Article pubs.acs.org/Langmuir
Membrane Remodeling Processes Induced by Phospholipase Action Susana Rocha,† Herlinde De Keersmaecker,† James A. Hutchison,§ Karen Vanhoorelbeke,∥ Johan A. Martens,‡ Johan Hofkens,†,⊥ and Hiroshi Uji-i*,† †
Molecular Imaging and Photonics, Faculty of Science and ‡Centre for Surface Chemistry and Catalysis, Faculty of Bioscience Engineering, KU Leuven, Belgium § ISIS & icFRC, Université de Strasbourg & CNRS UMR 7006, Strasbourg, France ∥ Laboratory for Thrombosis Research, KU Leuven Kulak, Kortrijk, Belgium ⊥ Nano-Science Center/Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark S Supporting Information *
ABSTRACT: Important cellular events such as division require drastic changes in the shape of the membrane. These remodeling processes can be triggered by the binding of specific proteins or by changes in membrane composition and are linked to phospholipid metabolism for which dedicated enzymes, named phospholipases, are responsible. Here widefield fluorescence microscopy is used to visualize shape changes induced by the action of phospholipase A1 on dyelabeled supported membranes of POPC (1-palmitoyl-2-oleolysn-glycero-3-phosphocholine). Time-lapse imaging demonstrates that layers either shrink and disappear or fold and collapse into vesicles. These vesicles can undergo further transformations such as budding, tubulation, and pearling within 5 min of formation. Using dye-labeled phospholipases, we can monitor the presence of the enzyme at specific positions on the membrane as the shape transformations occur. Furthermore, incorporating the products of hydrolysis into POPC membranes is shown to induce transformations similar to those observed for enzyme action. The results suggest that phospholipase-mediated hydrolysis plays an important role in membrane transformations by altering the membrane composition, and a model is proposed for membrane curvature based on the presence and shape of hydrolysis products.
1. INTRODUCTION
Flippases drive curvature by translocating phospholipid molecules unidirectionally across the membrane, generating a difference in the number of phospholipid molecules and, consequently, a difference in the area of the leaflets.5,6 Their participation in membrane budding has been clearly demonstrated.7,14−16 Phospholipid-modifying enzymes can induce membrane curvature by introducing an asymmetry in the composition of the two leaflets. Because each phospholipid type has a different shape, differences in the composition will lead to differences in the spontaneous curvature of the inner and outer leaflets and, consequently, induce a global curvature of the membrane. Recent studies have demonstrated the involvement of three different groups of phospholipid-modifying enzymes in the modulation of membrane curvature: lysophospholipidactyltransferases (LPAT),17 sphingomyelinase (SM),18 and phospholipase A (PLA).6,19−22 PLAs are classified as either phospholipase A1 (PLA1) or phospholipase A2 (PLA2), depending on the position on the glycerol backbone where hydrolysis occurs (sn-1 and sn-2, respectively).23
The ability of the cellular membrane to alter its shape is crucial for processes such as cell movement, division, and vesicle trafficking.1−3 These phenomena are associated with dynamic remodeling processes, such as budding, tubulation, fission, and fusion, that require changes in the curvature of the membrane. Despite their enormous importance, some of the specific mechanisms used by the cell to produce membrane curvature remain unclear. It is assumed that membrane curvature is achieved through the action of membrane proteins.4 Membrane proteins can generate curvature either by applying forces or mechanical constraints (scaffolds) to the membrane surface or by altering the lipid bilayer composition.5−7 Although proteins that control the membrane shape through scaffolding have been the focus of several studies,1,4,8,9 less attention has been paid to the role of membrane-associated enzymes, such as flippases and phospholipid-modifying enzymes, that can nevertheless cause pronounced reorganization of the membrane 3D structure.9−11 These enzymes modify the phospholipid composition of the membrane in such a way that the two leaflets differ in terms of the absolute number of phospholipid molecules or the relative concentrations of various phospholipid types or both.5−7,12,13 © 2014 American Chemical Society
Received: January 10, 2014 Revised: March 10, 2014 Published: April 2, 2014 4743
dx.doi.org/10.1021/la500121f | Langmuir 2014, 30, 4743−4751
Langmuir
Article
accomplished by the addition of a solid support. The support consisted of polystyrene beads with a low degree of cross-linking, onto which a poly(ethylene glycol) derivative with free terminal amino groups was grafted. The unreacted dye was captured on the surface of the beads after 5 min of shaking. The solid support was separated from the enzyme solution by filtration. The process was repeated three times to ensure total removal of the unreacted dye. 2.5. Wide-Field Microscopy. Single-color fluorescence imaging was performed using an inverted epi-fluorescence microscope (IX71, Olympus) equipped with a 20× air objective (NA 0.5) and a cooled electron-multiplying CCD (Cascade 512B, Princeton Instruments Inc.). The dye-labeled layers (DiI) were excited with a 1−5 W/cm2 532 nm laser line (CDPS 532M-50, JDS Uniphase Co.). The excitation light was guided onto the sample through a dichroic mirror (z532rdc, Chroma Technology, Inc.). The emission from the membrane was observed through a 545 nm long-pass filter (HQ545LP, Chroma Technology, Inc.). Movies acquired after the addition of PLA1 were recorded over a 500 ms integration time at a sampling rate of 2 Hz using continuous illumination. For the slow addition of lyso-PC, the images were acquired at 6.67 Hz using 150 ms of exposure time. For the simultaneous imaging of the labeled layers and enzyme, a two-color detection setup was used (Supporting Information Figure S1). The inverted microscope was equipped with a 60× TIRFM oil objective (NA = 1.49) and two electron-multiplying CCD cameras (ImagEM, Hamamatsu Corporation). The DiO-labeled layers were excited with a 1−5 W/cm2 491 nm laser line (Cobolt Calypso 100, Laser 2000) using continuous illumination while the Atto-PLA1 molecules were excited with a 1−5 kW/cm2 644 nm laser line (Excelsior, SpectraPhysics) using stroboscopic illumination. The laser lines were combined using a 505 dichroic mirror (505dclp, Chroma Technology, Inc.) and guided onto the sample through a 488/633 dichroic mirror (z488/633rdc, Chroma Technology, Inc.). The emission arising from the sample was split into the two detection paths using a 630 nm short-pass dichroic mirror (630dcsp, Chroma Technology, Inc.). The emission from the membrane was observed after passing through a 505 nm long-pass filter (HQ505LP, Chroma Technology, Inc.) and a 535/50 band-pass filter (HQ535/50m-2P, Chroma Technology, Inc.) while the emission from the labeled enzymes was observed after passing through a 655 nm long -pass filter (HQ655LP, Chroma Technology, Inc.). Movies were recorded at a 2 Hz sampling rate using 500 and 10 ms integration times to image the layer and enzyme, respectively. As a result, while the DiO-labeled layers were continuously excited, the labeled enzymes were excited for just 2 × 10 ms per minute to avoid photobleaching. Acquisition of the fluorescence images of the labeled enzymes was synchronized with the excitation using mechanical shutters (Oriel, NewPort). This ensured minimal crosstalk between the two channels. 2.6. Determination of Desorption Rates. To calculate the desorption rates, the amount of lipids desorbed was determined through the difference in the area of the supported layers between successive frames using a mean molecular area of 0.65 nm2 for POPC.28 Fluorescence images were analyzed using home-developed software written in Matlab.
In this work, the dynamic remodeling of supported lipid membranes by the action of PLA1 enzymes was investigated using wide-field fluorescence microscopy. The enzyme-induced changes in the morphology of the supported bilayers were visualized by employing fluorescently labeled 1-palmitoyl-2oleoly-sn-glycero-3-phosphocholine (POPC)-supported multilayers as the substrate. To enable the simultaneous visualization of the structural changes in the membrane and enzyme localization, PLA1 labeled with dye Atto 647N (Atto-PLA1) and fluorescently labeled POPC-supported membranes were studied using two-color fluorescence wide-field imaging with single-molecule detection sensitivity. Furthermore, the effect of the direct incorporation of a product of POPC hydrolysis by PLA1, the lysophospholipid lyso-PC, into the membrane in the absence of enzymes was also explored. In this way, deeper insights into the relationship between PLA1 enzyme action and phospholipid membrane remodeling are obtained, and a model for membrane curvature is proposed on the basis of the shape of the hydrolysis products.
2. EXPERIMENTAL SECTION 2.1. Multilayer Preparation. Phospholipid multibilayers were prepared using the rehydration method as described previously.24,25 Briefly, a POPC solution in chloroform (10 mg/mL, Avanti Polar Lipids) containing a fluorescent dye (1,1′-dioctadecyl-3,3,3′,3′tetramethylindocarbocyanine perchlorate (DiI) or 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO), Molecular Probes, 0.5−1% molar ratio) was spin-coated onto freshly cleaved mica (3000 rpm for 40 s). In this way, samples with up to five dye-doped bilayers could be prepared. The top bilayers were removed by gently pipetting buffer solution onto the top surface and then washing with fresh solution. On the basis of the contrasting regions of fluorescence intensity, each sample had between one and three bilayers (Supporting Information Figure S2). DiO-labeled multibilayers were used for the two-color fluorescence measurements because the fluorescence properties of DiO and the Atto 647N enzyme label allow for good spectral separation. To achieve single-molecule resolution imaging, the mica support was cleaved until it was thin enough to allow the use of a large numerical aperture objective. To avoid breaking the thin mica support, it was glued onto a glass coverslip using a thin layer of poly(dimethylsiloxane) (PDMS, Sylgard 184, Dow Corning Corporation, refractive index 1.4118). After being dried under vacuum for 2 h, the multibilayer samples were hydrated by immersion in buffer solution (section 2.2) followed by heating in an oven at 80 °C for 3 h to rehydrate the lipids. 2.2. Buffer Solution. Supported membrane rehydration, enzyme solution preparation, and lyso-phospholipid solution preparation used a buffer containing 0.01 M HEPES (Fluka, 99.5%), 30 μM calcium chloride (Sigma-Aldrich 99%), 10 μM EDTA (Sigma 95%), and 0.15 M sodium chloride (Sigma-Aldrich 99.99%). The pH of the buffer solution was adjusted to 8 using sodium hydroxide (Sigma-Aldrich 99.998%). 2.3. Addition of Lyso-phospholipid. A 0.5 nM solution of 1oleoyl-2-hydroxy-sn-glycero-3-phosphocholine (lyso-PC) was prepared by evaporating chloroform from the original solution (10 mg/mL, Avanti Polar Lipids) and suspending the dried powder in buffer solution (section 2.2). The lyso-PC solution (0.5 mL) was added to the sample during imaging at a rate of 2.5 mL/h using a syringe pump (Harvard Apparatus). 2.4. Enzyme Labeling. The PLA1 enzyme was a gift from Dr. Alan Svendsen.26 The labeling procedure was similar to one described previously.25,27 Briefly, Atto 647N-NHS (Atto-Tec) was added in 10fold excess to an enzyme solution in carbonate buffer (pH 7.5). The solution was then incubated at 4 °C for 2 h to produce the dye-labeled enzyme. Atto 647N-NHS binds to the enzyme through the amino groups of the lysine residues and does not interfere with enzyme activity.25,27 Removal of unreacted dye from the enzyme solution was
3. RESULTS To investigate the influence of PLA1-mediated hydrolysis on the shape of phospholipid membranes, PLA1 (approximately 10−7 M) was released on supported POPC multilayers. The multibilayers were rendered fluorescence by mixing with the a dye that preferentially partitions into the membrane layers during the rehydration process (Experimental Section). The intensity of membrane fluorescence can be used to quantify the stacking of the layers as reported previously.25 (Supporting Information Figure S2). Time-lapse fluorescence microscopy can then be used to follow membrane structural remodeling due to hydrolysis by enzymatic action.25,29−31 4744
dx.doi.org/10.1021/la500121f | Langmuir 2014, 30, 4743−4751
Langmuir
Article
3.1. Membrane Transformations upon Enzyme Activity. 3.1.1. Shrinkage of the Layer. As shown in a recent report,25 dark regions appear on the fluorescence image of the supported phospholipid multilayers after the addition of PLA1 (10−7 M) (Figure 1 and Supporting Information Movie 1). The
top of other bilayers (Supporting Information Figure S3). Because this bilayer island is hydrolyzed by PLA1, one of two phenomena is observed: either it retracts until its complete disappearance (Supporting Information Figure S4, Supporting Information Movie 2) or it is transformed into a vesicle (Figure 2, Supporting Information Movie 3). This second transformation, designated “vesicle formation” here, always starts from one side of the bilayer island. The boundary of the bilayer island starts to curl, bending itself until it starts to cover the remaining flat part of the island. As the hydrolysis proceeds, the bilayer folds on itself completely to form a vesicle. Note that the phenomenon of vesicle formation occurs from bilayer islands occupying areas with different sizes (ranging from 150 to 3300 μm2), indicating that there is no critical radius linked to this transformation. In contrast, the rate of lipid desorption seems to be crucial. Vesicle formation happens for very low (less than 3 × 10−18 mol/s) or very high (higher than 15 × 10−18 mol/s) desorption rates. Furthermore, the formation of vesicles is not instantaneous but occurs within a few minutes after the addition of PLA1 to the bilayers. This likely indicates a dependence on the local concentration of hydrolyzed products in the bilayer. 3.1.3. Other Transformations. During or after vesicle formation, the spontaneous formation of protrusions can be observed on the surface of the newly formed vesicle. Often these protrusions grow further, forming a bud as shown in the fluorescence images in Figure 3. These buds are usually formed within 1 min of vesicle formation. The size and number of buds vary from vesicle to vesicle. Once the bud is formed, it can form a spherical vesicle itself (Figure 3) or grow further, acquiring an elongated shape (tubulation, Figure 4, Supporting Information Movie 4). Compared to the original phospholipid bilayer, these tubular structures are less stable (i.e., their shape is continuously changing). Indeed, a careful examination of the recorded images shows the formation of necks along the entire extension of the tubules (Figure 4B, left). In the final state (i.e., when no more changes are observed), the tubule has acquired a form that resembles pearls on a string (Figure 4B, right). This phenomenon is therefore designated as pearling in the remainder of the text. Although sometimes observed only after the elongation process, the pearling phenomenon occurs mainly during the process of elongation (Figure 5A,B, Supporting Information
Figure 1. Time-resolved fluorescence images of POPC layers labeled with DiI at progressive stages of hydrolysis. Collapse of the multilayers proceeds via retraction of the top bilayer ([PLA1] ≈ 10−7 M). The increase in the darker area observed herein is thus attributed to the breakdown of the POPC layer caused by the desorption of reaction products, which results in the release of the incorporated DiI label.
formation of darker regions is attributed to the desorption of the DiI dye from the layer, as shown by its fluorescence quantum yield decreasing dramatically as a result of cis−trans isomerization when it enters the solution. Because the dark features appear in areas with higher enzyme concentration, as visualized previously in two-color single-molecule measurements,25 they are attributed to the breakdown of the layer caused by the partial desorption of hydrolyzed product molecules (and DiI) following hydrolysis by PLA1. As hydrolysis proceeds, in addition to the retraction of the layer, other phenomena, namely, vesicle formation, tubulation, vesicle pearling, and budding, are observed. It must be stressed that these phenomena are observed only after the addition of PLA1 and not when the layers are monitored by time-lapse fluorescence microscopy without the addition of the enzyme (data not shown). It is thus concluded that the observed membrane modifications are linked to the enzyme activity. 3.1.2. Vesicle Formation. When supported phospholipid layers are preparing using the rehydration method, multilayers are formed. Before imaging, the top bilayers are removed by gently pipetting buffer solution across the surface and washing extensively with fresh solution (Experimental Section). This process often creates round double-bilayer islands situated on
Figure 2. Process of vesicle formation. (A) Fluorescence time-lapse images of phospholipid multibilayers labeled with DiI while going through vesicle formation during hydrolysis by PLA1 ([PLA1] ≈ 10−7 M). Images were acquired with a 500 ms integration time at a sample rate of 2 Hz. (B) Schematic illustration of the vesicle formation process shown in A. 4745
dx.doi.org/10.1021/la500121f | Langmuir 2014, 30, 4743−4751
Langmuir
Article
Figure 3. Vesicle budding. Fluorescence time-lapse images of DiI-labeled POPC multibilayers while undergoing budding on the side of a bilayer during hydrolysis by PLA1 ([PLA1] ≈ 10−7 M).
Although rarely observed, the vesicle can return to its original shape after the formation of tubules (Figure 6, Supporting Information movie 7), showing that the process is completely reversible. Figure 6 shows time-lapse images of such reversible shape change. Until around 40 s after vesicle formation, several tubules grow from the mother vesicle. After 40 s, the elongated vesicle returns to its original shape with just a 3% reduction in size. 3.2. Membrane Transformations upon Addition of Lyso-phospholipid (Lyso-PC). To determine if the presence of hydrolyzed product could be responsible for the observed remodeling processes, a solution containing lyso-PC was slowly added to the supported POPC layers (details in the Experimental Section; lyso-PC is one of the products of hydrolysis of POPC by PLA1). Note that these experiments were conducted without the addition of enzymes. Depending on the concentration of lyso-PC in the solution above the supported bilayers, different changes in bilayer morphology were observed (Figure 7). At low lyso-PC concentrations (
