Actin polymerization counteracts prewetting of N-WASP on supported lipid bilayers Authors: Tina Wiegand, Jinghui Liu, Lutz Vogeley, Isabel LuValle-Burke, Jan Geisler, Anatol W. Fritsch, Anthony A. Hyman, Stephan W. Grill Published: 2024 ABSTRACT SignificanceLiquid–liquid phase separation has emerged as a key mechanism for intracellular compartmentalization. How proteins condense in cells under nonequilibrium conditions remains a fundamental question. While chemical reactions can drive the formation and dissolution of biomolecular condensates, the onset of the phase separation process is less well characterized. Here, we pursue a reconstitution approach to show that Neural Wiskott-Aldrich syndrome protein (N-WASP), a central protein component of cortical condensates, is capable of undergoing a prewetting transition to form surface condensates on supported lipid bilayers. We also find that actin polymerization counteracts N-WASP prewetting. Our work provides perspectives for understanding how phase separation can shape the intracellular environment. Surface transitions of biomolecules near membranes can serve as an important mechanism to localize biochemical reactions in the cell. The affinity of proteins to membranes can induce their phase separation on the membrane (1–4), far below the saturation concentration for bulk liquid–liquid phase separation (LLPS) (5, 6). Condensation at surfaces below the saturation concentration can arise through prewetting, a process related to wetting phenomena (7). The prewetting transition describes the abrupt switch of a single-molecule-thin layer of molecules adsorbed on a surface toward a thick layer (8). In the context of biological systems, prewetting transitions have for example been reported for sequence-dependent protein condensation processes on DNA (9). For biological membranes, prewetting transitions play a role in protein condensation on the lipid surface (10–14). However, how prewetting condensates form on membranes remains poorly understood. Furthermore, understanding wetting phenomena in biological systems that typically are far from equilibrium remains challenging. While the principles of LLPS are still applicable within localized regions within cells that remain equilibrated (15), chemical reactions can drive condensates out of equilibrium and significantly influence phase separation kinetics (16–18). Condensates can be dynamic in nature, and recruit additional components (19) that drive specific biochemical reactions, such as actin polymerization (2, 20–22), which can even lead to their dissolution (23). Examples for dynamic condensates forming in vivo are the actin fusion focus in budding yeast (24), and cortical condensates that transiently form prior to their self-organized disassembly at the oocyte to embryo transition (25). These two examples indicate that condensates that form in the cytoskeletal context can be highly dynamic. Indeed, the activation of the actomyosin cortex in the Caenorhabditis elegans embryo is accompanied by the transient formation of hundreds of cortical condensates that are rich in the actin nucleation pathway member Neural Wiskott-Aldrich Syndrome Protein (N-WASP), the branched nucleator Arp2/3, and actin. Cortical condensates exhibit unconventional chemical dynamics and a condensate dynamic instability that orchestrates a switch from condensate growth to shrinkage and dissolution (25). The molecular mechanisms that are at the heart of this collective switch remain unclear. Furthermore, N-WASP has been shown to undergo LLPS in vitro together with its adapter proteins NCK (non-catalytic region of tyrosine kinase adaptor protein 1) and Nephrin, but it remains unclear how this behavior impacts on cortical condensate formation (26, 27). Here, we set out to understand the mechanisms that drive the condensation of cortical condensates and that orchestrate their switch to disassembly. We employ an in vitro approach to investigate the formation of liquid-like condensates on supported lipid bilayers (SLBs) in order to reveal the spatiotemporal dynamics of key proteins that constitute cortical condensates dynamics. Our study encompasses two variants of N-WASP, human and C. elegans, in order to allow for comparisons with previously published findings, both in human and C. elegans systems. Results N-WASP Condenses in Absence of Adaptors In Vitro. In the C. elegans oocyte, typical N-WASP interaction proteins like NCK-1 or Cdc42 (Cell division control protein 42 homolog) are dispensable for the formation of cortical condensates (25). We first asked whether homotypic interactions between N-WASP proteins are sufficient to drive N-WASP condensation. We purified full-length N-WASP, both from human and C. elegans, recombinantly and performed phase separation assays in vitro (for quality controls of different protein constructs and activity assay toward actin polymerization see Fig. 1A and SI Appendix, Fig. S1 A–C). We find that purified full-length C. elegans N-WASP undergoes phase separation at physiological salt concentrations (150 mM KCl) in vitro without further adaptor proteins or crowding agents (Fig. 1D and SI Appendix, Fig. S1D). The critical concentration for phase separation in bulk solution is 947 ± 621 nM, as determined by the volume fraction of the condensed phase in bulk (SI Appendix, Fig. S1E). However, typical concentrations of N-WASP in C. elegans embryos are ∼100 nM (28), indicating that bulk phase separation of N-WASP should not take place in vivo. We next focused on human N-WASP (SI Appendix, Fig. S2 A and B), which similarly undergoes LLPS without additional proteins. We determined the associated saturation concentration to be 2.17 ± 0.44 μM (SI Appendix, Fig. S2C). We conclude that homotypic interactions are sufficient to drive bulk condensation of N-WASP. N-WASP undergoes phase separation without adapter proteins in vitro. (A) Alpha fold predicts large disordered regions for C. elegans N-WASP (WSP-1) (sequence-dependent color code). Sodium dodecyl sulfate gel shows recombinantly expressed and purified MBP-His6-WSP-1 (for activity controls and results of human N-WASP see SI Appendix). (B) Confocal time-lapse of His6-WSP-1 (5 μM, 10% 488-tagged, MBP-tag cleaved right before experiment; see SI Appendix) forming droplets in bulk upon lowering KCl concentration to 150 mM. Kymograph indicates growth of droplets and fusion (white arrows). (C) Fluorescence recovery after photobleaching (FRAP) images of condensate formed from 5 μM His6-WSP-1 between 5 to 60 min. Diffusion coefficients D inside the condensates were determined for n = 14 condensates. Data are the mean ± SD. (D) Phase separation assays of His6-WSP-1 reveal a critical concentration for phase separation of 500 nM in bulk (Upper row; see SI Appendix, Fig. S1E) and 50 nM on a supported lipid bilayer (SLB) with 1% Ni-NTA lipids (Lower row). (All scale bars: 10 μm.) N-WASP In Vitro Condensates Show Liquid Features. In vivo cortical condensates were found to form via a mechanism different from equilibrium LLPS, but display characteristics indicative of rapid internal mixing (25). Do in vitro N-WASP condensates also show liquid-like features? We performed time-lapse imaging and observed growth and fusion (Fig. 1B and Movie S1) of in vitro condensates of C. elegans N-WASP. We further conducted FRAP experiments showing exchange of molecules between the condensates and the surrounding dilute phase (Fig. 1C) and determined the diffusion rate within the condensates to be 0.006 ± 0.001 μm2/s (mean ± SD, Fig. 1D). Similarly, human N-WASP in vitro condensates also exchange material with the dilute phase. Compared to C. elegans N-WASP, we observed the diffusion coefficient of human N-WASP inside the condensates to be higher (0.098 ± 0.036 μm2/s), likely a consequence of differences in protein length and structure (SI Appendix, Fig. S2D). These diffusion coefficients are within reported ranges for reconstituted protein condensates (29). We conclude that N-WASP condensates show liquid features such as coalescence, fusion, and diffusion. N-WASP Forms Condensates on SLBs at Physiological Concentration. Typical in vivo concentrations of N-WASP are below the critical concentration (csat) of bulk condensation of N-WASP (SI Appendix, Fig. S1E). Do interactions with membranes allow for N-WASP condensation below csat? To answer this question, we performed phase separation assays in the presence of SLBs. We found that SLBs containing PIP2 (Phosphatidylinositol 4,5-bisphosphate) or Cdc42 (30, 31) could recruit N-WASP to form small surface puncta not displaying liquid features (SI Appendix, Fig. S17 and Movie S8). It is possible that PIP2 and Cdc42 need to act cooperatively to enable N-WASP to form surface condensates resembling in vivo conditions (30–33). To achieve surface condensation, we utilized His-tagged N-WASP, adjusting the surface interaction strength by tuning the concentration of DGS-Ni-NTA within the SLB. At physiological N-WASP concentration (100 nM), dense objects appear to form on SLBs at 0.5% DGS-Ni-NTA (SI Appendix, Fig. S3A). We used 1% DGS-Ni-NTA for subsequent experiments, as the condensates resemble in vivo appearance and show liquid features (SI Appendix, Fig. S3B and Movie S2). At 1% DGS-Ni-NTA content condensates start forming on the membrane at bulk His-N-WASP concentrations of 50 nM (Fig. 1D, Lower row and SI Appendix, Fig. S3D). We asked whether these objects are monolayered or multilayered structures. If they were monolayered, their integrated fluorescence intensity should scale linearly with covered area on the SLB. If they were multilayered without uniform thickness, the integrated N-WASP fluorescence intensity should scale with covered area with an exponent greater than one (SI Appendix). We find that at 100 nM bulk concentration, this scaling exponent is above one: 1.10±0.06 for C. elegans and 1.12±0.04 for human N-WASP (SI Appendix, Figs. S3C and S4, also see SI Appendix, section on multilayered condensates). We conclude that N-WASP can form surface condensates that are multilayered and of nonuniform thickness. These form at concentrations significantly lower than the critical concentration csat for bulk phase separation. Abrupt N-WASP Surface Condensation Following N-WASP Adsorption. Our results indicate that N-WASP at physiological concentrations can form condensates on surfaces. Next, we sought to understand the mechanism by which N-WASP forms such multilayered surface-associated condensates. Above the critical saturation concentration for LLPS, condensates formed in bulk can wet biological surfaces (Fig. 2 A and B, III). Affinity-induced adsorption leads to proteins having increased concentrations near biological surfaces (Fig. 2 A and B, I). Hence, surface condensates can form below the saturation concentration via a prewetting transition (Fig. 2 A and B, II). In our in vitro experiments, do N-WASP molecules undergo a prewetting transition on the supported lipid bilayer surface? The prewetting transition is discontinuous, and the height of the surface layer should increase abruptly as bulk concentration increases beyond the prewetting concentration (7). We anticipate to observe signatures of this critical behavior at the onset of in vitro cortical condensate formation, as seen for example for prewetting condensate formation of transcription factors on DNA (9). There, histograms of surface protein amounts underwent a split up with time post adsorption of proteins to the DNA surface, into an absorbed layer with low protein amounts and a condensed layer with higher protein amounts. N-WASP prewets on supported lipid bilayers. (A) Protein interactions with membranes depend on their bulk concentration. Three protein phases exist: a dissolved phase in the bulk, an adsorbed phase on the SLB, and a condensed phase on the SLB. At low concentrations, proteins adsorb as a thin layer (I, Left). Crossing the critical concentration for prewetting (cpw) condensed areas form (II, Middle). Above the saturation concentration (csat) for liquid–liquid phase separation (III, LLPS) droplets form spontaneously in bulk (Right). (B) Phase diagram depicting these three regions of a binary fluid in the presence of a surface. Prewetting line discriminates between areas of adsorption and surface condensation (prewetting). (C) Confocal images of human His6-N-WASP (10% AF488 labeled, 100 nM Upper row, 250 nM Middle, and 500 nM Lower) condensates on SLBs after t = 10 min. (Scale bar: 10 μm.) Kymographs show adsorption and condensation phase over 10 min (Movie S5). Histogram of the logarithm of pixel intensities (Right) shows enrichment of the condensed phase over time. Two distinct time points marked with light gray and dark gray arrows are chosen respectively for each time series. (D) Condensate number density growth rate (Upper) and condensed phase area (Lower, solid lines) and intensity (Lower, dashed lines) fractions quantified as a function of time from (C). Red filled circles show the time when maximum condensation rate is reached. (E) Left: Confocal images at the time point (red filled circles) identified in (D). Middle: Probability density kymographs of the logarithm of pixel intensities for time-lapse images from (C). Gray lines (single Gaussian) and blue lines (sum-of-two-Gaussians) represent overlays of probability density fits from either method. The time point of high- and low-intensity-branch splitting (red open circles) is determined by comparing fitting residues between the two methods. Right: Histograms and fits (gray dashed lines) for the 3 different N-WASP bulk concentrations each at 2 distinct time points (arrows). In the case of a sum-of-two-Gaussians fit, blue dashed lines show respectively the fitted lower and higher Gaussian peaks. (F) Upper: The time of abrupt switch inversely scales with initial bulk concentrations. Error bars represent the range extracted from 10% variation of the threshold used for determining tpw. For segmented images, the threshold is the growth rate of condensate number density; For splits of probability density branches, the threshold is the residue difference between single- and sum-of-two Gaussian fits (SI Appendix). Lower: The curated (SI Appendix) mean pixel intensity at the time of abrupt switch is comparable across different initial bulk concentrations. To check for signatures of a prewetting transition for our in vitro cortical condensates, we set out to examine the kinetics of N-WASP accumulation on supported lipid bilayers (Fig. 2C and SI Appendix, Fig. S7 A and B). We focused on the human N-WASP because of its higher mobility in bulk condensates and its reduced propensity of hardening (SI Appendix, Fig. S2D, compared to the C. elegans variant Fig. 1C). At time t=0, we exposed below-saturation (c0