Добавил:
kiopkiopkiop18@yandex.ru t.me/Prokururor I Вовсе не секретарь, но почту проверяю Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5927_Библиотеки_им_академика_М_И_Перельмана

.pdf
Скачиваний:
0
Добавлен:
15.09.2026
Размер:
14 Мб
Скачать
☆
220
Drug Repurposing and Computational Drug Discovery: Strategies and Advances
success of CANDO first version (v1), as it has predicted >1 billion interactions in humans (3733 approved compounds and 48,278 protein structures), and drug predictions have been done for 2030 indications.
32
• In-house interaction scoring protocol
—It facilitates faster assess ment of query compound–protein interaction score and their interac tion similarities.
• Benchmark accuracy
30–32
—The benchmarking validates the accuracy of the CANDO platform and opens the gateway for shotgun drug repur posing and novel discovery. It applies different types of benchmarking protocols that recognize the relationship between known drugs with respect to a particular disease or an indication for which they have received approval. In total, CANDO v1 has done benchmarking for
1439 indications (≥2 approved compounds). After bench validation,
the compounds move toward in vitro and in vivo screening assays. When the query compound passes this phase, then it is forwarded to the clinical trials. Besides v1 pipeline, ligand and structure-based pipelines are also applied for drug repurposing and calculating the benchmarking performance of putative drug candidates.
-
-
-
31
CANDO has been effective in predicting apernyl, prednisolone, predni sone, and cloquinate for autoimmune disorder systemic lupus erythematosus (SLE). The indications for Alzheimer’s disease and diabetes mellitus 2 have been obtained. for Ebola virus disease (EVD).
Virtual screening/traditional HTS CANDO
Closed system In virtual screening, the interaction of the
ligand with the active site of protein is given preference.
The binding interaction is possible with a limited set of biomolecules, cells, and tissues. Hence, the total performance of the test compound remains incomplete. So, the chances of failure are high during clinical trials or even after
30
33
Open system In CANDO, the interaction of the ligand
with the active site of protein is a first step toward determining the lead compound
The binding interaction is possible with all macro and micro biomolecules such as nucleic acid, RNA, proteins (enzymes, receptors), lipids, carbohydrates, etc. This provides a complete framework about the performance of the test compounds; hence, the chances of failure are low/minimal during a clinical trial or even after
-
Virtual screening/traditional HTS CANDO
Данная книга находится в списке для перевода на русский язык сайта https://meduniver.com/
Not based on polypharmacology Off- and anti-target effects of the test
compound remain unknown. So, the chances of drug failure increases
Focusses on single disease aetiologies
Molecular docking and simulation study play a prominent role in drug discovery via virtual screening
Virtual screening is one of the computational tools that is applied in drug repurposing
Based on polypharmacology Off- and anti-target effects are known as
wider space is provided for interaction. This surfaces the pleiotropic effect of the compound, as biologists and scientists become aware of other mechanisms of action as well. This not only kindles the beneficial effects but also safeguards from the adverse effects of the compound
Open system, big databases, docking/ simulations, and benchmarking study leverages CANDO to focus on multiple diseases aetiologies
Molecular docking and simulations are important computer-aided tools among several other software tools
CANDO provides a complete platform for drug repurposing
221 Drug Discovery for Aging and Neurological Disorders

Reverse screening, also known as in silico/computational target fishing, tries to find out the appropriate target structure from a query ligand structure. It comprises of shape screening, pharmacophore screening, and reverse docking. The first two methods are used by overall comparison of shape or pharmacophore when crystal structure of protein is not available. In reverse docking, the target structure is searched based on known ligand (crystal structure of protein available). Despite the development of Big Data, evolving computational tools and techniques, reverse screening has garnered attention because with every new drug discovery, new drug indica
­tions (20% of new drugs launched in 2013) pops-up. These new indications (majorly multi-target) need a potent target to be categorized as a drug. Here, reverse screening is the suitable choice, where these new indications will be repurposed as a novel drug. Besides this, after passing clinical trials, several compounds fail due to their off-target or anti-target effects. So, a new target can be assigned to those failed compounds and studying their proper mechanism of action will give a new life to failed or withdrawn drugs. This will save a lot of time, and money. Hence, reverse screening has a significant contribution in the drug repurposing/drug repositioning.
34
Drug Repurposing and Computational Drug Discovery: Strategies and Advances

The principle of shape screening is based upon two viewpoints- (1) 2D: the molecules that have a similar structure, target similar proteins, and show similar biological activity, (2) 3D: molecules having same volume can fit in the same space or volume of the active site present in the target protein. The screening involves two steps; first is to map the ligands present in the database with the query target molecules and then second is the validation step, where mapped ligands undergo further mapping with their annotated targets (proteins whose information is already present in the database). The selection of potent targets is based on similarity scores. The software or programs used in the shape screening are as follows: 2D: FingerPrint 2D (FP2) - extended connectivity fingerprints (ECFPs), and Molecular ACCess System (MACCS), the MDL structural key, ChemProt 3.0. 3D: gWEGA, WEGA, SHAFTS (encoded in ChemMapper). ROCS, Target Hunter, CSNAP3D, SEA, and Swiss Target Prediction are programs and algorithms used in the shape screening method.
34–37

The presence of chemical entities, chemical structures, or bonds such as hydrogen bond acceptor (HBA) or donor (HBD) vector, hydrophobic center (H), positively (P) or negatively (N) charged centers, create a space where ligands and target can interact effectively—space is known as pharmacophore. The pharmacophore screening follows a basic principle—pharmacophore directs the ligand and target interaction. The first step in the pharmacophore screening is to map pharmacophore models of a query (target) and ligands present in the database, and the second is to map the first step results with their annotated targets. Pharmacophore modeling – (1) ligand-based, (2) structure-based, (3) complex-based builds pharmacophore database. Ligand­based modeling incorporates QSAR; structure-based models is built upon by Pocket v.2, and Catalyst SBP in Discovery Studio (DS) (BIOVIA, 2017); complex-based models are constructed by PharmTargetDB (PharmMapper), and PharmaDB in Discovery Studio.
34

In reverse docking, generally, the active site of the target is known or derived from a co-crystal ligand (small-molecule). The steps involved in reverse
34,20,35
docking are the following
:
• For target fishing: Potential Drug Target Database (PDTD) sc-PDB,
Данная книга находится в списке для перевода на русский язык сайта https://meduniver.com/
Protein Data Bank (PDB) (Ligand Expo), Therapeutic Target Data
-
base (TTD), Pocketome, ZINC, DrugBank, ChEMBL, PubChem & PDSPKi (BindingDB), UniProt, and ChEBI are used.
• Reverse docking (RD) is performed using the following programs: INVDOCK, DOCK, AutoDOCK, AutoDOCK Vina, ACTP, TarFis­DOCK, idTarget, SELNERGY, GlamDock, GOLD, MDock (uses PDTD database and computes via ITScore), and MEDock are applied. The ligand is docked with the grid database of the protein targets.
• Docking score (GLIDE) and binding/docking energy (binding
strength/interaction energy) between small molecule ligand and query target are calculated.
• Docking energy helps in ranking the potent target molecules.
Machine learning methods, such as protein atom score contributions
derived interaction ngerprint (PADIF), support vector machine (SVM), and neural networks (NN) are also being explored in target shing to enhance the sensitivity and efcacy of reverse docking method.
20

Reverse screening has been successfully applied to identify target receptors for marine or terrestrial natural compounds, off- and anti-target effects (drug toxicity/adverse actions), drug repositioning, and in the investigation of molecular mechanisms against chronic infectious diseases and cancer.
34

• SELNERGY, a reverse docking tool was used to repurpose tofisopam, an antianxiety drug.
36
• Off-target effects of torcetrapib-cholesteryl ester transfer protein (CETP) inhibitor were identified using CDOCKER (Accelrys Soft ware Inc., Discovery Studio Modeling Environment, CA, USA).
against Helicobacter pylori.
• Identification of anticancer targets.
34
20
-
34
224
Drug Repurposing and Computational Drug Discovery: Strategies and Advances
 Applications of Shape and Pharmacophore Screening.
Screening method Applications
Shape screening Drug repurposing: identified receptors (target) for the
following drugs—Prozac, Validex, and Rescriptor. Identified targets for the following small molecules -
Plumbagin, Obacunone, 5-aza-dc, Sini decoction (aconitine, liquiritin, 6-gingerol), Salvinorin A, Lignan, and Wuweizi
Pharmacophore screening Identified targets for the following small molecules: BBR,
Phytoestrogens (genistein, daidzein, secoisolariciresinol), UA, NCI 748494/1, HSYA, Arctigenin, CT, and 5,7,
-dihydroxy 4 - methoxy -8-prenylflavanone
34

Based on structure or ligand-based programs or algorithms, the compounds are sorted on the grounds of genomic, transcriptomic, or proteomic signatures. There are several sensitive and efficient data repositories, databases, and soft­ware designed to accomplish the task of drug discovery and drug repositioning. These tools help us analyze how these compounds will operate in the human system by envisioning test compound–protein, test compound–protein–gene/ test compound–gene interaction, and activation of signaling pathways; hence, elucidating on-, off-, and anti-target effects of the test compound. There are public repositories, such as National Center for Biotechnology Information (NCBI), DNA Data Bank of Japan (DDBJ), European Bioinformatics Insti­tute (EBI), etc., that provide information about diseases related to changes produced in the transcriptome; yet, there is a need for the development of a transcriptomic-based data repository. The reasons are as follows:
37
• Preclinical: To determine the mechanism of action (MoA) and other effects of perturbagens, in vitro, and in vivo studies are performed. Clinical: Before administering the test compounds, a range of effec­tive doses are required which must be determined in preclinical stage so, preclinical requirements are costly and time-consuming, whereas clinical one is tedious and requires perfection. Hence, to minimize the preclinical issues and solve the clinical issues, transcriptomic-based data repository is required.
• Research related to lead discovery is carried out simultaneously in several laboratories, and the results of each lab may vary. So, a constant tool is required which is devoid of human errors, minimizes
225 Drug Discovery for Aging and Neurological Disorders
Данная книга находится в списке для перевода на русский язык сайта https://meduniver.com/
discrepancy and disputes, and provides a platform for benchmarking and validation of results.

The connectivity map establishes the connections between diseases, genes, and perturbagens (drugs, small molecules, shRNA, cDNA, or other biologics). The aim is to perturb the existing condition using perturbagens, modify the disease-associated gene expression, and study the downstream signaling pathways inside the cell. This first-generation data repository was constructed by treating different cancer cell lines—human prostate cancer (PC3), breast cancer (MCF-7), leukemia (HL-60), and melanoma (SKMEL5) with perturbagens. The cell lines were treated with varying doses of perturbagens at varying time points.
37–40
CMap approach has been extensively utilized in the eld of cancer; its
applications are listed below.
Lead discovery of the following drugs (in vitro validation):
 Connectivity map in drug discovery or drug repositioning.
39,40
226
Drug Repurposing and Computational Drug Discovery: Strategies and Advances
Explored mechanism of action (MoA) of the following drugs: VXL50 (ovarian cancer), b-AP15 (AML), thioridazine (ovarian cancer), epoxy anthraquinone derivatives (EAD) (neuroblastoma), celastrol and gedunin (prostate cancer), cisplatin (chemotherapy resistance), etc. Apart from cancer, it has also been explored in Down’s syndrome and obesity.
 Role of CMap in Drug Repositioning.
Drug Initial therapy Repurposed for Preclinical
validation
Piperazine
Fluphenazine
Topiramate
Phenothiazines
Cimetidine
Chlorpromazine and trifluoperazine
Ursolic acid
Phenoxybenzamine
Vorinostat
Antipsychotic CNS injury In vitro
Antipsychotic Hair growth In vivo (rodent)
Anticonvulsant Inflammatory
bowel disease (IBD)
Antipsychotic and antihistamine
Antiulcer Lung cancer In vitro and in
Schizophrenia/ psychotic disorders
Indications for anti-inflammatory, antioxidant, antiapoptotic, and anticarcinogenic
41
role
Antihypertensive Osteoarthritic pain
Cutaneous T-cell lymphoma
Breast cancer In vitro
Hepatocellular carcinoma (HCC)
Muscle atrophy In vitro and in
Gastric cancer In vitro
In vivo (rodent)
vivo (rodent)
In vitro and in vivo (rodent)
vivo (rodent)
In vivo (rodent)
Library Integrated Network-based Cellular Signatures (LINCS): For CMap, Lamb et al. tested 164 small molecules on four cancer cell lines. Further, to make this system diverse, more molecules and cell lines were
included. The sensitivity and efcacy of the system were enhanced by using uorescent-colored microspheres and the ow cytometry detection
technique. This advanced technique was named L1000 (or large-scale
connectivity map) containing 1058 probes. These probes were conrmed
227 Drug Discovery for Aging and Neurological Disorders
Данная книга находится в списке для перевода на русский язык сайта https://meduniver.com/
by targeting landmark transcripts; 955 shRNAs were developed for 978 landmark transcripts and 80 control transcripts. CMap-L1000v1 generated a 1000-fold large dataset in comparison to CMap. Total
1319138 L1000 proles and 473647 gene signatures were created by
testing 42,080 perturbagens (19,811 small molecule test compounds, 314 biologics, ~18,500 shRNAs, and 3462 cDNAs against 5075 genes in >100 cell lines) (experiments were performed in triplicate, treatment timeline was in between 6 and 24 h). L-1000 is now termed as Library of Integrated Network-Based Cellular Signatures (LINCS), which is also known as high-throughput reduced representation of transcriptome
proling method and its efciency is at par with RNA sequencing.
41–45
In the LINCS database primarily breast adenocarcinoma (MCF-7), pancreatic carcinoma (YAPC), colorectal adenocarcinoma (HT29), malignant lung melanoma (A375), prostate cancer cell lines – prostate adenocarcinoma (PC3), and metastatic prostate (VCAP), lung cancer cell lines – non-small cell carcinoma (NSCLC) (A549) and non-small cell adenocarcinoma (HCC515), and hepatocellular cancer cell line (HEPG2) are present, but NeuroLINCS has opened the gateway for the central nervous system (CNS) disorders as well. NeuroLINCS has been main­tained and established by culturing organoids and induced pluripotent stem cells (iPSCs) from the neuronal cells of patients suffering from CNS disorders, such as Alzheimer’s diseases, spinal muscular atrophy (SMA), and amyotrophic lateral sclerosis (ALS). NeuroLINCS encapsulates datasets for all “omics,” viz, genomic (epigenomics), transcriptomic,
37
proteomic, and imaging. candidates that can be repurposed for Alzheimer’s disease.
NeuroLINCS and CMap have explored drug
43
The tools and software designed to browse, visualize, and analyze the LINC database are Enrichr, LINCS Data Portal, Slicr (LINCS L1000 Slicer GSE70138), LIFE, L1000CDS, LINCS Canvas Browser, and iLINCS.
39,62
Genome-Wide Associated Studies (GWAS): Due to diverse alleles, the genomic variation is prominent in neuropsychiatric disorders (include Alzheimer’s disease and dementia, Parkinson’s disease, schizophrenia, Huntington’s disease, depression, and bipolar disorders). These alleles and single-nucleotide polymorphisms (SNPs) increase the risk of developing neuropsychiatric disorders. GWAS approach is based on SNPs and has a signicant contribution to drug discovery and drug repurposing for neuro­psychiatric disorders. In the meta-analysis study (796 studies), GWAS iden-
tied small-molecule targetable genes, and biopharmable genes (generate peptides/transmembrane domains modied into pharmaceutical agents
228
Drug Repurposing and Computational Drug Discovery: Strategies and Advances
or drugs).
46–50
GWAS has been intricately explored in drug discovery and
repurposing for neurological disorders, discussed in detail below.
Articial Intelligence and Machine Learning: Articial intelligence
has given a new dimension and horizon to the computational drug designing. This platform is based on the mathematical approach and provides new indications with precision and accuracy. Articial intel­ligence and machine learning (ML) is a combination of applied mathe-
matics and software/databases. In the computer science, ML is a sub-eld of articial intelligence (AI). Machine learning performs its operations
or tasks in three ways.
70–75
• Supervised learning algorithms: Supervised learning algorithms, such as SVM or (Deep) neural networks have been applied in the biological field.
• Unsupervised learning: Unsupervised learning determines the related relationships or patterns present in the unlabeled data. It is performed by dimension reduction methods, such as PCA (principal component analysis), collaborative filtering, clustering/grouping data, and density estimation.
• Sequential learning: In this task, the input for data generation is obtained from the previous interacting environment. The goal-oriented entity that interacts with its surrounding environment, makes the data selection choice based on input. In sequence learning, from the stream of data, only one data is processed at one time by algorithms. Further, the decision made by algorithms is a trial-and-error process. Multi­Armed Bandit (MAB) algorithms belong to the sequence learning family and are associated with recommender systems. Recommenders have a significant role in the prediction of drug–target interactions (DTI) and drug repurposing. In the recommender system, the goal agent recommends the test compound (or the compounds that can be chosen as a new drug candidate). The selection criteria are based either on disease or the chemical structure/chemical composition of the test compound.
50–53
The baseline regularization algorithm of ML repurposes drugs from
electronic health record (EHR) data.
45,77
Articial intelligence and ML has several implications in Alzheimer’s
disease,
46–48
Parkinson’s disease,
49,50
and multiple sclerosis.
51,52

Данная книга находится в списке для перевода на русский язык сайта https://meduniver.com/
SMALL MOLECULES FORAGING, IN NEUROLOGICAL, AND NEURODEGENERATIVE DISORDERS
 List of Drug Repurposed Small Molecules in Varieties of Neurological and Neurodegenerative Diseases.
Diseases Repurposed drugs, Initial Approach
identification of transcription intervention factors/targets/MoA
Alzheimer’s disease
•Drugs proposed for repurposing: Vorinostat, Trimethadione, Cyproterone, Metrizamide, Cefuroxime, and Dydrogesterone
53
•Identification of master regulators of AD-ATF2 and
53
PARK2
Valsartan Galantamine
10
10
Hypertension
Paralysis, Polio Amyloid precursor protein (n=14) Proteomics Dopamine D1 receptor (n=13) Metabolomics Acetylcholinesterase (n=10) Metabolomics Plasminogen (n=7) Proteomics CGMP-specific 3, 5- cyclic Metabolomics
phosphodiesterase (n=6) Myeloid cell surface antigen GWAS
CD33 (n=6) Proposed for anti-AD
54
:
Local anesthetic CMap, LINCS
Bupivacaine (target SCN10A). SCN10A associated with tau
protein (Microtubule Associated Protein Tau, MAPT) {hallmark of AD}
Topiramate (target SCN1). SCN1 is linked with presenilin 1 (PSEN1) {hallmark of AD}
Selegiline and iproniazid (target monoamine oxidase (MAO) inhibitors
Connectivity map
54
54
54
(L1000) (LINCS downloaded from Gene Expression Omnibus, GEO), STRING
229 Drug Discovery for Aging and Neurological Disorders
54
54
54