Source-linked AI summary
How to design multi-target drugs: Target search options in cellular networks
Tamas Korcsmaros, Mate S. Szalay, Csaba Bode, Istvan A. Kovacs, Peter Csermely
TL;DR
The paper addresses the shortage of successful single-target drugs and the challenge of selecting relevant target combinations. It reviews multi-target drug mechanisms, network attack strategies, and approaches for identifying target-sets, emphasizing indirect network effects and low-affinity binding. The review concludes that partial multi-target actions can expand pharmacologically relevant targets and druggability, although current network databases remain uncertain and the usefulness of network approaches for target-set selection is not yet established.
Problem
Successful drugs and novel targets have fallen behind expectations, while selecting clinically relevant combinations from complex cellular networks remains unresolved.
Method
The paper reviews multi-target drug progress, compares network attack strategies, and discusses network-based methods for identifying target-sets.
Results
The review reports that multiple partial attacks on carefully selected targets were more efficient than single-target knockout in modeled regulatory networks, and that indirect effects and partial binding expand target choices.
Takeaways & Limitations
Multi-target drugs can broaden pharmacologically relevant targets by combining indirect network effects with low-affinity, partial interactions.
Takeaways & Limitations
Current cellular-network databases contain substantial uncertainties, and whether network approaches can suggest relevant target-sets is currently unknown.
Abstract
from arXiv · showhide
Despite improved rational drug design and a remarkable progress in genomic, proteomic and high-throughput screening methods, the number of novel, single-target drugs fell much behind expectations during the past decade. Multi-target drugs multiply the number of pharmacologically relevant target molecules by introducing a set of indirect, network-dependent effects. Parallel with this the low-affinity binding of multi-target drugs eases the constraints of druggability, and significantly increases the size of the druggable proteome. These effects tremendously expand the number of potential drug targets, and will introduce novel classes of multi-target drugs with smaller side effects and toxicity. Here we review the recent progress in this field, compare possible network attack strategies, and propose several methods to find target-sets for multi-target drugs.
1. Introduction: emergence and rationale of the multi-drug concept
Conventional drug development has produced fewer successful drugs and novel targets than expected. Multi-target strategies address this shortfall by influencing several targets in parallel, while single-target interventions can be blunted by backup systems and network robustness.
- Recent drug development identifies clinically relevant targets, finds druggable binders, validates them experimentally, and advances them toward clinical applications.
- The number of successful drugs and novel targets fell significantly behind expectations despite substantial development efforts.
- Multi-target strategies seek to overcome target shortages by influencing multiple targets in parallel, including through combination therapies and multi-target lead discovery.
- Natural venoms, plants, and traditional remedies illustrate that multi-component strategies have long been used in therapeutics and defense.
- Single-target drugs may fail to produce desired system-level effects because backup systems compensate and cellular networks remain robust despite major constituent changes.
2. Examples for multi-target strategies
Multi-target strategies range from combination therapies and multi-component biologics to integrated molecules with overlapping pharmacophores. They expand drug discovery beyond a narrow overlap between pharmacologically relevant pathways and conventionally druggable proteins.
- Multi-target strategies include drugs and therapies that affect several targets simultaneously, including NSAIDs, metformin, antidepressants, kinase inhibitors, and multi-target antibodies.
- Multi-target ligands are described with terms such as dual, bivalent, mixed, or triple, spanning conjugates, overlapping pharmacophores, and highly integrated drugs.
- Multi-target drugs enlarge the drug-discovery sweet spot by adding indirect network effects to the overlap between pharmacologically interesting pathways and conventionally druggable proteins.
3. Cellular networks: drug target maps
Cellular networks represent interacting molecules as weighted links and provide several ways to identify potential drug targets, including signalling bridges and critical nodes. Their usefulness is constrained by uncertain, incomplete, and inconsistently interpreted interaction data.
- Cellular networks model molecules as interacting elements connected by weighted, sometimes directed links that represent interaction strength and influence.
- Signalling-network cross-talk identifies bridge elements and critical nodes, such as PI-3-kinase, Akt-kinase, and insulin-receptor substrate-family members, as potential targets.
- Current cellular-network databases contain false positives and averaged interaction probabilities that often omit simultaneous expression, co-localization, and protein-ratio information.
- Evidence-based databases also face nomenclature and interpretation problems, although linking interaction data with protein structures can improve validation and prediction.
- Curated databases improve accuracy but miss most low-affinity interactions, whereas retaining all information produces fuzzy databases requiring integrated error correction.
4. Multi-target drugs are often low-affinity binders
Multi-target drugs often bind several targets with lower affinity than single-target drugs, but low affinity does not necessarily imply low efficiency. Weak interactions are widespread in cellular networks and can produce substantial regulatory effects.
- Multi-target drugs are likely to bind individual targets with lower affinity because one small molecule rarely binds many different targets equally strongly.
- Low-affinity multi-target binding may reduce the prevalence and range of side effects compared with high-affinity single-target binding, as illustrated by memantine and related antagonists.
- More than 80% of cellular protein, signalling, and transcriptional network connections are low-affinity or transient weak linkages.
- Because weak links dominate cellular networks, low-affinity multi-target drugs may achieve significant modification, and imperfect binding can support cooperative switch-like activation.
5. Identification of drug targets using the network approach: attack strategies
The network approach models drug action as targeted damage to cellular networks and uses structural, weighted, directed, mesoscopic, and metabolic analyses to identify vulnerable targets. Initial modelling suggests that distributed partial attacks can damage regulatory networks more efficiently than single-target knockout.
- Drug-induced inhibition is modelled as eliminating a target’s interactions, while partial inhibition is modelled as partially removing them.
- Scale-free networks are robust to random damage but vulnerable to targeted attacks on hubs and other structurally important elements.
- Network attacks can target hubs, globally central links, weighted or directed interactions, long-range links, and metabolic bottlenecks.
- Betweenness centrality identifies links carrying shortest paths, while inverse geodesic length and whole-network performance measures assess damage after removals.
- Multiple partial attacks on carefully selected targets were more efficient than single-target knockout in bacterial and yeast regulatory-network models.In E. coli, the damage caused by removing one node with 72 connections could also be achieved by partially inactivating 3 to 5 nodes, even when the number of affected interactions was unchanged.
5. ‘Network diseases’
The paper extends network-based drug analysis to disease contexts in which pathological states involve changing network behaviour. It links signalling and metabolic network models to symptom fluctuations, ageing-related variability, and disease-specific target analysis.
- Bacterial and yeast gene-regulatory networks provide an initial model for multi-target antibiotics and fungicides, whereas disease-specific drugs require more specific signalling and metabolic models.
- Changing functional neuron assemblies may help explain daily fluctuations in neurodegenerative symptoms and support trials focused on short-term symptom attenuation.
- Ageing is framed as a network disease because its multiple causes, stages, and variable symptoms call for network-based analysis of accumulated data.
6. Target-sets of multi-target drugs – the help of networks
The paper asks how network methods can identify target-sets for multi-target drugs while recognizing that exhaustive combination screening is impractical. It proposes adapting single-target network analyses, complemented by improved in vivo models.
- Existing experimental and modelling approaches for single-target identification could be modified to narrow the search for multi-target sets.
- High-throughput screening of all possible target combinations may be a daunting task.
- More efficient in vivo testing may require animal models with humanized metabolic and signalling networks.
- Whether network approaches can currently suggest useful target-sets remains unknown, despite the availability of promising tools for assessing network knowledge in pathological states.
7. Conclusions
Multi-target drugs extend combinatorial therapy by integrating multiple agents into one chemical entity and exploiting indirect network effects. Their often low-affinity binding can expand the druggable proteome and the pool of potential target-sets.
- Multi-target drugs integrate the effects of several agents into a single chemical entity and introduce indirect, network-dependent effects.
- Low-affinity binding can ease druggability constraints and significantly increase the size of the druggable proteome.
- These properties expand potential drug targets and may support novel multi-target drug classes with smaller side effects and toxicity.
- Cellular-network hubs and bridges with high betweenness centrality are among the possibilities for finding multi-target drug sets.
8. Expert opinion: Network-based, smart multi-target drugs of the future
The review anticipates that multi-target drugs will become more common by combining carefully selected primary targets with indirect network effects and partial, low-affinity interactions. It proposes expanding network-analysis tools to identify relevant target sets as cellular datasets become more complex.
- Multi-target drugs are predicted to become much more common within 5-10 years.
- Partial effects on multiple targets expand drug choices by relaxing druggability constraints.
- Carefully selected primary targets can combine their effects on therapeutically relevant secondary targets through network-mediated interactions.
- Low-affinity multi-target drugs may stabilize cellular networks by converting strong links into weak links rather than eliminating them.
- Future target-set discovery will require more sophisticated network-analysis tools to navigate increasingly complex datasets.
Network elements Network links Potential drug targets
The section maps cellular-network elements and links onto possible drug targets. It presents network structures ranging from metabolic and transcriptional systems to pharmacophore and attack scenarios for identifying vulnerable points.
- Network links: Network links include transient or permanent bonds, enzyme reactions, gene-regulatory interactions, and highly specific interactions that can activate or inhibit cellular processes.
- Network elements: Cellular-network elements include metabolites, proteins, protein complexes, membrane structures, transcription factors, and DNA sequences.
- Potential drug targets: Potential targets include hubs, bridge proteins, modular centres, overlaps, metabolic switch enzymes, and regulatory proteins.
- Potential drug targets: Multi-target drug structures span a continuum from conjugates through partially overlapping pharmacophores to highly integrated designs.
- Potential drug targets: Network attacks can target high-degree hubs, hub-links, high-betweenness bridges, or bridges selected by weighted centrality.