CHAPTERS. CoMFA and CoMSIA Investigation of Diverse Pyrrolidine based Analogues as DPP-IV Inhibitors

Transcription

1 CHAPTERS CoMFA and CoMSIA Investigation of Diverse Pyrrolidine based Analogues as DPP-IV Inhibitors

2 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine Introduction Most drugs used in human therapy interact with certain macromolecular biological targets, e.g., with enzymes, receptors, ion channels and transporters etc.} It is generally accepted that the structure, composition and physical properties of a ligand directly affect its biological activity against a target. The attempt to transform this qualitative belief into a quantitative method of activity assessment is known as the Quantitative Structure Activity Relationships (QSAR) which took a formal shape with the work of Hansch and further developed by others. Although there are a number of methods, Hansch,2-5 Free-Wilson 6 and modified Free-wilson 7 approaches are widely practiced ones for modeling the biological response. (a) Hansch analysis Hansch model takes into account numerical information on Lipophilicity, electronic and steric effect. 2-5 The general form of Hansch equation is: Log BA =alogp + bcr + C Es + constant (linear) Log BA =alogp + b(logpi + c cr + d Es + constant' (non-linear) Hansch model correlates biological activity with physicochemical properties. The coefficients (a, b, c, d and constant) are determined by multiple regression analysis. (b) Free-Wilson analysis This method is based on the assumption that the introduction of a particular substituent at a particular molecular position always contributes in the same way to the biological potency of the whole molecule, as expressed by the equation: Log BA = contribution of unsubstituted parent compound + contribution of corresponding substituents where, aj = number of position at which substitution occurs, aj= number of substituents at that position and f.1 = overall average 142

3 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine.... The equation is solved by multiple linear regression using the presence (1) or absence (0) of the different substituents as independent parameters, while the measured activity serves as dependent variable. 6 (c) Mixed approach Mixed approach is the combination ofhansch & Free-Wilson approach. 8 Log BA =k l 1t + k20' + k3es + k (Hansch analysis) Log BA =f..1 + L aiaj (Free-Wilson approach) So, the mixed approach can be written as Where, L(aiaj) is the free-wilson part for the substituents <Pj =cr, 1t & Es contribution of the parent skeleton. Among the above-mentioned approaches, Hansch approach became the most popular approach in QSAR. The high dimensional QSAR analyses (3D, 4D and 5D) are developed to avoid pitfalls of classical method and to create the hypothetical drug receptor model. 5.2 Basic requirements for QSAR analysis Some basic requirements are very essential for best model development. They are mentioned below: 1. All analogues belong to a congeneric series (classical QSAR studies) exerting the same mechanism of action: This is a series of compounds with a similar basic structure but with varying substituents. Non-congeneric series are analyzed through higher dimensional (3D, 4D) studies. 2. Also, the set of compounds with same mechanism of action is essential. 3. Biological response should be distributed over a wide range. 4. Observed biological activity should be in specific units (concentration in molar units or ICso or percentage inhibition). 143

4 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine A simple rule is that the total number of compounds in the training set divided by the number of variables in the final model should be greater than -five or six. 6. This will assure that a data set will not be "over predicted" and that the model will have a better chance to retain the predictive value. 5.3 Steps involved in QSAR studies The QSAR studies enable the development of mathematical model to predict the biological activity of newly designed compounds. There are three steps involved in this procedure: i) Creation of database involving calculation of various physicochemical and structural parameters of a congeneric series. ii) iii) Regression analyses leading to model development between biological activities versus derived physiochemical descriptors. Validation of the models and prediction of the biological activity of the designed compounds. 5.4 Model development procedures (I) Classical or 2D QSAR analysis 2D descriptors are usually developed by using the atoms and connective information of the molecule but 3D coordinates and individual conformations are not considered. In 2D QSAR, physicochemical parameters such as hydrophobic (n), steric (Molar refractivity or MR), hydrogen acceptor (HA), hydrogen donor (HD) and electronic (Field effect or F, Resonance or R, Hammett's constant or 0) are normally used?-5 In addition to these parameters, de novo constants or indicator variables with 0 or 1 values denoting absence or presence of certain feature (cis/trans, ring atom and bridge atom or chain, different test model etc) are also used to adequately parameterize the compounds. 7 In this, many topological indices are also considered as parameters for analysis. 9, 10 Drug distribution and binding process are equilibrium process governed by the corresponding free energy differences, K = e-~g/rt = e-(~h-ns)irt, such relationships should use logarithmic scale. For these, the biological inhibitory values i.e. ICso or EDso or LDso or Ki must be converted into logarithmic form such as log (liicso) or log (liedso) or log (lildso) or log (liki) values to obtain appropriate 144

5 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine activity parameters for QSAR study. The logarithmic scales also ensure a nonnal distribution for the experimental error of biological tests, a requirement for regression-type statistical analyses. In some cases, the activity percent (%) values (A) are converted to Log {A/(100-A)} as a binding equilibrium constant which is physicochemically more meaningful than A alone for QSAR analysis. (II) Three Dimensional Quantitative Structure Activity Relationships (3D- QSAR) Three-dimensional quantitative structure-activity relationships (3D-QSARs) are quantitative models that relate the biological activity of small molecules with their properties calculated in 3D space. Hence 3D properties of a molecule are considered rather than considering individual substituents. The 3D structures are usually generated with configurational infonnation or 3D-structure database or X-ray crystallographic analysis or 2D NMR study. This structure is optimized to refine the geometry based on size of the molecule such as molecular mechanics (large systems; thousands of atoms) or semiempirical (Medium size systems; hundreds of atoms) or ab initio (small systems; tens of atoms), in order to obtain one lowest energy structure per molecule. There are many 3D-QSAR techniques used for various purposes. A few of them are: Comparative Molecular Field Analysis (CoMF A), II Comparative Molecular Similarity Indices Analysis (COMSIA),12 Molecular Shape Analysis (MSA),13, 14 The Distance Geometry Approach, The binding Site Model Approach, COMPASS, the Hypothetical Active Lattice Method, the Molecular Similarity Approach, Genetically Evolved Receptor Models. Some of these approaches to QSAR are based on the statistical analysis of the 3D interaction fields. These are generated by measuring over a regular 3D grid the interaction energy between a small probe atom or group and the ligands. Initially the 3D structures of the training set of compounds are aligned based on common molecular features so as to occupy the same volume of space. The interaction energies of the small probe, usually a methyl group and a proton is measured with each of the training set compounds at each grid co-ordinates in space. The interaction energy at each grid point in space becomes a descriptor in a QSAR analysis. It results in a data table containing several hundreds or even thousands of descriptors for the analysis. 145

6 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine Comparative Molecular Field Analysis (CoMFA) CoMF A is a 3D-QSAR technique employing both interactive graphics and statistical techniques for correlating the shapes and the biological properties of the molecules. The idea underlying CoMF A is that differences in a target property are often related to differences in the shapes of the non-covalent fields surrounding the tested molecules. To put the shape of a molecular field into a QSAR table, the magnitude of its steric (Lennard-Jones) and electrostatic (Coulombic) fields are sampled at regular intervals throughout a defined region. To do so, bioactive conformation of each compound is chosen and they are superimposed in a manner defined by the supposed mode of interaction with the target receptor. CoMF A then compares, in three dimensions, the steric and the electrostatic fields calculated around the molecules with various probe groups and extract the important features related to the biological activity. In doing so, CoMF A tries to identify the quantitative influence of specific chemical features of the molecules on their potencies. The results are then displayed in contour plots showing the important regions in three-dimensional space that are highly associated with biological activity. Advantages of CoMF A technique include prediction of activity of new compounds and representation of QSAR models in the form of contour maps. There are many important aspects that need to be considered for developing good CoMF A model including biological data, selection of compounds and series design, generation of 3D structure of ligand molecules, conformational analysis of each molecule, establishment of bioactive conformation of each molecule, binding mode and superimposition of the molecules, position of lattice points, choice of force field and calculation of interaction energies, statistical analysis of the data and selection of the 3D QSAR model, display of results in contour plots and interpretation of them, design and forecasting the activity of unknown compounds Comparative Molecular Similarity Indices Analysis (CoMSIA) The general methodology and crucial variables for Comparative Molecular Similarity Indices Analysis (CoMSIA) are same as for CoMF A. The primary difference between them is that in case of CoMF A, the contribution due to dispersion forces between molecules are described by Lennard-Jones potential and electrostatic properties are characterized by Coulomb-type potential while in CoMSIA a special Gaussian 146

7 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine.... function is considered for calculation of interaction energies. 12 CoMSIA avoids some of the inherent deficiencies arising from the functional fonn of the Lennard-Jones and Coulomb potentials used in the original version of CoMF A. Both potentials are very steep close to the van der Waals surface and produce singularities at the atomic positions. As a consequence, the potential energy expressed at grid points in the proximity of the surface changes dramatically. To avoid unacceptably large energy values, the potential evaluations are nonnally restricted to regions outside the molecules and require the definition of some arbitrarily determined cutoff values. Due to differences in the slope of the Lennard-Jones and Coulomb potentials, these cutoff values are exceeded at different distances from the molecules, requiring further arbitrary ~caling of the two fields in a simultaneous evaluation which can involve the loss of infonnation about one of the fields. To overcome such problems, CoMSIA evaluates molecular similarity in space. Furthermore, in addition to the steric and electrostatic fields, CoMSIA defines explicit hydrophobic and hydrogen bond donor and acceptor descriptor fields, which are not available with standard CoMF A. 5.5 Basis of work Glucagon-like peptide 1 (GLP-l) therapy is a pnme area of exploration for the treatment of type 2 diabetes. GLP-l is an incretin honnone released from the gut during meals and serves as an enhancer of glucose stimulated insulin release from pancreatic f3-cells. A continuous administration of GLP-l results in significantly lowering the fasting plasma glucose as well as HbAlc!5, 16 However, GLP-l is rapidly degraded in plasma by the serine protease dipeptidyl peptidase-n (DPP-N). Therefore, the inhibition of DPP-IV is rapidly emerging as an attractive target in the treatment of type 2 diabetes These inhibitors function as indirect stimulators of insulin secretion and this effect is mainly believed to be mediated by enhancing the action of the incretin honnone GLP_ Development of small molecules as selective inhibitors of DPP-IV is a major challenge because the different chemical prototypes inhibit the other members of DPP family namely DPP-II, DPP-VIII and DPP-IX. Moreover, the inhibition of DPP-II and DPP-VIIIIDPP-IX has been shown to result in several complications such as apoptosis of quiescent T cells and severe toxic reactions including alopecia, thrombocytopenia, anemia, enlarged spleen, multiple histological pathologies and increased mortality in animals respectively.25 In view of the side effects associated with the inhibition of DPP family, it has become 147

8 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine.... necessary to design compounds with selective affinity towards the DPP-IV over the other members of this class. In addition to focusing on potency and selectivity, development of long acting inhibitors is also desirable. This may potentially provide maximal efficacy particularly in patients suffering from severe diabetes (e.g. HbAlc >9%). Ligand-based 3D-QSAR techniques such as Comparative Molecular Field Analysis (CoMFA) and Comparative Molecular Similarity Indices Analysis (CoMSIA) are useful to correlate experimental biological activities with structural changes of ligands that affect the binding affinity to the target The earlier modeling studies on DPP-IV inhibitors are restricted to selected prototype compounds with limited variations?9,30 The present modeling study has involved diverse pyrrolidine based analogues as DPP-IV inhibitors. The aim of this work is to provide critical diagnosis of the DPP-IV's putative active site and inputs for the design of new and selective pyrrolidine based inhibitors for the same. 5.6 Materials and Methods Data Set and biological activity A chemical structure database of 190 pyrrolidine based analogs (Table 1) as Dipeptidyl Peptidase-IV Inhibitor was investigated using CoMFA and CoMSIA methods This database has seven different derivatives namely Biarylphenylalanine (S)- fluoropyrrolidine, 4-aminocyc1ohexylalanine pyrrolidine, 4- aminocyclohex ylgl ycine pyrrolidinelthiazolidide, 3 -substituted-cycloalkylgl ycine pyrrolidine/thiazolidide, /3-methyl phenylalanine pyrrolidine, Glutamate cyanopyrrolidine and Oxadiazole pyrrolidine derivatives. The diversity of this database may help to understand the potential binding poses of these analogues. Figure 1 shows the common structure space of this pyrrolidine database represented in three parts namely the head portion (pyrrolidine moiety), the central ring linker and the tail (aryl or heteroaryl moiety). The Merck laboratory has evaluated all these analogues for the DPP-IV inhibitory activity under same experimental protoco1. 40 For the CoMFA and CoMSIA study, the DPP-IV enzyme inhibitory activity (IC so) of the compounds were transformed into the logarithm of reciprocal inhibitory concentration and expressed as picso (Table 1). With due consideration to the structure space, all the compounds were distributed into training (140 compounds) and test (50 compounds) sets. Only training set compounds were used for the model development. 148

9 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine... Table 1. Structural classes of compounds with picso of Pyrrolidine based DPP-N inhibitors used in the CoMF A and CoMSIA analysis. A) Biarylphenylalanine (S)-F1uoropyrrolidine analogues R 0 Ac~Q X Compa Ar X R p(ii1cso) 1 4-FPh F CONMC I-methylpyridin-2( 1 H)-one-5-yl F CONMC imidazo[1,2-a]pyridin-6-yl F CONMe ,2,4-biazolo[4,3-a]pyridin-6-yl F CONMe ,2,4-biazolo[ 1,5-a ]pyridin-6-yl F CONMC ,2,4-biazolo[1,5-a]pyridin-7-yl F CONMe pyrazolo[1,5-a]pyridin-5-yl F CONMe imidazo[ 1,2-b ]pyridazin-6-yl F CONMe methyl-l,2,4-biazolo[ 1,5-a ]pyridin-6- yl yl 5-methyl-1,2,4-biazolo[ 1,5-a ]pyridin-6-8-methyl-l,2,4-biazolo[1,5-a]pyridin-6- yl F CONMe F CONMe F CONMe b 12 1,2,4-biazolo[1,5-a]pyridin-6-yl F CONMe ,2,4-biazolo[1,5-a]pyridin-6-yl H CONMe ,2,4-biazolo[ 1,5-a ]pyridin-6-yl -di-f CONMe ,2,4-biazolo[1,5-a]pyridin-7-yl -di-f CONMe pyrazolo[ 1,5-a ]pyrimidin-5-yl -di-f CONMe FPh F Me Ph-S02-NH- F CONMe Me- Ph-S02-N(Me)- F CONMe Me- Ph-CO-N(Me)- F CONMe F- Ph-CO-N(Me)- F CONMe F- Ph-CO-N(Me)- F Me Me-CO-NH- F CONMe

10 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine Me-CO-NH- Me-CO-N(Me)- Me-CO-N(Me)- F F F a compounds 1 to 16 taken from reference 38. Me 6.22 CONM~ 6.82 Me 7.68 b 3-fluoroazetidin-l-yl is placed instead of3-fluoropyrrolidin-l-yl head in the general structure ofbiarylphenylalanine (S)-Fluoropyrrolidine analogues. B) 4-aminocyclohexylalaninepyrrolidine analogues DY. NH2 Q R1 ~ F Compa Rl R2 p(liicso) 27 2,4-di-F-Ph-S02NH- H ,4-di-F-Ph-S02NH- CONMe ,4-di-F-Ph-S02NH- Me CF3-Ph-S02NH- CONMe CF3-Ph-S02NH- Me F-Ph-CON(Me)- CONM~ F-Ph-CON(Me)- Me Me-Ph-CONH- Me Me-Ph-CON(Me)- Me Me-CO-N(Me)- CONMe Me-CO-N(Me)- Me cpropyl-co-n(me )- Me 7.68 a compounds 17 to 38 taken from reference 39. C) 4-aminocyclohexylglycine pyrrolidine/thiazolidide analogues 150

11 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine... Comp 8 R X p(liicso) 39 H S IPhS02 S CF3OPhS02 S CNPhS02 S ,3,4-triFPhS02 S Naptbyl S02 S Me2NCONHPh S02 S IPhCO S CF30PhCO S ,4-diFPhCO S ,4-diCIPhCO S I-NapthylCO S NapthylCO S QuinolinylCO S PhCH2OCO S ,4-diCIPhCH2 OCO S I-NapthyICH2OCO S NapthylCH20CO S IPhNHCO S CF3OPhNHCO S ,4-diCIPhNHCO S ,4-diFPhNH CO S CF 3 OPhS02 CH ,4-diFPh S02 CH ,3,4-triFPh S02 CH Me2NCONHPh S02 CH CF 3CH2S02NHPhS02 CH ,4-diFPhCO CH I-NapthylCO CH QuinolinylCO CH PhCH20CO CH a compounds 39 to 69 taken from reference

12 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine... D) 3-substituted-cycloalkylglycine pyrrolidine/thiazolidide analogues Y~ a N) NH2 LX y0~jy) NH2 Lx Compounds Compa Configuration p(liicso) X y a 1 3 (S, S, R Derivatives) 70 S S R CH2 OH S S R CH2 OCONHPh-4-0Me S S R S OCONHPh-4-0Me S S R CH2 OCONHPh-4-I S S R S OCONHPh-3,4-Ch S S R S OCONHPh-3-F S S R CH2 OCONHPh-2-Ph S S R CH2 NH CO-I-naphthyl S S R CH2 NH CONH -i-naphthyl S S R CH2 NHS02Ph-4-0CF (R, S, R Derivatives) 80 R S R S OCONH -i-naphthyl R S R S OCONHPh-3-F 5.92 (S, S, S Derivatives) 82 S S S S NH CO-I-naphthyl S S S S NHCONH-l-naphthyl S S S CH2 NHS02Ph-4-C02H S S S CH2 NHS02Ph-4-S02NH S S S CH2 NHS02Ph-4-S02NMe S S S CH2 NHS02Ph-4-S02Me S S S CH2 NHS02Ph-4-S02CF S S S CH2 NHS02Ph-2-S02Me S S S CH2 NHS02(CH2)2NHS02Me S S S CH2 OH

13 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine S S S CH2 OCONHPh-4-0Me S S S CH2 OCONHPh-4-I 6.74 (S, R, S Derivatives) 94 S R S CH2 OCONHPh-4-I S R S CH2 NHCONH-1-naphthyl S R S CH2 NHS02Ph-4-S02NH S R S CH2 NHS02Ph-4-S02Me 6.38 (S, R, R Derivatives) 98 S R R CH2 NHCONH-1-naphthyl S R R CH2 NHS02Ph-4-0CF S R R CH2 NHS02Ph-4-C02H S R R CH2 NHS02Ph-4-S02NH S R R CH2 NHS02Ph-4-S02Me S R R CH2 NHS02(CH2hNHS02Me S R R CH2 OCONHPh-4-I S S R CH2 OH S S R CH2 OCONHPh-4-0Me S S R S OCONHPh-4-0Me S S R CH2 OCONHPh-4-I S S S CH2 OH S S S CH2 OCONHPh-4-0Me S S S S OCONHPh-4-0Me S S S CH2 OCONHPh-4-I 6.48 a compounds 70 to 112 taken from reference 34. E) fl-methyl phenylalanine pyrrolidine analogues - 0 : : X "- N ---.::::: NH2 / R o N Compa R X p(liicso) 113 H (S)-3-fluoropyrrolidin-l-yl F (S)-3-fluoropyrrolidin-l-yl

14 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine Me (S)-3-fluoropyrrolidin-l-yl 2-Me,5-Br (S)-3-fluoropyrrolidin-l-yl 5-F (S)-3-fluoropyrrolidin-l-yl 5-Br (S)-3-fluoropyrrolidin-l-yl 6-F (S)-3-fluoropyrrolidin-l-yl 6-Me (S)-3-fluoropyrrolidin-l-yl 6-CH20H (S)-3-fluoropyrrolidin-l-yl 6~OMe 5-Cl,6-F 5-F, 6- NHC02Et H H H H H NH NEt cpr-ch2-n CH2C02Et-N CF2C02Et-N 4-FPh-N (S)-3-fluoropyrrolidin-1-yl (S)-3-fluoropyrrolidin-1-yl (S)-3-fluoropyrrolidin-1-yl 3,3-difluoropyrrolidin-l-yl pyrrolidin-1-yl thiazolidin-3-yl 3-fluoroazetidin-l-yl 4-fluoropiperidin-1-yl (S)-3-fluoropyrrolidin-l-yl (S)-3-fluoropyrrolidin-l-yl (S)-3-fluoropyrrolidin-l-yl (S)-3-fluoropyrrolidin-l-yl (S)-3-fluoropyrrolidin-l-yl (S)-3-fluoropyrrolidin-l-yl a compounds 113 to 135 taken from reference F) Glutamate cyanopyrrolidine derivatives o N~ en Compound

15 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine.... Compa Rl R2 p(lilc so ) 136 H H ,7-(OMeh H ,7-(OMeh -(CH2hOH ,7-(OMeh Isopropyl ,7-(OMeh Benzyl ,7-(OMeh tert-butyl Me tert-butyl Me tert-butyl H -CH(4-FCJlsh H Nicotinonitrile H Benzoyl H H H Benzyl H Ethyl H Isopropyl H tert-butyl ,4-0Me H H -CH 2 OMe H Isopropyl 6.52 a compounds 136 to 154 taken from reference 35. G) Oxadiazole pyrrolidine derivatives ~V)0N CI NH2 0 x p(liic so ) ')--N}- N- O 6.69 o ~r 6.25 N-N 155

16 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine... " y-n y 5.92 O-N \[N~ yn HN-~> N X_ ~N" ~ 0 R2 N-O NH2 "-;. Compa R. R2 X p(lilcso) 160 Me 4-NHS02Me H Me 4-NHS02Me F"I Me 4-NHS02Me F Me 4-NHS02Me dif Me 2-CI, 4-S02Me F cpr-ch2 2-CI, 4-S02Me F ipr-ch2 2-CI, 4-S02Me F (CH2)20H 2-CI, 4-S02Me F cpr-ch2 4-S02Me F cpr-ch2 2-Me, 4-S02Me F cpr-ch2 2-F,4-S02Me F Me H H Me 2-CHF2 H Me 2-0CF3 H Me 2-F H Me 2-CF3 H Me 2-CI H Me 3-CF3 H Me 4-CI H Me 4-0CF3 H Me 4-CF3 H Me 4-S02CF 3 H 6.48 X 156

17 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine Me Me Me Me Me Me Me Me Me 4-S02NH2 4-S02Me 4-NHS02Me 3-Cl,5-Cl 2,4-diF 2-Cl,4-Br 2-Cl,4-F 2-Cl, 4-S02Me 2-F,4-S02Me acompounds 155 to 190 taken from reference 37. H 6.72 H 6.91 H 6.28 H 5.13 H 5.34 H 6.80 H 6.08 H 7.77 H 7.37 Pyrrolidine ~ X=F,CN &H 1) 2) 3) NH2 <j?0,n.,me-o- 4)~ Me NH2 5) ~ -0- NH2 6) )ly 7);; ~Ny 'I Biatylphenylalanine (S)- Fluoropyrrolidine analogues -HND -R 4-aminocyclohexylalanine pyrrolidine # analogues -HND 1 -R 4-aminocyclohexylglycine pyrrolidinel # thiazolidide analogues -HND -R 1 3-substituted-cycloalkylglycine # pyrrolidine/thiazolidide analogues ~N-R beta-methyl phenylalanine pyrrolidine ~ analogues o Tetrahydroisoquinoline, P iperazine and phenyl. < moieties ) R Glutamate cyanopyrrolidine derivatives O"d""',pyrrolidm,doriwH~ NH2 Figure 1. Common substructure and structural classes ofpyrrolidine based DPP-IV inhibitors. 157

18 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine Figure 3. The superposition of 190 pyrrolidine DPP-IV inhibitors based on atom based alignment method CoMF A and CoMSIA field generation For deriving the CoMF A indices, the molecular fields were sampled at the default density (i.e. lattice spacing) of 2A with an extension of 4A beyond the aligned molecules in all directions. CoMF A descriptors were calculated using an Sp3 carbon probe atom with a van der Waals radius of 1.52 A and a charge of to generate steric (Lennard-Jones 6-12 potential) field energies and electrostatic (Coulombic potential) fields with a distance-dependent dielectric at each lattice point. Steric and electrostatic fields generated were scaled by the CoMF A-Standard method in SYBYL with default cutoff energy of 30 kcallmol. The minimum column filtering was set to 2.0 kcallmol in order to minimize the influence of noisy columns. 43 The CoMSIA indices were derived according to Klebe et al. 44 The same lattice box as used for the CoMF A was adopted for the CoMSIA indices with a grid spacing of la and employing a Cl+ probe atom with a radius of l.oa as implemented in SYBYL. The CoMSIA similarity indices (AF) for molecule j with atoms i at a grid point q were calculated using equation 1. (1) 159

19 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine.... The CoMSIA method incorporates five different physicochemical properties (k) namely steric, electrostatic, hydrophobic, hydrogen bond donor and hydrogen bond acceptor (equation 1), which were evaluated using the probe atom. A Guassian type distance-dependence was used between the grid point q and each atom i in the molecule. A default value of 0.3 was used as the attenuation factor (a). In CoMSIA, the steric indices are related to the third power of the atomic radii; the electrostatic descriptors are derived from partial atomic charges; the hydrophobic fields are derived from atom based parameters;4s and the hydrogen-bond donor acceptor atoms within a putative protein environment are derived from experimental values CoMFA and CoMSIA Model Derivation The 3D QSAR models of CoMF A and CoMSIA descriptors were derived using PLS regression as implemented in the SYBYL. 47 The predictive ability of the model was quantitated in terms of the r cv which is defined as 1'2. cw = 1 _ I(VprM!!IteCI-~tndl l;(v.'l'vtd-yi)jfw Where Ypredicted, Yobserved and Y mean are predicted, actual and mean values of the target property (pic so), respectively. L (Ypredicted -Yobservedi is the predictive sum of squares (PRESS). The optimum number of components (ONC) which corresponds to the lowest standard error of prediction (SEP) were used to generate the final PLS regression models. CoMF A and CoMSIA contour maps were generated from the field energies at each lattice point calculated as the scalar results of the coefficient and the standard deviation associated with a particular column of the data. This was plotted as the percentage of the contribution to the CoMF A or CoMSIA equation. The contour plots help in identifying the important regions where changes may affect the binding preference. Furthermore, they will be helpful in recognizing the important features contributing to the interactions between the ligand and the active site of a receptor. The boot strapping analysis for 100 runs and the cross-validation analysis (e.g. leavehalf-out and leave 20% out; each 50 runs) were carried out and the models were confirmed by the averages from each simulation Predictive r2 value The test set compounds are not included in deriving the CoMF A and CoMSIA models. The activity of the test set was predicted by the CoMF A and CoMSIA models (2) 160

20 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine.... using the predict property command. The predictive (l was based only on molecules not included in the training set and is computed using the following equation r21'"<1 = (SD - PRESS)/SD (3) where SD is the sum of the squared deviations between the biological activity of molecules in the test set and the mean biological activity of the training set molecules and PRESS is the sum of the squared deviations between predicted and actual activity values for every molecule in the test set. 5.7 Results and discussion CoMF A and CoMSIA statistics The statistics emanating from the CoMF A and CoMSIA models of pyrrolidine based DPP-IV inhibitors for type 2 diabetes are shown in Table 2. The best models were selected on the basis of highest Q2 value and test set predictive (l value. 49 The CoMFA model has emerged from 6 components and showed a cross-validated Q2 of with a non-cross validated fl of This has explained 70.6 per cent variance in the DPP-IV enzyme inhibitory activity of the test set compounds (fl pred = 0.706). In this model, the stericand electrostatic fields have respectively contributed 59.8% and 40.2% to the explained enzyme inhibitory activity of the compounds. Table 2. Summary of CoM FA and CoMSIA investigation of the DPP-IV series. PLS parameters f2ncv a SEE b F test C (lcv d SEP e fl pred f PLS components g Contributions Steric Electrostatic H -bond donor H-bond acceptor CoMFAmodel CoMSIA model

21 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine.... -l'boot h SEE boot i l'lho j l'sd k l'scv I SDscv m a The conventional values correlation coefficient b Standard error of estimate c Ratio of-l' explained to unexplained = -l'/(1 --l') d Leave-one-out Predicted cross-validated correlation coefficient e Standard error of prediction f External predicted correlation coefficient for test set of compounds g Optimal number of principal components h Average of correlation coefficient for 100 samplings using bootstrapped method i Average standard error of estimate for 100 samplings using bootstrapped method j Average cross-validated correlation coefficient for 50 runs using leave-half-out (LHO) group k Standard deviation of average cross-validated correlation coefficient for 50 runs. ] Average cross-validated correlation coefficient for 50 runs using five crossvalidation group m Standard deviation of average cross-validated correlation coefficient for 50 runs The experimental and the CoMFA predicted DPP-IV enzyme inhibitory activity values for the training set and test set are reported in Tables 3 and 4 respectively. Figure 4 shows the plot of experimental and CoMF A predicted picso values of training and test sets. Table 3. Experimental and predicted binding affinities [(p(l/icso)] ofthe training set compounds. Exp CoMFA CoMSIA Compound Residual Residual p(liicso) Pred (SEDA) Pred

22 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine '

23 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine l

24 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine II

25 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine

26 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine Table 4. Experimental and predicted binding affinities [p(liicso)] of the test set Compounds. Compound CoMSIA Exp CoMFA Residual (SEDA) p(liicso) Pred Pred Residual

27 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine For the CoMSIA study different combinations of fields, from two to five were investigated in order to identify the potential molecular features to explain the DPP- IV enzyme inhibitory activity of the compounds. From this a four-field (steric, electrostatic, HB-donor and HB-acceptor) 3-component CoMSIA model has 168

28 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine optimally explained the activity of the compounds with a cross-validated Q2 value of and a non-cross validated? value of (Table 2). This model has predicted the external test set with a predictive? value of l Training set A Test set 8.00 <».e ~ ~ ': ~.~ -; ~ 7.00 J '.00. -' Experimental value Figure 4. The experimental pic so versus the predicted plc so binding affinity values for the training set and the test set based on CoMF A model. Training set A Test set 9 8 < - ~6 - f ~ , , , EXl)erimental value Figure 5. The experimental pic so versus the predicted pic50 binding affmjty values for the training set and the test set based on CoMSIA (SEDA) model. 169

29 Chapter 5 CoMF A and CoMSIA Investigation of Diverse Pyrrolidine The inclusion of hydrophobic field to this has reduced the predictive significance of resultant CoMSIA model. This may be viewed as that hydrophobic field has a modest importance to DPP-N enzyme inhibitory activity of the compounds. The predicted activities of the training and test sets compounds from the four field (steric, electrostatic, HB-donor and HB-acceptor) CoMSIA model are shown in Tables 3 and 4 respectively. Figure 5 shows the plot of experimental and CoMSIA predicted activities of the training and test sets. Both the CoMF A and CoMSIA models have showed the robustness in bootstrapping and cross-validation analysis (Table 2). In two-group cross-validation study, the average ~cv values for CoMFA and CoMSIA were respectively 0.55 and 0.59 (standard deviations 0.05 and 0.04). In five-group cross-validation study, for CoMFA model the average (lcv and SD values were 0.61 and 0.03 respectively and for CoMSIA the same were 0.64 and 0.02 respectively. The bootstrapped ~ were for CoMF A and for CoMSIA. These results suggest the consistency and robustness of the models CoMF A steric contour plot In CoMF A steric contour map (Figure 6) the green color contours indicate the position Qfbulky groups necessary for the DPP-N enzyme inhibitory activity in these pyrrolidine based analogues. The yellow color contours of this map indicate the unfavorable (non-tolerant) sites of bulky group in pyrrolidine based analogues for the active site of the DPP-N enzyme. Among the yellow steric contours, two are located below the pyrrolidine moiety (head part) which indicates necessity of minimum steric occupancy of this region. This further suggests that five and three membered rings of pyrrolidine and azetidine containing compounds show higher affinity (compounds 5, 6, 12, 15, 65, 160 etc.) to the enzyme than the compounds with six membered ring of piperidine moiety (compound 129). In the series of oxadiazole based pyrrolidine analogues, one large yellow contour cavity and two small yellow contours enclose the phenyl ring. This indicates the bulky substitutents present at meta and ortho position lead to decrease the binding affinity (compounds 172, 174, 177 and 185) culminating in low biological activity. The presence of green contour at para position of the phenyl ring shows the favorable steric occupancy of methanesulphonyl and methanesulfonamide substitutents leading to better inhibition (compound 160, 161, 166 and 170). Most of cyclohexylglycine pyrrolidine analogues (compounds 43, 45, 170

30 Chapter 5. CoMFA and CoMSIA Investigation of Diverse Pyrrolidine , SO, 52, 53 and 65} have shown the highest enzyme affinity in the dataset. This was indicated by the large green contour wrapping from cyclohexyl ring to aromatic carbonyl and sulphonyl moiety of tail portion. The upper and lower portion of the tail part (triazolopyridine, pyridone and other large size moieties) ofpyrrolidine inhibitors are covered by two sterically disfavored stretched yellow contours. It explains that increasing the steric group on these moieties decrease the enzyme inhibitory activity. Accordingly, the following compounds 29, 30, 115, 124, 129 and 132 showed moderate to lower biological activity CoMF A electrostatic contour plots In the CoMF A electrostatic field contour map (Figure 7) areas where electronegative groups increase the binding affinity are represented in red, while areas where electropositive groups increase the binding affinity are represented in blue. The red contour in between 2 nd and 3rd position of pyrrolidine ring reflect the electronegative substitutions at this position improve the binding affinity in compounds (e.g. -F, S and -CN in compounds 2, 5, 45, 52, 57, 83, 136, 138 etc.), while the lack of such substitutions in compounds (e.g. compounds 86, 87, 88, 110, 157, 158 etc.) will reduce the binding affinity toward the DPP-IV enzyme. Two red contours near to the aryl tail position of the all type of pyrrolidine based analogues suggest that placement of electron rich groups such as -F, =0, -OCF3, triazole, S02Me, -OMe and -CH20H at this position (e.g. compounds 5, 6, 27, 32, 90, 107, 121 etc.) would improve the binding affinity. A large hyperbolic shaped blue contour located below the phenyl, cyclohexane and cyclopentane linkers indicate that positively charge containing rings make them more hydrophobic and enhances the biological potency. It suggests the necessity of electropositive ring linker for the DPP-IV enzyme inhibitory activity. In oxadiazole analogues, two blue contours closed to the upper and lower portion of the oxadiazole ring. This type of contour in oxadiazole pyrrolidine analogues suggests that electronegative charges of oxadiazole's nitrogen and oxygen reduce the enzyme inhibition property. One large blue and two small red contours placed at the end of the phenyl ring (tail portion) of oxadiazole analogues indicates that dual nature substituents containing electropositive as well as negative groups (methanesulfonamide and methanesulphonyl) are required in this prototype for good binding affinity with the DPP-IV enzyme (e.g. compounds 160, 161 and 166). 171

31 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine Figure 6. Steric CoMFA contour maps shown in the presence of the compounds 5 and 160 (green: bulky group favorable; yellow: bulky group unfavorable). Figure 7. Electrostatic CoMFA contour maps based on the compounds 5 and 160 (Red: negative charges favorable; blue: positive charges favorable). 172

32 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine CoMSIA steric and electrostatic contour plots In CoMSIA contour map of steric field (Figure 8a), the favorable steric areas for the DPP-IV enzyme binding are shown in green contour whereas the disfavored steric areas for the enzyme are indicated in yellow contour. In the oxadiazole substituted pyrrolidine analogues, a large yellow contour is located around the oxadiazole moiety. Most of these analogues (compounds 156, 157, 174, 175, 185, 186 etc.) have shown moderate to low biological activity because of the central ring linker (oxadiazole moiety) has entirely turned away from the most active location (e.g. compound 5). A V-shaped yellow contour is present in the vicinity of cyc10pentyl portion of the S,S,Rand R,S,R- configuration of cyc10pentylglycine based pyrrolidine analogues. Compounds with these configurations (compounds 70, 73 and 80) have shown moderate to low activity due to the change in the orientation of the aryl substitution on the cyc10pentane ring. The other configuration such as SSS, SRS, SRR and SSR (compounds 83, 84, 96, 98, 107 etc.) have shown higher binding affinity due to correct adaptation of aryl substitution in the active site of DPP-IV. The extended green contour from ring linkers (cyc1opentane, cyc10hexane and phenyl rings) to aryl tail substituents suggest that a wide steric bulk in this portion of the molecule is essential for the inhibitory activity. The CoMSIA electrostatic contour map (Figure 8b) indicates that the introduction of electronegative substituents in red regions may favor the binding with the DPP-IV enzyme while the same in blue regions decrease the binding affinity. A stretched out blue contour spread from the head pyrrolidine moiety to the tail aryl moiety indicates that elongated electropositive potential surface is essential for the enzyme inhibitory activity. A small blue contour has mapped the para position of the phenyl ring of oxadiazole based pyrrolidine derivatives. It has provided the electropositive substituents for better enzyme inhibitory activity in these analogues (compounds 160, 162 and 166). A star shaped red contour is located at the end of the aryl moiety. It indicates that electron rich substitutents such as -CI, -F, triazole and pyridone in the tail portion ofthe.analogues enhance their enzyme inhibitory activity. 173

33 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine Figure 8a. CoMSIA contour maps shown in the presence of the compounds 5, 83 and 160 for steric field (green: bulky group favorable; yellow: bulky group unfavorable). Figure 8b. CoMSIA contour maps shown in the presence of the compounds 5, 83 and 160 for electrostatic field (Red: negative charges favorable; blue: positive charges favorable). 174

34 ChapterS CoMFA and CoMSIA Investigation of Diverse Pyrrolidine CoMSIA donor and acceptor contour plots In CoMSIA donor contour plot (Figure 9), the cyan color indicates the regions where hydrogen donor group is favorable for enzyme inhibitory activity, while the purple color indicates the regions where the hydrogen donor group is unfavorable for the activity. In cyclohexylglycine pyrrolidine derivatives, one large cyan contour is close to the sulphonamide group (-S02NH-) attached to the 4th position of the cyclohexane ring linker. It indicates that hydrogen bond donor functionalities in this region enhance the DPP-IV enzyme inhibitory activity of the compounds (e.g. compound 45 and 65). Five purple contours are positioned around the 3rd position of the cyclopentylglycine pyrrolidide derivatives. This Contour disposition signifies that hydrogen donor groups are not favorable at 3rd position of the cyclopentane ring. This explains the poor inhibitory activity of most of the compounds in this series against thedpp-n enzyme. The CoMSIA acceptor contour plot is shown in Figure 10 with magenta color notifying the regions where hydrogen bond acceptor groups increase the DPP-IV enzyme binding affinity and red color notifying the regions where hydrogen bond acceptor group decrease the affinity. A small magenta contour close to the 2 nd and 3rd positions of the pyrrolidine moiety suggest that electron rich acceptor groups such as - F, Sand -CN in this region would enhance the DPP-IV enzyme binding affinity of the compounds (e.g. compounds 2, 5, 45, 52, 57, 83, 136, 138 etc.). Absence of the disposition of such electron rich acceptor groups can make the compounds less active ones (e.g. compounds 86, 87, 88, 110, 157, 158 etc.). This explains the importance of hydrogen bond acceptor groups in the vicinity of pyrrolidine moiety for enhancing the DPP-IV binding affinity. The large magenta contour near the para position of phenyl ring of oxadiazole based DPP-IV inhibitors indicates the location of hydrogen bond acceptor groups in these compounds. It has well explained the high inhibitory activity of compound 160 which has methanesulfonamide acceptor group at the para position of the phenyl ring. Two large red contours are located on the top and bottom of oxadiazole moiety of oxa4iazole based pyrrolidine derivative. They indicate the location of unfavorable hydrogen bond acceptor nitrogen and oxygen functions in the central ring linker region of the analogues. They are detrimental to the activity of these compounds. In glutamate cyanopyrrolidine derivative, two red contours (one small and one big) are in the proximity of amide (-CONH-) linker. They indicate the 175

35 Chapter 5 CoMFA and CoMSIA Investigation of Diverse Pyrrolidine unfavorable hydrogen bonding character of amide linker in this region for the DPP-IV binding affinity. Accordingly, the glutamate cyanopyrrolidine derivatives (e.g. compounds 143, 144, 148 and 149 etc.) have shown moderate DPP-IV inhibitory activity when compared to the other ring linkers such as phenyl, cyclohexane and cyclopentane in the investigated analogues. Figure 9. CoMSIA H-bond donor contour map shown in the presence of compounds 5 and 160 (cyan : H-bond donor favorable; purple: H-bond donor unfavorable). Figure 10. CoMSIA H-bond acceptor contour map shown in the presence of compounds 5 and 160 (magenta: H-bond acceptor favorable; red: H-bond acceptor unfavorable). 176