Prediction and Experimental Confirmation of Novel ... - ACS Publications

Subscriber access provided by UNIV OF DURHAM

Pharmaceutical Modeling

Prediction and Experimental Confirmation of Novel Peripheral Cannabinoid-1 Receptor Antagonists Shayma El-atawneh, Shira Hircsh, Rivka Hadar, Yossi Tam, and Amiram Goldblum J. Chem. Inf. Model., Just Accepted Manuscript • DOI: 10.1021/acs.jcim.9b00577 • Publication Date (Web): 21 Aug 2019 Downloaded from pubs.acs.org on August 23, 2019

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 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 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.

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 29 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

Prediction and Experimental Confirmation of Novel Peripheral Cannabinoid-1 Receptor Antagonists Shayma El-Atawneh1,#, Shira Hirsch2,#, Rivka Hadar2, Joseph Tam2, *, Amiram Goldblum1, *

1Molecular

Modeling Laboratory; 2Obesity and Metabolism Laboratory, Institute for Drug

Research, School of Pharmacy, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem 9112001, Israel

#

These authors contributed equally to this work

*Address

correspondence to: Joseph Tam, [email protected]; Amiram Goldblum,

[email protected]

1 ACS Paragon Plus Environment

Journal of Chemical Information and Modeling 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 Small molecules targeting peripheral CB1 receptors have therapeutic potential in a variety of disorders including obesity-related, hormonal and metabolic abnormalities, while avoiding the psychoactive effects in the CNS. We applied our in house algorithm, Iterative Stochastic Elimination, to produce a ligand-based model that distinguishes between CB1R antagonists and random molecules, by physico-chemical properties only. We screened ~2 million commercially available molecules, and found that about 500 of them are potential candidates to antagonize CB1R. We applied a few criteria for peripheral activity and narrowed that set down to 30 molecules, out of which 15 could be purchased. Ten out of those 15 showed good affinity to CB1R and two of them with nanomolar affinities (Ki of ~400 nM). The eight molecules with top affinities were tested for activity: two compounds are pure antagonists, and five others are inverse agonists. These molecules are now being examined in vivo for their peripheral vs. central distribution, and subsequently will be tested for their effects on obesity in small animals.


Page 2 of 29

Page 3 of 29 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


1. Introduction The cannabinoid receptors (CBRs), belong to the superfamily of G-protein coupled receptors (GPCRs), include two members: the CB1 and CB2 receptors (CB1R and CB2R, respectively). They mainly signal through the heterotrimeric Gi/o type G proteins, regulating the activity of many plasma membrane proteins and signal transduction pathways, including ion channels, enzymes producing cyclic nucleotide second messengers and various kinases.1–3 When Gi/o protein-coupled receptors are concomitantly activated, CB1Rs couple to Gs 4 or Gq/11 proteins.5 CB1R is mainly expressed in the central nervous system (CNS),6–10 but also in many peripheral tissues like the gastrointestinal (GI) tract, liver and others.2,11–16 CB2R is mainly associated with immune cells and may be found at a lower concentration in the CNS. Both receptors regulate a variety of central and peripheral physiological functions, including neuronal development, neuromodulatory processes, energy metabolism as well as cardiovascular, respiratory and reproductive functions, proliferation modulation, motility, adhesion and apoptosis of cells.17 They are therefore potential therapeutic targets for many disorders.18–23 However, the pharmaceutical industry hesitates to develop agents that target these receptors due to the CNS-mediated psychoactive effects induced by cannabinoids including disruption of short-term memory, cognitive impairments, a sense of time dilation, mood alterations, sedation, enhanced body awareness and discoordination.24–29 CB1R is involved in the overall homeostatic balance and regulation of food intake, fat accumulation, lipid, and glucose metabolism in the CNS as well as in peripheral tissues (adipocytes, skeletal muscle cells, liver, kidney, pancreas and gastrointestinal tract). Stimulation of the hypothalamic CB1Rs affects neuropeptides that regulate energy homeostasis, food intake and lipogenesis in visceral tissues.30 However, stimulation of the CB1R in the nucleus accumbens excites the dopaminergic reward pathway and thus increases the motivation to eat,31 as well as to smoke or intake drugs of abuse.32 CB1R activation in adipose tissue increases the activity of the lipoprotein lipase, promoting the hydrolysis of triglycerides into non-esterified fatty acids and their subsequent uptake.30 It also enhances fat storage within adipocytes,33 and regulates adipogenesis by increasing the expression of the nuclear receptor peroxisome proliferator-activated receptor gamma, which promotes adipocyte differentiation.34 Hepatic CB1R leads to an increase in lipogenesis and to a decrease in insulin sensitivity.35 Pancreatic CB1R modulates insulin secretion,36 and in the GI tract it increases ghrelin secretion and fat ingestion.37Accumulating evidence suggests that endocannabinoids can affect glucose homeostasis by also acting onto the skeletal muscle, where CB1R activation decreases basal and insulin-mediated glucose uptake, an effect blocked by pharmacological inhibition of CB1R.37,38 Therefore, it was not surprising to find that its blockade would inhibit food intake.39–42 In fact, all of the above-mentioned observations provided the motivation for testing such compounds as a potential treatment for obesity. Indeed, the first-in-class CB1R antagonist, Rimonabant (Acomplia®, Sanofi-Aventis), proved very effective not only in reducing food intake and body weight, but also in improving the obesity-induced insulin and leptin resistance, restoring glucose homeostasis and dyslipidemias, as well as ameliorating hepatic fat accumulation in obese/overweight people with the metabolic syndrome.43–49 3 ACS Paragon Plus Environment


However, rimonabant and its synthetic analogues (such as Taranabant, Otenabant, Ibipinabant, Surinabant, Drinabant, and Rosonabant) were withdrawn from pre-clinical development and clinical testing due the growing documented evidence of anxiety, depression, and/or suicidal ideation that was reported in a small but significant fraction of humans treated with rimonabant.50 It is assumed that the central effects of rimonabant are responsible for the short-term reduction of food-intake, whereas the more sustained effects on body weight and the improvement of insulin resistance and blood lipids are more due to its peripheral actions.51 Thus, a second generation of peripherally restricted CB1R antagonists was developed as a safer therapeutic strategy for treating obesity.52,53 CB1R antagonists such as JD5037 52 and AM6545 54 reduce body weight, adiposity, insulin resistance and dyslipidemia in obese animal models 37 and lack the CNS side effects.52,54 Peripheral CB1R antagonists could also be used for the prevention and management of conditions related to obesity like cardiovascular disease, insulin resistance, dyslipidemia, hypertension, chronic inflammation, and a hypercoagulable/prothrombotic state, chronic kidney disease, also known as the metabolic syndrome.38,55 Peripheral CB1R antagonists were found to prevent the development of type 2 diabetic nephropathy,56 and a combined treatment of peripheral CB1R antagonist and CB2R agonist was shown to abolish diabetes-induced albuminuria, inflammation, tubular injury, and renal fibrosis.57 In this paper, we present our discovery of novel peripheral CB1R antagonists by applying our inhouse machine learning “Iterative Stochastic Elimination” (ISE) algorithm,58,59 which already served for the discovery of highly active molecules in a few projects.60,61 The “peripheral” preference was examined by specific criteria (see METHODS). Many of the predicted active compounds were confirmed by in vitro studies, which are also presented here.

2. Results 2.1.

ISE activity models

Several Models were built by ISE for CB1R antagonist activity, and four were selected for virtual screening (VS). The models contained filters with five ranges of descriptors each; the models differed from each other by the number and composition of filters (for detailed occurrences of the descriptors in the different models; see supporting information Figure S1). All the models had quite similar quality, as it was evident from MCC values for the top filter, from the mean MCC in each model and from the high AUC (> 0.9) of each. These numbers, taken together, indicated successful classifications by all four models (Table 1). Each of the actives and the inactives got an index for its success in each of the models. Scatter plots and tables for the learning set in each model helped to determine how to analyze results for the subsequent VS of millions of molecules. A cutoff index for VS in each model was required for deciding that molecules above that index would be further examined as potential hits. Scatter plots of the internal test sets distribution in each model are shown in Figure S2. We chose a different cutoff for each model based on the TP/FP rate, the larger, the better. We present those cutoffs (lowest line) together with results for the external set (Table 2).


Page 4 of 29

Page 5 of 29 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 1. Parameters of the different models

ISE models

Model A

Model B×

Model C

Model D

Number of filters MCC of the top filter Mean MCC

1399 0.78 0.75

1895 0.72 0.67×

1960 0.75 0.7

995 0.75 0.69

AUC 0.91 0.95 0.91 0.92 EF 60* 94** 82* 2* Sensitivity 0.54* 0.37** 0.42* 0.42* Specificity 0.99* 0.99** 0.99* 0.97* TP/FP 1.15* 16** 3.5* 16.2* ×Only top 1000 filters used for screening. *Above index 0.8, ** above index 0.7. MCC- Mathew correlation coefficient, AUC- area under the ROC curve, EF-Enrichment factor.

2.2.

Test set screening

We screened the external test set of active and inactive molecules collected from CHEMBL database (2970 molecules, see METHODS) through the four models (Table 2). Table 2. External test set validation results

2.3.

ISE models

Model A

Model B

Model C

Model D

AUC EF Sensitivity Specificity TP/FP Index cutoff used for VS

0.76 1.41 0.06 1 ∞ 0.779

0.87 1.41 0.09 1 ∞ 0.764

0.84 1.38 0.08 0.99 42.2 0.796

0.82 1.41 0.02 1 ∞ 0.8

Screening the Enamine database through the four models

The Enamine database (2,170,859 molecules)62 was screened through each of the four models. The results are summarized in Table 3. Different numbers of hits (above the cutoff index of each, line 2) were found for each of the models. Combining them gave 626 hits, which were reduced to 498 after removing duplicates.



Page 6 of 29

Table 3. Screening results of the Enamine database through the four models

Models

Model A

Model B

Model C

Model D

Number of hits

238

237

13

138

Index cutoff for VS Total hits

0.779

2.4.

0.764 0.796 Total = 498 unique hits

0.8

Applying the criteria for peripheral action

Applying our CNS-avoidance criteria (see METHODS) to the 498 hits from the four models narrowed this number to only 33 candidates. These include 11 enantiomers so that only 15 molecules have unique chemical formulae (reference name SH 1-15). SH2 was found to be unstable, so only 14 molecules were tested in vitro.

2.5.

CB1R Binding

The actual affinity of the 14 candidates was tested in a CB1R competitive binding assay. Table 4 lists the molecular structures and their binding assay results. Ten compounds showed good affinity for the CB1R. The most potent compounds, SH9 and SH11, had a CB1R Ki of ~400 nM. Moreover, rimonabant was tested under the same conditions, and its Ki values for the CB1R was 4.7 nM, in line with our previously reported values.63

Table 4. Chemical properties of the VS novel compounds Name

Structure

ISE

Index

model

Score

MW

cLogP

Hydrogen

PSA

bond

Ki (µM)

(donor)

SH1

NH

O N H Cl

H N

S

N H N H

A (D)

(0.829)

Model

0.853

A (B)

(0.791)

480.39

6.1

3

73.98

0.0

483.51

5.5

3

94.4

2.1

Cl

F

OF HN

0.853

O

F

SH3

Model

O


Page 7 of 29 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


Model

HN

SH4

H N

H N

O O

3

108.13

1.8

0.806

463.9

4.9

3

87.1

1.1

0.804

469.4

4.28

3

87.3

0.0

0.791

502.4

5.78

3

91.0

1.0

0.804

460.9

4.9

4

86.0

3.9

0.853

497.5

5.6

3

94.4

0.408

0.791

484.4

6.4

3

78.9

0.0

Model

0.853

499.4

6.8

0

84.2

0.414

A (B)

(0.791)

Model

0.847

496.4

6.1

3

82.0

1.3

S

Model

NH

SH5

4.5

O

O

Cl

524.01

D

O

O

0.812

A

H N

N H

O

Cl

Model SH6

O

O N H

F

H N

N H

F F

O O

HN

S

SH7

A

O

Cl

Model

H N

B

O Cl

N H

Model O

SH8

N H

Cl H N

N H

O

F O

S

NH

F

Model

F

HN

SH9

D

A

N H N H O

Cl

Model Cl S

B

O

N

SH10

N

NH2 OH

N Cl N

SH11

Cl

S N N Cl O

A

S NH

SH12

HN

O HN

Cl



Cl

Model

O

S

O

0.791

474.0

5.9

3

102.3

3.4

Model

0.791

482.0

5.1

3

91.0

6.5

B (D)

(0.8)

Model

0.812

474.0

5.4

3

73.9

0.938

A (D)

(0.821)

B

HN

SH13

Page 8 of 29

H N

H N S

O HN Cl HN

SH14

O S O

N H

Cl H N

SH15 HN

2.6.

H N O

O

Testing the activity of the selected compounds

By using [35S]GTPγS binding assay, we next evaluated the activity (agonism, antagonism, inverse agonism) properties of 8 out the 14 molecules that showed the highest affinity for the CB1R. The test was performed for each compound with and without the CB1R agonist HU-210 (100 nM). Whereas two compounds (SH4 and SH5) showed neutral antagonism properties, five others - SH3, SH9, SH12, SH13, and SH15 were defined as inverse agonists (Table 5). While SH11 could be a positive allosteric modulator (PAM).

Table 5. The property of the VS novel compounds (agonist, antagonist, inverse agonist). *P

Prediction and Experimental Confirmation of Novel ... - ACS Publications

Recommend Documents