To identify new hits, a screening strategy (or method) must be adopted. A set of specific assays must be carried out to identify and optimize potential drugs that then become drug candidates for clinical trials. Two major kinds of approaches exist (Figure 1), namely, those that require the identification of a target and validation of the relationship between that target and a particular disease–called target-based approaches–and those that work in a target agnostic fashion known as phenotypic approaches. The latter consist in observing the effect(s) of a new potential therapy at the level of cells or whole organisms. Phenotypic approaches require an experimental model as close as possible of the pathology and symptoms observed in human.
2.1 Target-based screening
Target-based screening relies primarily on the identification of a disease-relevant target; typically, for example, proteins and nucleic acids. This type of screening can be performed in vitro using biochemical and biophysical methods, or in cellulo using cellular models to assess the activity of the compound towards the target in a cellular context. The assays developed to perform the screening are designed either to measure the interaction between a potential drug and the chosen target, or the ability of a drug to modulate a cellular function through its interaction with the target. The aim of these methods is to develop an assay that produces a detectable signal in order to visualize, primarily through the emission of light (in the visible or fluorescence spectra), the activity of a given compound towards the target. The development of an assay is not trivial and can take time as it must be sensitive, reliable and reproducible enough to provide comparative results when thousands of compounds are screened.
2.1.1 High Throughput Screening (HTS)
High Throughput Screening consists in the screening of large libraries of compounds (from thousands to sometime millions) in order to identify hits (Blay et al., 2020). This approach is based on the automation of biological and biophysical assays that can be miniaturized and must be a highly sensitive method to identify active compounds. At its inception, HTS screening campaigns were carried out using 96-well plates, now screens are conducted in 384-well or 1,536-well plates of the same dimension as their predecessor. This means that the number of reactions that can be performed in parallel has significantly increased over the last 2 decades, and that the time needed to screen large libraries has thus been considerably reduced. However, it still takes several weeks to months to complete a screen, and the typical hit rate is around 1%. The success of an HTS screening largely depends on the design of the assay and the achieved statistical performance. The robustness of the assay can be assessed using a statistical parameter like the Z’ score (Zhang, 1999). This parameter has been widely used to determine the suitability of an HTS assay but other parameters like the distribution of standard deviation have been described (Hanley, 2019)
HTS was the gold standard in the 1990s and gave good results. Today it is used alongside other approaches, like structure-based drug design or other computational techniques (Macarrón and Hertzberg, 2011). In order to be robust and not too expensive, an HTS assay should not comprise too many steps. To that effect, several homogeneous-phase assays have been developed, in which all reagents are mixed and no washing step is required. Among the homogeneous-phase techniques, HTRF (Homogeneous Time Resolved Fluorescence) is widely used (Gotoh et al., 2010; Shin et al., 2023). This technique is based on the transfer of energy between two fluorophores, a donor and an acceptor. This transfer occurs when both fluorophores are in close proximity, resulting in a measurable fluorescent signal. This kind of methodology is used for a number of applications, such as the detection of protein-protein interactions (each fluorophore being linked to one of the proteins), enzymatic activities or receptor binding. Another technology named ALPHAscreen (Amplified Luminescent Proximity Homogenous Assay) is also based on a signal obtained when two entities are in close proximity or linked. In this assay the donor and the acceptor are microbeads that are brought together by the molecular interaction of the binding partners that are linked to these beads. Fluorescence polarization which measures the rate of rotation of a fluorescent-labeled ligand is also a powerful method to identify hits and to obtain information during the optimization process (Lea and Simeonov, 2011; Hua et al., 2023).
2.1.2 Cell-based assay
Cell-based assays also played a crucial role in the identification and validation of bioactive compounds serving as versatile tools to assess cellular responses to various stimuli and compounds. These assays use living cells to investigate drug efficacy on cell viability, proliferation or specific cellular functions. These assays differ from phenotypic screening in terms of complexity of the readouts. While phenotypic assays generally involve the simultaneous analysis of multiple cellular parameters (shape, size, surface, biomarkers of specific pathways), cell-based assays focus on a single parameter. One notable example is the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide) assay, where the reduction of a yellow tetrazolium salt to purple formazan product by metabolically active cells is measured, providing an indirect assessment of cell viability.
Specific target-based screens like protein-protein interactions can also be conducted using cell-based assays. For example, bimolecular complementation assays, also known as PCA (Protein-fragments Complementation assays) have been developed in the last decade (Kodama and Hu, 2012; Sharma and Anand, 2016; Bellón-Echeverría et al., 2018). In these assays, a fluorescent reporter protein is divided into two non-functional fragments. Each fragment is then fused to two proteins of interest. When these proteins are in close proximity, the split fragments of the reporter protein reassemble, restoring its functionality and resulting in a measurable signal. One of the limitations of the system is that complex reassembly is irreversible; hence, more dynamic systems have recently been developed (Tebo and Gautier, 2019).
Protein-protein interactions within cells can also be monitored, for instance using energy transfer as in FRET (Fluorescence Resonance Energy Transfer) (Song et al., 2011) or BRET (Bioluminescence Resonance Energy Transfer) (Machleidt et al., 2015; Cho and Dalby, 2021). Both methodologies rely on the transfer of energy between a donor and an acceptor, one being a bioluminescent protein in the case of BRET. The major advantage of cell-based assays is that it addresses the activity of a candidate compound in a specific cellular context. If an effect is detected this means that the compound is able to cross the cellular membrane and to reach a target. Additionally, if cells tolerate the compound, this is a first indication that the compound is not toxic for the cell.
2.1.3 Structural and biophysical methods
Structural an biophysical methods are now systematically integrated into the hit identification and validation process, as well as in the subsequent steps of candidate molecule optimization. Their use for hit identification dates back to the 1990s, via Nuclear Magnetic Resonance (NMR) and X-ray crystallography coupled with computational analyses. Since then, several techniques have been developed or adapted particularly in terms of throughput, and they have become complementary to biochemical or cellular biology methods fostering a positive selection of the compounds. These technologies have provided scientists with important information for the development of compounds, such as evidence that the compound binds to the target, the kinetics of the binding, the affinity (measurement of the strength) of the binding, or thermodynamic parameters. In addition, these techniques can also help to identify the binding mode and the binding pocket of molecules. This information is essential for developing a molecule with the right mode of action. Indeed, as an example an enzyme can be inhibited by a molecule that binds in the active site or at a distal site (allosteric inhibition).
A wide range of techniques is now available and their use in drug discovery has been reviewed elsewhere (Renaud et al., 2016; Holdgate and Bergsdorf, 2021). A majority of these methods focus on analyzing isolated targets, which implies producing and purifying the target, albeit more recent methods can now be performed using cellular extracts. Some of them require labeling of the target with a fluorochrome or use a native unmodified target, the ultimate goal for all these techniques being to demonstrate that a candidate compound binds to the target.
Among the most frequently employed methods, we can mention:
(i) Calorimetry techniques (like Isothermal Titration Calorimetry (ITC) or Differential Scanning Calorimetry (DSC) provides thermodynamic data about the protein-ligand complex. For example, ITC measures the consumption or generation of heat when a compound binds to the protein (Falconer et al., 2021). (ii) Temperature-related intensity change that measures the modification of fluorescence intensity of a fluorochrome when the target and the compound are bound (Jerabek-Willemsen et al., 2014). (iii) NMR that relies on the behavior of certain atomic nuclei when placed in a strong magnetic field and exposed to a specific frequency of radiofrequency radiation (Shimada et al., 2019). (iv) Surface plasmon resonance detects changes in the refractive index near a metal surface. (v) Mass spectrometry that determines the mass-to-charge ratio of ions (Gavriilidou et al., 2022) (vi) X-ray diffraction that measures the diffraction angles and changes of intensities of X-rays can be applicable to crystals (X-ray crystallography) (Maveyraud and Mourey, 2020) but also to proteins in solution like enzymes (Byer et al., 2023) (vii) cryo-electron microscopy (cryo-EM) is a powerful technique used for imaging macromolecules at near atomic resolution. This technique is now complementary to NMR or X-ray diffraction in small molecule drug design ((Vénien-Bryan et al., 2017; Renaud et al., 2018)
These techniques all rely on high-standard equipment, and depending on the method employed the throughput can vary from a few compounds a day to a few compounds a week or month. These methodologies are part of the drug discovery process from the early phases to the selection of the preclinical candidate therapy.
2.2 Phenotypic-based screening
Historically, the discovery of medicines relied on phenotypic approaches, however with the advent of genomics in the 1980s and the sequencing of the human genome in 2001, these approaches were neglected. Nevertheless, over the last decade there has been renewed interest in phenotypic approaches, as they are valuable at identifying novel therapeutic agents (Ege et al., 2021; Vincent et al., 2022). One of the advantages of phenotypic assays is that they explore a broader spectrum of biological responses than target-based approaches, elucidating complex biological pathways and uncovering unforeseen interactions, thus offering a holistic perspective of the potential effect of a new agent. Technological advances have played a pivotal role in boosting phenotypic screening. Assay miniaturization, development of high-throughput screening platforms (gathering automated equipment to rapidly test a huge quantity of samples rapidly), automated imaging (microscopy technology) and data analysis systems have opened new avenues to perform phenotypic analyses. One such technique is fluorescence imaging, which enables scientists to visualize and quantify various biological processes at the cellular and subcellular levels. This technique uses fluorescent probes, markers or genetically-encoded fluorescent proteins to highlight specific cellular structures, proteins, or functional activities. High content screening (a combination of powerful imaging tools and biochemical/molecular biology assays) captures dynamic cellular events in real-time; for instance, the monitoring of processes like cell migration, proliferation, modification of the cellular morphology (shape, size…) and cell death. Additionally, it allows the concomitant assessment of multiparametric data including protein localization, analysis of subcellular organelles or responses to external stimuli. Numerous approved therapies for cardiovascular diseases, viral infections, neurodegenerative disorders and cancers originate from phenotypic screening (Blay et al., 2020). Despite, many advantages that have led to the identification of innovative therapies that could not have been identified without this approach, phenotypic screening has one major drawback–this global approach makes it difficult to decipher the molecular mechanisms of action of a drug and to identify its target(s), both necessary to optimize the potency of a drug and for its further development in the clinic.