2D-QSAR Study of Fluoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents

Transcription

1 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 559 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents Mahesh B. Palkar 1*, Malleshappa. oolvi 2, Harun M. Patel 2, Veeresh S. Maddi 1 and L.V.G. argund 3 1 Department of Pharmaceutical Chemistry, K.L.E.Soceity s College of Pharmacy, Vidyanagar, Hubli , Karnataka, India. 2 Department of Pharmaceutical Chemistry, ASBASJS Memorial College of Pharmacy, Bela (Ropar) , Punjab, India 3 Department of Pharmaceutical Chemistry, SET s argund College of Pharmacy, Dattatreyanagar, Bangalore , Karnataka, India. ABSTRACT: QSAR studies were performed on a set of 39 analogs of fluoroquinolone using MDS vlife science QSAR plus module by using. Multiple Linear Regression (MLR), Principal Component Regression (PCR) and Partial Least Squares (PLS) Regression methods. Among these, MLR method has shown a very promising result as compared to other two methods and a QSAR model was generated by a training set of 28 molecules with correlation coefficient (r2) of , significant cross validated correlation coefficient (q2) of and test of In the selected descriptors, estate contribution, chi, path cluster and alignment independent descriptors were the most important descriptors in predicting antitubercular inhibitory activity. KEYWRDS: Anti-tubercular activity; luoroquinolones; Mycobacterium tuberculosis; Multiple Linear Regression (MLR); Principal Component Regression (PCR); Partial Least Squares (PLS) ITRDUCTI Tuberculosis (TB) continues to be a major health concern. The World Health rganisation (WH) declared TB a global emergency in 1993 [1]. The emergence of AIDS and decline of socioeconomic standards have contributed to the disease s resurgence in industrialized countries. In most developing countries, although the disease has always been endemic, its severity has increased because of the global HIV pandemic and extensive social restructuring due to rapid industrialization and conflicts. There is now recognition that new drugs to treat TB are urgently required, specifically for use in shorter treatment regimens than are possible with the current agents and which can be employed to treat multi-drug resistant and latent disease. During the past decade, several of the fluoroquinolone antibacterial drugs have been examined as potential chemotherapeutics for Mycobacterium tuberculosis infection because of their favorable pharmacokinetic * or correspondence: Mahesh B. Palkar, Tel.: (off), (res) ax: Mob.: bpmahesh1@yahoo.com, bpmahesh1@gmail.com profiles such as easily absorbed after oral administration and readily penetrated into mammalian cells [2]. luoroquinolones exhibit potent in vitro and in vivo antimycobacterial activity [3]. Quinolones inhibit DA gyrase, which is essential enzyme catalyzing DA supercoiling and deactivation reactions, respectively [4]. There is also a considerable effort to discover and develop newer fluoroquinolones, and some of them might have value in the treatment of TB [5]. The key to improving therapy is to develop new agents with potent sterilizing activity that will lead to a shortening of the duration of chemotherapy. Several of the quinolone antibacterial drugs have been examined as inhibitors of MTB as well as other mycobacterial infections [6]. As a result, gatifloxacin and moxifloxacin (8-methoxy quinolones) are drugs in pipeline and to be approved by DA for the treatment of TB by the year Considering this panorama several fluoroquinolone derivatives have been designed and synthesized as attractive anti-tubercular drug candidates. Due to the hypothesis that the introduction of a lipophilic character at C7 of the 4-oxoquinoline-3-carboxylic acid heterocyclic scaffold of fluoroquinolone analogues could modulate or improve the anti-tubercular properties in vitro [7, 8]. 559

2 560 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 ne could not, however, confirm that the compounds designed would always possess good inhibitory activity to Mycobacterium tuberculosis, while experimental assessments of inhibitory activity of these compounds are time-consuming and expensive. Consequently, it is of interest to develop a prediction method for biological activities before the synthesis [9]. Quantitative structureactivity relationship (QSAR) searches information relating chemical structure to biological and other activities by developing a QSAR model. Several molecular descriptors are used to quantify the structural feature of lead molecule. The purpose of using QSAR-Descriptors is to calculate the properties of molecules that serve as numerical descriptions or characterizations of molecules in other calculations such as diversity analysis or combinatorial library design. Using such an approach one could predict the activities of newly designed compounds before a decision is being made whether these compounds should be really synthesized and tested. Moxifloxacin and gatifloxacin are highly active compounds against Mycobacterium tuberculosis, and have become a core structure for the design of new anti-tubercular drug candidates. To date few studies have been undertaken to optimize the fluoroquinolones against M. tuberculosis [10, 11]. In view of the above facts QSAR study is performed on structurally-related fluoroquinolone derivatives in order to get a better understanding of their structural features and anti-tubercular activity. Table 1 Structure, Experimental and Predicted Activity of Quinazolines Used in Training and Test Set Using Model 1 (MLR). S.. Compounds IC 50 a (µg/ml) pic 50 b Residual Exp. Pred C H 2 T H C H H T

3 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 561 S.. Compounds C H IC 50 a (µg/ml) Exp. pic 50 b Pred. Residual 6 H C H 7 T H 3 C

4 562 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 S.. Compounds IC 50 a (µg/ml) pic 50 b Residual Exp. Pred. C H 12 H 3 C T H C H C H 14 H

5 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 563 S.. Compounds IC 50 a (µg/ml) pic 50 b Residual Exp. Pred C H 19 T H T C H C 2 H 5 H

6 564 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 S.. Compounds IC 50 a (µg/ml) pic 50 b Residual Exp. Pred. 24 T C H 26 H T C H H 2 C H H 2

7 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 565 S.. Compounds IC 50 a (µg/ml) pic 50 b Residual C H Exp. Pred H 2 C H 31 T H 2 C H H

8 566 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 S.. Compounds IC 50 a (µg/ml) pic 50 b Residual Exp. Pred T C H 38 T H CH Expt. = Experimental activity, Pred. = Predicted activity a = Compound concentration in micro mole required to inhibit growth by 50% b = -Log (IC ): Training data set developed using model 1 (MLR) T = Test Set Computational Methods Chemical Data A series of 39 molecules belonging to fluoroquinolone derivatives as Mycobacterium tuberculosis inhibitors were taken from the literature and used [12-14]. The 2D-QSAR models were generated using a training set of 28 molecules. The observed and predicted biological activities of the training and test set molecules are presented in Table 1. Predictive power of the resulting models was evaluated by a test set of 11 molecules with uniformly distributed biological activities. The observed selection of test set molecules was made by considering the fact that test set molecules represents a range of biological activity similar to the training set. Data Set All computational work was performed on Apple workstation (8-core processor) using Vlife MDS QSAR plus software developed by Vlife Sciences Technologies Pvt Ltd, Pune, India, on windows XP operating system. All the compounds were drawn in Chem DBS using

9 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 567 fragment database and then subjected to energy minimization using batch energy minimization method. Conformational search were carried out by systemic conformational search method. Biological Activities The negative logarithm of the measured IC 50 (µm) against Mycobacterium tuberculosis as pic 50 [pic50 = log (IC 50 X 10-6 )] was used as dependent variable, thus correlating the data linear to the free energy change. Since some compounds exhibited insignificant/no inhibition, such compounds were excluded from the present study. All the values had been obtained using the trypomastigote form of Mycobacterium tuberculosis and the assays were performed in Dulbecco s modified Eagle medium [12-14]. The IC 50 values of reference compounds were checked to ensure that no difference occurred between different groups. The pic 50 values of the molecules under study spanned a wide range from 2 to 6. Molecular Descriptors Various 2D descriptors (a total of 208) like element counts, molecular weight, molecular refractivity, log P, topological index, Baumann alignment independent topological descriptors etc., were calculated using VlifeMDS software. The preprocessing of the independent variables (i.e., descriptors) was done by removing invariable (constant column) and cross-correlated descriptors (with r = 0.99) which resulted in total 156, 125 and 162 descriptors for MLR, PCR and PLS respectively to be used for QSAR analysis. Selection of Training and Test Set The dataset of 39 molecules was divided into training and test set by Sphere Exclusion (SE) method for MLR, PCR and PLS model with pic 50 activity field as dependent variable and various 2D descriptors calculated for the molecules as independent variables. Model Validation This is done to test the internal stability and predictive ability of the QSAR models. Developed QSAR models were validated by the following procedure: Internal Validation Internal validation was carried out using leave-one-out (q 2, L) method. or calculating q 2, each molecule in the training set was eliminated once and the activity of the eliminated molecule was predicted by using the model developed by the remaining molecules. The q 2 was calculated using the equation which describes the internal stability of a model. Where y i and ˆy i are the actual and predicted activity of the ith molecule in the training set, respectively, and y mean is the average activity of all molecules in the training set. External Validation or external validation, the activity of each molecule in the test set was predicted using the model developed by the training set. The pred_r 2 value is calculated as follows. Where y i and ˆy i are the actual and predicted activity of the ith molecule in the training set, respectively, and y mean is the average activity of all molecules in the training set. Both summations are over all molecules in the test set. Thus, the pred_r 2 value is indicative of the predictive power of the current model for external test set. Randomization Test To evaluate the statistical significance of the QSAR model for an actual dataset, one tail hypothesis testing was used. The robustness of the models for training sets was examined by comparing these models to those derived for random datasets. Random sets were generated by rearranging the activities of the molecules in the training set. The statistical model was derived using various randomly rearranged activities (random sets) with the selected descriptors and the corresponding q 2 were calculated. The significance of the models hence obtained was derived based on a calculated Z score. A Z score value is calculated by the following formula: Where h is the q 2 value calculated for the actual dataset, µ the average q 2, and s is its standard deviation calculated for various iterations using models build by different random datasets. The probability (a) of significance of randomization test is derived by comparing Z score value with Z score critical value as reported in reference, if Z score value is less than 4.0; otherwise it is calculated by the formula as given in the literature. or example, a Z score value greater than 3.10 indicates that there is a probability (a) of less than that the QSAR model constructed for the real dataset is random. The randomization test suggests that all the developed models have a probability of less than 1% that the model is generated by chance.

10 568 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 QSAR by Multiple Linear Regression (MLR) Analysis Multiple regression is the standard method for multivariate data analysis. It is also called as ordinary least squares regression (LS). This method of regression estimates the values of the regression coefficients by applying least squares curve fitting method. or getting reliable results, dataset having typically 5 times as many data points (molecules) as independent variables (descriptors) is required. The regression equation takes the form Y = b1*x1 + b2*x2 + b3*x3 + c..(4) Where, Y is the dependent variable, the b s are regression coefficients for corresponding x s (independent variable), c is a regression constant or intercept. In the present study QSAR model was developed using multiple regression by forward-backward variable selection method with pic 50 activity field as dependent variable and physicochemical descriptors as independent variable having crosscorrelation limit of 20 and number of variable in final equation was 5. Selection of test and training set was done by sphere exclusion method. QSAR by Principal Component Regression (PCR) Method Principal components analysis rotates the data into a new set of axes such that the first few axes reflect most of the variations within the data. By plotting the data on these axes, we can spot major underlying structures automatically. The value of each point, when rotated to a given axis, is called the principal component value. Principal Components Analysis selects a new set of axes for the data. These are selected in decreasing order of variance within the data. They are also perpendicular to each other. Hence the principal components are uncorrelated. Some components may be constant, but these will be among the last selected. The problem noted with MLR was that correlated variables cause instability. This process gives the modeling method known as Principal Components Regression. Rather than forming a single model, as with MLR, a model can be formed using 1, 2 components and a decision can be made as to how many components are optimal. If the original variables contained collinearity, then some of the components will contribute only noise. So long as these are dropped, the models can be guarantee that our model will be stable. The QSAR model was developed using principal component regression by forward-backward variable selection method with pic 50 activity field as dependent variable and physico-chemical descriptors as independent variable having crosscorrelation limit of 18.9 and number of variable in final equation was 4. Selection of test and training set was done by sphere exclusion method. QSAR by Partial Least Squares (PLS) Regression Method PLS is an effective technique for finding the relationship between the properties of a molecule and its structure. In mathematical terms, PLS relates a matrix Y of dependent variables to a matrix X of molecular structure descriptors, i.e., a latent variable approach to modeling the covariance structures in these two spaces. PLS have two objectives: to approximate the X and Y data matrices, and to maximize the correlation between them. Whereas the extraction of PLS components is performed stepwise and the importance of a single component is assessed independently, a regression equation relating each Y variable with the X matrix is created. PLS decomposes the matrix X into several latent variables that correlate best with the activity of the molecules. PLS can be done using IPALS or SIMPLS iterative algorithm, with consecutive estimates obtained using the residuals from previous iterations as the new dependent variable The QSAR model was developed using partial least squares by forward-backward variable selection method with pic 50 activity field as dependent variable and physicochemical descriptors as independent variable having crosscorrelation limit of 19.7 and number of variable in final equation was 4. Selection of test and training set was done by sphere exclusion method. Result and Discussion Training set of 28 and 11 of test set of fluoroquinolones having different substitution, were employed. ollowing statistical measure was used to correlate biological activity and molecular descriptors; n,number of molecules; k,number of descriptors in a model; df,degree of freedom; r 2,coefficient of determination; q 2, cross validated r 2 ; pred_r 2, r 2 for external test set; pred_r 2 se, coefficient of correlation of predicted data set; Z score, Z score calculated by the randomization test; best_ran_r 2 ; best_ran_q 2,highest q 2 value in the randomization test; α, statistical significance parameter obtained by the randomization test. Generation of QSAR Models Model 1 (Multiple Linear Regression (MLR) Analysis) After 2D QSAR study by Multiple Linear Regression method using forward-backward stepwise variable selection method, the final QSAR equation was developed having 5 variables as follows. pic50 = (SaaE-index) (SsssCHEindex) (T_2_T_4) (ChiV6chain) (chiv4pathcluster)

11 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 569 Table 2 Statistical parameters of MLR, PCR and PLS Parameters MLR PCR PLS Df r q test best_ran_r best_ran_q Z score_ran_r Z score_ran_q α_ran_r α_ran_q MLR = Multiple Linear Regression, PCR = Principal Component Regression, PLS = Partial Least Squares, n = number of molecules of training set, df =degree of freedom, r2 = coefficient of determination, q2 = cross validated r2, pred_r2 = r2 for external test set, pred_r2se = coefficient of correlation of predicted data set. Model 1 has a correlation coefficient (r 2 ) of , significant cross validated correlation coefficient (q 2 ) of , test of and degree of freedom 15. The model is validated by α_ran_r 2 = , α_ran_q 2 = , best_ran_r 2 = , best_ran_q 2 = , Z score_ran_r 2 = and Z score_ran_q 2 = The randomization test suggests that the developed model have a probability of less than 1% that the model is generated by chance. The observed and predicted pic 50 along with residual values are shown in Table 1.Statistical data is shown in Table 2.The plot of observed vs. predicted activity is shown in igure 1. The descriptors which contribute for the pharmacological action are shown in igure 2. The above study leads to the development of statistically significant QSAR model, which allows understanding of the molecular properties/features that play an important role in governing the variation in the activities. In addition, this QSAR study allowed investigating influence of very simple and easy-to-compute descriptors in determining biological activities, which could shed light on the key factors that may aid in design of novel potent molecules. The present QSAR model reveals that Estate Contribution descriptor has a major contribution in explaining variation in activities. The definition and interpretation of the descriptors that were found to be dominating in the developed QSAR models is given below. An estate contribution descriptor SaaE -index (60.94%), which represents electro-topological state indice for number of nitrogen atom connected with two aromatic bonds is directly proportional to the activity. It reveals that presence of nitrogen in aromatic ring (piperidine, piperazine, pyrolidine) at 7 th position of quinoline and naphthyridine is favourable for the activity (compound 1, 2, 3 and similar analogues) SsssCHE-index (17.17%), this is also an estate contribution descriptor which represents electrotopological state indices for number of CH group connected with three single bond. This type of descriptor shows that the presence of cyclopropyl ring at -1 position is essential for the activity (compound 1, 2, 3 and similar analogues). Alignment independent descriptor T_2_T_4 (10.52%) and chiv6chain (4.69%) both are directly proportional to the activity and explains the importance of aromaticity in anti-tubercular activity. Path cluster count descriptor chiv4pathcluster (-6.69%) is inversely proportional to the activity. Model 2 (Partial Least Squares (PLS) Analysis) Model - 2 is having following QSAR equation with 4 variables. pic50 = (chiv0) (T 6) (SaaE-Index) (T_2 5) The model -2 gave correlation coefficient (r 2 ) of , significant cross validated correlation coefficient (q 2 ) of , test of and degree of freedom 28. The model is validated by α_ran_r 2 = , α_ran_q 2 = , best_ran_r 2 = , best_ran_q 2 = , Z score_ran_r 2 = and Z score_ran_q 2 = The randomization test suggests that the developed model have a probability of less than 1% that the model is generated by chance. Statistical data is shown in Table 2. The plot of observed vs. predicted activity is shown in igure 3.The descriptors which contribute for the pharmacological action are shown in igure 4. An estate contribution descriptor SaaE -index (21.08%), which represents electro-topological state indice for number of nitrogen atom connected with two aromatic bonds is directly proportional to the activity. It reveals that presence of nitrogen in aromatic ring (piperidine, piperazine, pyrolidine) at 7 th position of quinoline and naphthyridine is favourable for the activity (compound 1, 2, 3 and similar analogues) The next important chiv descriptor is chiv0, which signifies atom valency connectivity index is directly proportional to the activity and shows that presence of heavy atom such as fluorine on quinoline and naphthyridine ring has favourable effect (compound 1, 2, 3 and similar analogues). egative contribution of alignment independent descriptor T 6 (-25.22%) shows that fluorine group should not be directly attached at 6 th position of quinoline and naphthyridine ring. Similarly inverse relation of T_2 5 (-19.35%) reveals that carboxylic acid group should be at distance from quinoline and naphthyridine ring (compound 1, 2, 3 and similar analogues).

12 570 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 Predicted activity Actual activity ig.1 Graph of Actual vs. Predicted activities for training and test set molecules from the Multiple Linear Regression model. (A) Training set (Red dots) (B) Test Set (Blue dots). ig. 2 Plot of percentage contribution of each descriptor in developed MLR model explaining variation in the activity.

13 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 571 ig. 3 Graph of Actual vs. Predicted activities for training and test set molecules by Partial Least Square model. (A) Training set (Red dots) (B) Test Set (Blue dots). ig. 4 Plot of percentage contribution of each descriptor in developed PLS model explaining variation in the activity.

14 572 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 Model 3 (Principal Component Regression (PCR) Analysis) By using model 3 the final QSAR equation was developed having 4 variables as follows. pic50 = (chi5) (T_C 3) (T 7) (SaasCcount) The model -3 gave correlation coefficient (r 2 ) of , significant cross validated correlation coefficient (q 2 ) of , test of and degree of freedom 22. The model is validated by α_ran_r 2 = , α_ran_q 2 = , best_ran_r 2 = , best_ran_q 2 = , Z score_ran_r 2 = and Z score_ran_q 2 = The randomization test suggests that the developed model have a probability of less than 1% that the model is generated by chance. Statistical data is shown in Table 2. The plot of observed vs. predicted activity is shown in igure 5.The descriptors which contribute for the pharmacological action are shown in igure 6. Alignment independent descriptor T 7, signifies the number of nitrogen atom connected with other nitrogen atom by seven bond distance is directly proportional to the activity. It reveals that presence of nitrogen in aromatic ring (piperidine, piperazine, pyrolidine) at 7 th position of quinoline and naphthyridine is favourable for the activity (compound 1, 2, 3 and similar analogues). The next important alignment independent descriptor T_C 3 means the count of number of carbon atom seprated from any nitrogen atom by three bond distance. The direct relationship of this descriptor shows that presence of cyclopropyl ring at -1 position of quinoline and naphthyridine is essential for the activity. The negative involvement of Estate contribution descriptors SaasCcount, which can be defined as the total number of carbon atom connected with one single bond along with two aromatic bonds, reveals that carboxylic acid group should be at distance from quinoline and naphthyridine ring (compound 1, 2, 3 and similar analogues) for maximal activity. ig. 5 Graph of Actual vs. Predicted activities for training and test set molecules by Principal Component Regression model. A) Training set (Red dots) B) Test Set (Blue dots).

15 Mahesh B. Palkar, et al : 2D-QSAR Study of luoroquinolone Derivatives: An Approach to Design Anti-tubercular Agents 573 ig. 6 Plot of percentage contribution of each descriptor in developed PCR model explaining variation in the activity. Conclusion In conclusion, the model developed to predict the structural features of fluoroquinolone to inhibit DA gyrase, reveals useful information about the structural features requirement for the molecule. In all three optimized models, Multiple Linear Regression method is giving very significant results. The estate contribution, chi, path cluster and alignment independent descriptors were major contributors. The result obtained from QSAR study consider not only wide range of structures, but also various physic-chemical interactions involved in enzyme inhibitor complex. The present study is more versatile than the earlier reported methods. The QSAR results obtained are in agreement with the observed SAR of fluoroquinolone studied. Hence the model proposed in this work is useful and can be employed to design new derivatives of fluoroquinolones with specific inhibitory activity. Acknowledgements The authors are thankful to Dr. B. M. Patil, Principal, K.L.E. University's College of Pharmacy, Hubli, for providing necessary facilities to carry out this research work. The authors sincerely acknowledge AICTE, ew- Delhi (India), for financial support under RPS Scheme (ile o. 8023/BR/RID/RPS-170/ ). References [1] Ang, D.; Hsu, A. A. L.; Tan, B. H. Singapore Med. J. 2006, 47, 747. [2] Ball, P.; ernald, A.; Tillotson, G. Exp. pin. Investig. Drugs 1998, 7, 761. [3] Shandil, R. K.; Jayaram, R.; Kaur, P.; Gaonkar, S.; Suresh, B. L.; Mahesh, B..; Jayashree, R.; andi, V.; Bharath, S.; Balasubramanian, V. Antimicrob. Agents Chemother. 2007, 51, 576. [4] Maxwell, A. Trends Microbiol. 1997, 5, 102. [5] Anquetin, G.; Greiner, J.; Mahmoudi,.; Santillana- Hayat, M.; Gozalbes, R.; arhati, K.; Derouin,.; Aubry, A.; Cambau, E.; Vierling, P. Eur. J. Med. Chem. 2006, 41, [6] Sato, K.; Tomioka, H.; Sano, C.; Shimizu, T.; Sano, K.; gasawara, K.; Cai, S.; Kamei, T. J. Antimicrob. Chemother. 2003, 52, 199 [7] Sriram, D.; Yogeeswari, P.; Basha, J.S.; Radha, D.R.; agaraja, V. Bioorg. Med. Chem. 2005, 13, [8] Sriram, D.; Aubry, A.; Yogeeswari, P.; isher, L.M. Bioorg. Med. Chem. Lett. 2006, 16,

16 574 International Journal of Drug Design and Discovery Volume 2 Issue 3 July September 2011 [9] Bridges, A. J. Chem. Rev. 2001, 101, [10] Renau, T.E.; Sanchez, J.P.; Gage, J.W.; Dever, J.A.; Shapiro, M.A.; Gracheck, S.J. J. Med. Chem. 1996, 39, [11] Anquetin, G.; Greiner, J.; Mahmoudi,.; Santillana- Hayat, M.; Gozalbes, R.; arhati, K. Eur. J. Med. Chem. 2006, 41, [12] Kiely, J.S.; Hutt, M.P.; Culbertson,T.P.; Bucsh,R.A.; Worth,D..; Lesheski,L.E.; Gogliotti, R.D.; Sesnie, J.C.; Solomon, M.; Mich, T.. J. Med. Chem, 1991, 34 (2), [13] Laborde, E.; Kiely, J.S.; Culbertson, T.P.; Lesheski, L.E. J. Med. Chem, 1993, 36 (14), [14] Sanchez, J.P.; Domagala, J.M.; Hagen, S.E.; Heifetz, C.L.; Hutt, M.P.; ichols, J.B.; Trehan, A.K. J. Med. Chem, 1988, 31 (5),