|Home | About | Journals | Submit | Contact Us | Français|
Many of the more than 20 mammalian proteins with N-BAR domains1-2 control cell architecture3 and endocytosis4-5 by associating with curved sections of the plasma membrane (PM)6. It is not well understood whether N-BAR proteins are recruited directly by processes that mechanically curve the PM or indirectly by PM-associated adaptor proteins that recruit proteins with N-BAR domains that then induce membrane curvature. Here, we show that externally-induced inward deformation of the PM by cone-shaped nanostructures (Nanocones) and internally-induced inward deformation by contracting actin cables both trigger recruitment of isolated N-BAR domains to the curved PM. Markedly, live-cell imaging in adherent cells showed selective recruitment of full length N-BAR proteins and isolated N-BAR domains to PM sub-regions above Nanocone stripes. Electron microscopy confirmed that N-BAR domains are recruited to local membrane sites curved by Nanocones. We further showed that N-BAR domains are periodically recruited to curved PM sites during local lamellipodia retraction in the front of migrating cells. Recruitment required Myosin II-generated force applied to PM connected actin cables. Together, our study shows that N-BAR domains can be directly recruited to the PM by external push or internal pull forces that locally curve the PM.
N-BAR domains are versatile membrane-binding regulatory elements that function in a wide range of cellular processes including regulation of cortical actin structures3 and Clathrin-mediated endocytosis4-5. Although the cellular functions of many N-BAR domain proteins have been extensively studied, the fundamental chicken and egg causality dilemma whether they primarily sense or form curvature is not resolved. This question applies to all curvature binding domains but is particularly relevant for N-BAR domain proteins where in vitro assays showed that they selectively bind to6 as well as form7 membranes of high positive curvature. During endocytosis, where N-BAR domains have been most extensively studied, the N-BAR domain proteins Amphiphysin1, Endophilin2, and Bin2 are only recruited after the assembly of Clathrin-coated pits in a step that preceded vesicle scission8. This raised the question whether N-BAR domain proteins are recruited directly by curved membranes created by Clathrin-mediated formation of a tubular neck or indirectly by first binding to adaptor proteins that then recruit N-BAR domains to curve the membrane. Similarly, it is not clear how N-BAR domain proteins that bind N-WASP9-10 and other actin binding proteins11-12 are recruited to actin rich regions of the PM to then in turn regulate the actin architecture.
Our study aimed to distinguish between sensing and forming curved membranes by investigating recruitment of N-BAR domain protein to the PM in living cells. We focused on two examples of N-BAR domains, the one from Nadrin23,13, a regulator of actin polymerization, and the one from Amphiphysin14, a regulator of endocytosis. We developed a method to directly deform the PM in live cells using cone-shaped tinoxide nanostructures14 (Nanocones). The top panels in Fig. 1a show top and side scanning electron microscopy views of the formed Nanocones and the bottom left panel shows a cell growing on top of the Nanocone surface.
We tested an array of cone sizes that can be formed by this process and found an optimal height of 200 nm with a diameter of 50 nm at the base. 200 nm Nanocones had a 90% transparency compared to glass without Nanocones (Fig. 1a, bottom right). When cultured on 200 nm but not on 600 nm Nanocones, cells showed no significant preference between Nanocone and flat surfaces (Fig. 1b) and freely migrated across 20 μm wide bands of Nanocones (data not shown). When culturing cells on these Nanocones, we found small diameter positively curved sections (inward bend) in the basal PM of ‘unroofed’ cells in scanning electron micrographs (Fig. 1c), and in transmission electron micrographs of cross-sections of the basal PM (Fig. 1d).
To compare the effect of Nanocone-induced PM curvatures on N-BAR domain recruitment within the same cell, we designed a pattern of 3 μm-wide stripes, alternating between smooth and Nanocone-coated surfaces, plated 3T3 cells expressing fluorescently tagged N-BAR proteins on these patterned nanostructures, and imaged them with a confocal microscope. Markedly, we found puncta selectively over Nanocone stripes when we expressed the YFP-conjugated N-BAR protein Nadrin2 (Fig. 2a). Control experiments showed that the average local increase in Nadrin2 intensity in puncta was 113 +/− 9% over a uniform cytosolic background while a PM marker increased at the same sites only by 45 +/ −3%, arguing that Nadrin2 is recruited to local sites within the PM. A correlation analysis resulted in the same conclusion (Supplementary Figs. S1a, b).
We then showed that the N-BAR domain of Nadrin2 is sufficient to mediate the punctuate recruitment to the regions with Nanocone stripes (Fig. 2b). In a quantitative single cell analysis, full-length protein and isolated N-BAR-domain generated significant more puncta in the PM over Nanocone stripes compared to the PM over the flat surfaces (Fig. 2b, right panel). This recruitment by Nanocone stripes was not restricted to the N-BAR domain of Nadrin2, but also applied to the isolated N-BAR domain of Amphiphysin1 (Fig. 2c).
Using electron micrographs, we then validated the precise localization of the N-BAR domain over Nanocones. Staining cells transfected with the YFP-tagged N-BAR domain of Nadrin 2 with an antibody directed against the fluorescent tag showed a significant enrichment of gold particles over Nanocones (Fig. 2d; Supplementary Fig. S1c). Next, we measured the persistence of N-BAR domain puncta over Nanocones in live cells. We found the puncta themselves to be stable for periods longer than 1 minute (Fig. 2e) whereas about half of the BAR domains within the puncta exchanged with a time constant of 14 seconds, as shown in fluorescence recovery after photobleaching (FRAP) experiments (Fig. 2f).
Since many N-BAR domain proteins have roles in endocytosis4-5, we determined, as a control, whether the observed puncta reflect endocytotic sites that might be induced by Nanocones. We did not observe a Nanocone-dependent increase in the average endocytosis rate (Supplementary Fig. S2a). Although, Clathrin light chain and Dynamin both enriched over stripes of Nanocones (Supplementary Figs. S2b, c), co-expression experiments with Nadrin2 and Clathrin light chain showed only minimal overlap (Supplementary Fig. S2d), arguing that N-BAR domain recruitment by Nanocones can be induced independent of endocytosis. This low correlation was confirmed in a spatial cross-correlation analysis where the co-localization of the near identical CFP-tagged Nadrin2 and YFP-tagged Nadrin2 was used as a reference (Supplementary Figs. S2e, f). Tracking individual CFP-Clathrin light chain puncta over Nanocones over time provided no evidence for sequential co-localization, ruling out that N-BAR domain recruitment follows Clathrin accumulation with a delay (Supplementary Fig. S2g). Together, this argues that Nanocones provide an external local push force that induces stable membrane deformations and that an inward curved PM is sufficient to recruit N-BAR domains.
We then considered that internal forces that pull on a local membrane sites during actin reorganization or endocytosis might be equally capable in inducing the high inward curvatures necessary for N-BAR domain recruitment. We tested this hypothesis by focusing on migrating cells that experience significant membrane deformation at the leading edge15. As shown in Fig. 3a, the PM-localized N-BAR domain proteins Nadrin2 and Amphiphysin1 were enriched in the actin-rich front of polarized migrating 3T3 cells. This localization only required the N-BAR domain, since the same polarized localization was observed for full length and isolated N-BAR domains of Nadrin2 and Amphiphysin1 (Fig. 3b). Deletion of the amphipatic helix, a sequence motif required for binding and stabilization of curved membranes7,16-17, prevented enrichment of the N-BAR domain protein, arguing that localization critically depends on this structural feature required for membrane binding and present in all N-BAR domain proteins (Fig. 3c).
We observed an overlap between membrane regions with high cortical actin and either Nadrin2 or Amphiphysin1 N-BAR domains, suggesting that curved membrane sites are created above cortical actin networks (Fig. 3d). Control experiments with markers for the cytoplasm and the PM showed that this partial co-localization of N-BAR domains with cortical actin was not caused by variability in cell thickness or membrane accumulation (Figs. 3a, b and e). Furthermore, the N-BAR domain of Nadrin2 did not co-localize with a marker for Clathrin-mediated endocytosis at the leading edge (Supplementary Fig. S2h). However, we observed transient co-localization with Clathrin puncta behind the leading edge, suggestive of temporary recruitment of the isolated N-BAR domain of Nadrin2 to these internal sites during endocytosis8,18, (Supplementary Fig. S2i). Together, this argued that N-BAR domain proteins are recruited to the actin-rich leading edge membrane in a process that does not involve endocytosis and can be mediated by their N-BAR domains alone.
To more directly test the effect of peripheral actin polymerization on the localization of N-BAR domains in the front, we transiently changed actin dynamics in live cells. First, we monitored N-BAR domain puncta at the leading edge of 3T3 cells upon depolymerization of actin filaments. N-BAR domain puncta disappeared within less than 5 minutes after addition of 4μM Latrunculin A (Fig. 3f) or 5μM Cytochalasin D (Fig. 3g) and reappeared once the drug was washed out. In a complementary experiment, we rapidly triggered actin polymerization taking advantage of an assay based on inducible translocation of TIAM1, an exchange factor for Rac19. 3T3 cells were co-transfected with a fluorescently tagged N-BAR domain together with a construct containing FKBP fused to TIAM1 and a membrane-anchored FRB domain. Addition of rapamycin triggered dimerization of FRB and FKBP, rapidly recruiting TIAM1 to the PM, and causing the activation of the small GTPase Rac1. This led to increased levels of Rac-GTP at the PM causing extensive actin polymerizations and, with a delay, the translocation of the isolated N-BAR domain to the PM (Supplementary Fig. S3). A comparable result was observed when a photoactivatible version of the small GTPase Rac1 (LOV-Rac120) was used to trigger actin polymerization (Supplementary Fig. S4). Together, these experiments argue that increases in dynamic cortical actin polymerization result, with a delay, in the recruitment of curvature-sensing N-BAR domain proteins.
Cortical actin polymerization often occurs in cycles where a polymerization phase that extends the lamellipodia outward is followed by a retraction phase whereby the actin motor Myosin II pulls lamellipodia partially back before the next extension21-23. Markedly, time lapse imaging of cyclic lamellipodia in the front of polarized 3T3 cells showed that N-BAR recruitment occurs primarily during the retraction phase rather than the extension/polymerization phase (Fig. 4a). A time-course analysis over multiple expansion-retraction cycles showed that the concentration of the N-BAR domain (normalized to the total amount of actin filaments) remained low as the lamellipodia moved outward and only increased as it was retracted (Fig. 4b, Supplementary Movie S1). This observation was confirmed using an autocorrelation analysis comparing normalized N-BAR domain concentration and lamellipodia position (Fig. 4c, d). This cyclic nature of N-BAR domain recruitment during extension and retraction is schematically shown in Fig. 4d.
We hypothesized that Myosin II-mediated lamellipodia retraction involves pulling forces applied to the PM via membrane-anchored actin cables that cause local inward membrane curvature while the distributed surrounding actin meshwork resists the retraction. Consistent with such a role of Myosin II, we observed during lamellipodia retraction a parallel increase of the N-BAR domain of Nadrin2 at the front and of the heavy chain of non-muscle Myosin II (MHC2a) in a region closely behind where the lamellipodia transitions into the lamella (Fig. 4e, Supplementary Movie S2). To further validate the hypothesis that Myosin II may pull on actin cables to promote retraction, we used ML-7, an inhibitor of Myosin light chain kinase (MLCK) that controls Myosin II activity of non-muscle cells. Consistent with a key role of MLCK and Myosin II in the process of N-BAR domain recruitment, the patches of the respective N-BAR domains of Nadrin2 and Amphiphysin1 rapidly disappeared upon addition of the inhibitor (Fig. 4f, g).
This suggested that Myosin II-mediate force on actin cables induces local inward deformations of the PM at sites where the cables contact the leading edge PM. We used electron microscopy24 to analyze the topography of the PM of migrating cells and observed significant inward curved sections along the leading edge (Fig. 5a). Actin cables oriented perpendicular to the PM often pointed to the inward curved section of the PM (Fig. 5b, Supplementary Fig. S5a), suggesting that force applied to these actin cables was responsible for the deformation. Furthermore, immune gold localization showed that the isolated N-BAR domain of Nadrin2 locally enriched at these inward curved membrane sites (Fig. 5c, Supplementary Fig. S5b). Together, these findings introduce a recruitment mechanism whereby MLCK and Myosin II-mediated actin-contraction pulls on local PM-actin cable contact sites to create local inward curved membrane sites that then recruit N-BAR domain proteins. Notably, N-BAR puncta disappeared when PM tension was increased by lowering the external osmotic strength (Supplementary Fig. S5c). Correspondingly, lowering membrane tension with high osmotic pressure or the surface relaxant Deoxycholate25 both triggered puncta formation (Supplementary Figs. S5d, e), suggesting that N-BAR recruitment can further be regulated by changes in the global PM tension parameter.
Our results argue that different modes of mechanical deformation of the PM exist that trigger the recruitment of N-BAR domain containing regulatory proteins. First, the ability of external Nanocones to directly trigger N-BAR domain recruitment suggests that a receptor-independent mechanical signaling mechanism may exist whereby extracellular matrix components of sufficient stiffness can trigger local deformations of the PM to directly recruit and activate intracellular signaling proteins with N-BAR domains (Fig. 5d, left). Second, our results with actin and Myosin II argued that internal force applied to membrane connected actin cables can exert sufficient pull force to curve the PM and cause N-BAR domain recruitment (Fig. 5d, middle), a finding that has broad implications in actin-dependent processes such as cell polarization and migration. Third, our results also likely applies to endocytosis where Clathrin coats may provide the force that recruits N-BAR domains by creating a neck with a curved membrane tube that connects the forming Clathrin vesicle to the rest of the PM8 (Fig. 5, right). While these different recruitment mechanisms argue that local force application and induction of membrane curvature are needed to trigger the initial N-BAR domain recruitment in live cells, we would like to note that it is likely that N-BAR domains, once recruited, have a complementary role in stabilizing the curved membrane section.
In summary, our study shows that cytosolic N-BAR domains continuously scan the PM for highly inward curved sections and accumulate at sites where the membrane is sufficiently deformed by local forces. Our study further suggests that any type of push force applied to local PM sites from the outside or pull force applied to local sites from the inside will induce N-BAR domain recruitment as long as the local force is sufficiently strong and is counteracted by distributed forces on the membrane in the opposing direction. Finally, different force generating mechanism involving Clathrin, microtubules, actin cables or extracellular matrix may create in different contexts either inward (positive) or outward (negative) curved PM that likely provide only two types of distinct recruitment signals. To gain specificity, additional co-regulatory mechanisms are likely needed to link specific mechanical deformations to distinct signaling pathways and cell functions.
Methods and any associated references are available in the online version of the paper at http://www.nature.com/naturecellbiology.
Movie S1 Recruitment of the curvature-sensor Nadrin2 to the leading edge of migrating 3T3 cells. 3T3 cells transfected with the isolated N-BAR domain of Nadrin (red) and the actin marker f-tractin (green) are shown. Interval between frames is 10 seconds. Playback rate is 25 frames per seconds. Scale bar is 10μm.
Movie S2 3D reconstruction of 3T3 lamella with actin (blue), N-BAR domain of Nadrin2 (red) and MHC2a (green). Playback rate is 25 frames per seconds. Scale bar is 1 μm.
Movie S3 Recruitment of the N-BAR domain of Nadrin2 (red) and MHC2a (green) to the retracting lamella in 3T3 cells. Interval between frames is 10 seconds. Playback rate is 10 frames per seconds. Scale bar is 1 μm.
The authors thank the members of the Meyer lab for comments and discussion. M.G. was supported by Swiss National Science Foundation (No. PBBSP3-123159), Novartis Jubilaeumsstiftung and Stanford Deans Postdoctoral Fellowship. S.J. was supported by the Korea Foundation for Advanced Studies graduate fellowship. Y.C. acknowledges the partial support from a DOE-EFRC at Stanford: Center on Nanostructuring for Efficient Energy Conversion (No. DE-SC0001060). T.M. acknowledges funding from the National Institute of Health, MH064801, MH095087 and GM063702.
Author Contributions: M.G. performed all experiments and analyzed the data. FC T. developed the temporal cross-correlation analysis. S.J. and Y.C. designed the Nanocones. LM J helped with the SEM. YI W and KM H developed the PA-Rac construct. M.G. and T.M. designed the experiments, interpreted the results and wrote the manuscript. All authors discussed results and manuscript. T.M. supervised the study.
Competing Financial Interests: The authors declare no competing financial interests.