Medicinal chemistry is a discipline that encloses the design, development, and synthesis of pharmaceutical drugs. The discipline combines expertise from chemistry, especially synthetic organic chemistry, pharmacology, and other biological sciences. It is also part of medicinal chemistry the evaluation of the properties of existing drugs.

The use of plants, minerals, and animal parts as medicines has been recorded since the most ancient civilizations. With the evolution of the knowledge the means for drug discovery also evolved. New molecules with potential pharmaceutical interest, "hits', are natural products, or compounds generated by computational chemistry, or compounds from a screening of chemical libraries, from combinatorial chemistry, and from pharmaceutical biotechnology. The “hit” compound is improved for its pharmacologic, pharmacodynamic and pharmacokinetic properties by chemical or functional group modifications, transforming it into a lead compound. A lead compound should have a known structure and a known mechanism of action. The lead compound is further optimized to be a drug candidate that is safe to use in human clinical trials (Figure 1) [1].

[enlarge]

Figure 1. A time frame for drug discovery.

Natural products have been major sources of lead compounds in the discovery of new drugs for the treatment of infectious diseases, lipid disorders, neurological diseases, cardiovascular and metabolic diseases, immunological, inflammatory and related diseases, and oncologic diseases (Table 1) [2-6]. From the pharmaceutical entities approved worldwide between 1981 and 2006, 5.7% were natural products and 27.6% were natural-product derived molecules, whereas from 2005 to 2010, seven natural products and 12 natural product derivatives were approved for use in clinical practice [7-11]. Traditional medicine, however, can also produce false leads, as reviewed recently for curcumin [12].

Natural compounds have high chemical diversity. They come from different organisms. The choice of plants, microorganisms, fungi or other organisms for investigation for new compounds tend to be based on random screening, selection of specific taxonomic groups, a chemotaxonomic group of secondary metabolites like alkaloids, database surveillance of a species collection, or ethnomedical approach. Several drugs trace their origin to ethnobotanical use [13].

0.3 to 1 Kg of dried plant or 1 Kg of wet weight of marine samples should be collected from different parts of the organism. If the sample needs to be re-collected, it should be performed in the same localization and at the same time during the day, since different habitats can lead to different secondary metabolites. Of importance is also the deposit of a representative voucher specimen in a central repository, for later access [14]. Besides plants, in the last decade the interest in fungi, cyanobacteria of marine origin and the evolution of the “genome mining” techniques for culturing microorganisms in the laboratory led to the discovery of new natural products [15-18].

The extraction protocols differ from laboratory to laboratory. The initial extraction of plants usually is done with a polar solvent, like methanol or ethanol. It is then subject to a defatting process through partition with a nonpolar solvent, such as hexane or petroleum ether. The extract is partitioned between a semipolar solvent, such as chloroform or dichloromethane, and a polar aqueous solvent. Marine and aquatic organisms are extracted fresh with methanol or dichloromethane. Vegetable tannins have to be removed from the extracts. Their presence can lead to protein precipitation and enzyme inhibition, interfering with the biologic assays [14].

Separation of the active compound is done using several chromatographic techniques. High pressure liquid chromatography (HPLC) and its coupling with high-throughput screening (HST) assays simplify the purification and isolation of active compounds. The compounds are identified by nuclear magnetic resonance (NMR) and mass spectrometry (MS). These techniques, when coupled with liquid chromatography (LC), allow simultaneous purification and structure elucidation of biologically active compounds [19-21]. Purification and isolation of the active compound are measured by biologic assays. HTS assays can evaluate a large number of extracts or compounds in a cell-based or non-cell-based context.

In addition to HPLC, a new method combining high-speed counter-current chromatography (HSCCC) and size exclusion chromatography with a Sephadex LH-20 has been effectively applied to extract neuroprotective compounds from the marine brown algae, Ecklonia maxima [22]. The structure of the isolated compounds was verified by NMR and MS.

The major challenge in natural compounds development is to obtain significant amounts for further development. Re-collection of the species of origin, or plant tissue culture, or cultivation and fermentation in large scale for microbes of terrestrial origin can produce the compounds in large scale [16, 23]. The natural compounds can also be synthesized. For example, Kawano S et al achieved the total synthesis of one of halichondrins, a group of structurally complex natural products isolated from various marine sponges, and addressed the issue of limited material supply [24].

Molecular modeling, or more generally, computational chemistry, has become a well-established part of drug development. Molecular modeling searches new molecules, based on a theoretical platform or by screening a library. Crystallographic and/or NMR information on receptors and specific targets, such as that of cystic fibrosis transmembrane conductance regulator (CFTR) [25], allows through molecular modeling techniques, the design of new molecules for the target [26, 27]. Other approaches use known active molecules as a target, and design new molecules or search the libraries for similar molecules (Figure 2).

Modeling of protein-protein interactions (PPI) is becoming a valuable tool for the development of new therapeutic strategies targeting selective PPI modifiers. For instance, a new method combines cross-linking mass spectrometry with modeling approaches to generate antibody inhibitors of the interplay between R7BP, a regulator of itching sensation, and RGS7/Gβ5 [28]. Inhibitory activity was evaluated by surface plasmon resonance spectroscopy.

[enlarge]

Figure 2. Computational chemistry approaches for the development of new drugs.

The found “hit” molecules can either be synthesized in the laboratory or purchased. After evaluation and structure-activity-relationship (SAR) disclosure, new studies of molecular modeling can be performed to find a more active molecule or to optimize the found molecule (Figure 2).

Computational chemistry is based on the visualization and manipulation of three-dimensional molecular models. Rotation of bonds, structure building, molecular mechanisms and/or dynamics, conformational analysis, electronic properties, molecular surface displays and calculation of various physical properties are possible through molecular modeling. The techniques used are molecular mechanisms and molecular dynamics simulations, Monte Carlo techniques, ligand docking, and virtual screening methods.

The current state-of-the art systems allow working with more than 20 molecules and thousands of molecular surface points in real time. Each molecule can be color labeled and controlled in three dimensions, where the intramolecular distances and the noncontiguous dihedral angles can be adjusted and monitored. Shape, charge and hydrophobicity of the atoms in the molecule can also be simulated. The electrostatic potential gradient or electrical field can also be displayed graphically using short vectors. Molecular modeling requires make-bond, break-bond, fuse rings, delete-atom, add-atom, add-hydrogens, invert chiral center, etc. These operations should allow a refined structure for a selected target.

A fragment-based reinforcement learning method has been implemented to design new compounds with specific functions [29]. This method, deep fragment-based multi-parameter optimization, uses long short-term memory networks to create new molecules with required properties from lead compounds by modifying their parts. An actor-critic model, the base of this approach, implements temporal difference learning.

Several systems were developed for storing and retrieving the information generated by molecular modeling. The number of new molecules generated by molecular modeling can number thousands for a target, making the synthesis and biological evaluation of all these new molecules a challenge. The development of virtual screenings allowed to overcome the previous problem. After virtual screening evaluation only the molecules with the required biological activity would be synthesized or purchased for further biological evaluation.

Click chemistry is a new method to synthesize drug-like compounds that can potentiate the drug discovery by utilizing practical and straightforward reactions. For instance, a biomimetic hydrogels have been synthesized for hydrogel gelation-based cell encapsulation [30]. The synthesis has been performed by linking hyaluronic acid and chondroitin sulfate with polyethylene glycol diacrylate and modifying thiol degrees of glycosaminoglycans. Encapsulation has shown the high cell viability and has been effective to study 3D cellular responses.

Also, click chemistry techniques combined flow cytometry-based screening have been applied to generate aptamers [31]. This approach shows several advantages, such as the ability to perform a variety of chemical modifications and to screen a large number of aptamers. In particular, aptamers with high affinity to epinephrine and concavaline A have been generated.

An abundance of compounds are synthesized and biologically evaluated, almost daily. In the past years around 2000-3000 compounds were published in the main medicinal chemistry journals.

PubChem, CheMBL and BindingBD are public databases of compounds and their bioactivity. Other databases such as ChemBank [32] and IUPHARDB [33] are also available (Table 2). Drug Repurposing Hub [34] and ZINC15 [35] are used for virtual screening, for example, for new antibiotics [36].

The National Institute of Health (NIH) founded in 2004 PubChem database, a public library containing more than 33 million compounds. The main purpose of this database is to collect and disseminate information on the biological activities of small molecules. It started by collecting the bioassays performed in NIH, today it accepts data from other sources such as depositions. PubChem does not include information extracted from literature, however the incorporation of data from CheMBL and BindingBD allows the access to several sets of curated literature data [37, 38]. As of September 2019, the database contains 96,324,655 compounds, 235,712,478 substances, 1,067,644 bioassays, 268,177,463 bioactivities, 17,847 protein targets and 58,029 gene targets.

CheMBL was initiated with a set of commercial products, it became public in January of 2010. This library captures data from the literature in medicinal chemistry. In the scope of this database are protein-ligand affinities and cell-based data. More recently ChEMBL incorporated therapeutic proteins and other drug types besides the data on small molecules [37, 39]. ChEMBL24_1 release contains 1,879,206 compounds, 1,125,387 assays, 12,481 targets, and 72271 documents, as of September, 2019.

BindingBD originated in an academic environment, in the late 1990s. This database initially focused on small molecules with reported biological activity. BindingBD focuses on the assay conditions and factors reported to influence the outcome of the assay, such as pH, temperature and substrates. Virtual screening can be performed directly using the BindingBD website tools [37, 39]. As of September 2019, BindingDB contains 1,714,438 binding data, for 7,336 protein targets and 761,317small molecules.

With the increasing number of compounds being published every year, the importance of chemical libraries with free access in drug discovery is critical.

Combinatorial chemistry is defined as the laboratory synthesis or computational aided design of a large number of molecules, starting from one scaffold. The scaffold should have diverse points for modification, through combination with know molecules or molecules derived from a molecular modeling study. Combinatorial chemistry has been used to optimize a lead compound [40].

One of the best approaches in combinatorial chemistry is to use a central scaffold with several substituents, which can be independently modified. This approach increases the possibility of finding a “hit” molecule, since the synthesized molecules have a higher molecular diversity [40].

Parallel synthesis, a method of combinatorial chemistry, allows the formation of a large set of compounds. The formed mixture can be tested for biological activity. If the mixture does not have activity it can be archived and later tested for other biological activities. When the mixture turns out to be active, the challenge becomes the isolation of the active compound. One of the disadvantages of synthesized molecules in combinatory chemistry is that it has poor diversity, compared to natural compounds [8].

The majority of the combinatorial synthesis is performed through solid phase techniques. The starting material is bound, directly or through a linker, to a bead. The reagents are added and the product is formed. This procedure can be repeated several times using the previously formed product as a starting material (Figure 3). The product can be removed from the bead or directly tested with the bead attached, for biologic activity. The bead reduces side reactions and formation of by-products. The linkers should be resistant to reaction conditions and easily removed after synthesis.

[enlarge]

Figure 3. Simplified scheme of combinatorial chemistry.

The simple techniques of combinatorial chemistry allow the synthesis of one product per vessel. More complex methods, such as mixed combinatorial synthesis, allow the synthesis of complex mixtures. Manual techniques for the synthesis of one compound per vessel, such as the Houghton’s tea bag procedure, are still used. Automated parallel synthesis is currently the most used method.

The synthesis of large quantities of diverse compounds often makes use of the mix and split technique. In this method, the first mixture of compounds synthesized is divided in n parts, the n parts are again subject to a new modification, synthesizing z new mixtures (z = n × number of different modifications). This procedure can be repeated until the number of desired modifications is reached. The mixtures can be tested before a new modification is performed. This will exclude mixtures where no biological activity is observed, avoiding further modification.

When a mixture is biologically active, micromanipulation, recursive deconvolution and sequential release techniques can be used to separate the active(s) compound(s). Identification and structural elucidation are performed after isolation of the active(s) compound(s).

The evaluation of the libraries generated by combinatorial chemistry is usually made by HTS [41]. Screening methods using fluorescence and chemoluminescence are being developed; which allow the simultaneous identification of the active compounds.

Combinatorial chemistry also brings clear advantages to the nanomaterials research. In particular, it helps to perform chirality-controlled synthesis. A recently developed method allows assembling carbon nanotube precursors with any possible chirality in one single step [42]. The precursor generation is followed by the fixation on the metal surface, cyclodehydrogenation and elongation of carbon nanotubes with established chirality.

Pharmaceutical biotechnology is a recent area of medicinal chemistry, producing new therapeutic and diagnostic products. The common products are peptides and proteins, hormones of different origins, and enzymes, including vaccines and monoclonal antibodies (Table 3). The discovery of new drugs in pharmaceutical biotechnology is made through genomics, transcriptomics, proteomics, pharmacogenomics and metabolics.

The optimization of a lead compound can be made after a “hit” compound is found through biological evaluation. The optimization aims to improve the absorption, distribution, metabolism and excretion of the drug (ADME properties), reduction of the toxicity, and improvement of the efficacy.

The optimization can be made through chemical synthesis, computational chemistry or/and combinatorial chemistry. These should take into account the SAR studies and the preliminary mechanism of action. Lead optimization can direct the research to the synthesis of a new pharmacophore or a more active molecule (Figure 1). The techniques used for lead optimization overlap with drug discovery.

A lead compound can be modified through its functional groups to achieve better absorption, to avoid enzymatic degradation and to improve the excretion profile. The improvement of absorption can be done by the synthesis of a pro-drug. To avoid enzymatic degradation the target group can be modified to block the action of a key metabolic enzyme. The excretion profile can be improved through chemical modification of the lead compound in order to reduce the binding to albumin, for example. Organic synthesis can also improve the pharmacologic properties of a lead molecule, by increasing the bioactivity and reducing side effects.

The lead compound can be modified by combination with a series of molecules from chemical libraries or from molecular modeling. This can be achieved through the previously described combinatorial chemistry techniques. This procedure is especially useful in leads with different functional groups, which can be independently modified through mix and split techniques. Modifications performed by combinatorial chemistry can improve the ADME properties, and increase the specificity and efficacy of the lead molecule.

Computational modulation of a lead molecule can increase the specificity when the structural 3D image of the target is available, fitting the molecule for the target. The use of computational chemistry to produce almost perfect fits for a target also reduces the probability of side effects and toxicity. Computational chemistry is also useful to create and screening virtually lead compounds with better absorption properties and less metabolic degradation.

Natural compounds can be subject to the lead optimization, with chemical modifications, formulation optimization, and pharmacokinetics improvement. It is relevant to refer that some important SAR conclusions can be achieved by the observation of the biological activity of the compounds isolated in parallel with the lead compound [8, 43].

The compounds selected from molecular modeling and screening of chemical and combinatorial libraries can be subject to optimization by the described techniques. This aims to improve the pharmacokinetic and pharmacologic properties of the compounds. The SAR between the different compounds can also be drawn in this phase.

Studies around a potential drug candidate can always return to the development of the lead molecule at any giving point of the pre and clinical studies. The interaction of diverse fields such as chemistry, biology, biochemistry, pharmacology, and medicine contribute to successful drug design.

Dr. Konstantin Yakimchuk updated the article and added sections on click chemistry in September 2019.