Water-protein interactions as a driver of phase separation, biology, and disease

Main text

Molecular interactions between water and proteins enable many essential biological functions and processes. In this issue of Biophysical Journal, Martina Havenith and co-workers use a new method they call “terahertz calorimetry” to reveal how the thermodynamics of water-protein interactions contribute to liquid-liquid phase separation (LLPS) and liquid-solid phase separation (LSPS), including the formation of protein fibrils that may be relevant to the pathogenesis of neurodegenerative diseases, such as Alzheimer’s and Parkinson’s diseases (1).

Standard calorimetric methods measure heat exchange to quantitate enthalpic, entropic, and free energy changes but do not distinguish how different molecular populations contribute to thermodynamics. For instance, water can hydrogen bond to hydrophilic protein surfaces or terminate the hydrogen-bonding network at hydrophobic protein surfaces. Standard calorimetry cannot differentiate which of these water populations drive biological processes, such as LLPS and LSPS. Resolving this molecular-level ambiguity will not only inform our ability to tune the physical properties of biomolecular systems through environmental factors (e.g., salt, pH, temperature) or protein engineering but will also provide fundamental insights correlating molecular interactions with the thermodynamics of biological processes.

Rather than measuring heat exchange, terahertz (THz) spectroscopy records the absorbance of far-infrared light. Havenith and co-workers reported in a series of publications (2,3,4,5,6,7,8) that different water populations absorb THz radiation at distinct THz frequencies. Based on combined experimental THz spectroscopy and molecular dynamics studies of water around alcohols (3), Havenith and co-workers assign distinct THz signals to “bound water” and “cavity-wrap water” populations. They define bound water as water molecules hydrogen bonded to hydrophilic solute surfaces and cavity-wrap water as water molecules surrounding hydrophobic solute surfaces, where the three-dimensional hydrogen-bonding network of water is interrupted. Examining the temperature-dependent behavior of THz water absorbances, Havenith and co-workers propose a direct correlation between THz absorbances and thermodynamic quantities of enthalpy and entropy as measured by standard calorimetry (2). They suggest that the THz absorbances not only report on changes in the number of water molecules belonging to bound water and cavity-wrap water populations but are also directly proportional to the free energy changes of each of these water populations. On this basis, they propose that THz spectroscopy can be used as “THz calorimetry” to track the free energy changes due to bound water or cavity-wrap water during LLPS and LSPS (Fig. 1). Havenith and co-workers have applied THz calorimetry to study LLPS and LSPS for a variety of proteins, including intrinsically disordered proteins (4,5,7), globular proteins (8), and mammalian prions (6).

Figure 1.

Terahertz spectroscopy signals can be correlated with the thermodynamics of water populations at protein surfaces during liquid-liquid phase separation (LLPS) and liquid-solid phase separation (LSPS). (Top) Quantifying enthalpy (ΔH) and entropy (ΔS) contributions of “bound water,” “cavity-wrap water,” and protein-protein interactions may help scientists to understand and predict fundamental properties of biological phase separation, such as upper or lower critical solution temperature. (Bottom) Terahertz spectroscopy supports a two-step model of water release during LLPS (reversible) and LSPS (irreversible). For illustration purposes only, the ACC_1–13K₂₄ peptide was modeled using AlphaFold3 (9), the parallel β-sheet structure was adapted from PDB: 6ZCG (10), and both structures were rendered using PyMol.

The team’s latest work (1) focuses on a synthetic peptide (11), ACC_1–13K₂₄, composed of amino acids 1–13 of the insulin A chain and a tail of 24 positively charged lysines. On its own, the ACC_1–13K₂₄ peptide solution exists as a single phase. However, when ATP is added to the peptide solution, the lysine-ATP interactions drive the formation of liquid condensates via LLPS. The insulin A chain later promotes fibrillization of ACC_1–13K₂₄-ATP at condensate interfaces via LSPS.

By monitoring THz absorbances, Havenith and co-workers demonstrate that LLPS and LSPS of the ACC_1–13K₂₄-ATP system take place in two steps (Fig. 1). In the first step, cavity-wrap water is released from the hydrophobic protein surfaces, while the bound water is retained at the hydrophilic protein surfaces. Releasing cavity-wrap water drives the system from a single phase to the condensation of protein-rich liquid droplets via LLPS. In the second step, the bound water is released from the hydrophilic protein surfaces, leading to formation of solid protein fibrils at condensate interfaces via LSPS. The research team proposes that this two-step “drying” of the ACC_1–13K₂₄-ATP complex defines the LLPS and LSPS transitions. These results suggest an important conclusion—maintaining protein condensates in a liquid state depends on retaining bound water molecules at the protein surface.

This conclusion likely bears on a key unresolved question in the field of biological LLPS: why do some biomolecular condensates “age”? Aging here means that the condensates transition over time from having more liquid-like properties to more gel-like or solid-like properties. The findings may suggest that the tendency of condensates to remain liquid correlates with a greater propensity to retain bound water molecules at the protein surface. Accordingly, the tendency of condensates to harden over time could indicate a reduced propensity to retain bound water molecules at the protein surface.

Exploiting the proposed proportionality of THz absorbances and free energy changes (1,2,3,4,5,6,7,8), Havenith and co-workers also connect the thermodynamics of water-protein interactions to LLPS and LSPS. They find that the release of cavity-wrap water from hydrophobic protein surfaces is entropically favorable and drives LLPS. The subsequent release of bound water from hydrophilic protein surfaces during LSPS is also entropically favorable but counterbalanced by the enthalpy costs of breaking water-protein hydrogen bonds. The results imply that the major free energy gain to drive LSPS is likely due to large, enthalpically favorable protein-protein interactions, supporting that the LSPS transition is irreversible and relatively insensitive to temperature. At present, the proposed approach of THz calorimetry cannot yet resolve the free energy contributions of protein-protein interactions in LLPS and LSPS. To do so, researchers might combine THz with other spectroscopies, such as NMR, infrared, or Raman. This multifaceted approach could potentially prove powerful for quantitating the subtle thermodynamic balances between water-protein and protein-protein interactions, allowing researchers to predict whether a biomolecule can undergo LLPS or LSPS under various conditions.

Computational modeling of THz spectroscopic response remains a high priority for the continuous development of THz calorimetry, which relies on assigning THz absorbances to distinct hydration water populations and proteins (12). One particular challenge for studying LLPS and LSPS is that these THz absorbances are sensitive to the refractive index, which can depend on both the phase of protein samples and the frequency of THz radiation (13). Computational modeling is expected to help validate THz absorbance assignments, potentially resolving hydration water populations beyond the level of cavity-wrap or bound water (e.g., strongly bound versus weakly bound water), and distinguishing protein from water signals. Parallel advances in computational and experimental methods will guide researchers on how to mine the information about structures and dynamics of water and protein during LLPS and LSPS that is embedded in THz data.

Spectroscopists continue to develop methods for novel biophysical investigations of water-protein interactions in vitro (14), at interfaces (15), and even inside living cells (16). THz spectroscopy may also be extended to probe important biophysics beyond LLPS and LSPS, including, for example, the role of hydration water thermodynamics in chromatin remodeling, membrane-bound organelle biogenesis and function, or viral replication. In their latest report in Biophysical Journal, Havenith and co-workers lay the groundwork for future advances in THz spectroscopy, which promise greater recognition of the biological and disease relevance of water-protein interactions, interfaces, LLPS, and LSPS. The evidence is already stacking up: hydration water molecules shape protein structure, influence protein function and dysfunction, and are an essential part of protein surfaces.

Acknowledgments

E.A.P. acknowledges support from Schmidt Science Fellows, in partnership with the Rhodes Trust. T.S. was supported by the National Science Foundation MPS-Ascend Postdoctoral Research Fellowship (CHE-2402247). E.C.Y.Y. acknowledges support from the National Science Foundation (CHE- 2108690).

Declaration of interests

The authors declare no competing interests.

Editor: Samrat Mukhopadhyay.