexamic acid (TXA), which, as a lysine mimetic, inhibits binding of plasminogen to fibrin. However, the daily dose of TXA is high due to its modest potency and pharmacokinetic properties. In this study, we report a computational approach, where the focus was on finding electrostatic potential similarities to TXA. Coupling this computational technique with a high-quality low-throughput screen identified 5-(4-piperidyl)-3-isoxazolol (4-PIOL) as a potent plasminogen binding inhibitor with the potential for the treatment of various bleeding disorders. Remarkably, 4-PIOL was found to be more than four times as potent as the drug TXA.

■

INTRODUCTION

The identification of novel high-quality lead compounds is the foundation for research efforts aimed at the development of a new drug. Computational approaches can reveal difficult to perceive similarities between molecules known to bind to a given protein and are thus valuable methods for lead identification. In the current work, we describe and apply an approach for comparing electrostatic potentials of molecules aligned according to their similarity in molecular shape, with the aim of identifying a small molecule capable of disrupting a specific protein−protein interaction. Protein−protein interaction is a common mechanism involved in numerous physiological processes.1 One example is binding of the zymogen plasminogen to fibrin.2 Plasminogen contains a serine protease domain and five kringle domains.3,4 As plasminogen circulates in blood plasma, kringles 2−5 are wrapped around the protease domain and only kringle 1 can bind to fibrin. Once anchored to fibrin, plasminogen unfolds and is activated to plasmin by the tissue-plasminogen activator (tPA).3,5 The result is a localized action of plasmin cleaving fibrin to fibrin degradation products (FDPs), see Figure 1. Although recent progress with regard to small molecule inhibition of protein−protein interaction has been reported,6 it is generally considered a challenge to discover compounds that modulate protein−protein interactions due to large protein− protein interfaces.6−8 Nevertheless, two small generic drugs, tranexamic acid (TXA) and epsilon aminocaproic acid (EACA), exert their effect by inhibiting the protein−protein interaction between plasminogen and fibrin. TXA and EACA work by reversibly preventing the protein−protein interaction via blockade of the well-defined lysine binding site (LBS, Figure 1D) present in the kringle domains of plasminogen.9 © XXXX American Chemical Society

Figure 1. A schematic illustration of the normal fibrinolytic system (A−C) and the intervention of the system with a small molecule (D). (A) Plasminogen (Plg) initially binds with kringle 1 (K1) to fibrin (represented by the red ribbon). (B) Plasminogen unfolds and is activated by tissue-plasminogen activator (tPA) to plasmin, which cleaves fibrin to fibrin degradation products (FDPs) in (C). The mechanism of drug action shown in (D) is the prevention of plasminogen binding to fibrin by inhibition of the lysine binding site of K1 by a small molecule (TXA), and fibrinolysis is inhibited.

TXA and EACA are well-established clinical agents used to reduce blood loss following surgery and trauma and to treat heavy menstrual bleeding, mild hemophilia, and certain forms of von Willebrands disease.2,10 However, these existing Received: December 10, 2012

A

dx.doi.org/10.1021/jm301818g | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

program based on OpenEye software, principally the ZAP Toolkit,14 for this purpose (see Methods). The aim of the ItsElectric program is to increase the probability of finding structures which have similar shape and electrostatics to that of a template structure but which are less intuitively related according to medicinal chemistry notions of chemical functionality. Due to the nature of the calculations being performed, the ItsElectric program is computationally relatively demanding. Since the calculations described in what follows were completed, it has become straightforward to carry out at least an order of magnitude greater number of calculations.15,16 In particular, the electrostatic comparison program EON16 has shown promise in identifying leads and tool compounds17 and recently incorporates approximations to enable fast calculation of the integral in eq 2.18 It should be noted that ItsElectric and EON are based on the same toolkit (ZAP) and theory (Poisson−Boltzmann electrostatic potentials19). The rank order results obtained by ItsElectric can readily be reproduced by EON. The computational expense of ItsElectric calculations meant that it was preferable to use faster methods prior to the more timeconsuming electrostatic comparisons in the virtual screening strategy to reduce the number of such calculations. The ionization state is an important factor in calculating the electrostatic potential field associated with a molecule. In this study, the pragmatic decision that the calculations should be run using neutral forms of all molecules was taken. The reason for this was made first because of the difficulty in accurately assigning correct ionization states to a large collection of molecules. Second, the atoms bearing the charge has a tendency of dominating electrostatic comparisons, making the method less sensitive to subtle atomic changes in other areas on the molecule. In addition, comparisons between molecules of different ionization states are challenging. To illustrate, Figure 3 shows the electrostatic potentials for TXA, in its zwitterionic state, and 4-PIOL, in its positively charged state. Third, molecules that have similar electrostatic characteristics in their neutral forms often share similar characteristics in their ionized forms. This observation was made by Churchill and co-workers in the identification of the first known small molecule inhibitor of the Ca2+-releasing second messenger NAADP (nicotinic acid adenine dinucleotide phosphate).17 Virtual Screen. In the current study, a fast prescreening step was used in the virtual screening strategy to reduce the number of computationally expensive electrostatic comparisons. This step comprised a variety of graph-based similarity methods to generate a relevant subset from the AstraZeneca screening collection (see Methods). Unlike the method of electrostatic comparison, there is no underlying physical model to graphbased searches, and all methods will have differing notions of

medicines have modest potency and nonoptimal pharmacokinetic properties, leading to inconvenient dosing (2−4 doses, up to 6 g per day for TXA, the most potent agent). The agents are also associated with side-effects such as nausea and vomiting (http://www.drugs.com/sfx/tranexamic-acid-side-effects. html).2 This clearly limits their use, and there is consequently an unmet need for new inhibitors of plasminogen with a more convenient dosing and a more acceptable side effect profile. A consideration for this drug project was that although the assay for testing putative inhibitors of fibrinolysis was accurate it had a limited capacity. In contrast to modern highthroughput screening (HTS) methods that screen several hundred thousands of compounds, our low-throughput assay could only screen a few hundred compounds during the same time frame. Consequently, the search for novel plasminogen binding inhibitors is an area that benefits greatly from the combination of appropriate computational tools together with prior information about molecules that bind to the target protein, plasminogen. This facilitates the selection of a small set of compounds for experimental evaluation. It will be shown in the present study that the method based on using electrostatic comparisons identified, from a large compound collection, the small molecule 5-(4-piperidyl)-3isoxazolol (4-PIOL) as a potent and high-quality lead compound for a new series of plasminogen inhibitors. The molecular structures of TXA, EACA, and 4-PIOL are depicted in Figure 2, from which it can be seen that 4-PIOL is structurally different to the existing drugs.

Figure 2. Structures of plasminogen inhibitors, the clinical agents TXA and EACA, and the fibrinolysis inhibitor 4-PIOL discovered in the current study.

Shape and Electrostatic Calculations. Shape-based approaches, that model some physical aspects of molecular recognition, have been successful in virtual screening for lead generation strategies.11 However, it was obvious that shapebased screening alone would not be sufficiently discriminatory due to the simple molecular framework of the reference compound TXA. This fact together with the prominent electrostatic features of TXA motivated us to compare electrostatic potentials in addition to shape complementarity.12,13 We thus developed and employed the program ItsElectric, which now is an AstraZeneca in-house electrostatic comparison

Figure 3. The electrostatic potentials for TXA (left) modeled in its zwitterionic state and 4-PIOL (right), where the aliphatic nitrogen is positively charged and the 1,2-oxazol-3-ol fragment is not charged. This illustrates the fact that charged molecular fragments can dominate the electrostatics and may obscure comparisons. Red color denotes the electronegative area and blue color the electropositive area. The electrostatic Tanimoto is 0.1. B



Article

problematic in biochemical assays (reactive groups, frequent hitters, etc.) were removed using a set of substructure rules26−28 to leave roughly a thousand compounds. The filtered output from the first stage of the virtual screen was compared with the ligand-bound X-ray structure of TXA (pdb code: 1ceb)29 using ItsElectric to assess the electrostatic similarity between each compound as follows. An OMEGA30 multiconformational database was generated for all compounds according to standard procedure.31 Each conformation was subsequently aligned on the basis of maximizing their volume overlap, using a Gaussian representation of molecular shape11−13 onto the bioactive conformation of TXA. The electrostatic Tanimoto values reflecting the similarity of the molecules aligned to the bioactive conformation of TXA were obtained. This enabled the ranking of compounds whose Tanimoto was greater than a threshold value of 0.2, a value in our experience corresponding to useful similarity. As an aside, when running TXA against a 10k subset of molecules of similar size present in the MDL Drug Data Report (MDDR)32 the mean Tanimoto was found to be 0.08, with a standard deviation of 0.04. It should be noted that the average value of 0.08 is on the low side, as compared to other commonly used similarity measures.21−25 The virtual screen resulted in a 77 compound set of which we subsequently were able to test 68 based on compound availability at the time, in the Clot-Lysis buffer assay. Electrostatic Calculations. The ZAP Toolkit14 enabled us to compute potentials in water modeled as a dielectric, with a smooth boundary at the water−molecule interface. In essence, the electrostatic potentials are obtained by numerical solution of the Poisson equation33 1, viz:

chemical similarity. On the other hand, graph- and shape-based descriptors are essentially independent estimates of similarities and the value of combining information from such orthogonal descriptors have been discussed in more detail recently.20 A schematic view of the virtual screening strategy is depicted in Figure 4.

∇{ε(r )∇ϕ(r )} = − ρmol (r )

(1)

where ϕ(r) is the electrostatic potential, ε(r) is the dielectric constant, and ρmol(r) is the molecular charge distribution (usually modeled as a set of point charges). Electrostatic potentials between aligned molecules are compared by determining EAB: ⎯⎯⎯⇀

EAB =

∫ ϕ A(r)ϕB(r)Θ A(r)ΘB(r) dr ≈ h3 ∑ ϕijkAϕijkB ΘijkA ΘijkB ijk

Figure 4. A schematic view of the discovery of 4-PIOL as a highquality lead structure.

(2) where Θ is a masking function to ensure potentials interior to the molecule are not considered as part of the comparison. The integral appearing in eq 2 is a volume integral, computed using a grid-based quadrature. The accuracy of the quadrature is determined by the numerical grid-spacing parameter, h. Finally, the electrostatic similarity is obtained by the standard Tanimoto eq 3:

The ranked list obtained from the virtual screen was examined to obtain a final set of compounds for assay testing, omitting a number of compounds not consistent with the leadgeneration strategy for this project. This includes the intellectual property (IP) position, the tractability of the synthetic chemistry that would be required in lead optimization, and the removal of compounds with functional groups that had historically caused difficulties in later stage development for clinical compounds. This level of visual inspection is only practical and accurate when applied to a small subset produced by a computational approach. Applying such experience in medicinal chemistry ensured that the best prior information available from the known experimental data but not encapsulated by the similarity principle was utilized.

■

Tanimotoelectrostatic = AB

EAB EAA + E BB − EAB

(3)

where the quantities in eq 3 are computed from eq 2, for molecule A overlaid on to B (EAB) and for molecule A and B overlaid on to each other (EAA and EBB, respectively). The interested reader is referred to the EON manual for more details regarding the underlying theory (http://www.eyesopen.com/docs/eon/current/html/theory.html). Tautomeric Preferences of 4-PIOL. Tautomers should not be forgotten.34 Theoretically, there are two tautomeric forms of the neutral heteroaromatic ring in 4-PIOL, the keto (1,2-oxazol-3-one) and the enol (1,2-oxazol-3-ol). Quantum chemical calculations were used to rationalize which is the most prevalent species. That is, gasphase density functional calculations (B3LYP) at the 6-31G** level of theory using the Jaguar program35 were performed. The calculations showed that 4-PIOL exist predominantly in the enol form (Figure 2), which is the energetically most stable tautomer by 4 kcal/mol. These results are in agreement with previously published work on lowtemperature single-crystal structure determinations carried out on derivatives of 4-PIOL.36 Clot-Lysis Assay. The primary in vitro assays used to determine efficacy were the human Clot-Lysis buffer and the human Clot-Lysis plasma assays (Figure 5). In the former assay containing plasminogen, fibrinogen, and tPA in buffer, the reaction was initiated by thrombin,

METHODS

Virtual Screen. A set of 10 query structures with differing scaffolds: TXA, EACA, and eight other compounds based on relevant patents were identified (see Supporting Information). These structures were used to search the corporate compound collection (∼1 M compounds) using four conventional types of fingerprints (GhoseCrippen,21 Daylight,22 ALFI,23 Unity24) and a maximum common substructure similarity measure,25 retrieving ranked lists of the top 50 compounds for each query and method. This resulted in an initial hitlist of 2500 compounds, from which any duplicate compound was removed. In addition, compounds with functional groups known to be C



Article

Table 1. Clot-Lysis Buffer and Plasma (IC50), pKa Acid and Base,a Caco-2 Papp A−B Datab compd TXA 4-PIOL 5-(piperidin4-yl) tetrazole

Clot-lysis buffer (μM)c 11.5 ± 2.3 (n = 6) 2.8 ± 2.1 (n = 4) >100.00

Clot-lysis plasma (μM)c 3.1 ± 0.5 (n = 6) 0.8 ± 0.2 (n = 4) >100.00

pKa acid

pKa base

Caco-2 Papp A−B (10−6 cm/s)

4.1

9.7

7.83

4.0

10.5

0.2) were investigated. There are factors in the current approach that influence the output, where improvements may be made. The rigid-fit molecular alignment is one; another factor is the assumption that partial charges are conformationally invariant. Work along these lines is in progress. Nevertheless, and important to realize, the main purpose of virtual screens is to maximize the odds of finding true positives, as well as ranking true inactives as low as possible.39 We have found that for biological targets for which there is a well-validated assay with high signal and little noise, shape and electrostatic comparisons facilitates minimal experimental screening to discover novel lead structures. This alternative to the relentless high-throughput screening of compounds in relatively noisy assays was found useful in the work by Churchill and co-workers.17 In the current study, only 68 compounds had to be screened to find the potent and structurally diverse plasminogen binding inhibitor 4-PIOL. For practical reasons, we structured the graph-based and the electrostatic similarity measures as sequential screens, see Figure 4. An important aspect of the visual inspection of the results is that it gives an interpretation of the underlying reason for biological response observed in the assay for 4-PIOL. In essence, the complete molecule of 4-PIOL is a bioisostere for TXA, in which the piperidine nitrogen of 4-PIOL corresponds to the primary amine of TXA, and the 4-PIOL isoxazolol mimics the effect of the carboxylic acid functionality of TXA. Overall, the isosterism can be interpreted as the carbonyl and

Figure 5. An illustration of the effects of TXA and 4-PIOL on fibrinolysis in the Clot-lysis experiment. The Clot-lysis time is defined as the difference in time between fibrin formation (positive Vmax) and dissolution (negative Vmax).

which cleaves fibrinogen leading to fibrin polymerization.37 Plasminogen and tPA assemble on the fibrin surface and plasminogen is activated to plasmin. Plasmin then cleaves fibrin to FDPs. The ClotLysis plasma assay use pooled citrated platelet-poor human plasma, which contains all proteins necessary for the formation and degradation of fibrin particles. In this assay, coagulation is initiated by addition of calcium ions and tPA. The formation and dissolution of fibrin particles is, in both assays, followed by measuring absorbance at 405 nm. The Clot-Lysis time is defined as the time difference between fibrin particle formation (positive Vmax) and particle dissolution (negative Vmax), see Figure 5. For each concentration of the compound tested, the time difference between maximal velocity of clot formation and maximal velocity if clot break down were calculated. The change in clot lifetime were then related to the lifetime without inhibition (vehicle) according to: % = 100*((TnegVmax − TposVmax)vehicle/(TnegVmax − TposVmax)compound). The IC50 values were calculated as the concentration of the compound giving a 50% effect on the Clot Lysis. Determination of pKa Values. The ionization constants were measured using a high throughput pKa screening assay by pressureassisted capillary electrophoresis (CE) and mass spectrometry (MS) as described by Wan et al.38 This method offers reproducibility with an accuracy ±0.2 of pKa units.38 Determination of Permeability in Caco-2 Cell Monolayers. The apical to basolateral transport experiments (Papp A−B) was measured in a Caco-2 membrane assay at pH 6.5. The Caco-2 reproducibility is compound-dependent and often lower for low permeable compounds. The experimental protocol is described in more detail in the Supporting Information.

■

RESULTS The virtual screening strategy adopted identified 4-PIOL as a potent (IC50: 2.8 μM) compound in the Clot-Lysis buffer assay. Although the remaining compounds selected for the assay were either weakly active or inactive, it was surprising that such an active compound from such as a small set of experimental measurements was discovered. Further, it is remarkable that the potency of the lead compound 4-PIOL was found to be four times that of TXA (Table 1), the current drug widely used in the clinic. A similar result was observed in the more complex human Clot-Lysis plasma assay. In this case, the 4-PIOL compound has a potency of 0.8 μM compared to that of 3.1 μM for TXA, see Table 1. D



Article

NH of the 3-isoxazolone in 4-PIOL are “displaced” by two bonds from the 1-position of the cyclohexane ring in TXA, while the nitrogen of the primary amine in TXA “shifts” two bonds by transforming into a secondary amine of the piperidine ring in the 4-PIOL. The shape and electrostatics of TXA and 4PIOL are therefore similar despite the significant difference in their chemical structures. Figure 6 illustrates the high-quality

heterocycles by computation as well as supports our current approach of modeling all compounds in their neutral state. Replacement of the hydroxyisoxazole of 4-PIOL with tetrazole, perhaps the commonest carboxylic acid isostere, gives 5(piperidin-4-yl)tetrazole.41 This compound has pKa values very similar to TXA and 4-PIOL but had no measurable activity in either of the Clot-Lysis assays (Table 1). This lack of biological activity is partly accounted for by inspecting the electrostatic potentials as shown in Figure 6, which shows a clear similarity between TXA and 4-PIOL but not as apparent for the tetrazole analogue. This is also quantified by the electrostatic Tanimoto values. Thus despite the overall shape and the pKas of 4-PIOL and the tetrazole compound compared to TXA both being similar, their different electrostatic characteristics can be used to discriminate and rationalize their biological activity. The value of comparing the shape and the electrostatic potentials gives some insight as to why tetrazole is an imperfect surrogate for carboxylic acid. There is a crystal structure of the recombinant kringle 1 domain of human plasminogen in complex with TXA.29 This provided the evidence for the bioactive conformation used in the shape and electrostatic matching as mentioned above. Figure 7 shows the ItsElectric alignment of 4-PIOL onto TXA

Figure 7. A depiction showing the shape and electrostatic alignment of 4-PIOL onto the crystal structure of TXA, in their zwitterionic forms, in complex with the recombinant kringle-1 domain of human plasminogen.

Figure 6. A depiction showing the shapes and the electrostatic potentials for TXA, 4-PIOL, and the tetrazole analogue in their neutral forms. Red color denotes electronegative areas, whereas blue color shows electropositive areas. The calculated Tanimoto values show that 4-PIOL is electrostatically very similar to (the bioactive conformation of) TXA, whereas the tetrazole analogue is not.

in the LBS of kringle 1. The electrostatic nature of the binding site is evident with the acid functionalities of the ligands directed toward a pair of arginine residues (ARG:34 and ARG:70) and the ligands amines interacting with a pair of aspartic acids (ASP:54 and ASP:56) of the protein. Visual inspection of the hydrogen-bonding network in the ligand− protein complex indicates that 4-PIOL and TXA bind in their zwitterionic forms (Figure 7). The well-defined shape of the binding site is also clear, with both ligands filling the binding site, which also accounts for the lack of affinity for the tetrazole analogue, given it only partially occupies the pocket. Guided by computational chemistry, the identification of 4PIOL as a fibrinolysis inhibitor made a major impact on this drug discovery effort into bleeding disorders. Given that 4PIOL was demonstrably more active than TXA ensured that the project could begin to address the issues with the high dose

shape and electrostatic complementarity between TXA and 4PIOL. Note also that a Tanimotoelectrostatic of 0.35 between TXA and 4-PIOL corresponds to a relatively high similarity. Experimental determination of the pKa of the acid and base functions in the molecule confirms some aspects of the inferred electrostatic similarity. Table 1 show that the pKa of the basic amines in TXA and 4-PIOL are similar (9.7 and 10.5) and that the acid pKas are nearly identical (4.1 and 4.0). In addition, calculated pKa values were performed with the frequently used ACD/pKa DB program.40 Calculated and experimental values were in disagreement. That is, the 1,2-oxazol-3-ol fragment in 4-PIOL was not calculated to be acidic (pKa = 12.4). This highlights difficulties in assigning correct ionization states for E



■

of TXA by improving some combination of the efficacy, bioavailability, and clearance. The role of lipophilicity in determining the overall quality of candidate drug molecules is important because it affects many other compound properties relevant in lead optimization.42 The lipophilicity of 4-PIOL was determined to be low (log D < 0), and this may be viewed as an atypical liability. Hydrophilicity can be disadvantageous, mainly due to poor cell permeability (indicated by low Caco-2 assay measurement, see Table 1), resulting in unfavorable drug metabolism and pharmacokinetics (DMPK) characteristics in general and suggestive of difficulties with creating a successful oral formulation. Having established 4-PIOL as a potent fibrinolysis inhibitor, we subsequently investigated the literature to retrieve other relevant pharmacology. This showed that 4-PIOL was identified in 1987 as a low affinity GABA-A agonist43 and has been extensively studied within the GABA-A field. For example, 4PIOL has been used as a lead and tool structure and in the discovery of other potent GABA-A antagonists.44,45 Further characterization46,47 concluded 4-PIOL to act as a low efficacy nondesensitizing partial agonist with mainly an antagonstic profile.45,48 Due to the fact that 4-PIOL exhibited low-efficacy partial GABA-A agonism, it was flagged as a potential safety concern. It should be noted that Furtmüller et al.49 recently attributed a side effect (convulsions) of TXA to a GABA-A antagonistic effect. As an initial test for checking putative parallel structure− activity relationships between GABA-A and plasminogen, a phenyl-substituted analogue of 4-PIOL, 4-phenyl-5-piperidin-4yl-1,2-oxazol-3-one, was acquired and tested in the Clot-Lysis assays. Previous reports had shown that lipophilic 4-substituted analogues of 4-PIOL generally displayed increased affinity for the GABA-A receptor, and 4-phenyl-5-piperidin-4-yl-1,2oxazol-3-one was one example providing a 40-fold increase in affinity for the GABA-A receptor as compared to 4-PIOL.50 The analogue compound was found to be inactive in the ClotLysis assays (>100 μM), thus, indicative of differences of the binding sites of the two proteins. This information provided us confidence for achieving the goal of selectivity. In the subsequent lead optimization program, the abovementioned issues were addressed. By solving challenging synthetic chemistry issues, compounds selective against GABA-A, with good cell permeability and retained potency with respect to plasminogen inhibition, could be obtained.51,52 More details on the lead optimization work will be reported in a separate publication.52

Article

ASSOCIATED CONTENT

S Supporting Information *

Detailed descriptions of the experimental methods used to measure Caco-2 permeability, pKa, Clot-Lysis buffer and plasma, and LC/MS data for compound characterization, as well as the molecular structures for the virtual screening hits and the reference compounds in SMILES format are included. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +46 31 7065251. E-mail: jonas.bostrom@astrazeneca. com. Address: AstraZeneca CVGI iMed, Department of Medicinal Chemistry, Pepparedsleden 1, Mölndal S-43183, Sweden. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. J.B., O.F., initial virtual screening; A.T., biological testing; J.B., J.A.G. method development (ItsElectric); J.B., detailed virtual screening; D.G., planned the project; J.B., handled the manuscript. Notes

The authors declare no competing financial interest. ∥ J.A.G. is deceased.

■ ■

DEDICATION In memory of J. Andrew Grant (1963−2012), a great friend and scientist. †

ABBREVIATIONS USED TXA tranexamic acid, 4-PIOL; 5-(4-piperidyl)-3-isoxazolol, EACA; epsilon aminocaproic acid, tPA; tissue-plasminogen activator, PLG; Plasminogen, FDPs; fibrin degradation products, LBS; lysine binding site, HTS; high-throughput screening, PB; Poisson−Boltzmann, NAADP; nicotinic acid adenine dinucleotide phosphate, IP; intellectual property, CE; capillary electrophoresis, MS; mass spectrometry, DMPK; drug metabolism and pharmacokinetic, B3LYP; 3-parameter hybrid Becke exchange/Lee−Yang−Parr correlation functional, GABA; γ-aminobutyric acid.

■

REFERENCES

(1) Nooren, I. M. A.; Thornton, J. M. Diversity of protein−protein interactions. EMBO J. 2003, 22, 3486−3492. (2) Dunn, C. J.; Goa, K. L. Tranexamic acid: a review of its use in surgery and other indications. Drugs 1999, 57, 1005−1032. (3) Xue, Y.; Bodin, C.; Olosson, K. Crystal structure of the native plasminogen reveals an activation-resistant compact conformation. J. Thromb. Haemost. 2012, 10, 1385−1396. (4) Law, R. H. P.; Caradoc-Davies, T.; Cowieson, N.; Horvath, A. J.; Quek, A. J.; Encarnacao, J. A.; Steer, D.; Cowan, A.; Zhang, Q.; Lu, B. G. C.; Pike, R. N.; Smith, A. I.; Coughlin, P. B.; Whisstock, J. C. The X-ray crystal structure of full-length human plasminogen. Cell Rep. 2012, 1, 185−190. (5) Rathore, Y. S.; Rehan, M.; Pandey, K.; Sahni, G. Ashish. First structural model of full-length human tissue−plasminogen activator: a SAXS data-based modeling study. J. Phys. Chem. B 2012, 116, 496− 502. (6) Fry, D. C. Small-molecule inhibitors of protein−protein interactions: how to mimic a protein partner. Curr. Pharm. Des. 2012, 18, 4679−4684.

■

CONCLUSIONS 4-PIOL was identified as a 4-fold more potent fibrinolysis inhibitor than TXA using a low-throughput screen where the compound selection was made using computational techniques. The key computational approach to our contribution for finding the right lead compound was shape and electrostatic comparisons between molecules. Other aspects included selecting and establishing an accurate experimental assay as well as a suitable screening collection. Accordingly, it is now clear that the computational approach described in the present work identified a nontrivial bioisostere of TXA as a high-quality lead for a subsequent lead optimization program. 4-PIOL served as an excellent starting point for subsequent lead optimization.51,52 F



Article

(30) OMEGA, version 1.8; OpenEye Scientific Software: Santa Fe, NM. (31) Boström, J.; Greenwood, J. R.; Gottfries, J. Assessing the performance of OMEGA with respect to retrieving bioactive conformations. J. Mol. Graphics. Modell. 2003, 21, 449−462. (32) MDDRA Structural Database; MDL Information Systems Inc.: San Leandro, CA. (33) Böttcher, C. J. F. Theory of Electric Polarization: Dielectrics in Static Fields: Elsevier: New York, 1973; Vol. 1. (34) Martin, Y. C. Let’s not forget tautomers. J. Comput.-Aided Mol. Des. 2009, 23, 693−704. (35) Jaguar, v7.8; Schroedinger, LLC: Portland, OR. (36) Frydenvang, K.; Matzen, L.; Norrby, P-.O.; Sløk, F. A.; Liljefors, T.; Krogsgaard-Larsen, P.; Jaroszewski, J. W. Structural characteristics of isoxazol-3-ol and isothiazol-3-ol, carboxy group bioisosteres examined by X-ray crystallography and ab initio calculations. J. Chem. Soc., Perkin Trans. 2 1997, 1783−1791. (37) Kim, P. Y.; Stewart, R. J.; Lipson, S. M.; Nesheim, M. E. The relative kinetics of clotting and lysis provide a biochemical rationale for the correlation between elevated fibrinogen and cardiovascular disease. J. Thromb. Haemost. 2007, 5, 1250−1256. (38) Wan, H.; Holmén, A. G.; Wang, Y.; Lindberg, W.; Englund, M.; Någård, M. B.; Thompson, R. A. High throughput screening of pKa values of pharmaceuticals by pressure-assisted capillary electrophoresis and mass spectrometry. Rapid Commun. Mass Spectrom. 2003, 1, 2639−2648. (39) Scior, T.; Bender, A.; Tresadern, G.; Medina-Franco, J. L.; Martínez-Mayorga, K.; Langer, T.; Cuanalo-Contreras, K.; Agrafiotis, D. K. Recognizing pitfalls in virtual screening: a critical review. J. Chem. Inf. Model. 2012, 52, 867−881. (40) ACD/pKa DB, ACD/Labs release 12.0; Advanced Chemistry Development, Inc.: 110 Yonge Street, Toronto, Ontario, Canada. (41) McManus, J. M.; Herbst, R. M. Tetrazole analogs of pyridinecarboxylic acids. J. Org. Chem. 1959, 24, 1462−1464. (42) Waring, M. J. Lipophilicity in drug discovery. Expert Opin. Drug Discovery 2010, 5, 235−248. (43) Byberg, J. R.; Labouta, I. M.; Falch, E.; Hjeds, H.; KrogsgaardLarsen, P.; Curtis, D.; R.; Gynther, B. D. Synthesis and biological activity of a GABAA agonist which has no effect on benzodiazepine binding and of structurally related glycine antagonists. Drug Des. Delivery 1987, 1, 261−274. (44) Møller, H. A.; Sander, T.; Kristensen, J. L.; Nielsen, B.; Krall, J.; Bergmann, M. L.; Christiansen, B.; Balle, T.; Jensen, A. A.; Frølund, B. Novel 4-(piperidin-4-yl)-1-hydroxypyrazoles as γ-aminobutyric acid A receptor ligands: synthesis, pharmacology, and structure−activity relationships. J. Med. Chem. 2010, 53, 3417−3421. (45) Krogsgaard-Larsen, P.; Frølund, B.; Liljefors, T. Specific GABA(A) agonists and partial agonists. Chem. Rec. 2002, 2, 419−430. (46) Kristiansen, U.; Lambert, J. D.; Falch, E.; Krogsgaard-Larsen, P. Electrophysiological studies of the GABA-A receptor ligand, 4-PIOL, on cultured hippocampal neurons. Br. J. Pharmacol. 1991, 104, 85−90. (47) Kristiansen, U.; Lambert, J. D. Benzodiazepine and barbiturate ligands modulate responses of cultured hippocampal neurons to the GABA-A receptor partial agonist, 4-PIOL. Neuropharmacology 1996, 35, 1181−1191. (48) Frølund, B.; Kristiansen, U.; Brehm, L.; Hansen, A. B.; Krogsgaard-Larsen, P.; Falch, E. Partial GABA-A receptor agonists. Synthesis and in vitro pharmacology of a series of nonannulated analogs of 4,5,6,7-tetrahydroisoxazolo[4,5-c]pyridin-3-ol. J. Med. Chem. 1995, 38, 3287−3296. (49) Furtmüller, R.; Schlag, M. G.; Berger, M.; Hopf, R.; Huck, S.; Sieghart, W.; Redl, H. Tranexamic Acid, a widely used antifibrinolytic agent, causes convulsions by a γ-aminobutyric acid a receptor antagonistic effect. J. Pharm. Exp. Ther. 2002, 301, 168−173. (50) Frølund, B.; Jensen, L. S.; Guandalini, L.; Canillo, C.; Vestergaard, H. T.; Kristiansen, U.; Nielsen, B.; Stensbøl, T. B.; Madsen, C.; Krogsgaard-Larsen, P.; Liljefors, T. Potent 4-aryl- or 4arylalkyl-substituted 3-isoxazolol GABAA antagonists: synthesis,

(7) Smith, S. E. P.; Bida, A. T.; Davis, T. R.; Sicotte, H.; Patterson, S. E.; Gil, D.; Schrum, A. G. IP-FCM measures physiologic protein− protein interactions modulated by signal transduction and smallmolecule drug inhibition. PLoS One 2012, 7 (9), e45722. (8) Arkin, M. R.; Wells, J. A. Small-molecule inhibitors of protein− protein interactions: progressing towards the dream. Nature Rev. Drug Discovery 2004, 3, 301−317. (9) Hochschwender, S. M.; Laursen, R. A. The lysine binding sites of human plasminogen. J. Biol. Chem. 1981, 256, 11172−11176. (10) CRASH-2 trial collaborators, effects of tranexamic acid on death, vascular occlusive events, and blood transfusion in trauma patients with significant haemorrhage (CRASH-2): a randomised, placebocontrolled trial. Lancet 2010, 376, 23−32. (11) Nicholls, A.; McGaughey, G. B.; Sheridan, R. P.; Good, A. C.; Warren, G.; Mathieu, M.; Muchmore, S. W.; Brown, S. P.; Grant, J. A.; Haigh, J. A.; Nevins, N.; Jain, A. N.; Kelley, B. Molecular shape and medicinal chemistry: a perspective. J. Med. Chem. 2010, 53, 3862−86. (12) Grant, J. A.; Gallardo, M. A.; Pickup, B. A fast method of molecular shape comparison. A simple application of a Gaussian description of molecular shape. J. Comput. Chem. 1996, 17, 1653− 1666. (13) ROCS; OpenEye Scientific Software: Santa Fe, NM. (14) OEChem Toolkit; OpenEye Scientific Software: Santa Fe, NM. (15) Brood; OpenEye Scientific Software: Santa Fe, NM. (16) EON; OpenEye Scientific Software: Santa Fe, NM. (17) Naylor, E.; Arredouani, A.; Vasudevan, S. R.; Lewis, A. M.; Parkesh, R.; Mizote, A.; Rosen, D.; Thomas, J. M.; Izumi, M.; Ganesan, A.; Galione, A.; Churchill, G. C. Identification of a chemical probe for NAADP by virtual screening. Nature Chem. Biol. 2009, 4, 220−226. (18) Nicholls A., unpublished results. (19) Gilson, M. K.; Rashin, A.; Fine, R.; Honig, B. On the calculation of electrostatic interactions in proteins. J. Mol. Biol. 1985, 184, 503− 516. (20) Muchmore, S. W.; Debe, D. A.; Metz, J. T.; Brown, S. P.; Martin, Y. C.; Hajduk, P. J. Application of belief theory to similarity data fusion for use in analog searching and lead hopping. J. Chem. Inf. Model. 2008, 48, 941−948. (21) Viswanadhan, V. N.; Ghose, A. K.; Revankar, G. R.; Robins, R. K. Atomic physicochemical parameters for three-dimensional structure-directed quantitative structure−activity relationships. 4. Additional parameters for hydrophobic and dispersive interactions and their application for an automated superposition of certain naturally occurring nucleoside antibiotics. J. Chem. Inf. Comput. Sci. 1989, 29, 163−172. (22) Daylight Toolkit 4.91; Daylight Chemical Information Systems, Inc.: Aliso Viejo, CA. (23) Blomberg, N.; Cosgrove, D. A.; Kenny, P. W.; Kolmodin, K. Design of compound libraries for fragment screening. J. Comput.-Aided Mol. Des. 2009, 23, 513−525. (24) UNITY v4.4.2; Tripos Inc.: St. Louis, MO, 2004. (25) Boström, J.; Hogner, A.; Schmitt, S. Do structurally similar molecules bind in a similar fashion? J. Med. Chem. 2006, 49, 6716− 6725. (26) Seidler, J.; McGovern, S. L.; Doman, T. N.; Shoichet, B. K. Identification and prediction of promiscuous aggregating inhibitors among known drugs. J. Med. Chem. 2003, 46, 4477−4486. (27) Baell, J. B.; Holloway, G. A. New substructure filters for removal of pan assay interference compounds (PAINS) from screening libraries and for their exclusion in bioassays. J. Med. Chem. 2010, 53, 2719− 2740. (28) Bruns, R. F.; Watson, I. A. Rules for identifying potentially reactive or promiscuous compounds. J. Med. Chem. 2012, 55, 9763− 9772. (29) Mathews, I. I.; Vanderhoff-Hanaver, P.; Castellino, F. J.; Tulinsky, A. Crystal structures of the recombinant kringle 1 domain of human plasminogen in complexes with the ligands ε-aminocaproic acid and trans-4-(aminomethyl) cyclohexane-1-carboxylic acid. Biochemistry 1996, 35, 2567−2576. G



Article

pharmacology, and molecular modeling. J. Med. Chem. 2005, 48, 427− 439. (51) Boström, J.; Cheng, L.; Fex, T.; Karle, M.; Pettersen, D.; Schell, P. Isoxazole-3(2H)-one analogs as therapeutic agents. Patent WO2010117323, 2010. (52) Pettersson, D.; Schell, P.; Boström, J.; Ohlsson, B.; Evertsson, E.; Gustafsson, D.; Cheng, L. unpublished results.

H


Potent Fibrinolysis Inhibitor Discovered by Shape and Electrostatic

Recommend Documents