Co-reconstitution of actomyosin networks and the MinDE system inside GUVs
To achieve the co-reconstitution of actomyosin networks and Min oscillations inside vesicles, we carried out encapsulation experiments via double emulsion transfer to identify optimal experimental conditions for the dynamic and functional interplay of both systems’ components.
First, since G-actin and Min protein self-organization depend on factors like salt concentration, supply of ATP and the presence of divalent cations in solution, we tuned the inner environment of the vesicles to simultaneously facilitate actin filament polymerization and MinD dimerization, critical for MinD interaction with negatively charged amphiphiles and its cooperative binding to membranes. Given the importance of the membrane as catalytic matrix for the spatiotemporal organization of MinDE proteins, we generated vesicles containing negative charge in the bilayer to enable the self-organization of Min proteins into different oscillation modes22,23. In addition, we incorporated biotinylated lipids to link biotinylated actin filaments to the inner leaflet of the GUVs. Anchoring actin assemblies to the membrane through neutravidin-biotin bonds allowed us to exploit the diffusiophoretic capabilities of Min proteins, which require membrane-bound cargo to induce the ATP-driven transport of molecules on membranes17.
Once optimal buffer conditions and membrane composition were identified, we chose fascin as the crosslinking protein to generate high-order actin bundle structures. As previously reported15, by binding fascin-assembled bundles to the membrane via neutravidin-biotin bonds we obtained long and curved bundles that robustly bound to the membrane adapting to the vesicle curvature. Additionally, to accelerate actin polymerization kinetics and decrease MinDE wavelength and oscillation velocity, we employed Ficoll70 as macromolecular crowder9,24,25, which also facilitated vesicle production. Finally, to make our actin assemblies contractile and render membrane deformations14,15, the motor protein myosin II was added to the inner solution mix (Fig. 1a).
a Schematic illustration of the GUV content and the two macromolecular reactions at membrane level: the MinDE self-assembly mechanism behind pattern formation and the diffusiophoresis-mediated transport of neutravidin-bound actomyosin bundles by Min proteins. The active flux of MinDE proteins on the vesicle membrane interacts non-specifically via frictional forces with membrane-bound neutravidin inducing the transport and positioning of these molecules, and consequently the actomyosin bundles linked to them, towards areas of low MinD density. b 3D projections of confocal images showing the 4 phenotypes of actin architectures obtained after encapsulating 2.4 µM actin, 0.6 µM fascin (fascin/actin molar ratio = 0.25), 0.05 µM myosin II, 50 g/L Ficoll70, 3 µM MinD, 3 µM MinE and 5 mM ATP. Scale bars: 10 µm. c Bar graphs with the frequencies of the four actomyosin phenotypes observed at different vesicle diameters when encapsulation experiments were performed at 0.25 and 0.5 fascin/actin molar (M/M) ratio in the presence and absence of Min proteins and protein/crowding conditions specified in (b). Total number of GUVs analyzed per condition = 150. d 3D projections of time-lapse confocal images depicting the reorganization and stacking of actomyosin bundles towards the vesicle equator driven by the diffusiophoretic transport of Min pole-to-pole oscillations. Yellow arrows indicate the perpendicular orientation of MinDE oscillations with respect to actomyosin bundles, which get antagonistically positioned at mid-cell. Kymographs generated at the vesicle equator (blue dashed circle) are meant to visually define the position of fluorescent features at this region over time. Orange dotted lines depict the approximate distribution of actin bundles on the membrane at two time points. Vesicle content as specified in (b). Scale bars: 10 µm. Source data are provided as a Source Data file.
To investigate the effects of MinDE oscillations on the formation of actin-bundle architectures under crowding conditions, we encapsulated actin and fascin at different molar ratios together with myosin II in the presence and absence of Min proteins. To this end, we analyzed actin architecture types and quantified their frequency of appearance in terms of GUV sizes. Similar to recent cytoskeletal reconstitutions in GUVs16,26, we observed four main actin phenotypes in both the presence and absence of Min proteins: soft bundle webs, actomyosin asters, flexible rings and stiff-straight bundles (Fig. 1b).
Although phenotype yields differed, Min oscillations supported the bundling and assembly of actomyosin architectures on GUV membranes. In particular, at a 0.25 fascin/actin molar ratio (Fig. 1c), when Min proteins were part of the reaction mix, we detected an increase in flexible ring yields, as well as a lower probability of stiff bundle formation, irrespective of vesicle size. As previously reported, the size of a spherically confining environment impacts actin-bundle architecture due to the persistence length of actin filaments and the spontaneous equatorial assembly of bundles to minimize their bending energy14,26. In both samples containing or in absence of Min proteins, the probability of flexible ring formation was significantly higher in small diameter vesicles (diameter < 15 µm), whereas for medium and big vesicles the predominant phenotype was aster, reaching almost 50% formation probability in vesicles between 20–25 µm in diameter and 80% for vesicles bigger than 25 µm. Similarly, when a 0.5 fascin/actin molar ratio was employed and a higher fascin/actin concentration was encapsulated, we observed the four aforementioned phenotypes in the presence and absence of Min proteins (Fig. 1c). Under these conditions, however, only the frequency of stiff bundle formation decreased upon addition of Min proteins and the probability of flexible ring formation drastically decreased for GUVs bigger than 15 µm diameter in samples containing Min proteins or in their absence.
Furthermore, to study the evolution of this co-reconstituted system over time, we performed experiments in the presence of MinDE at 0.25 fascin/actin ratio and quantified the frequency of actin phenotypes observed on the sample at three timepoints: right after encapsulation, 7 h and 24 h after vesicle production (Supplementary Fig. 1a). Interestingly, we found that aster formation was immediate after vesicle generation, as the frequency of these star-like condensates remained markedly similar at 7 and 24 h: 34% and 38%, respectively. Conversely, vesicles initially presenting no distinguishable phenotype (but filled with G-actin on their lumen) showed a frequency as high as asters right after encapsulation (40%), which progressively decreased over time reaching 6% at 24 h. Concomitantly, we observed that the number of vesicles containing rings and soft webs increased after 7 h, the latter phenotype showing progressive increment in number as we prolonged sample incubation time to 24 h.
Taken together, we show the successful co-reconstitution of the actomyosin system together with Min proteins and demonstrate that MinDE oscillations are compatible with the assembly of membrane-bound actomyosin architectures inside GUVs, and vice versa. Notably, addition of Min proteins promotes the formation of flexible actomyosin rings in all vesicle sizes encapsulated with a 0.25 fascin/actin ratio.
Diffusiophoresis-mediated positioning of actomyosin bundles at mid-cell by the MinDE system
Having established the conditions to reconstitute dynamic Min oscillations together with actomyosin-bundle assemblies inside GUVs, we then investigated whether Min proteins could effectively reorganize these assemblies and position them at mid-cell via their diffusiophoretic mechanism of molecular transport. Since flexible rings and soft bundle webs are the two types of actin architectures that could efficiently transmit contractile forces to the membrane, we studied the spatiotemporal organization of these two phenotypes by Min oscillations with time-lapse microscopy.
In agreement with past studies27, we observed three main Min oscillation modes resultant from the reaction-diffusion fluxes of Min proteins on the inner leaflet of vesicles (Supplementary Fig. 1b): pulsing (oscillation characterized by the consecutive binding and unbinding of MinD to the entire vesicle membrane), pole-to-pole (sequential binding of MinD to the hemispheres of the vesicle), and circling waves (MinDE waves revolving around the inner leaflet of the membrane).
As MinDE pole-to-pole oscillations are the desired phenotype to actively transport molecules to the mid-cell region via diffusiophoresis9,21, we first scrutinized actomyosin-containing vesicles exhibiting this dynamic pattern. Strikingly, in vesicles containing actomyosin bundles isotropically distributed all over the membrane, MinDE pole-to-pole oscillations yielded an anticorrelated and directional movement of the bundles perpendicular to the oscillation axis, reducing bundle interdistance and accumulating them at mid-cell (Fig. 1d, Supplementary Movie 1). Subjected to the highly dynamic MinDE pattern, the actomyosin bundles still showed positional fluctuations at the GUV equator over time, but maintained a perpendicular orientation to the oscillation axis.
Subsequently, to test the robustness of the MinDE diffusiophoretic transport in our actin-based encapsulation system, we varied the experimental conditions from our standard inner solution mix. We found that, in the absence of myosin II, under varying Ficoll70 concentrations (10-50 g/L), and employing different molar ratios for fascin/actin (0.25 or 0.5) as well as MinD/MinE ratios (1 or 2), Min proteins were capable of actively arranging membrane-bound actin bundles via diffusiophoresis when patterns different from pulsing developed on the vesicle membrane. Importantly, and as expected from our previous experiments, MinDE pole-to-pole patterns rotated and positioned fascin-assembled actin rings in the absence of myosin II, maintaining ring orientation perpendicular to MinDE oscillations (Supplementary Fig. 2b). In addition, when MinDE circling patterns emerged on vesicles containing fascin-bundled actin rings, we observed that the frictional forces induced by the directional MinD protein flux promoted the circular displacement of one end of the ring towards the opposite end (Supplementary Fig. 2c), resulting in the complete folding of the ring in less than 15 minutes (Supplementary Fig. 2d). Moreover, we detected that chaotic MinDE patterns—a dynamic mode in which Min proteins bind and unbind membrane areas in stochastic direction—could also alter the distribution of actomyosin bundles and, in some instances, buckle and collapse the network (Supplementary Figs. 1c, d).
Thus, our data demonstrate that the MinDE system can be used to regulate the spatiotemporal localization of membrane-bound actomyosin bundles and most importantly, position contractile actomyosin architectures at mid-cell via its characteristic pole-to-pole oscillation mode.
Equatorial constriction of vesicles induced by positioned actomyosin architectures
In our system, myosin II not only acts as a crosslinking agent but also provides the contractile force required to induce membrane deformations. Having shown that the MinDE system localized actomyosin assemblies at the vesicle equator, we next investigated whether the positioned contractile assemblies could generate furrow-like membrane deformations.
First, to explore the contractile effect of myosin II on positioned actin structures, we carried out encapsulation experiments at 0.25 and 0.5 fascin/actin molar ratio with 50 g/L Ficoll70 and examined the vesicles presenting pole-to-pole oscillations with time-lapse confocal microscopy. Interestingly, when we employed a 0.5 fascin/actin molar ratio, mid-cell deformation induced by actomyosin ring constriction could be observed (Fig. 2a). This furrow-like invagination of the membrane, sustained over time (Supplementary Movie 2), generated two lobes where MinD proteins continued oscillating in a pole-to-pole pattern maintaining the localization of the ring at mid-cell.
a Schematic illustration behind the mechanism of membrane deformation. Contractile actomyosin bundles positioned by MinDE proteins at mid-cell induce furrow-like membrane invaginations. 3D projections and 2D confocal images show an actomyosin ring constricting the vesicle at its equator. Orange lines indicate the major (a) and minor (b) axes measured to calculate the aspect ratio of the deformed vesicle (for spherical vesicles: aspect ratio = 1). Inner solution mix: 4 µM actin, 2 µM fascin (fascin/actin molar ratio = 0.5), 0.05 µM myosin II, 50 g/L Ficoll70, 3 µM MinD, 3 µM MinE and 5 mM ATP. Scale bar: 10 µm. b Schematic illustration, 3D projections and 2D confocal images of a vesicle containing a soft web of actomyosin bundles at the vesicle center being positioned by pole-to-pole Min oscillations. The contractile actomyosin band formed causes the deformation of the vesicle (aspect ratio < 1). Inner solution mix: 2.4 µM actin, 0.6 µM fascin (fascin/actin molar ratio = 0.25), 0.05 µM myosin II, 50 g/L Ficoll70, 3 µM MinD, 3 µM MinE and 5 mM ATP. Scale bar: 10 µm. c Schematic illustration, 3D projection and 2D confocal image of a vesicle with a non-positioned contractile actomyosin assembly due to the loss in pole-to-pole MinDE oscillations. Constriction of the actomyosin bundles results in the deformation of the vesicle membrane into an asymmetric dumbbell shape. Scatter plot depicts the aspect ratio of the vesicle at different time points. Inner reaction mix: 4 µM actin, 2 µM fascin (fascin/actin molar ratio = 0.5), 0.05 µM myosin II, 20 g/L Ficoll70, 3 µM MinD, 3 µM MinE and 5 mM ATP. Scale bar: 10 µm. Source data are provided as a Source Data file.
In line with these observations, we also detected vesicle deformation at 0.25 fascin/actin molar ratio. Contrary to single rings, soft actomyosin bundle networks positioned at the vesicle equator formed a constriction band resulting in the loss of spherical vesicle shape (Fig. 2b). In addition to this large-scale membrane deformation, the vesicle presented an actin-filled membrane out-bud at the constriction site (Supplementary Fig. 3, Supplementary Movie 3). Although we could not acquire the formation process of this bud, Litschel et al. already reported on this type of membrane deformation, which results from the sliding of actomyosin bundles along the membrane to one side of the vesicle and their collapse into a condensate15. Notably, the membrane bud we observed was localized closely to the actomyosin band constricting the vesicle at the equator, while MinDE pole-to-pole oscillations continued at the two hemispheres generated at each side of the actomyosin band. To quantitatively assess vesicle deformation, we calculated the vesicle aspect ratio (AR) as the ratio between their major (a) and minor (b) axes. In both examples presented, aspect ratios indicated a vesicle deformation towards a rod-like shape (AR < 0.9), with the lowest AR obtained (0.75) 24 h after encapsulation, where a vesicle with a centered actomyosin network exhibited a distinct elongated shape (Supplementary Fig. 4a).
Furthermore, given that macromolecular crowders can impact both actin bundle architecture and the mechanical properties of the vesicle membrane28,29, we subsequently performed encapsulation experiments at lower crowding concentration (20 g/l Ficoll70) and found membrane deformations in line with those already showed. Furthermore, under these experimental conditions, we detected an example of eccentric membrane deformation caused by an actomyosin ring when the established MinDE oscillation pattern inside the vesicle was different from the pole-to-pole mode (Fig. 2c). More specifically, the MinDE pattern at the vesicle membrane transitioned into a circling and less dynamic MinDE oscillation (possibly due to ATP depletion). Contrary to vesicles presenting mid-cell constriction, the membrane deformation observed induced a characteristic asymmetric dumbbell shape with two differently sized sub-compartments. Nevertheless, this asymmetric deformation, sustained over time (average AR over 20 minutes = 0.79), did not collapse after more than an hour of imaging (Supplementary Movie 4). Myosin II and ATP concentration added were 0.05 µM and 5 mM, respectively. Further experiments are therefore required to find the optimal encapsulating conditions enabling MinDE-stabilized rings positioned at mid-cell to undergo progressive contraction and controllably reduce their diameter by the action of myosin II motors.
In summary, we show that MinDE pole-to-pole oscillations can target the constriction of actomyosin architectures at the vesicle equator, resulting in the generation of sustained furrow-like membrane deformations.
MinDE-induced blebbing in reconstituted actomyosin vesicles
When thicker and more abundant actomyosin bundles developed at the membrane in the form of soft webs, we detected the establishment of more chaotic and static MinDE patterns. Thus, to further characterize the system, we performed time-lapse imaging on vesicles presenting chaotic or static-like MinDE patterns.
Interestingly, a large number of vesicles exhibiting these patterns developed membrane deformations similar to bleb-like morphologies (Fig. 3a, Supplementary Movie 5). To get further insights into the underlying mechanism behind bleb formation, we analyzed the interaction between our co-reconstituted protein systems and the vesicle membrane. Closer inspection of the blebs’ cross-sections revealed that the actomyosin soft web inside the vesicle compartmentalized the membrane into areas delimited by the peripheral attachment of bundles (Figs. 3b, Supplementary Fig. 5a). Due to this compartmentalization, container symmetry was lost and MinDE proteins generated a chaotic oscillation capable of deforming the membrane. More specifically, we found that the initial spontaneous curvature induced on these compartments by MinDE binding increased as more MinD molecules were recruited to the membrane, resulting in the outward growth of the dynamic bleb-like protrusions (Fig. 3c, Supplementary Fig. 5b). Subsequently, after MinDE-driven membrane deformations, we observed that the reduction in bilayer tension and the recovery of the initial vesicle shape was accompanied by the generation of a membrane out-bud (Fig. 3a, blue arrows), which was not reabsorbed into the mother vesicle (Supplementary Fig. 5c)30.
a Schematic illustration depicting the change in vesicle shape due to MinDE chaotic oscillations. Min proteins attach to areas delimited by soft actomyosin bundles and deform the membrane generating dynamic bleb-like protrusions. Fluorescence and brightfield confocal time-series show a blebbing vesicle. After bleb retraction, the reduction in bilayer tension generates an outward lipid bud (blue arrows). Encapsulation conditions: 2.4 µM actin, 0.6 µM fascin, 0.05 µM myosin II, 50 g/L Ficoll70, 3 µM MinD, 3 µM MinE and 5 mM ATP. Scale bars: 10 µm. b Confocal cross-section images at two time points of the vesicle in section a. Peripheral actomyosin anchoring creates a delimiting area which deforms upon MinDE binding. Additionally, MinDE diffusiophoretic transport changes the position of actomyosin bundles and the shape of the membrane area available for Min protein recruitment (blue arrows). Fluorescence intensity line plots of EGFP-MinD (green) and ATTO647-actin (magenta) demonstrate the demixing of both protein systems at the membrane perimeter (orange dotted line). Scale bars: 10 µm. c Schematic illustration of the proposed mechanism behind MinDE-induced blebbing. The recruitment of MinDE proteins to the compartmentalized inner leaflet of the bilayer generates the effect of a membrane outward protrusion in bleb form. d Schematic illustration depicting the radius of curvature RC used to calculate the curvature (Κ = 1/RC) of the blebs. 3D projection and 2D time-lapse confocal images show a vesicle with diverse bleb-like deformations emerging over time. Orange arrow points at a bleb with Κ = 0.73 µm−1. Blue arrow, Κ = 0.27 µm−1. Magenta arrow, Κ = 0.10 µm−1. Encapsulation mix: 4 µM actin, 2 µM fascin, 0.05 µM myosin II, 50 g/L Ficoll70, 3 µM MinD, 3 µM MinE and 5 mM ATP. Scale bars: 20 µm. Source data are provided as a Source Data file.
Furthermore, consistent with our previous observations and simultaneous to this membrane remodelling effect, the diffusiophoretic transport of actomyosin bundles reorganized the network at the membrane. As a result, membrane compartments changed in size and the oscillations maintained a chaotic mode inducing dynamic blebs in other areas of the vesicle (Fig. 3b).
To further scrutinize the membrane-remodelling capabilities of Min proteins along with our actomyosin architectures, we performed encapsulation experiments at 0.5 fascin/actin molar ratio and calculated the curvature of the blebs observed. Similar to our non-deflated vesicles encapsulated with 0.25 fascin/actin ratio, MinDE binding induced blebbing in a subset of vesicles (Supplementary Movie 6). Time-lapse imaging revealed that Min proteins can induce blebs with a wide range of curvatures (Κ, calculated as the inverse of the radius of a circle that fits the bleb). As MinDE established a chaotic oscillation inside the vesicle, small blebs (Κ = 0.73 µm−1), medium size (Κ = 0.27 µm−1) and big (Κ = 0.10 µm−1) deformations emerged (Fig. 3d, orange, blue, and magenta arrows, respectively).
Hence, our results show that, when co-reconstituted with actomyosin bundle networks, the MinDE system can generate dynamic bleb-like outward protrusions in vesicles encapsulated at iso-osmolar conditions, confirming its capabilities as a membrane remodelling protein system as previously observed27,31.
Co-reconstitution of actomyosin bundle networks and the MinDE system in phase-separated vesicles show remodelling of membrane domains
Lastly, to increase the complexity of the system and test its compatibility with other shape-remodelling strategies, we first tested its reconstitution in vesicles of ternary lipid mixtures. Due to their tuneable mechanical and biochemical properties, phase-separated lipid membranes constitute another approach to aid in the remodelling of biomimetic systems by altering membrane curvature, fluidity, etc32,33,34. Furthermore, two-phase vesicles constitute an additional strategy to study the reorganization and deformation of free-standing lipid domains by actomyosin networks35,36,37.
To achieve the co-reconstitution of the two protein systems in GUVs with phase-separated lipid domains we again employed the double emulsion transfer method. At room temperature (25 °C), GUVs demixed into coexisting Liquid-ordered (Lo) and Liquid-disordered (Ld) domains, where Ld domains consisted of DOPE-Biotin to facilitate actin binding (Fig. 4a). To study the successful reconstitution of the system inside phase-separated vesicles, we again performed time-lapse confocal imaging and observed that Min proteins could oscillate by binding to Ld domains on the vesicle membrane. Notably, due to the high frequency of soft actomyosin bundle webs formed, MinDE proteins also established chaotic oscillations on the Ld domains of the vesicle. The flexible bundles, however, spanned and crossed both lipid domains.
a Schematic illustration (top) and 3D confocal image (bottom) show the membrane composition employed to generate phase-separated vesicles and the domains obtained. Scale bar: 10 µm. b 3D projections and 2D confocal images depict a blebbing phase-separated vesicle. MinDE proteins bind and oscillate on Ld domains. Actomyosin bundles remain at lipid-phase boundaries as Min proteins transiently deform Ld domains (orange arrows). Inner encapsulation mix: 2.4 µM actin, 0.6 µM fascin (fascin/actin molar ratio = 0.25), 0.05 µM myosin II, 20 g/L Ficoll70, 3 µM MinD, 3 µM MinE and 5 mM ATP. Scale bars: 20 µm. c Schematic illustration of the proposed mechanism behind the dynamic deformation of Ld domains by MinDE protein oscillations.
Interestingly, MinDE-driven diffusiophoretic transport displaced the actomyosin bundles bound at Ld/Lo boundaries, reorganizing the actomyosin network inside the vesicle. Moreover, consistent with our previous experiments performed on single-phase GUVs, MinDE binding to areas delimited by actomyosin bundles at domain boundaries deformed Ld domains into dynamic bleb-like protrusions (Figs. 4b, Supplementary Movie 7). In contrast to studies in which bulging or budding of phase-separated GUVs is externally induced by tuning membrane composition or changing osmotic conditions, and where deformations arise by an imbalance between surface tension and interfacial line tension34,38, our phase-separated GUVs remained spherical over time in the absence of Min proteins, and no deformations in the form of blebbing or budding were observed (Supplementary Figs. 6, Supplementary Movie 8). Only in the presence of MinDE oscillations in GUVs containing actomyosin networks, timelapses showed the active—i.e., energy-consuming—deformation of Ld domains into outward blebs (Fig. 4c). During the course of these transient outward deformations, we also observed the remodelling of domains at the membrane such as their maneuvering followed by splitting (Supplementary Movie 7). As a result of MinD binding to Ld domains, the initial demixing of lipid phases on the vesicle thus changed, and domain reorganization occurred on the entire vesicle. Interestingly, while the resultant dynamic protrusions in the membrane comprised of Ld domains, Lo domains stayed intact.
In addition, to test the compatibility of our system with other membrane-based strategies for the enhancement of GUV deformation, we generated vesicles containing DOPE, a lipid with intrinsic negative curvature due to its inverse-cone shape39. Although the yield of vesicle production dramatically decreases at 15% DOPE molar ratio, successful encapsulation of our co-reconstituted system is possible when 10% DOPE is added to the lipid mix together with 60% POPC and 30% POPG. Importantly, in vesicles generated under these conditions, we observed generation of the three standard Min oscillations, actomyosin-driven equatorial deformation of vesicles and the formation of highly protruding blebs due to MinDE chaotic oscillations (Supplementary Fig. 4b).
In conclusion, we demonstrate that the co-reconstitution of Min proteins and actomyosin can deform DOPE-containing and complex ternary vesicle membranes, and dynamically remodel phase-separated lipid domains by rearranging and reshaping them on the membrane.
- SEO Powered Content & PR Distribution. Get Amplified Today.
- PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
- PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
- PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
- PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
- Source: https://www.nature.com/articles/s41467-024-54807-9