Biomolecular condensation meets membrane biophysics: from interfacial physical principles to emerging therapeutic opportunities

1 Introduction

Biomolecular condensates have reshaped cell biology by revealing a dynamic, self-organizing layer of compartmentalization beyond classical membrane-bound organelles. Liquid-liquid phase separation (LLPS) is driven by multivalent protein-protein and protein-nucleic acid interactions, enabling the rapid assembly of concentrated, non-stoichiometric phases. These phases modulate reaction kinetics, sequester enzymes and regulatory factors, and respond sensitively to physicochemical changes. Over the past decade, reconstituted systems and cells have helped define the sequence features, interaction networks, and environmental parameters that govern condensate formation, material properties, and functional outputs (Banani et al., 2017; Hyman et al., 2014; Shin and Brangwynne, 2017; Alberti and Dormann, 2019).

Within this framework, an emerging view emphasizes condensates that assemble and function at membrane interfaces, where condensates and lipid bilayers reciprocally remodel each other (Zhao and Zhang, 2020; Ditlev, 2021; Botterbusch and Baumgart, 2021; Snead and Gladfelter, 2019; Kim et al., 2024; Mondal and Baumgart, 2023). Minimal in vitro systems and theoretical approaches have established key concepts such as wetting and prewetting, condensate-membrane adhesion, and lipid-protein phase coupling, which together help rationalize how condensates can nucleate at membranes, tune local membrane organization, and drive curvature generation and complex remodelling (Mangiarotti and Dimova, 2024; Lipowsky, 2025). These ideas provide a useful physical language for thinking about membrane-associated condensates in cells without prescribing a single universal mechanism.

In the following sections, we first summarize these physical concepts for understanding how biomolecular condensates interact with and remodel membranes, and then use two concrete examples, viral replication assemblies and synaptic condensates, to illustrate how these principles manifest in cells and to explore how condensate-membrane interfaces may constitute a future target space for therapeutic intervention.

2 Fundamentals of condensate-membrane interactions

2.1 Wetting, prewetting, and membrane remodelling

Several pieces of evidence show that phase-separated liquids in contact with membranes follow a set of general physical principles that are largely independent of molecular identity (Zhao and Zhang, 2020; Ditlev, 2021; Botterbusch and Baumgart, 2021; Snead and Gladfelter, 2019; Mondal and Baumgart, 2023). At the mesoscale, the interaction can be described in terms of wetting regimes (dewetting, partial and complete wetting) governed by the balance between condensate-membrane adhesion, membrane bending elasticity and condensate interfacial tension, which together determine the apparent contact angles, the extent of spreading and the emergence of highly curved morphologies (Zhao and Zhang, 2020; Botterbusch and Baumgart, 2021; Mangiarotti and Dimova, 2024; Lipowsky, 2025; Ma et al., 2025). Changes in parameters such as polymer concentration, buffer salinity, or membrane charge, can promote wetting transitions; from dewetting to partial wetting to complete wetting. These transitions can lead to mutual remodelling processes, including spreading, tubulation, partial engulfment, budding, and interfacial ruffling (Figure 1A) (Lipowsky, 2025; Mangiarotti et al., 2023a). A complementary theoretical framework based on fluid elasticity and capillarity, together with combined experimental and simulation approaches, extends this picture, predicting and observing complex remodelling routes such as nanotubulation, vesicle-within-vesicle structures, closure of membrane necks around condensate droplets, and distinct endocytic pathways (Mondal et al., 2025; Zinga et al., 2025; Satarifard and Lipowsky, 2023; Mondal and Cui, 2024; Anila et al., 2023). Thermodynamic theory and experiments further indicate that surface binding and prewetting can shift the onset of condensate formation far below the bulk saturation concentration, with membranes acting as low-dimensional nucleation platforms for phase separation (Zhao and Zhang, 2020; Ditlev, 2021; Snead and Gladfelter, 2019; Bagheri et al., 2026). When phase-separating proteins bind to membranes, two coupled “phases” coexist at the surface: a two-dimensional bound layer and an adjacent three-dimensional enriched layer, or laterally phase-separated domains within the bilayer coupled to an overlying condensate (Ditlev, 2021). This coupling lowers the apparent wetting angle and promotes the formation of thick, protein-rich layers on membranes, even when bulk concentrations remain in the one-phase regime (Zhao and Zhang, 2020). Such surface phase transitions, termed prewetting, provide a powerful means for cells to nucleate and restrict condensates at specific membranes without wholesale cytoplasmic condensation (Zhao and Zhang, 2020; Snead and Gladfelter, 2019).

2.2 Lipid-protein phase coupling and local restructuring

Dimova´s group revealed the effect of condensate-membrane interactions at the nanoscopic scale using environment-sensitive dyes and phasor analysis (Mangiarotti et al., 2023a). Across multiple condensate-membrane systems (protein and polymer), stronger wetting correlates with local increased lipid packing and decreased membrane hydration at the condensate-membrane interface, revealed by reduced water dipolar relaxation (Mangiarotti et al., 2023a; Mangiarotti et al., 2025). Thus, condensates do not merely adhere; they locally densify and dehydrate the membrane, likely altering lateral pressure profiles, hindering lipid diffusion, and influencing embedded proteins (Mangiarotti et al., 2023a; Mondal et al., 2025).

A subsequent study dissected how membrane composition feeds back onto wetting affinity. Increasing lipid chain length, saturation, or cholesterol content enhances lipid packing and systematically reduces condensate affinity for membranes, independent of the specific condensate identity, whereas more loosely packed, highly hydrated membranes are more readily wetted and undergo extensive remodelling, such as tubes and double-membrane sheets (Mangiarotti et al., 2025). Coarse-grained simulations and multiscale experiments similarly show that condensate adsorption reduces membrane fluidity, orders lipid tails, and can drive lateral lipid demixing at the interface (Mondal et al., 2025; Zinga et al., 2025; Mondal and Cui, 2024). Together, these studies establish lipid packing, rather than nominal phase state alone, as a central regulator of condensate-membrane affinity and remodelling (Mangiarotti et al., 2023a; Mangiarotti et al., 2025).

2.3 A unifying physical view

Together, these findings support a unified framework in which condensate-membrane interfaces are governed by interfacial energies and protein-lipid coupling (Figure 1A). These transitions set the spatial organisation of condensates at membranes, while local packing and hydration tune condensate affinity, membrane mechanics, and protein activity. The resulting remodelling behaviours, ranging from budding and tubulation to multilamellar structures, emerge from shared elastic and interfacial principles (Mangiarotti et al., 2023a; Mondal et al., 2025; Zinga et al., 2025; Mondal and Cui, 2024; Mangiarotti et al., 2025). Sequence-encoded and charge-regulated interactions further bias this landscape, favouring either curvature generation and lipid demixing or extensive wetting with limited deformation, thereby providing a mechanistic basis for the morphological diversity observed across biological membranes (Zinga et al., 2025; Mondal and Cui, 2024; Anila et al., 2023).

Importantly, these principles are not restricted to minimal systems; they provide a predictive framework for interpreting membrane-associated condensates in cells. In the following sections, we apply this interfacial perspective to two biologically and clinically relevant context-viral replication organelles and synaptic condensates-to illustrate how common physical rules give rise to distinct functional architectures at membrane surfaces.

3 Viral condensates and membrane remodelling

3.1 Viroplasms and replication organelles as condensate-membrane systems

RNA and DNA viruses remodel intracellular membranes to form viroplasms/viral factories and replication organelles that concentrate viral RNA, polymerases, and host factors (de Beijer et al., 2024; Lopez et al., 2021; Zhang et al., 2023; Wang Z. et al., 2025). Despite recent progress, the molecular steps and unifying mechanisms by which viral proteins hijack and remodel host membranes remain poorly understood. The interactions holding this new organization inside infected cells are normally stabilized by weak interactions of LLPS proteins and those of the proteins and the membranes (de Beijer et al., 2024; Lopez et al., 2021; Zhang et al., 2023; Wang Z. et al., 2025; Qin et al., 2025). In positive-strand RNA viruses (coronaviruses, enteroviruses, flaviviruses, arteriviruses), these structures often appear as double-membrane vesicles (DMVs) or invaginated spherules that are continuous with organelles such as the endoplasmic reticulum (ER) (de Beijer et al., 2024; Wolff et al., 2020; van der Hoeven et al., 2016; Snijder et al., 2020). Although DMVs and spherules are fully enclosed membrane structures, they originate from surface-bound viral protein assemblies that behave as condensate-like phases adhering to and remodelling membranes (Lin and Nagy, 2024; Ambroggio et al., 2021; Glon et al., 2024). Functionally, they are hybrid condensate-membrane organelles in which membrane-anchored condensates tightly interface with the bilayer, creating privileged microenvironments for genome replication and assembly and shielding viral RNA from immune surveillance (de Beijer et al., 2024; Etibor et al., 2021).

Coronaviruses and arteriviruses generate extensive DMV networks in which viral non-structural proteins (nsp3, nsp4, and nsp6 or their arterivirus homologs) oligomerize in the cytosolic leaflet to form crown- or pore-like complexes at DMV necks; these multimeric assemblies both drive membrane curvature and provide selective routes for RNA flux, illustrating how a protein “shell” at the membrane-cytosol interface functions as a condensate-like organizing center for replication (van der Hoeven et al., 2016; Zimmermann et al., 2023; Huang et al., 2024; Laurent et al., 2022). In the alphavirus chikungunya virus, the capping enzyme nsP1 forms a membrane-bound dodecameric ring at the neck of plasma-membrane spherules that acts as a base for a larger replication complex. Modelling and cryo-ET indicate that pressure from polymerizing viral RNA inside the spherule is sufficient to remodel the membrane, directly linking a crowded RNA-protein interior to budding/invagination of the enclosing membrane (de Beijer et al., 2024; Laurent et al., 2022). In this view, DMVs and spherules are end-points of strong wetting of cellular membranes by viral protein-RNA condensates that nucleate and grow at the membrane surface. Flaviviruses, including dengue and Zika virus, further illustrate the coupling of condensate behaviour with membranes, specifically at assembly sites. Their capsid proteins act as scaffolds that bind preferentially to liquid-disordered, ER-like membranes, undergo LLPS with nucleic acids, and form RNA-protein droplets that dock onto liposomes; these condensates are proposed to seed genome recruitment and nucleocapsid formation at the ER, from which budding and envelopment proceed (Ambroggio et al., 2021; Barrows et al., 2018; Surya et al., 2024; Chaudhuri et al., 2025).

3.2 Lipid flux and organelle contact sites

Beyond their structural role, membrane contact sites (MCSs) act as supply hubs that sustain membrane-associated condensates of viral proteins and RNA. These zones, where organelles are closely apposed without fusion, enable non-vesicular lipid transfer and metabolic coordination (Wu et al., 2018; Khaddaj and Kukulski, 2023; Wong et al., 2021; Balla et al., 2019) through specialized lipid transfer proteins that mediate the exchange of sterols, glycerophospholipids, and sphingolipids, while fuelling local membrane expansion (Khaddaj and Kukulski, 2023; Wong et al., 2021; Balla et al., 2019; Rakotonirina-Ricquebourg et al., 2022). In infected cells, +RNA viruses rewire lipid metabolism and exploit MCSs to channel specific lipids toward replication organelles, supporting condensate-like compartments and shaping their membrane composition and biogenesis (Laufman et al., 2019; Strating and van Kuppeveld, 2017; van der Schaar et al., 2016; Nagy et al., 2016).

Replication compartments of enteroviruses, for example, are generated through novel contact sites that recruit lipid droplets (LDs) to the replication membranes. Viral proteins act as tethers that physically link LDs and replication compartments and engage host lipolysis machinery, allowing transfer of fatty acids from LDs to support membrane synthesis and replication organelle expansion (Laufman et al., 2019; van der Schaar et al., 2016). In this framework, replication organelles function as condensate-like concentrated phases of viral proteins and RNA anchored on remodelled membranes, while inter-organelle contact sites channel lipid flux required for their formation, growth, and maintenance. Disruption of these LD-replication organelle contacts or of the associated lipolysis pathways impairs replication organelle biogenesis and strongly reduces viral replication (Laufman et al., 2019). More generally, hijacking ER-based and other MCSs to redirect lipids has emerged as a common strategy among diverse + RNA viruses to generate and sustain membranous replication organelles with optimized lipid composition and physical properties (Wong et al., 2021; Strating and van Kuppeveld, 2017; van der Schaar et al., 2016; Nagy et al., 2016).

3.3 Connecting to wetting and remodelling principles

Viral replication organelles thus recapitulate key physical behaviours of minimal condensate-membrane systems (Figure 1B) (de Beijer et al., 2024; Wolff et al., 2020; den Boon and Ahlquist, 2010; Stancheva and Sanyal, 2024). Membrane-associated viral protein assemblies (e.g., nsP1 rings, nsp3/4/6 crowns, NS4A oligomers, capsid-RNA droplets) act as surface-bound condensates nucleated by multivalent interactions and membrane binding, consistent with prewetting and surface phase-transition concepts (de Beijer et al., 2024; Zhang et al., 2023; Snijder et al., 2020; Sagan and Weber, 2023). DMVs and spherules can be viewed as end-points of condensate-driven invagination and budding, analogous to condensate-induced membrane deformations in reconstituted systems (de Beijer et al., 2024; Wolff et al., 2020; Snijder et al., 2020; den Boon and Ahlquist, 2010; Paul and Bartenschlager, 2015). Virus-induced contact sites (e.g., LD-replication organelle interfaces, ER-connected double-membrane sheets) then function as junctions where condensate-like assemblies rewire lipid flux and organelle geometry (Wong et al., 2021; Laufman et al., 2019; Strating and van Kuppeveld, 2017; van der Schaar et al., 2016; Nagy et al., 2016). In this framework, viral factories and replication organelles are pathogen-exploited condensate-membrane constructs that co-opt cellular wetting and remodelling principles to enhance replication by spatially organizing viral reactions on and within remodelled membranes (de Beijer et al., 2024; Lopez et al., 2021; Zhang et al., 2023; Snijder et al., 2020; Sagan and Weber, 2023).

4 Neuronal condensates at synaptic membranes

Neurons are enriched in specialized biomolecular condensates, with synapses displaying particularly high organizational complexity. Synapses harbour neuron-specific assemblies: presynaptic active zones (AZ), synaptic vesicle (SV) clusters, and postsynaptic densities (PSD) (Figure 1C), that behave as phase-separated protein networks arranged at or near curved, protein-decorated membranes. These assemblies are formed via multivalent interactions of scaffold proteins and SVs or plasma-membrane associated proteins. This multilayered synaptic condensate-membrane system provides a physical framework for spatially confined control of vesicle mobilization, SNARE complex assembly, and receptor organization by clustering synapsin-SV condensates, active-zone release machinery, and postsynaptic receptor nanodomains into distinct yet communicating phases (Chen et al., 2020; Sansevrino et al., 2023; Milovanovic and De Camilli, 2017; Wu et al., 2025; Wu et al., 2023). It also constitutes a privileged interface where disease-associated IDPs, such as α-synuclein and Tau, undergo LLPS that contribute to neurodegenerative disease pathogenesis (Nam and Gwon, 2023; Cruz Rodríguez et al., 2024).

4.1 Presynaptic condensates

At presynaptic boutons, SVs and associated scaffolding proteins assemble into liquid-like multilayer condensates that organize vesicle pools required for neurotransmitter release. Synapsin-1, an abundant phospho-IDP, undergoes LLPS with SVs to form a condensate phase that contributes critically to the presynaptic vesicle reservoir, with vesicles remaining partially mobile yet spatially restricted and poised for rapid mobilization upon stimulation (Chen et al., 2020; Sansevrino et al., 2023; Milovanovic and De Camilli, 2017; Wu et al., 2025). The highly curved, protein-decorated SV membrane acts as a catalyst and specificity factor for condensate formation: by presenting a dense surface for synapsin binding, vesicles promote synapsin condensation and tune condensate material properties, including viscoelasticity and inter-vesicle spacing, as shown in reconstitution systems (Milovanovic et al., 2018; Hoffmann et al., 2023; Neuhaus et al., 2024). These findings align with general wetting-controlled remodelling principles derived from model condensate-membrane systems (Mangiarotti and Dimova, 2024; Mangiarotti et al., 2023b).

In parallel, AZ scaffolds including RIM, RIM-BP, and liprin-α form distinct but functionally coupled condensate phases at the plasma membrane that tether vesicles and organize Ca²⁺ channels and exocytic machinery (Emperador-Melero et al., 2021; Wu et al., 2021; Wu et al., 2019). These AZ-associated condensates interact with SV-synapsin phases, generating a mesoscale organization in which vesicle membranes, interaction valency, and local geometry jointly tune condensate nucleation, stability, and spatial arrangement (Chen et al., 2020; Wu et al., 2025).

Thus, presynaptic condensates can be viewed as partially spreading on and bridging highly curved SV membranes and the AZ interface in a curvature- and composition-dependent manner, organizing vesicle reservoirs and positioning them for efficient coupling to exocytic release.

4.2 Postsynaptic condensates

Beneath the postsynaptic plasma membrane lies the PSD, a protein-rich, electron-dense subcompartment that interfaces with the dendritic spine cytoplasm and mediates synaptic signalling. Multivalent scaffold proteins such as PSD-95, Shank, Homer and GKAP assemble LLPS-driven condensates that spatially organize glutamate receptors and signalling complexes into nanoscale PSD domains (MacGillavry et al., 2013; Nair et al., 2013; Zeng et al., 2016; Zeng et al., 2018; Feng et al., 2019).

Reconstituted PSD-like assemblies on membranes form liquid-like phases whose properties are shaped by membrane anchoring, which constrains diffusion and tunes scaffold interactions (Zeng et al., 2018; Shen et al., 2023). Simulations further indicate that membrane geometry favours finite-sized clusters and modulates condensate behaviour, linking these assemblies to synaptic plasticity (Yamada et al., 2025; Yamada and Takada, 2023).

Together, these findings indicate that the postsynaptic condensate-membrane system is therefore dynamic, with material state, composition, and membrane coupling regulating signalling and receptor turnover, and remaining susceptible to modulation by disease-linked proteins. In turn, physical parameters known to govern condensate wetting in minimal systems (interfacial tension, adhesion energy, lipid order) are expected to tune how PSD condensates nucleate, reorganize, and potentially transition toward more arrested, pathological states at excitatory synapses.

4.3 α-Synuclein and Tau in synaptic condensates

α-Synuclein and Tau are key proteins in Parkinson’s and Alzheimer’s diseases, respectively, and both engage synaptic condensates and membrane interfaces in ways relevant to disease pathogenesis.

α-synuclein is a curvature-sensing protein with high affinity for anionic vesicles similar in size to SVs, supporting its physiological role in the synaptic vesicle cycle (Antonny, 2011). Notably, this curvature sensitivity is altered by pathological oligomerization (Gallea et al., 2018). Consistent with this behaviour, α-synuclein acts as a membrane-associated client of synapsin/SV condensates (Brodin et al., 2022), partitioning into the synapsin/SV phase while remaining highly mobile and resistant to immediate aggregation, with SVs further promoting heterotypic coacervation (Hoffmann et al., 2021). α-Synuclein and SVs increase synapsin condensate viscosity, while only SVs alter interfacial tension; together, they promote time-dependent maturation under cytosolic conditions (Wang H. et al., 2025). Notably, excess α-synuclein slows the assembly kinetics of synapsin/SV condensates, suggesting that balanced stoichiometry between the two proteins is required to maintain SV clusters architecture (Hoffmann et al., 2021).

A key membrane-anchored regulator of α-synuclein phase behaviour is the SV R-SNARE VAMP2, whose interaction with α-synuclein promotes co-condensates that support vesicle clustering and SNARE assembly while limiting aggregation-prone states (Ag et al., 2024; Wang et al., 2024). Notably, a disease-associated α-synuclein variant with impaired lipid binding show reduced condensate formation (Ag et al., 2024). Additional regulators, including N-WASP and presynaptic actin networks, further modulate the material properties of condensates and can bias them toward more solid, pathogenic states (Jackson et al., 2024; Chhabra et al., 2025).

On the other hand, Tau, traditionally considered an axonal microtubule-binding protein, is now recognized as a synaptic protein present in both pre- and postsynaptic compartments, where a relevant pool contributes to synaptic function and plasticity (Robbins et al., 2021). Presynaptically, pathological Tau binds SVs and interacts with transmembrane partners such as synaptogyrin-3, forming a membrane-associated interface. Tau-SV association reduces vesicle mobility and neurotransmission, while disruption of Tau-synaptogyrin-3 interactions restores synaptic function, highlighting this interface as a key pathogenic node (Zhou et al., 2017; McInnes et al., 2018). In parallel, Tau undergoes LLPS at presynaptic terminals to form activity-dependent nanocondensates that tune the nanoscale organization of recycling vesicle pools (Longfield et al., 2023). Postsynaptically, Tau integrates into PSD-like condensates through multivalent interactions with PSD-95, reducing condensate dynamics and promoting a more arrested material state, effectively “hardening” PSD assemblies. This effect is enhanced by disease-associated phosphorylation, which strengthens Tau-PSD interactions and further decreases condensate fluidity (Shen et al., 2023).

Collectively, these observations support a framework in which α-synuclein and Tau act as regulated clients and modulators of synaptic condensates at membrane interfaces, where protein-membrane interactions and scaffold networks, tune condensate behaviour and susceptibility to pathological remodelling.

Of note, a number of biophysical studies show that Tau (Wegmann et al., 2018; Hochmair et al., 2022; Rai et al., 2021; Boyko et al., 2019; Foressi et al., 2025) and α-synuclein (Ray et al., 2020; Dada et al., 2023; Cruz et al., 2023; Cruz et al., 2025) form condensates that can evolve over time from dynamic, liquid-like states into more solid- or gel-like assemblies closely coupled to amyloid fibril formation (Wegmann et al., 2018; Ray et al., 2020; Dada et al., 2023; Siegert et al., 2021; Foressi et al., 2023). Importantly, heterotypic Tau/synuclein condensates are also highly susceptible to pathological maturation (Siegert et al., 2021; Foressi et al., 2023; Gracia et al., 2020; Lu et al., 2020). In this context, these transitions reflect not only intrinsic aggregation propensity but also a disruption in the regulatory balance at condensate-membrane interfaces. Moreover, the co-existence of Tau and α-synuclein within the same synaptic condensates offers a plausible mechanistic basis for the clinical and neuropathological overlap between tau- and synucleinopathies (Cruz Rodríguez et al., 2024).

5 Therapeutic opportunities at condensate-membrane interfaces

Classical condensate-modifying therapeutics (c-mods) have largely been conceptualized as dissolvers, inducers, localizers, or morphers that act on bulk properties, shifting phase boundaries, rewiring interaction networks, or tuning material states within the condensed phase. Rooted in multivalent protein-protein and protein-RNA interactions, and their modulation by small molecules or post-translational modifications, these approaches treat condensates as volumetric targets within a homogeneous cytoplasmic or nucleoplasmic environment (Mitrea et al., 2022; Patel et al., 2022; Usman et al., 2025; Martin et al., 2024).

By contrast, membrane-associated condensates are governed not only by bulk LLPS thermodynamics but also by interfacial free energies and lipid-protein phase coupling that dictate wetting transitions, contact angles, and membrane remodelling (Mangiarotti and Dimova, 2024; Mangiarotti et al., 2023a; Satarifard and Lipowsky, 2023; Mangiarotti et al., 2025; Mangiarotti et al., 2023b). In vitro and theoretical studies show that condensates undergo composition- and salinity-dependent prewetting and wetting transitions on membranes, spanning regimes from dewetting to complete wetting and driving membrane tubulation, ruffling, and budding. These morphologies are further modulated by lipid packing, where increased acyl-chain length, saturation, or cholesterol tightens packing and suppresses wetting. Conversely, stronger condensate adhesion correlates with local membrane dehydration and increased packing at the interface, highlighting a bidirectional coupling between condensate affinity and membrane physical state (Mangiarotti et al., 2023a; Satarifard and Lipowsky, 2023; Mangiarotti et al., 2025; Mangiarotti et al., 2023b). This additional layer of control opens a distinct therapeutic design space in which interfacial tension, lipid packing, and lateral lipid order become primary levers, defining an interface-centric therapeutic framework comprising interfacial dissolvers that reduce adhesion or promote dewetting, wetting inducers that shift systems across prewetting thresholds, and lipid-phase morphers that tune membrane organization, without necessarily disrupting bulk condensation.

Viral condensates present a promising new paradigm in drug discovery (Qin et al., 2025; Martin et al., 2024; Yang et al., 2025). Classical c-mods that target viral condensates have been described as dissolvers (Wang et al., 2021; Quek et al., 2023) or morphers (Etibor et al., 2023; Risso-Ballester et al., 2021). However, viral replication organelles provide a first class of translational examples in which condensate-membrane coupling creates actionable vulnerabilities rather than merely structural complexity. As described above, many viruses build replication factories or inclusion bodies whose formation and maintenance rely on membrane association, lipid flux through contact sites, and coupling to host membranes (de Beijer et al., 2024; Wolff et al., 2020; van der Schaar et al., 2016). Thus, membrane contact sites and the lipid flux between LDs and replication membranes expose membrane coupling as a potential vulnerability for antiviral intervention. In this context, interface c-mods would aim not to dissolve viral condensates per se, but to disrupt the specific interfacial processes, such as contact-site formation or lipid-phase remodelling.

Synaptic condensates illustrate a second druggable interface in which condensate-membrane coupling can be therapeutically modulated, beyond classical c-mods acting on bulk Tau/α-synuclein condensates (Foressi et al., 2024). Here, therapeutic levers emerge from how α-synuclein and Tau engage SVs and postsynaptic membranes to tune condensate composition, material state and maturation. At presynaptic boutons, α-synuclein in SV condensates operates within a narrow stoichiometric and interfacial window that supports vesicle clustering and SNARE assembly while limiting aggregation. Disease-linked perturbations, such as protein overabundance, altered lipid binding, or changes in SV curvature and packing, shift this balance toward slower assembly, increased viscosity, and enhanced maturation (Brodin et al., 2022). This highlights interface localizers/morphers that could (i) normalize α-synuclein binding to proteins and SVs, (ii) retune SV lipid composition or curvature to restore a permissive wetting regime, or (iii) bias synapsin/SV condensates back toward a fluid, rapidly exchanging state without abolishing LLPS.

Tau provides complementary interfacial targets. At presynapses, Tau interactions with SVs and membrane-anchored proteins reduce vesicle mobility and neurotransmission, while postsynaptically it hardens PSD-like condensates and impairs receptor trafficking (Shen et al., 2023; Zhou et al., 2017; McInnes et al., 2018), supporting these interfaces as druggable targets. Here, interfacial morphers could weaken Tau-PSD-95 binding or adjust spine membrane lipid order and nanoscale geometry to keep PSD condensates in a more dynamic wetting regime, preserving receptor nanodomain plasticity while leaving bulk Tau condensates elsewhere relatively untouched. Given that heterotypic Tau/α-synuclein condensates are particularly prone to pathological maturation (Cruz Rodríguez et al., 2024), strategies that selectively modulate their membrane recruitment, rather than globally dissolving condensates, may be especially effective in mixed proteinopathies.

Interfacial parameters and protein-lipid phase coupling in ordered domains can be tuned pharmacologically or via engineered lipids to prevent disease-induced changes in condensate localization, morphology, and function. Rather than merely extending c-mod categories, membrane-interacting condensates call for an explicitly interface-aware subclass of c-mods that operate at the level of surface thermodynamics and lipid–protein phase behaviour, offering targetable axes that are largely inaccessible when focusing exclusively on bulk condensate phases.

6 Future directions

Future directions in condensate-membrane biology hinge on closing the gap between simplified reconstitutions and crowded, organelle-rich cells. A key challenge is to quantitatively map wetting, prewetting, and lipid-packing changes at endogenous membranes in vivo, and to link these to specific functional outputs such as viral RNA flux, synaptic vesicle recycling, or receptor signalling. Mechanistically, we still lack a unified description of how curvature, lipid phase behaviour, and active forces (polymerization, motor activity, ion gradients) reshape condensate phase diagrams at surfaces. Finally, translating interfacial thermodynamics into drug design will require high-throughput, quantitative assays for interface selectivity and material state, together with lipid- and curvature-specific perturbations. Systematically developing and validating this interfacial c-mod framework, targeting condensate-membrane coupling rather than bulk phase behaviour, represents a central opportunity for the coming decade.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. While preparing this work, the authors used the following AI-assisted tools: Grammarly (Grammarly Inc.) for grammar and style checking; ChatGPT (GPT-5.3, OpenAI) and Claude (Claude Sonnet 4.6, Anthropic) to improve language and readability; and Gemini (Gemini 3.1 Pro, Google) to generate illustrations used in panels B and C of Figure 1. After using these tools, the authors reviewed and edited all content as needed and take full responsibility for the accuracy and integrity of the published work.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Diana Mitrea, DeMix Solutions, LLC, United States

Reviewed by: Michael Hughes, St. Jude Children’s Research Hospital, United States