3D-QSAR - ACS Publications - American Chemical Society

Subscriber access provided by CMU Libraries - http://library.cmich.edu

Article

Energy Based Pharmacophore and 3D QSAR Modeling Combined with Virtual Screening to Identify Novel Small Molecule Inhibitors of SIRT1 Venkat K Pulla, Dinavahi Saketh Sriram, Srikant Viswanadha, Dharmarajan Sriram, and Perumal Yogeeswari J. Chem. Inf. Model., Just Accepted Manuscript • DOI: 10.1021/acs.jcim.5b00220 • Publication Date (Web): 04 Dec 2015 Downloaded from http://pubs.acs.org on December 9, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Chemical Information and Modeling is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 64

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Chemical Information and Modeling

Energy Based Pharmacophore and 3D QSAR Modeling Combined with Virtual Screening to Identify Novel Small Molecule Inhibitors of SIRT1 Venkat Koushik Pullaa, Dinavahi Saketh Srirama,b, Srikant Viswanadhab, Dharmarajan Srirama, c, Perumal Yogeeswaria, c* a

Computer-Aided Drug Design Lab, Department of Pharmacy, Birla Institute of Technology and Science-Pilani, Hyderabad Campus, Hyderabad-500078, Telangana, India b

c

Incozen Therapeutics Private Limited, 450, Alexandria Knowledge Park, Phase-I, Shameerpet, Hyderabad-500078, Telangana, India

Yogee’S Bioinnovations Private Limited, Room 5, Technology Business Incubator, BITSPilani, Hyderabad campus, Shameerpet, Hyderabad-500078, Telangana, India

*For correspondence Prof. P. Yogeeswari ([email protected]) Telephone: +91-40-66303515 Fax: +91-40-66303998

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract Silent mating type information regulation 2 homolog 1 (SIRT1) being homologous enzyme of silent information regulator-2 gene in yeast has multifaceted functions. It deacetylates wide range of histone and non histone proteins hence it has good therapeutic importance. SIRT1 was believed to be overexpressed in many cancers (prostate, colan) and inflammatory disorders (rheumatoid arthritis). Hence designing inhibitors against SIRT1 could be considered valuable. Both structure based and ligand based drug design strategies were employed to design novel inhibitors utilizing high-throughput virtual screening of chemical databases. Energy based pharmacophore was generated using the crystal structure of SIRT1 bound with a small molecule inhibitor and compared with ligand-based pharmacophore model which showed four similar features. 3D QSAR model was developed and validated to be employed in the virtual screening protocol. Among the designed compounds, lead 17 emerged as promising SIRT1 inhibitor with IC50 of 4.34 µM and at nanomolar concentration (360 nM) attenuated proliferation of prostate cancer cells (LnCAP). In addition, lead 17 significantly reduced production of reactive oxygen species, thereby reducing pro inflammatory cytokines like IL6 and TNF-α. Further the anti-inflammatory potential of the compound was ascertained using animal paw inflammation model induced by carrageenan. Thus the identified SIRT1 inhibitors could be considered as potent leads to treat both cancer and inflammation.


Page 2 of 64

Page 3 of 64

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Graphical Abstract



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Introduction The sirtuin family proteins are nicotinamide adenine dinucleotide (NAD) dependent class III histone deacetylases. The mammalian genome has seven sirtuins (SIRT1-7) which are orthologues to yeast Sirt2 protein1. Among the seven sirtuins, SIRT1 that shows multiple functions in pathological disorders, has highest similarity to Sir22. It deacetylates both histone and non histone proteins3 and several transcription factors are deacetylated by SIRT1, including members of the forkhead box class O (FoxO)4 family, nuclear factor κB (NFκB)5, peroxisome proliferator-activated receptor-γ (PPARγ)6, and transcriptional co-activator PPARγ coactivator-1α (PGC-1α)7 regulating homeostasis in energy intensive tissues. SIRT1 has been found associated with cancer, as it deacetylates p538 and repress the transcriptional activity and apoptosis. Silencing SIRT1 could induce apoptosis as SIRT1 also deacetylates cell cycle regulator E2F1 and could further inhibit the apoptosis during the DNA damage9. In contrast to the above findings SIRT1 was shown to augment apoptosis in response to TNF-α in HEK293 cells10. In addition to its role in cancer it has also been shown to play a critical role in inflammation as many HDACs were shown to be overexpressed in inflammatory conditions like rheumatoid arthritis11, specifically SIRT1 was found to be upregulated in synovial tissues of rheumatoid arthritis patients. It was reported that TNF-α might stimulate the expression of SIRT1 in rheumatoid arthritis synovial fibroblasts12 (RASFS). Further, inhibition of SIRT1 through inhibitors or siRNA was reported to reduce levels of TNFα in monocytes, besides knockdown of SIRT1 also resulted in reduction of pro-inflammatory IL-6 and IL-8. As SIRT1 is involved in myriad of biological functions it is considered to be an active target for design of small molecule inhibitors. The first few small molecule inhibitors identified were sirtinol13 and splitomycin14, which showed micromolar potencies. The other sirtuin inhibitors identified were cambinol15 and tenovins16, with moderate potency and reduced


Page 4 of 64

Page 5 of 64

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


tumor cell growth. Earlier our research group had reported on various small molecule inhibitors based on lead optimization approach17, 18 and virtual screening19. In continuation to our efforts, we aimed at identifying novel SIRT1 small molecule inhibitors through virtual screening utilizing energy based and ligand based pharmacophore models. The resulted lead compounds identified from virtual screening of chemical libraries were tested in in-vitro and in-vivo studies. Material and methods: Computational details 3D QSAR and pharmacophore features were developed using PHASE 3.4 module in Maestro 9.3 software package (Schrödinger, LLC) and docking studies were executed using Glide 5.8 module (Maestro 9.3). Finally ADME properties were predicted using QikProp 3.5 in Maestro 9.3 software package. Protein structure preparation Protein with PDB code 4I5I having 2.50 Å resolution was considered for this study and prepared using the protein preparation wizard (Maestro, 9.3). To the hetero groups of the protein bond orders and formal charges were added along with addition of hydrogens. Water molecules in all the system were removed and energy was minimised using OPLS_2005 force field. Generation of epharmacophore Using Glide module (Maestro, 9.3) energy grids were generated for the prepared complex, and the binding site was defined by a rectangular box surrounding the crystal ligand. Bound ligand (EX527 analogue) was refined using the refine tool in Glide module, and then option for output Glide XP descriptor information was selected. All the default settings were used for scoring and refinement. Pharmacophore features were generated by default using Phase module, which used default set of six chemical features containing hydrogen bond acceptor



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(A), hydrogen bond donor (D), hydrophobic (H), negative ionisable (N), positive ionisable (P) and aromatic ring (R). As per the hybridization of acceptor atom hydrogen bond acceptor sites were depicted in red sphere and vectors along the axis of hydrogen bond, whereas the hydrogen bond donors were represented in blue sphere and vector arrow to define directionality. Each pharmacophore feature was assigned specific energy value equal to the sum of the Glide XP contributions of all the atoms in the site. As a result this allowed the sites to be quantified and ranked with respect to energy values. Human SIRT1 crystal structure with PDB ID 4I5I bound to known inhibitor (EX527 analogue) was retrieved from protein databank. Although there were other crystal structures (4KXQ, 4IG9 and 4IF6) reported for human SIRT1, there were no known inhibitors bound in the catalytic pocket and hence not utilized in our study. Energy based pharmacophore modelling was carried out on the crystal structure (4I5I) with 2.5 Å resolution. The protein was prepared to correct missing hydrogen, ambiguous protonation states and flapped residues. After removing water molecules protein structure was minimized and refined using OPLS_2005 force field. The prepared protein was later used for docking.

Pharmacophore validation In this study we have generated pharmacophores from both structure and ligand based drug designing strategies. The generated pharmacophore performance was evaluated using benchmark dataset from Schrodinger (http://www.schrodinger.com/glidedecoyset website), This set is a set of decoys that have similar properties to the active compounds but are topologically dissimilar. Further, we also evaluated ranking of actives using a variety of wellestablished methods including Enrichment factor (EF), RIE, ROC and BEDROC.


Page 6 of 64

Page 7 of 64

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The validation set composed of 1055 compounds, containing 1000 decoys and 55 known inhibitors wes employed to calculate enrichment and good ness of fit values. The drug-like decoy set of 1000 compounds was obtained from glide module. The ability of the pharmacophore to retrieve the known actives were determined using various parameters. Enrichment factor 1% (EF(1%))20 was one of the parameter considered, that determined the fraction of actives recovered after 1% of decoy database (1055 molecules) was screened.

EF =

(1)

/

Where Ya is yield of actives {Ya =

X100} and D is the total

number of database compounds Despite the early recognition problem the EF has some problems ignoring complete ranking of the whole dataset molecules . A method i.e. superior to random selection of compounds has EF > 1. To address the problem of EF. RIE metric was developed which gives heavier weight in “early recognized” hits

RIE = !

$%& ' ∑! () # $,$-

+ . * ,- $ *

(2)

/'

Where Xi= is the relative rank of the ith active compound and α is the tunning parameter. 0

Changing the parameter α, one can control the early ranking of hits. BEDROC values ranges in between [0,1] and can be defined as the probability that an active is ranked before a randomly selected compound was exponentially distributed with parameter α. BEDROC and RIE have a linear relationship

BEDROC= RIE 5

! 678 : * 9 ! ;78 :?@8 < A : 9 9 *

+

C 0.5.

The r2 value was calculated by the following formula:

R=

p )( q o 0.5 Close to 1 r2 > 0.5 0.85 ≤ k ≤ 1.15 0.85 ≤ k′ ≤ 1.15 R[k or R′[ k close to r2 R[k or R′[ k close to r2 2 [ [(r - Rk ) /r2] < 0.1 2 [(r2- R′[ k ) /r ]] < 0.1 [ rn (LOO) > 0.5 [ rn >0.5 [ R >0.5

[ r;

Cross validated coefficient.

R (or r2)

Correlation coefficient between the actual and predicted activities.

k and k′

Slope values of regression lines.

R[k and R[k :

Correlation coefficients for the regression lines through the origin.

2 [(r2- R[k ) /r2]and [(r2- R′[ k ) /r ]

[ rn (LOO) method.

To calculate the relation between r2, R[k , and R′[ k. Modified squared correlation coefficient for the “Leave One Out”



1 2 3 4 5 6 7 8 S.NO 9 1 10 2 11 12 3 13 4 14 5 15 6 16 17 7 18 8 19 20 9 21 10 22 23 11 24 12 25 26 13 27 14 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 50 of 64

Table 6: Docking score, important aminoacids, hydrogen bonds and %enzyme inhibition at 40µM concentration of lead compounds Compound

Important aminoacids

Fitness

Docking score

Hydrogen bonds

%Inhibition at 40µMa

Cry lig Lead 1

Gln-345, Ile-347, Asp-348 , Phe-297 Gln-345, Asp-348, Ile-347, Phe-297, Phe-273

2.77 1.09

-9.02 -13.69

4 4

92.24± 5.25 20.73±1.82

Lead 2 Lead 3

Gln-345, Ile-347, Asp-348, Phe-321, Phe-297 Ile-347, Asp-348 , Phe-273

1.20 1.76

-9.23 -9.18

4 4

11.63±0.86 29.88±2.61

Lead 4 Lead 5

Ile-347, Asp-348, Gln-345, Phe-297 Asp-348, Ile-347, Phe-273, Phe-297

2.61 2.68

-11.12 -10.66

4 4

10.25±0.85 50.22±3.11

Lead 6

Ile-347, asp-348, Gln-345, Phe-297 Ile-347, Asp-348, Asn-346, Gln-345, Phe-273, Phe321 Gln-320, Asp-348, Ile-347, Gln-345, Phe-31, Phe297 Asp-348, Ile-347, Gln-345, phe-297

2.17

-11.97

4

57.14±1.14

2.04

-11.53

6

48.69±1.32

1.94 2.00

-10.20 -13.45

6 4

59.88 ±1.46 77.32 ± 3.42

1.91 1.71

-12.04 -10.28

6 5

81.87 ± 2.51 83.71 ± 4.14

1.60

-9.51

6

Lead 7 Lead 8 Lead 9 Lead 10 Lead 11 Lead 12

Ile-347, Asp-348, Asn-346, Gln-345, Phe-273 Ile-347, Asp-348 , Gln-345 , Phe-294 Asp-348, Ile-347, Gln-345, Ile-316, Phe-273, Phe297

Lead 17

Gln345, Arg282, Phe273

1.61 -8.10 2 a Enzyme inhibitions studies were initially performed at 40µM concentration. Leads showing more 60% inhibition were further diluted to lower concentration to calculate IC50 values. Each reaction was performed in triplicates and standard deviation values were represented.


85.9 ± 2.84 91.13 ± 3.12

Page 51 of 64

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 7 Predicted ADME properties of lead compounds Molecules

Molecular weight a

QPlogP o/wb

QPPCacoc

QPlogHERGd

Percent human oral absorptione

Rule of fivef

Lead9 Lead10 Lead11 Lead12

243.30 283.32 253.29 247.24

2.18 1.40 2.78 -0.26

1052.43 155.851 1352.41 65.82

-4.19 -4.96 -5.11 -2.94

93.83 74.41 100 57.96

0 0 0 0

Lead17

322.2

1.14

32.21

-5.71

58.66

0

a

Molecular weight (acceptable range

3D-QSAR - ACS Publications - American Chemical Society

Recommend Documents