3D QSAR studies, molecular docking and ADMET evaluation, using thiazolidine derivatives as template to obtain new inhibitors of PIM1 kinase

Accepted Manuscript Title: 3D QSAR studies, molecular docking and ADMET evaluation, using thiazolidine derivatives as template to obtain new inhibitors of PIM1 kinase Authors: Adnane Aouidate, Adib Ghaleb, Mounir Ghamali, Abdellah Ousaa, M’barek Choukrad, Abdelouahid Sbai, Mohammed Bouachrine, Tahar Lakhlifi PII: DOI: Reference:

S1476-9271(18)30036-7 https://doi.org/10.1016/j.compbiolchem.2018.03.008 CBAC 6813

To appear in:

Computational Biology and Chemistry

Received date: Revised date: Accepted date:

16-1-2018 3-3-2018 10-3-2018

Please cite this article as: Aouidate, Adnane, Ghaleb, Adib, Ghamali, Mounir, Ousaa, Abdellah, Choukrad, M’barek, Sbai, Abdelouahid, Bouachrine, Mohammed, Lakhlifi, Tahar, 3D QSAR studies, molecular docking and ADMET evaluation, using thiazolidine derivatives as template to obtain new inhibitors of PIM1 kinase.Computational Biology and Chemistry https://doi.org/10.1016/j.compbiolchem.2018.03.008 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

3D QSAR studies, molecular docking and ADMET evaluation, using thiazolidine derivatives as template to obtain new inhibitors of PIM1 kinase Adnane Aouidate*, MCNSL, School of Sciences, Moulay Ismail University, Meknes, Morocco Corresponding Author: E-mail: [email protected]; Tel. 00212638076982

Meknes, Morocco. E-mail: [email protected]

IP T

Adib Ghaleb, MCNSL, School of Sciences, Moulay Ismail University,

Mounir Ghamali, MCNSL, School of Sciences, Moulay Ismail University,

SC R

Meknes, Morocco. E-mail: [email protected]

Abdellah Ousaa, MCNSL, School of Science, Moulay Ismail University, Meknes, Morocco. E-mail: [email protected]

U

M’barek Choukrad, MCNSL, School of Sciences, Moulay Ismail

N

University, Meknes, Morocco. E-mail: [email protected]

A

Abdelouahid Sbai, MCNSL, School of Sciences, Moulay Ismail University,

M

Meknes, Morocco. E-mail: [email protected] Mohammed Bouachrine, High School of Technology, Moulay Ismail

A

CC E

PT

Graphical Abstract

ED

University, Meknes, Morocco. E-mail: [email protected]

SC R

IP T

Thiazolidine derivatives



N

A



The comparative molecular similarity indices analysis (CoMSIA) was performed to study 3Dquantitative structure activity relationship (3D-QSAR) of the studied compounds. A Molecular docking study was carried out to evaluate the interaction between these kind of compounds and PIM1 enzyme. Potent PIM1 inhibitors were designed based on the QSAR and molecular docking studies, and they were evaluated for their in silico ADMET.

M



U

Highlights

ED

Tahar Lakhlifi, MCNSL, School of Sciences, Moulay Ismail University, Meknes, Morocco. E-mail:

[email protected]

Abstract: Proviral Integration site for Moloney murine leukemia virus-1 (PIM1) belongs to the serine/threonine

PT

kinase family of Ca2+-calmodulin-dependent protein kinase (CAMK) group, which is involved in cell survival and proliferation as well as a number of other signal transduction pathways. Thus, it is regarded as a promising

CC E

target for treatment of cancers. In the present paper, a three-dimensional quantitative structure activity relationship (3D-QSAR) and molecular docking were performed to investigate the binding between PIM1 and thiazolidine inhibitors in order to design potent inhibitors. The comparative molecular similarity indexes analysis (CoMSIA) was developed using twenty-six molecules having pIC50 ranging from 8.854 to 6.011 (IC50 in nM). The

A

best CoMSIA model gave significant statistical quality. The determination coefficient (R2) and Leave-One-Out cross-validation coefficient (Q2) are 0.85 and 0.58, respectively. Furthermore, the predictive ability of this model was evaluated by external validation using a test set of eleven compounds with a predicted determination coefficient R2test of 0.72. The graphical contour maps could provide structural features to improve inhibitory activity. Furthermore, a good consistency between contour maps and molecular docking strongly demonstrates that the molecular modeling is reliable.

Based on these satisfactory results, we designed several new potent PIM1 inhibitors and their inhibitory activities were predicted by the molecular models. Additionally, those newly designed inhibitors, showed promising results in the preliminary in silico ADMET evaluations, compared to the best inhibitor from the studied dataset. This study would be of great help in lead optimization for early drug discovery of highly PIM1 inhibitors. Keywords: QSAR; Molecular docking; PIM1; Drug design; Thiazolidine; in silico ADMET. 1

Introduction

PIM1 is a serine/threonine kinase belongs to the Proviral Integration site for Moloney murine leukemia virus

IP T

protein kinase subfamily. PIM1 kinase is involved in cell survival and proliferation as well as a number of other signal transduction pathways (Nawijn et al., 2011; Santio et al., 2010), such as it phosphorylates, and regulates the activity of many proteins involved with cellular division, differentiation and apoptosis (Wang et al., 2001).

SC R

However, in terms of cancer a preponderance of evidence indicates that PIM1 playing an important role in the inhibition of apoptosis and the regulation of cell proliferation (Amaravadi and Thompson, 2005; Möröy et al., 1993). Owing to its association to cancer, the specific molecular mechanism by which PIM1 influences cell survival and proliferation have attracted extensive attention of many research’s group. These findings have led

U

to a concerted effort to develop therapeutics that directly target the kinase activity of PIM1 for cancer treatment (Magnuson et al., 2010).

N

Several studies based on heterocycles have been undertaken by research groups to try to inhibit the PIM1(Wu et

A

al., 2015; Wurz et al., 2015), while this process is difficult, expensive and takes a long time. Computer Aided Drug Design (CADD) can circumvent these difficulties and increases number of lead molecules available for

M

further research as well as to reduce time and coasts, while taking advantage of the spectacular improvement in computer speed and capacity of computing molecular properties, which can be easily obtained from available

ED

software’s. Those can be exploited to build a quantitative structure activity relationship (QSAR) model to enable prediction of the biological activity and the efficacy of newly proposed compounds by means of cheminformatics methods. Recently, a series of some potent PIM1 inhibitors have been designed and reported by Bataille et al.

PT

(Bataille et al., 2017), so far, this study represents the first predictive model for the PIM1 binding site based on the reported activities of this series of substituted thiazolidine. That prompted us to aim an in silico study based on this series to highlight the structural factors controlling different active and inactive sites in the thiazolidne

CC E

scaffold to design new molecules with improved PIM1 inhibitory activity. CADD including structure-based and ligand-based have proven to be valuable approaches in drug discovery and drug design in medicinal chemistry. Structure-based approach, which includes molecular docking is based on the evaluation the interactions between the ligand and active site of the receptor. While, ligand-based approach,

A

which includes the 3D-QSAR models, Comparative molecular similarity indices analysis (CoMSIA) (Klebe et al., 1994), it correlates changes in 3D structural factors of chemical groups in molecules such as steric, electrostatic and hydrophobic properties with the biological activity (Damale et al., 2014). Thus, it becomes necessary to perform theoretical studies on the interaction mechanism between protein and inhibitors before the synthesis of new chemical entities. Whereas, those studies are not only help understand relationships between the molecular properties and biological activity of any class of molecules, but also help to provides researchers a deep vision about the lead molecules to be used in further studies to discover new drugs (Gupta et al., 2003),

investigate the pharmacological action of inhibitors involved in modern drug design and narrow down the library of derivatives for design of new chemical entities with enhanced biological activity (Roy et al., 2009). In the present paper, 3D-QSAR (CoMSIA) study following by docking molecular were mainly performed on a series of thirty-seven substituted derivatives (Eun et al., 2016) to gain deeper insight and identify the key substituents and mode of action of thiazolidines to design new compounds of this class of chemical entities acting as ATP-competitive inhibitors of PIM1. The main objectives of this study are to relate structure requirements of thiazolidines to PIM1 inhibitory activity and provide an explanation of the mechanism of thiazolidines inhibiting PIM1. Thus, several potent PIM1 inhibitors were designed, optimized and their inhibitory activities were

2 2.1

IP T

demonstrated by 3D-QSAR, molecular docking and ADMET. Material and methods Data collection

SC R

For molecular docking and 3D-QSAR studies a series of thirty-seven molecules with their PIM1 reported inhibitory IC50 were compiled from a recently published study (Bataille et al., 2017). For QSAR analysis, in vitro biological activities IC50 (nM) were converted into the corresponding pIC50 values (i.e. pIC50 is the negative logarithm of IC50 (pIC50 = −log(IC50)) and are listed with their corresponding structures in table 1. The data set was split randomly

to test the performance of the proposed model (Test set). Molecular modeling

N

2.2

U

into a training set (twenty-six molecules) to build the quantitative model and the remaining molecules were used

A

All modeling studies were performed using the SYBYL-X 2.0 molecular modeling package (Tripos Inc., St. Louis, USA) running on a windows 7, 32 bit workstation. Three-dimensional structures of the studied compounds

M

were built using the SKETCH option in SYBYL, then they were minimized under the Tripos standard force field (Clark et al., 1989) with Gasteiger-Hückel atomic partial charges (Purcell and Singer, 1967) by the Powell method

2.3

CoMSIA studies

ED

with a gradient convergence criterion of 0.01 kcal/mol Å.

The Comparative molecular similarity indices analysis (CoMSIA) (Klebe et al., 1994) model was carried out in

PT

SYBYL-X 2.0. All analyses were performed in a 3D regularly spaced grid of 2.0 Å in all Cartesians directions. A sp3 carbon with a Van Der Waals radius of 1.52 Å, net +1.0 charge, hydrophobic interaction, hydrogen-bond donor

CC E

and acceptor properties of +1.0 was used as a probe at every lattice point of the grid box to generate respectively, the steric, electrostatic, hydrophobic, hydrogen bond donor and acceptor fields from similar actives molecules, to develop a CoMSIA model. In the present study, the value of attenuation factor, which controls the Gaussian function’s steepness, was set by default to0.3 (Zheng et al., 2011). Partial least square (PLS) analysis

A

2.4

Because of the enormous variables obtained from the fields’ calculations, the PLS regression method (Wold, 1991), was performed to evaluate a linear correlation between the 3D-QSAR descriptors and the biological activity values. In the first step, the cross (Q2) validation was performed by using Leave-One-Out (LOO) (Kubinyi, 2003) method where one compound is eliminated from the training set and its activity is predicted from the developed model using the residual (N-1) compounds. The same way is repeated until all compounds have been eliminated once. The model with the highest value of Q2 with the lowest standard error of estimate (Scv) and a minimal number of components was accepted. In order to reduce noise and increase the speed up the analytical

process, the column filtering value (σ) was set to 2.0 kcal/mol. In the next step after determining the optimum number of components, they were used to derive the final PLS model with no validation method [19,20] to create the maximum determination coefficient (R2). 2.5

Validation and predictive power of the model

The main objective of any QSAR study is to obtain a model with the highest predictive and generalization abilities. So to evaluate the predictability of the developed 3D-QSAR models, eight compounds were used as a testing set (Golbraikh and Tropsha, 2002). These molecules were aligned using the same methods described above, and then their inhibitory activities were predicted using the generated 3D-QSAR model from the training

2.6

IP T

set. Y-Randomization test

The obtained model was further validated by the Y-Randomization method (Rücker et al., 2007). The activities

SC R

of the studied molecules (pIC50) are randomly shuffled many times and after every iteration, a new QSAR model is developed. The new QSAR models are expected to have lower Q2 and R2 values than those of the original model. This technique is performed to eliminate the possibility of chance correlation. If higher values of the Q2 and R2 are obtained, it means that an acceptable 3D-QSAR can’t be generated for this data set because of

2.7

U

structural redundancy and chance correlation. Model acceptability criteria

N

According to Alexander Tropsha and Alexander Golbraikh, a predictive model must satisfy a set of statistical

A

criteria. A QSAR model was considered predictive if the following conditions are satisfied: (i) Q2>0.50; (ii) R2>0.60 (Golbraikh and Tropsha, 2002; Tropsha et al., 2003). Molecular docking

M

2.8

Molecular docking is regarded among the most important method in discovering novel small-molecule drugs

ED

(Jain, 2007, 2003; Wang et al., 1999). In our study, this technique was performed using Surflex-dock implemented in SYBYL-X.2.0. The ligands and protein preparation steps for the docking protocol were carried out in SYBYL-X 2.0, then results were analyzed using Discovery Studio 2016 software (“Dassault Systèmes

A

CC E

PT

BIOVIA, Discovery Studio Modeling Environment, Release 2017, San Diego: Dassault Systèmes.,” 2016) .

I N U SC R

2* 3 4*

CC E

5

ED

1

pIC50(o pIC50(pr Res No bs) ed) 2-Cl phenyl 6.495 6.809 0.314 thien-2-yl 6.339 6.642 17 0.303 thiazol-4-yl 6.205 6.640 0.435 thiazol-5-yl 6.011 6.672 0.661 naphth-1-yl 7.208 6.907 0.301 18* naphth-2-yl 6.947 6.948 0.001 benzothien7.260 7.133 0.127 3-yl benzothien7.168 7.028 0.140 19* 7-yl benzofuran7.174 6.873 0.301 2-yl benzofuran6.807 7.166 3-yl 0.359 indol-3-yl 6.735 6.784 20 0.049 quinolin-27.260 7.101 0.159 yl R1

PT

No

6* 7

R1

pIC50(o pIC50(pr Res No bs) ed) 22* 8.174

7.602

7.836

7.169

9*

10 11 12

7.071

6.963

pIC50(o pIC50(pr Res bs) ed)

2-CF3

8.036

7.952 0.084

0.33 23 8

3-CF3

7.310

7.910 -0.600

24

4-CF3

7.796

8.159 -0.363

7.959

7.840 0.119

7.678

7.918 -0.240

7.959

7.900 0.059

2-OH

8.097

8.130 -0.033

1.00 29* 5

3-OH

8.301

8.080 0.221

30*

4-OH

8.620

7.938 0.682

31

4-SO2Me

8.444

8.133 0.311

3-Cl

7.444

7.748 -0.304

37.602 NHSO2Me

7.989 -0.387

28 8.174

R1

25 2-OMe 0.21 7.392 26* 3-OMe 0 27 4-OMe

A

8

M

A

Table 1: Chemical structures and PIM1 inhibitory activities of substituted thiazolidine derivatives

0.10 32 8 33

I quinoxalin6.900 2-yl

7.090

14

7-azaindol6.757 3-yl

6.642 0.115

38.155 piperidine 335 NHSO2NM 8.658 e2 34

0.001 21

8.854

M

6.948 0.496

PT CC E A

8.802

0.05 36 2

A

6.112

0.190

ED

1,815 naphthyridi 6.111 n-2-yl 4-methyl4Hthieno[3, 16 7.444 2b]pyrrol5-yl

N U SC R

13*

37

4-OCF3

8.351 -0.196 8.272 0.386

8.538

8.195 0.343

4-CH28.168 piperidine

8.173 -0.005

* Test set

2.8.1

Macromolecule preparation

The crystal structure of PIM1 was downloaded from the protein data bank, (PDB entry code: 3c4e). No one of the understudy ligands is complexed with this protein in PDB, so, its original ligand was removed. The PDB file was prepared using Discovery Studio 2016, such as all ligands, co-factors and solvent molecules were removed from the model. Before docking, hydrogen atoms of the receptor were added to the prepared structure. The definition of active site definition was performed based on the original ligand in the crystal. We chose compound 21 as the

2.8.2

IP T

subject to dock into the active pocket under the conditions previously cited. Ligand preparation

The compounds selected for docking were modeled in the same way as for the 3D-QSAR studies, Threedimensional structures were built using the SKETCH option in SYBYL, then they were minimised under the Tripos

SC R

standard force field (Clark et al., 1989) with Gasteiger-Hückel atomic partial charges (Purcell and Singer, 1967) by the Powell method with a gradient convergence criterion of 0.01 kcal/mol Å. 2.8.3

Applicability domain

3D-QSAR models were used to calculate the predicted activity of newly designed hits. The utility of a QSAR

U

model is its accurate prediction ability for new chemical compounds. So, once the QSAR model is built, its

N

domain of applicability (AD) must be defined. A model is regarded valid only within its training domain and only the prediction for new compounds falling within its applicability domain can be considered reliable and not

A

model extrapolations. Applicability domain was defined through the degree of similarity of the predicted

M

compound to those in the training set molecules. Assessment was done by comparing the value for each descriptor of the compound against the range of descriptor values for the compounds in the training set, using SYBYL software. The number of the out-of-range descriptors found in the test compound and the total contribution of

ED

such extrapolated points to the prediction is calculated. The sum of extrapolated terms (SUM) is the total contribution made to the prediction for the compound by the out-of-range descriptors. Additionally, standard error of prediction (SEP) was calculated during the 3D-QSAR model generation. If SUM of the test or predicted

PT

compound is larger than the SEP for the developed cross-validated model, then the extrapolation is probably too far outside the model to get a reliable prediction (Gupta et al., 2012). SUM was calculated for all newly designed compounds.

Synthetic accessibility

CC E

3

The best designed hits were screened for their synthetic accessibility using SwissADME (Daina et al., 2017). This provides a score on a scale from 1 (Very easy to synthesise) to 10 (Challenging and complex to synthesise). To measure this score for each compound, many criteria such as the number of stereocenters, complexity of the

A

molecular structure, complexity of the ring system, similar to a huge number of already synthesised compounds and the potential for using important synthetic reactions are taken in consideration. 4

In silico Pharmacokinetics (PK)/Pharmacodynamics (PD) evaluation

The success of a drug journey through the body is measured regarding his pharmacokinetics parameters (Adsorption, distribution, metabolism, excretion and toxicity; ADMET). For this purpose, the best-designed hits were evaluated for their in silico pharmacokinetics parameters, in addition, their expected metabolic products and sites of metabolism for Phase-I and Phase-II metabolisms were also estimated through MetaPrint2D-React

P a g e 8 | 26

software, to prevent the failure of those compounds during clinical studies and enhance their chances to reach the stage of being candidates drugs in future. 5

Results and discussion

5.1

Molecular alignment

Molecular alignment is a sensitive step in the development of any 3D-QSAR study (AbdulHameed et al., 2008). The figure 1 depicts the proposed alignment; all molecules were aligned on the common core using SYBYL software, based on the most potent inhibitor 21 as a template molecule to fit the other molecules in training and test sets. The best-docked conformation of compound 21 was chosen to align the data set in 3D-QSAR studies and to serve

IP T

as a template molecule to visualize the contour maps.

The predicted and experimental activity values and their residual values for both the training and test sets

5.2

CoMSIA results

Based

on

explain field

CoMSIA

and

effects

predict of

descriptors

available

quantitatively,

substituents

on

the

in

the

SYBYL,

a

3D-QSAR

hydrophobic,

PIM1

inhibitory

model

electrostatic, activity

of

a

was

steric, series

proposed and of

to

acceptor

thirty-seven

U

thiazolidine derivatives.

SC R

from the 3D-QSAR (CoMSIA) model are given in table 1.

N

Different combinations of the five fields were generated and results are listed in table 2. They showed Q2 ranging from 0.220 to 0.586, and R2 values of 0.573 to 0.94. In general, statistical results with Q2 value ≥ 0.5

A

and R2 value ≥ 0.6 are considered reasonable and meaningful (Golbraikh and Tropsha, 2002). These results

M

indicate that CoMSIA was robust in generating statistically significant models. The model CoMSIA/SHDA, which contains four fields (Steric, hydrophobic, donor and acceptor) and exhibits good statistical keys, was considered as the best one. Hence, all the following studies were based on the combination of S, H, D and A fields. The

ED

cross-validated determination coefficient Q2 of the training set and non-cross-validated determination coefficient R2 are 0.586 and 0.85 respectively. The optimal number of principal components used to generate

PT

the CoMSIA model is four (Hickey and Passino-reader, 1991), which is reasonable considering the number of molecules used to build the model. The standard error of estimate is 0.317. Finally, the prediction ability of the proposed model was confirmed using external validation, the R2test value obtained is 0.72. These statistics

CC E

results indicated the good stability and the powerful predictive ability of the proposed CoMSIA model. CoMSIA model.

Experimental and predicted activities of both the training set and test set are shown in figure 2, the CoMSIA model gave a determination coefficient (R2) value of 0.85, respectively, which demonstrated that the internal

A

robustness and external high prediction of the QSAR models. Table 2: PLS Statistics of CoMSIA models Model CoMSIA SEH SEA SED SHD SHA SDA

Q

2

0.484 0.432 0.327 0.484 0.571 0.569

2

R

Scv

F-t

N

R2test

0.660 0.861 0.761 0.925 0.934 0.860

0.447 0.298 0.383 0.230 0.212 0.314

46.643 45.452 36.564 49.495 56.526 24.668

1 3 2 5 5 5

0.77 0.71 0.73 0.53 0.53 0.71

Ster 0.373 0.402 0.492 0.355 0.333 0.440

Elec 0.310 0.364 0.420 -

Fractions Acc 0.233 0.272 0.371

Don 0.087 0.177 0.189

P a g e 9 | 26

Hyd 0.317 0.468 0.394 -

5.2.1

SC R

IP T

EHD 0.262 0.874 0.291 36.325 4 0.45 0.404 0.155 0.442 EHA 0.390 0.944 0.285 64.213 2 0.78 0.312 0.345 0.343 HDA 0.478 0.798 0.368 20.781 4 0.70 0.339 0.129 0.533 DAE 0.220 0.697 0.451 12.089 4 0.74 0.485 0.351 0.164 SEHD 0.383 0.573 0.501 32.255 1 0.81 0.240 0.19 0.204 0.357 SEHA 0.488 0.863 0.297 46.126 3 0.67 0.251 0.237 0.243 0.269 SEDA 0.449 0.812 0.275 33.197 5 0.77 0.328 0.253 0.285 0.133 EHDA 0.408 0.932 0.213 72.317 4 0.51 0.318 0.167 0.085 0.430 SHDA 0.586 0.850 0.317 29.806 4 0.72 0.301 0.255 0.138 0.305 SEHDA 0.487 0.930 0.218 55.390 5 0.63 0.196 0.241 0.220 0.111 0. 231 Q2 : Cross-validated determination coefficient. N : Optimum number of components obtained from cross-validated PLS analysis and used in final non-crossvalidated analysis. R2 : Non-cross-validated determination coefficient. Scv : Standard error of the estimate. F-t : F -test value R2test : External validation determination coefficient. Graphical interpretation of 3D-QSAR models

3D-QSAR contour maps were generated to visualise the data contents of the derived CoMSIA model, which

U

provide the information about the favorable and unfavorable regions for the biological activity in the studied

N

compounds. Changes in the structure of the molecule lead to changes in its physico-chemical properties, which might increase or decrease the biological activity. Steric, electrostatic, hydrophobic and hydrogen bond

A

acceptor contour maps of CoMSIA are shown in figures 3 . Compound 21 is the most active of the

5.2.2

CoMSIA Contour Map

M

series; therefore it was taken as reference structure for the generation of contour maps.

The most active molecule in the series (Molecule 21) is displayed superimposed with CoMSIA steric contour

ED

maps in figure 3 (a). In the CoMSIA steric contour map, we could see a large yellow contour covers the pyridine ring and the piperazine moiety, in addition to a medium-sized contour, inside the big yellow one, covers the N of the pyridine ring, suggest that inhibitors with moderate groups at the 4 position of the pyridine ring might

PT

have better inhibitory activity than those with no groups. For exemple, compounds 17 (pIC50 = 8.174), 29 (pIC50 = 8.301) and exhibit high activities. However, compounds, 34 (pIC50 = 8.155) and 35 (pIC50 = 8.658), with bulky

CC E

substituents, which outside the green contour, exhibit high inhibitory activity, which suggest that the steric effet may not be the vital key of the inhibitory activity. Therfore the steric effect is complemoantory to the hydrophilic, hydrogen donor and acceptor effects. In the hydrophilic contour maps depicted in figure 3 (b), it is shown two large gray areas, the first covers the

A

piperazine moiety (located at the 3 position of pyridine ring) and the second located nearby the sulfur atom of the thiazole moiety, with another small one near the sulfur atom of the thiazolidine moiety. This observation would tend to suggest that hydrophilicity at these parts of the molecule plays significant role in biological activity.that can be interpreted by the fact that compounds 31(pIC50 = 8.444), 34 (pIC50 = 8.155) and 35 (pIC50 = 8.658) withhydrophilic groups at position 3 of pyridine, show higher activities than compounds 23 (pIC50 = 7.310) and 32 (pIC50 = 7.444) with lipophilic groups at the same position, and if we observe that the low activity compound series as compared to reference, the reason behind that may be these compounds represented close

P a g e 10 | 26

to gray contours near the sulfur atom of thiaziole moiety and hydrophilic substituents at this position are absent as compared to template. Additionally there is a yellow contour locates between the thiazole and pyridine ring specifies that presence of hydrophilic moieties at this area are not suitable for activity. It is additionally supported after a comparison of the most potent compound 21 (pIC50 = 8.802), with less active compounds (2

A

CC E

PT

ED

M

A

N

U

SC R

IP T

(pIC50 = 6.339), 14 (pIC50 = 6.757) and 15 (pIC50 = 6.111).

P a g e 11 | 26

The CoMSIA contour maps of the H-bond donor and acceptor descriptors are shown in in figure 3 (c) A large cyan contours nearby the piperazine suggests that the hydrogen bond donor groups can benefit the inhibitory activity. It can be explained by the example of the compounds (28 and 29) with an OH at the 2 and 3 position on the pyridine ring leading to higher activities than molecules (25 and 26) with OMe at the same positions. The magenta contour maps figure 3 (d) indicate the areas where hydrogen bond accepting groups increased activity and red contour maps indicate areas where hydrogen bond accepting groups decreased inhibitory activity. A magenta contour located between the piperazine and pyridine ring suggests the requirement of hydrogen bond accepting groups at this position to enhance the inhibitory activity. This can be expounded by

IP T

the example, that is, the compounds 31(pIC50 = 8.444) and 37 (pIC50 = 8.168), with sulfonyl, OCF3 and piperidine groups at the 4 position of the pyridine ring, which contain Sulfre, Oxygen and Nitrogen atoms have higher activities than that of compounds 24 (pIC50 = 7.796), with CF3 group at the same position.

SC R

From the above discussions, we can infer the following conclusions: (1) hydrophilic hydrogen donor groups added to the 3 position of the pyridine ring could improve the inhibitory activity.

(2) Hydrophilic bulky hydrogen acceptor groups near the 4-position of the pyridine ring would help for the PIM1 inhibitory activity. (3) Hydrophilic groups near the sulfur atoms of thazolidine and thiazole moieties would

Summary of the structure-activity relationships

N

5.3

U

be beneficial for the bioactivity. The conclusion above can offer a way to design highly effective PIM1 inhibitors.

Some key structural factors between PIM1 and thiazolidine inhibitors derived from the all above analyses were

A

illustrated in figure 4. To be specific, bulky flexible substituents as or CH2(Cyclopropyl)2 groups was substituted

M

in the place of butyl at R5 position and it penetrated into the hydrophobic pocket respectively as predicted. Hydrophilic groups were introduced in the benzene ring near the oxime and is exposed to the solvent as expected. It also maintained the hydrogen bond interactions as that of the reference molecule (inhibitor 21).

CC E

PT

ED

The dock pose of inhibitor 21 and newly designed molecule A5 are shown in figures 5 and 6, respectively.

A

Figure 4: Structural requirements for PIM1 inhibitors obtained from CoMSIA contour map and molecular

5.3.1

docking analysis.

External validation

Validation of the developed model is an essential part of any QSAR study. Thus, a true and trustworthy model should be able to predict a precise activity in the external test set (Golbraikh and Tropsha, 2002). That is why the final developed 3D-QSAR model from a training set of twenty-six thiazolidine derivatives was used to predict the activity of eleven remaining molecules; the parameters of the performance of the generated models are shown in table 2.

P a g e 12 | 26

5.3.2

Y-Randomization Table 3 : Q2 and R2 values after several Y-randomization tests CoMSIA Iteration Q2 1 0.14 2 0.23 3 0.25 4 0.064 5 -0.110 6 -0.342 7 0.35

IP T

R2 0.67 0.79 0.66 0.81 0.82 0.68 0.53

The Y-Randomization method was carried out to validate the 3D-QSAR model. Several random shuffles of the dependent variable were performed then after every shuffle, a 3D-QSAR was developed and the obtained

SC R

results are shown in table 3. The low Q2 and R2 values obtained after every shuffle indicate that the good result in our original 3D-QSAR model are not due to a chance correlation of the training set. 5.4

Docking results

To better understand as well as to support the in vitro activity of the studied compounds for the rational design

U

of new potent molecules, a molecular docking study of the most active compound (21) was carried out to clarify

N

the probable binding modes between thiazolidine derivatives and PIM1 kinase (PDB ID : 3C4E), which provides straightforward knowledge for further structural optimization.

5.4.1

ED

M

A

Table 4: The hydrogen bond interactions between the most active compounds , HY7, 0F5 and T2 with PIM1 kinase Interaction type Inhibitor 21 HY7 T1 0F5 Lys 67 and Asp Lys 67, Glu 89 Hydrogen bonds Lys 67 and Glu 89 Lys 67 186 and Asp 186 Identification of binding modes

Binding interactions of most active molecule, compound 21 with PIM1 binding pocket are

PT

shown in figure 5, compound 21 forms two strong hydrogen bonds, the first between the -NH3+ of the Lys 67 residue and the oxygen of the thiazolidine moiety (NH---O at distance of 2.88 A) and the second between Hydrogen of the NH in the thiazolidine moiety and the oxygen of the Glu 89 residue (O---NH at distance of

CC E

2.93A), which can be seen in green dotted lines. The pyridine ring and thiazole moiety, which occupies the middle of the molecule are embedded in a lipophilic area madden by Leu 44, Phe 49, Val 53 and Ile 185. This observation is compatible with CoMSIA yellow contours found between pyridine and thiazole moieties.

A

While the 2, 3 and 4 positions of the pyridine ring, in addition to the sulfur of the thiazole moiety are located in a exposed solvent area. Additionally, the CoMSIA suggests that bulky hydrogen bond acceptor groups on 4 position, and hydrophilic hydrogen bond donor groups on 3 position are supposed to enhance the activity. These observations between docking and CoMSIA model are in concordance, because substituents on these positions are exposed to the solvent and tolerated to be bulky and hydrophilic. From the analysis of the various properties of inhibitor 21 in tables 4 and 7, it can be concluded that it fulfills the Lipinski’s rule (Lipinski, 2004). Additionally, inhibitor 21 binds to the active site of the PIM1 kinase and shares largely homogeneous in binding mode( especially hydrogen bond with Lys 67) to several PIM1 inhibitors

P a g e 13 | 26

reported in the literature, namely, HY7 in (PDB : 4alw) and 0F5 in (PDB : 3vbq). Therefore, that can prove our docking process was reasonable. While the main cause of its mediated PIM1 inhibition is considered due to that the compound 21 is aligned adequately within the ATP binding site, which allows it to make important interactions with this pocket. 5.5

Design for New PIM1 Inhibitors

Overall, the rationale behind the foregoing study is to design of novel PIM1 inhibitors, so, by extracting few structural features from the proposed 3D-QSAR (CoMSIA) model and molecular docking, and by exploring the scaffold of the thiazolidine derivatives. Here, three potent substituted thiazolidine analogs have been designed

IP T

in order to improve the PIM1 inhibitory activity as well as the drug-like properties of the compounds including pharmacokinetics and toxicity. All newly designed molecules have good predicted activities. These newly designed molecules were aligned to the database using inhibitor 21 as template and their pIC50 values were

SC R

predicted in addition to their SUM to check if they fall in the AD of the proposed model. For the proposed model to predict the PIM1 inhibitory activity, all newly designed compounds have SUM < SEP, so their predicted values are regarded reliable.

R4

ED

No

A

CC E

PT

T1

T2

T3

M

A

N

U

Table 5 : Structures and chemical properties of newly designed molecules and their predicted pIC50 based on CoMSIA 3D-QSAR models (SEP = 0.612)

pIC50(Pred)

SUM

8.865

0.483

8.856

0.229

8.872

0.091

To ensure the viable drug designing, predicted compounds were evaluated for synthetic accessibility. Later, results of their synthetic accessibility were compared with that of the original synthesised PIM1 inhibitors. The SwissADME scores (Daina et al., 2017) of newly designed and preexisted compounds were found similar, which indicate the synthetic ability of designed leads. Additionally, these molecules were further screened for their in

P a g e 14 | 26

silico ADMET properties using the pkCSM online tool and results were compared with that of the most potent compound from the studied series (inhibitor 21). According to the parameters, the designed compounds have obvious advantage on the inhibitor 21. 5.6 Binding model for designed inhibitors The newly designed molecules were docked into the PIM1 active site. The docking of the proposed molecule T1 as depicted in figure 6 reveals that it makes interactions with the essential residues at the active site as they do all reported PIM1 inhibitors in literature, such as it keeps the same conformation at the binding pocket of PIM1 as the inhibitor 21. Owing to the modification at the R4 position on the benzene ring, in addition to its hydrogen

IP T

bond interactions with the key residue Lys 67, T1 can form a hydrogen bond with Glu 89 and Asp 128, respectively. We can see from figures 5, 6 and table 4 that the residues interaction with designed inhibitor and inhibitor 21 are approximately same but the hydrogen numbers and the interactions between PIM1 and T1 were

5.7

In silico pharmacokinetic, metabolism and toxicity analysis.

SC R

better than the original compound.

The metabolism refers to the chemical biotransformation of a drug by the body; consequently, drugs produce several metabolites, which may have different pharmacological and physicochemical properties. Therefore, the

U

metabolism plays a critical role in the bioavailability of drugs and drug-drug interactions, and as known, the

N

cytochrome P450 (CYP450) enzymes are probably the most important class of enzyme to study this. Thus, exploring those parameters may prevent their failure in clinical studies and have a deep insight on how they

A

react in the body, which allows medicinal chemist to introduce new functional groups on the molecule to dodge

M

the metabolic pathways susceptible to give toxic or very polar compounds that can eliminated very easily from the body. Accordingly, that can help to synthesise metabolically stable drugs, as well as to avoid drug-drug interactions. For this purpose, the compound were submitted to learn the potential of compounds having

ED

substrate or inhibitor of CYPs along with CYP Human Liver Microsomes (HLM) (table 6). All new designed compounds were found to be the substrate of CYP3A4. Details of CYP450 mediated metabolites of the top ranked compound T1 and inhibitor 21 were summarised in figure 7. Additionally, possible metabolic sites and

PT

metabolites were predicted for different biochemical reactions through oxidation, hydroxylation, glucuronidation, dealkylation, hydroxydation, desulfuration, AmideHydrolysis and Demethylation. Results

CC E

indicate that both inhibitors may undergo a desulfuration and an AmideHyrolysis potentials. While inhibitor 21 may undergo a glucuronidation potential, which have the ability to transform the small molecules to water soluble form and may increases its excretion process. These metabolites may be helpful in designing new thiazolidine

A

derivatives, as well as to optimise their pharmacological effects.

P a g e 15 | 26

Each time when a lead compound with high inhibitory activity is identified, there is no guarantee that this compound with the best interactions with a target is necessarily a good medicine. Thus, it is essential to predict the ADMET parameters of the leads and optimise these parameters early on to reduce potential problems later during clinical studies. Accordingly, the ADMET parameters of the newly designed compounds and the most potent inhibitor from the studied dataset (inhibitor 21) were calculated using the pkCSM (Pires et al., n.d.) and SwissADME (Daina et al., 2017) online tools. Water solubility is given in log (mol/l) (Insoluble < -10 < poorly soluble < -6 < Moderately < -4 < soluble < -2 < very soluble < 0 < highly soluble) , Intestinal absorbance value below 30% indicates poor absorbance. Low value of total clearance (logCLtot) means high drug half lifetime. For

IP T

a given compound a logBB < -1 considered poorly distributed to the brain. Positive result in AMES test suggests that compound could be mutagenic. Prediction of ADMET parameters are listed in table 6

In their journey through the body, drug molecules encounter several different membrane barriers such as

SC R

gastrointestinal epithelial cells, blood-brain barrier and the target cell. Prediction of this permeability can assist to interpret the pharmacokinetics results and in the understanding the behavior of such kind of chemicals in the body.

An important parameter is the octanol–water partition coefficient (log p) that shows the drug hydrophobicity and

U

expresses that compounds with higher hydrophobicity have an increased metabolism and low absorption that may inadvertently be increased probability of binding to undesired hydrophobic macromolecules, hence increase the

N

potential toxicity. Similarly, rapid renal clearance is associated with small and hydrophilic compounds. Results

A

of ADMET and bioavailability (Rule of Five) (Lipinski, 2004) as depicted in tables 6 and 7, were within the acceptable limits for the designed compounds in boths lipophilicity and solubility.

M

Results show that all the compounds have a good calculated intestinal absorbance. The brain/blood partition coefficient of all compounds was predicted (in logarithm), thus, newly designed compounds showed little chance

ED

to cross the blood-brain barrier compared to inhibitor 21. As known the safety of the compounds is an important parameter to develop a successful drug, for this reason, the compound-induced toxicity was studied for Ames test. As shown in table 6, results specify that with the

PT

exception of T3, which is mutagenic, inhibitor 21 and all other newly designed compounds were non-mutagenic. According to the in silico ADMET parameters, the newly designed compounds showed obvious advantage on the inhibitor 21, in terms of intestinal adsorption, toxicity and Blood-brain barrier permeability; thus, it can be

CC E

concluded that compound T1 presents better biological activity, drug-like characteristics and in silico ADMET parameters than the most potent inhibitor of the series (inhibitor 21), consequently, it has more favorable

A

properties as such and on further lead optimization may guide to novel PIM1 inhibitors.

P a g e 16 | 26

I N U SC R

Table 6: In silico ADMET prediction and synthetic accessibility values of newly designed compounds.

21

T1

-4.41

93.274

-4.42

94.902

-4.185

92.955

CC E

T2 T3

-4.161

A

Inhibitor

(logBB)

2C 19

2 C 9

2 D 6

92.612

LogP

Inhibitor 21

2.4457

T1 T2 T3

2.6621 2.5773 2.2351

3 A 4

Toxi city

Total Clearanc e

AME S toxic ity

Numeric (log ml/min/ kg)

Synthe tic Accessi bility

Docki ng score

(Yes /No)

Numeri c

(-log ki)

0.546

No

3.69

0.185

No

0.597

No

0.723

Yes

inhibitor

A

Numeric (% Absorbe d)

1 A 2

Categorical (Yes/No)

M

(log mol/L)

Excretio n

CYP 2 3 D A 6 4 substr ate

inhibitor 21 from the studied dataset 0.112 N Y N N N No o es o o o Newly designed compounds -0.978 N Y N Ye N N o es o s o o -0.297 N Y N N N o es o No o o -0.87 N Y N N N o es o No o o

PT

Na me

Water solubil ity

Intestinal absorpti on (human)

Metabolism

ED

Absorption

Distributi on Blood brain barrier Permeab ility

Y es N o N o Y es

3.46 3.61 3.50

7.465

7.270 5.715 5.965

Table 7: Lipinski’s properties of inhibitor 21 and newly designed compounds Property H-bond acceptor

H-bond donor

Polar surface area (A2)

inhibitor 21 from the studied dataset 8 1 164.864 Newly designed compounds 7 1 162.795 7 1 169.806 7 2 163.231

Rotatable Bonds

Molecular weight (g/mol)

3

403.558

3 3 3

403.51 416.553 402.526

P a g e 17 | 26

I N U SC R

6

Conclusion

In this paper, computer-aided drug design techniques, i.e ligand-based and structure-based analyses were conducted based on thirty-seven thiazolidine derivatives, in order to explore the binding mode between this class of chemicals and the PIM1 kinase, also to identify the critical structural features affecting the inhibitory activity and exploit them to design new potent inhibitors. The best CoMSIA (Q2=0.58, R2=0.85) model displayed significant statistical quality and excellent predictive ability. The graphical contour maps obtained from 3D-QSAR indicated significant hydrophobic and steric potential contributions and suggested sufficient information for understanding the structure-

A

activity relationship, thus aided in the further design and development of novel thiazolidine derivatives with PIM1 improved inhibitory profile. Hence, molecular docking was used to better understand the binding mode and produce the binding poses of these compounds into PIM1 enzyme; in addition to confirm,

M

those suggested information from 3D-QSAR studies. Further, all those outcomes showed insight into the key structural features required for the PIM1 inhibitory behavior in the studied thiazolidine derivatives. Docking results indicates the the hydrogen bonds interactions with Lys67 and Asp 186 contribute mainly to the inhibitory activity.

ED

Thus, those obtained key structural factors were used to design three inhibitors by modifying the thiazolidine scaffold, and then got a better result in terms of biological activity, in silico ADME and toxicity. The combination of these molecular modeling results will provide the information required for better understanding of structural features

CC E

PT

necessary for PIM1 inhibitory activity and are expected to be useful for further design of novel active PIM1 inhibitors. Competing Interests "The authors declare that they have no competing interests." Acknowledgment

We are grateful to the “Association Marocaine des Chimistes Théoriciens” (AMCT) and “Moroccan Centre of Scientific and Technique research” (CNRST) for their pertinent help concerning the programs. References

A

AbdulHameed, M.D.M., Hamza, A., Liu, J., Zhan, C.-G., 2008. Combined 3D-QSAR Modeling and Molecular Docking Study on Indolinone Derivatives as Inhibitors of 3Phosphoinositide-Dependent Protein Kinase-1. J. Chem. Inf. Model. 48, 1760–1772. doi:10.1021/ci800147v

Amaravadi, R., Thompson, C.B., 2005. The survival kinases Akt and Pim as potential pharmacological targets. J. Clin. Invest. 115, 2618–2624. doi:10.1172/JCI26273 Baroni, M. Clementi, S. Cruciani, G. Costantino, G. Riganelli, D. Oberrauch, E., 1992. Predictive ability of regression models. Part II: Selection of the best predictive PLS model. J. Chemom. 6, 347–356. Bataille, C.J.R., Brennan, M.B., Byrne, S., Davies, S.G., Durbin, M., Fedorov, O., Huber, K.V.M., Jones, A.M., Knapp, S., Liu, G., Nadali, A., Quevedo, C.E., Russell, A.J.,

P a g e 18 | 26

I N U SC R

Walker, R.G., Westwood, R., Wynne, G.M., 2017. Thiazolidine derivatives as potent and selective inhibitors of the PIM kinase family. Bioorg. Med. Chem. 25, 2657– 2665. doi:10.1016/j.bmc.2017.02.056

Clark, M., Cramer, R.D., Van Opdenbosch, N., 1989. Validation of the general purpose tripos 5.2 force field. J. Comput. Chem. 10, 982–1012. doi:10.1002/jcc.540100804 Cruciani, G., Baroni, M., Clementi, S., Costantino, G., Riganelli, D., Skagerberg, B., 1992. Predictive ability of regression models. Part I: Standard deviation of prediction errors (SDEP). J. Chemom. 6, 335–346. doi:10.1002/cem.1180060604

Sci. Rep. 7, 42717. doi:10.1038/srep42717

A

Daina, A., Michielin, O., Zoete, V., 2017. SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules.

M

Damale, M.G., Harke, S.N., Kalam Khan, F.A., Shinde, D.B., Sangshetti, J.N., 2014. Recent advances in multidimensional QSAR (4D-6D): a critical review. Mini Rev. Med. Chem. 14, 35–55.

ED

Dassault Systèmes BIOVIA, Discovery Studio Modeling Environment, Release 2017, San Diego: Dassault Systèmes. [WWW Document], 2016. URL http://accelrys.com/products/collaborative-science/biovia-discovery-studio/ (accessed 2.25.17). Eun, M., Jin, B., Kim, H., Yun, J., Park, J., Lee, N., Hwa, Y., Un, S., Yang, K., Kim, S., Lee, K., Kim, Y., Choi, G., 2016. Bioorganic & Medicinal Chemistry Letters

PT

Discovery of indirubin derivatives as new class of DRAK2 inhibitors from high throughput screening. Bioorg. Med. Chem. Lett. 26, 2719–2723. doi:10.1016/j.bmcl.2016.03.111 Golbraikh, A., Tropsha, A., 2002. Beware of q2! J. Mol. Graph. Model. 20, 269–276. doi:10.1016/S1093-3263(01)00123-1

CC E

Gupta, P., Garg, P., Roy, N., 2012. Identification of Novel HIV-1 Integrase Inhibitors Using Shape-Based Screening, QSAR, and Docking Approach. Chem. Biol. Drug Des. 79, 835–849. doi:10.1111/j.1747-0285.2012.01326.x

Gupta, S.P., Mathur, A.N., Nagappa, A.N., Kumar, D., Kumaran, S., 2003. A quantitative structure-activity relationship study on a novel class of calcium-entry blockers: 1[(4-(aminoalkoxy)phenyl)sulphonyl]indolizines. Eur. J. Med. Chem. 38, 867–73.

A

Hickey, J.P., Passino-reader, D.R., 1991. Linear Solvation Energy Relationships : “ Rules of Thumb ” for Estimation of Variable Values 25, 1753–1760. Jain, A.N., 2007. Surflex-Dock 2.1: robust performance from ligand energetic modeling, ring flexibility, and knowledge-based search. J. Comput. Aided. Mol. Des. 21, 281– 306. doi:10.1007/s10822-007-9114-2

Jain, A.N., 2003. Surflex: fully automatic flexible molecular docking using a molecular similarity-based search engine. J. Med. Chem. 46, 499–511. doi:10.1021/jm020406h Klebe, G., Abraham, U., Mietzner, T., 1994. Molecular Similarity Indices in a Comparative Analysis (CoMSIA) of Drug Molecules to Correlate and Predict Their Biological Activity. J. Med. Chem. 37, 4130–4146. doi:10.1021/jm00050a010 Kubinyi, H., 2003. Comparative Molecular Field Analysis (CoMFA). Handb. Chemoinformatics 1555–1574. doi:10.1002/9783527618279.ch44d

P a g e 19 | 26

I N U SC R

Lipinski, C.A., 2004. Lead- and drug-like compounds: the rule-of-five revolution. Drug Discov. Today Technol. 1, 337–341. doi:10.1016/j.ddtec.2004.11.007 Magnuson, N.S., Wang, Z., Ding, G., Reeves, R., 2010. Why target PIM1 for cancer diagnosis and treatment? 1461–1478. Möröy, T., Grzeschiczek, a, Petzold, S., Hartmann, K.U., 1993. Expression of a Pim-1 transgene accelerates lymphoproliferation and inhibits apoptosis in lpr/lpr mice. Proc. Natl. Acad. Sci. U. S. A. 90, 10734–8. doi:10.1073/pnas.90.22.10734

Nawijn, M.C., Alendar, A., Berns, A., 2011. For better or for worse: the role of Pim oncogenes in tumorigenesis. Nat. Rev. Cancer 11, 23–34. doi:10.1038/nrc2986

A

Pires, D.E. V, Blundell, T.L., Ascher, D.B., n.d. pkCSM: Predicting Small-Molecule Pharmacokinetic and Toxicity Properties Using Graph-Based Signatures. doi:10.1021/acs.jmedchem.5b00104

doi:10.1021/je60033a020

M

Purcell, W.P., Singer, J.A., 1967. A brief review and table of semiempirical parameters used in the Hueckel molecular orbital method. J. Chem. Eng. Data 12, 235–246.

ED

Roy, P.P., Paul, S., Mitra, I., Roy, K., 2009. Models 1660–1701. doi:10.3390/molecules14051660 Rücker, C., Rücker, G., Meringer, M., 2007. Y-randomization and its variants in QSPR/QSAR. J. Chem. Inf. Model. 47, 2345–2357. doi:10.1021/ci700157b Santio, N.M., Vahakoski, R.L., Rainio, E.-M., Sandholm, J.A., Virtanen, S.S., Prudhomme, M., Anizon, F., Moreau, P., Koskinen, P.J., 2010. Pim-selective inhibitor

PT

DHPCC-9 reveals Pim kinases as potent stimulators of cancer cell migration and invasion. Mol. Cancer 9, 279. doi:10.1186/1476-4598-9-279 Tropsha, A., Gramatica, P., Gombar, V.K., 2003. The importance of being earnest: Validation is the absolute essential for successful application and interpretation of QSPR models. Qsar Comb. Sci. 22, 69–77. doi:10.1002/qsar.200390007

CC E

Wang, J., Kollman, P.A., Kuntz, I.D., 1999. Flexible ligand docking: a multistep strategy approach. Proteins 36, 1–19. Wang, Z., Bhattacharya, N., Weaver, M., Petersen, K., Meyer, M., Gapter, L., Magnuson, N.S., 2001. Pim-1: a serine/threonine kinase with a role in cell survival, proliferation, differentiation and tumorigenesis. J. Vet. Sci.

Wold, S., 1991. Validation of QSAR’s. Quant. Struct. Relationships 10, 191–193. doi:10.1002/qsar.19910100302

A

Wu, B., Wang, H.L., Cee, V.J., Lanman, B.A., Nixey, T., Pettus, L., Reed, A.B., Wurz, R.P., Guerrero, N., Sastri, C., Winston, J., Lipford, J.R., Lee, M.R., Mohr, C., Andrews, K.L., Tasker, A.S., 2015. Discovery of 5-(1H-indol-5-yl)-1,3,4-thiadiazol-2-amines as potent PIM inhibitors. Bioorganic Med. Chem. Lett. 25, 775–780. doi:10.1016/j.bmcl.2014.12.091

Wurz, R.P., Pettus, L.H., Jackson, C., Wu, B., Wang, H.L., Herberich, B., Cee, V., Lanman, B.A., Reed, A.B., Chavez, F., Nixey, T., Laszlo, J., Wang, P., Nguyen, Y., Sastri, C., Guerrero, N., Winston, J., Lipford, J.R., Lee, M.R., Andrews, K.L., Mohr, C., Xu, Y., Zhou, Y., Reid, D.L., Tasker, A.S., 2015. The discovery and optimization of aminooxadiazoles as potent Pim kinase inhibitors. Bioorganic Med. Chem. Lett. 25, 847–855. doi:10.1016/j.bmcl.2014.12.067 Zheng, J., Xiao, G., Guo, J., Zheng, Y., Gao, H., Zhao, S., Zhang, K., Sun, P., 2011. Exploring QSARs for 5-Lipoxygenase (5-LO) Inhibitory Activity of 2-Substituted 5-

P a g e 20 | 26

I N U SC R

Figure 1: 3D-QSAR structure superposition and alignment of training set using molecule 21 as a template

A

CC E

PT

ED

M

A

Hydroxyindole-3-Carboxylates by CoMFA and CoMSIA. Chem. Biol. Drug Des. 78, 314–321. doi:10.1111/j.1747-0285.2011.01146.x

P a g e 21 | 26

I N U SC R CoMSIA model

9

8

A

7.5 7

Training set Test set

M

Predicted pIC50

8.5

y = 0,8502x + 1,1271 R² = 0,8502

6 6

6.5

7

7.5

8

8.5

9

Observed pIC50

Figure 2: Experimental versus predicted activity of the training and test set based on the

A

CC E

PT

ED

6.5

P a g e 22 | 26

I N U SC R A M ED PT CC E

A

Figure 3: Std* coeff. contour maps of CoMSIA analysis with 2 Å grid spacing in combination with compound 16. (a) Steric contour map: green contours (80% contribution) indicate regions where bulky groups increase activity, while yellow contours (20% contribution) indicate regions where bulky groups decrease activity. (b) Hydrophobic contour map. Yellow contours (80% contribution) indicate regions where hydrophobic substituents are favored, gray contours (20% contribution) refer to regions where hydrophilic substituents are favored. (c) Hydrogen-bond donor contour map. The cyan contours (80% contribution) for donor-bond groups increase activity; purple contours (20% contribution) indicate the disfavored region. (d) Hydrogen-bond acceptor contour map. The magenta contours (80% contribution) for hydrogen-bond acceptor groups increase activity; red contours (20% contribution) indicate the disfavored region

P a g e 23 | 26

I N U SC R A M ED

A

CC E

PT

Figure 5: 2D view of the binding conformations and ligand interactions of the most active inhibitor 21 at the active site of PIM1

Figure 6: 3D view of the binding conformations and hydrogen bond interactions of the proposed T1 inhibitor at the active site of PIM1

P a g e 24 | 26

I N U SC R

E D

E

G

F

A

D

A

A

C

H Inhibitor 21

T1

B

B

A

M

C

ED

Figure 7: The possible phase I and phase II metabolites of candidate compound T1 and inhibitor 21 developed through different biochemical reactions. (A)

A

CC E

PT

AmideHydrolysis, (B) Oxidation, (C) Desulfuration, (D) Dealkylation, (E) Hydroxylation, (F) Glucuronidation, , (G) Demethylation, (H) Hydroxydation

P a g e 25 | 26

P a g e 26 | 26

A ED

PT

CC E

IP T

SC R

U

N

A

M