- Email: [email protected]
Cell-derived plasma membrane vesicles are permeable to hydrophilic macromolecules
Journal Pre-proof Cell-derived plasma membrane vesicles are permeable to hydrophilic macromolecules A.D. Skinkle, K.R. Levental, I. Levental PII:
S0006-3495(20)30068-0
DOI:
https://doi.org/10.1016/j.bpj.2019.12.040
Reference:
BPJ 10274
To appear in:
Biophysical Journal
Received Date: 9 August 2019 Accepted Date: 23 December 2019
Please cite this article as: Skinkle A, Levental K, Levental I, Cell-derived plasma membrane vesicles are permeable to hydrophilic macromolecules, Biophysical Journal (2020), doi: https://doi.org/10.1016/ j.bpj.2019.12.040. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Biophysical Society.
1
TITLE: Cell-derived plasma membrane vesicles are permeable to hydrophilic macromolecules
2
AUTHORS: AD Skinkle1,2, KR Levental1, I Levental1,*
3
AFFILIATIONS:
4
1
5
Pharmacology, Houston, Texas 77030
6
2
7
*
University of Texas Health Science Center at Houston, Department of Integrative Biology and
University of North Carolina at Chapel Hill, Biological and Biomedical Sciences Program
corresponding author; [email protected]
8 9
ABSTRACT: Giant Plasma Membrane Vesicles (GPMVs) are a widely used experimental platform for
10
biochemical and biophysical analysis of isolated mammalian plasma membranes (PM). A core advantage
11
of these vesicles is that they maintain the native lipid and protein diversity of the plasma membrane while
12
affording the experimental flexibility of synthetic giant vesicles. In addition to fundamental investigations of
13
PM structure and composition, GPMVs have been used to evaluate the binding of proteins and small
14
molecules to cell-derived membranes, and the permeation of drug-like molecules through them. An
15
important assumption of such experiments is that GPMVs are sealed; i.e. that permeation occurs by
16
diffusion through the hydrophobic core rather than through hydrophilic pores. Here we demonstrate that
17
this assumption is often incorrect. We find that most GPMVs isolated using standard preparations are
18
passively permeable to various hydrophilic solutes as large as 40 kDa, in contrast to synthetic Giant
19
Unilamellar Vesicles (GUVs). We attribute this leakiness to stable, relatively large, and heterogeneous
20
pores formed by rupture of vesicles from cells. Finally, we identify preparation conditions that minimize
21
poration and allow evaluation of sealed GPMVs. These unexpected observations of GPMV poration
22
important for interpreting experiments utilizing GPMVs as plasma membrane models, particularly for drug
23
permeation and membrane asymmetry.
24 25
STATEMENT OF SIGNIFICANCE: A central assumption in using Giant Plasma Membrane Vesicles to
26
study membrane penetration and interactions is that these vesicles maintain the permeability barrier of the
27
native membrane from which they originate. Using large fluorescently-labeled hydrophilic probes, we
28
demonstrate that this assumption is often incorrect and conclude that macromolecular solutes permeate
29
GPMVs through stable pores formed during shear-induced rupture of vesicles from cells. Using these
30
insights into the mechanisms of poration, we demonstrate an approach to isolate sealed GPMVs.
31 32
INTRODUCTION
33 34
Model membranes are among the most widely used tools to investigate the physicochemical behaviors of
35
biological lipids, interactions between lipids and proteins, and the properties of lipid assemblies. While
36
synthetic lipid membranes effectively recapitulate certain aspects of biological membranes, it is not
37
experimentally feasible to construct model systems with the lipid diversity and protein density of living
38
cells. In this context, an important development in membrane biophysics has been the discovery (1) and
39
characterization of Giant Plasma Membrane Vesicles (GPMVs), which are isolated from live mammalian
40
cells as intact PM vesicles (2). These GPMVs fill a unique niche, maintaining the approximate lipid and
41
protein content of living membranes while offering the experimental tractability of synthetic models (3).
42
This combination has allowed investigation of PM composition (4), physical properties (5–7), dynamics (8,
43
9), and potentially signaling and transport (8, 10–13) in isolation from the complexity of living cells.
44 45
Because of these advantages, and despite several important caveats distinguishing isolated vesicles from
46
living membranes (3), GPMVs have become an important and widely used tool for membrane biophysics.
47
Perhaps their most notable feature is their spontaneous separation into coexisting liquid-ordered and
48
disordered membrane phases, allowing investigations into the properties and compositions of ordered
49
membrane domains, as models for lipid rafts (14–21). These vesicles have also been used in combination
50
with mass spectrometric lipidomics to characterize the detailed and comprehensive lipid profiles of PMs
51
and their remodeling during various cellular processes (22–24). Recently, GPMVs have also been
52
employed to study mechanisms of viral binding and fusion to the PM (25), as well as toxin binding (26).
53 54
In addition to their considerable utility for basic research, GPMVs have recently gained attention for
55
possible translational applications, particularly for their potential as drug delivery vehicles (27, 28) or for
56
studies of membrane permeability and poration (29–31). An important premise for these and other similar
57
studies is that GPMVs are sealed to extravesicular compounds. This assumption stems from analogy to
58
synthetic Giant Unilamellar Vesicles (GUVs), which are unambiguously sealed to permeation of
59
hydrophilic solutes (26, 32). Here, we report the surprising finding that GPMVs are permeable to large,
60
hydrophilic macromolecules. Probes as large as 40 kDa rapidly accumulate inside GPMVs, apparently
61
through stable aqueous pores. We investigate the mechanism for this surprising permeability and suggest
62
that stable pores are formed during shear-induced rupture of GPMVs from cells. Based on these
63
observations, we develop a technique to limit pore formation and image sealed GPMVs.
64 65
MATERIALS AND METHODS
66 67
Materials: The following lipids were purchased from Avanti Polar Lipids (Alabaster, AL) and used without
68
further purification: dipalmitoyl phosphatidylcholine (DPPC), dioleoyl PC (DOPC) and cholesterol. Fast DiI
69
(Life Technologies, Carlsbad, CA), fluorescein- and TMR-labeled dextrans (Sigma, St. Louis, MO),
70
AlexaFluor-647-ATP (Fisher Scientific, Hampton, NH). Other chemicals, buffers, and media were
71
purchased from Sigma.
72 73
Cell culture: Rat basophilic leukemia (RBL) cells were maintained in medium containing 60% modified
74
Eagle’s medium (MEM), 30% RPMI, 10% fetal calf serum, 2 mM glutamine, 100 units/mL penicillin, and
75
100 µg/mL streptomycin at 37oC in humidified 5% CO2.
76
77
GPMV isolation and imaging: GPMVs were prepared as previously described (2). Briefly, cells were
78
washed in isolation buffer (10 mM HEPES, 150 mM NaCl, 2 mM CaCl2, pH 7.4) and then incubated with
79
isolation buffer supplemented with 25 mM paraformaldehyde (PFA) and 2 mM dithiothreitol (DTT) (or 2
80
mM N-ethylmaleimide (NEM) as indicated) for 1 h at 37oC (unless otherwise indicated). Before isolation,
81
cell membranes were labeled with 5 µg/mL of the membrane marker FAST DiI for 10 min on ice. After 1 h
82
incubation, the GPMV-rich cellular supernatant was transferred to a microcentrifuge tube, then
83
concentrated GPMVs were collected after 30 min sedimentation on ice. All imaging was done by confocal
84
microscopy on a Nikon A1R.
85 86
Quantification of GPMV leakiness: GPMVs were either prepared with (or incubated after isolation with)
87
250 nM AlexaFluor-647-ATP and/or 250 ng/mL fluorescein-labeled 3 kDa dextran. Confocal images were
88
then taken with the fluorescent molecules still in solution. In these images, the lumens of “sealed” vesicles
89
appear dark relative to the background fluorescence, whereas leaky vesicles have bright fluorescent
90
lumens (see Fig. 1). These images were analyzed using a custom-built algorithm in the open-source
91
software Cell Profiler. GPMVs were identified and masked as circular objects, then the median
92
fluorescence intensity (to eliminate edge-effects) of the vesicle lumen was calculated for each vesicle. The
93
fluorescence intensity of the bulk/background was calculated as the median intensity of all pixels outside
94
GPMVs. The “relative vesicle intensity” was calculated as the ratio of the intensity of the vesicle lumen to
95
the bulk intensity.
96 97
Although most “leaky” vesicles had an internal fluorescence that was approximately equal to bulk
98
fluorescence, some were slightly dimmer while others slightly brighter than the bulk fluorescence. Thus, a
99
“leaky” vesicle was defined as one whose intensity was at least 70% of bulk fluorescence (relative vesicle
100
intensity >0.7). Although this value is somewhat arbitrary, it was chosen because it generally segregated
101
the two populations of GPMVs that emerged in our analyses (e.g. Fig. 1D). Sealed vesicles also had non-
102
zero “relative vesicle intensities”, which would appear to suggest that some permeation still occurred.
103
Although we cannot absolutely rule out some leakage into apparently “dark” vesicles, it is much more likely
104
that the non-zero lumenal fluorescence comes from analyzing images with a very bright fluorescent
105
background. Even in confocal mode, some out-of-focus signal is detectable from above and below the
106
vesicle, and the magnitude of this signal depends on the size of the vesicle. Thus, there are no truly “dark”
107
pixels in our images, which prohibits accurate background subtraction. An example of this effect (non-zero
108
normalized intensity in putatively sealed vesicles) is shown in Fig. S1. Because of the large contrast
109
between the “sealed vesicle” lumens and the bright bulk fluorescence, and because most putatively sealed
110
vesicles had similar lumenal intensities, we believe they allow minimal permeation of the fluorescent
111
solutes.
112 113
Preparation of Giant Unilamellar Vesicles: GUVs were prepared by electroformation, as previously
114
described (33). Briefly, a lipid mixture of dipalmitoyl phosphatidylcholine (DPPC), dioleoyl PC (DOPC) and
115
cholesterol was prepared in chloroform with a molar ratio of 5:3:2 (lipids purchased from Avanti). The lipid
116
mixture was applied to Pt-wires in a custom electroformation chamber and solvent was evaporated in a
117
vacuum chamber. The electroformation chamber was filled with sucrose solution (0.1 M) and heated to
118
53°C. An alternating sinusoidal current was applied across the cell unit with 2.5 V, 10 Hz at 53 °C fo r 2
119
hours. GUVs were then tranferred to a microcentrifuge tube and maintained at room temperature, before
120
imaging and analysis as described above for GPMVs. We observed no effect of liquid-liquid phase
121
separation in GUVs or GPMVs on their leakage (Fig. S6).
122 123
RESULTS AND DISCUSSION
124 125
Isolated GPMVs are permeable to hydrophilic macromolecules
126 127
GPMVs are widely used models of cellular plasma membranes and as such have been presumed to retain
128
the barrier function of the native PM. This assumption has been reinforced by GPMVs’ superficial
129
resemblance to synthetic GUVs, which generally do not have large hydrophilic pores. Surprisingly, we
130
observed that cell-detached GPMVs exhibited considerable heterogeneity in their permeability to
131
extravesicular solutes in the form of fluorescently-labeled hydrophilic probes. When GPMVs were
132
prepared in the presence of an Alexa647-modified ATP analog (2 kDa) or FITC-modified dextran (3 kDa),
133
a major fraction of isolated vesicles showed accumulation of these probes in the intravesicular space (Fig.
134
1A), a phenomenon hereafter referred to as permeation.
135 136
While fluorescent dextran was generally slightly less concentrated in the lumen of permeable GPMVs
137
compared to the bulk extravesicular solution, fluorescent ATP was often clearly enriched (Fig. 1A inset).
138
We presume this enrichment reflects the interaction with intravesicular proteins. In striking contrast,
139
synthetic giant unilamellar vesicles (GUVs) prepared by electroformation were almost uniformly
140
impermeable to the same solutes, as exemplified in Fig. 1B and quantified in Fig. 1C. Within any given
141
isolated GPMV population, there was considerable variation in the accumulation of probes inside the
142
vesicle, ranging from impermeable to slightly brighter than the bulk fluorescence of the extravesicular
143
solution (Fig. 1D). Notably, all permeable GPMVs were similarly permeable to both fluorescent solutes,
144
while GPMVs that were impermeable to dextran were also impermeable to fluorescent ATP (Fig. 1D).
145
These observations allowed us to separate the GPMVs in a given preparation into clearly distinct sealed
146 147
and permeable populations (Fig. 1D).
148
The fraction of permeable GPMVs was somewhat variable between repeated preparations (the reasons
149
underlying this variability are discussed below); however, on average, the majority of vesicles were
150
permeable to both fluorescent solutes (Fig. 1E). GPMVs were similarly permeable to FITC-modified
151
dextran in the absence of fluorescent ATP (Fig. S2) revealing that permeation is not driven by the ATP
152
analogue. These observations were not dependent on the chemicals chosen for GPMV production, as
153
GPMVs prepared with N-ethylmaleimide (NEM) were similarly permeable to those prepared with
154
PFA+DTT (Fig. S3)
155 156
Time-lapse imaging revealed that permeation occurred abruptly, and after the formation of an initially
157
sealed cell-attached vesicle. Fig. 2A and Video S1 show the temporal evolution of one such vesicle, which
158
is sealed at an arbitrary starting point (approximately 30 mins after induction of vesiculation), but then
159
rapidly fills with both macromolecular solutes within approximately one minute of the permeation-inducing
160
event (the putative nature of the permeation-inducing event is discussed below). This permeation was not
161
associated with any change in the gross morphology of the vesicle or its membrane, as shown by the lipid
162
dye DiI (Fig. 2; bottom row). The kinetics of filling can be fit to a first-order rate constant, with both solutes
163
reaching the ‘permeability threshold’ (rel. vesicle intensity of 0.7) in <1 min. This rapid accumulation is
164
inconsistent with the permeation of large hydrophilic probes through an intact membrane hydrophobic
165
core, suggesting alternative permeation pathways. To test the mechanisms underlying these surprising
166
observations, we evaluated the size, temperature, time, and mechanical perturbation dependence of
167
GPMV permeability.
168 169
Large solutes permeate GPMVs via stable pores
170 171
To determine whether permeation through the GPMV membrane was occurring via well-defined pores, we
172
tested permeation to fluorescent dextrans of increasing molecular weight. We observed a clear solute size
173
dependence, with larger substrates permeating less efficiently. This dependence is demonstrated by
174
incubating a vesicle preparation with two differently sized dextran molecules labeled with two different
175
fluorophores. As shown in Fig. 3A, some vesicles are completely sealed (dark in both channels), while a
176
large fraction are completely permeable to 3 kDa dextran but not 60 kDa dextran (red only). This
177
dependence of permeation on size of the solute was observed up to ~40 kDa, with decreasing permeation
178
for larger substrates (Fig. 3B). However, the size-dependence was not as sharp as might be expected for
179
very consistent, homogeneous pores. In that case, we would expect a sharp cut-off at a particular
180
molecular weight, with all vesicles being permeable to smaller substrates and completely sealed to larger
181
ones. Instead, we observed a gradual decrease in the fraction of permeable vesicles to larger and larger
182
substrates, suggesting that a distribution of pore sizes exists in any given population of vesicles.
183 184
Finally, the possibility of stable pores was supported by observations that permeable GPMVs retained
185
their permeability well after isolation. GPMVs that were isolated in the presence of FITC-dextran alone
186
were incubated for 30 mins at 37oC, then exposed to fluorescent ATP. After this incubation, vesicles that
187
were permeable to one substrate were also permeable to the other, whereas vesicles that remained
188
impermeable to dextran after 30 mins were also sealed to the second substrate (Fig. 3C). These
189
observations suggest that stable pores (some of which can accommodate substrates up to 40 kDa) are
190
responsible for accumulation of solutes within GPMVs.
191
Cell-attached GPMVs are not permeable during their formation
192 193
To elucidate the conditions that caused stable pores in GPMVs, we first examined whether GPMVs were
194
permeable immediately upon their generation or became permeable due to handling during/after isolation.
195
GPMV formation was imaged from its initiation, allowing observation of vesicles as they began to bud,
196
inflated, then ultimately detached from cells. In these observations, both budding vesicles and almost all
197
detached vesicles remained sealed to the small molecule probes (Fig. 4A), revealing that cell-attached
198
GPMVs are generally not permeable but rather become leaky at some point after their generation or
199
isolation.
200 201
To test the possibility that GPMVs become permeable over time regardless of handling, we observed
202
GPMVs up to 2.5 hrs after isolation. We also compared GPMVs that were isolated from cells to those that
203
were left undisturbed in the original plate in which they formed. Isolated GPMVs were characteristically
204
heterogeneous in their accumulation of the extravesicular probes, with the majority being leaky to both
205
solutes (Fig. 4B and S4A). Notably, the general distribution of permeable vesicles remained consistent
206
throughout the three-hour time course, suggesting that the permeation-inducing event is not time-
207
dependent. In striking contrast, GPMVs that were not isolated from cells, but rather imaged without
208
manipulation in the same plate in which they formed, remained sealed throughout the time course, up to 3
209
h after induction of blebbing (Fig. 4B, S4A). These results suggested that the permeation-inducing event
210
occurs during the process of isolating GPMVs from the cells.
211 212
Neither temperature changes, phase separation, nor shear from pipetting contribute to GPMV
213
permeability
214 215
Given that GPMVs remain sealed until isolation from cells, we hypothesized that either temperature
216
changes or the mechanical stress of pipetting during isolation was responsible for eliciting vesicle
217
permeation. To test the effect of temperature changes, GPMVs were prepared at 37˚C (as usual) then
218
observed either at 37˚C or 23˚C. Both cases showed similar and characteristic distributions of permeable
219
vesicles, while samples that were left undisturbed remained sealed regardless of temperature changes
220
(Fig. 4C, S4B). Similarly, heating of the samples after isolation was not sufficient to “heal” pre-formed
221
pores. Thus, temperature changes occurring during isolation are insufficient to elicit permeation.
222 223
The permeability of GPMVs was also unaffected by their phase separation into Lo and Ld phases. We
224
observed both sealed and “leaky” vesicles in microscopically homogeneous and phase-separated
225
populations, with no notable difference in leakiness between them. GUVs remained sealed to
226
macromolecules regardless of whether they were mixed or phase-separated (Fig S6).
227
228
To test the possibility that the mechanical stress of pipetting causes lipid bilayer disruptions large enough
229
to allow permeation of solutes into the vesicle, GPMVs were isolated using pipette tips of varying bores
230
and pipetted >20 times (Fig 4D, S4C). In all conditions, GPMVs displayed a characteristic distribution of
231
permeable vesicles, suggesting that pipette shear during handling had little effect on GPMV permeability.
232
Additionally, further pipetting after removal from the plate did not affect permeability (Fig S7). We also
233
minimal mechanical stress conditions, by transferring the vesicles one time using 1000 µL pipette tips from
234
which the ends were cut to minimize shear stress on the vesicles. Vesicles isolated in this fashion were
235
similar to those pipetted many times using small bore pipettes (Fig 4D), suggesting that the mechanical
236
perturbations during transferring were not inducing permeation.
237 238
GPMVs become permeable upon mechanical rupture from cells
239
During the course of these investigations, we observed that plates of GPMV-forming cells that were
240
mechanically jostled during preparation contained many fewer cell-attached GPMVs and many more
241
permeable GPMVs. These observations suggested that shear from flow of the bulk solution surrounding
242
GPMVs may be inducing permeability. To test this hypothesis directly, GPMV production was imaged
243
while cells were immobilized on the microscope stage. As previously observed, unperturbed cells
244
produced exclusively sealed GPMVs, most of which remained attached to the cells. Immediately after
245
observation, the same plate of cells was jostled to induce flow of the bulk solution. After this perturbation,
246
the previously sealed population of GPMVs now contained a significant subset of permeable, free-floating
247
vesicles (Fig. 5A).
248 249
Given that extensive pipetting of isolated GPMVs did not induce permeation but bulk fluid flow around
250
blebbing cells did, it is unlikely that shear forces from fluid flow damage intact isolated GPMVs. Rather, it
251
seems more likely that shear flow results in mechanical rupture of GPMVs from cells, and it is this rupture
252
event that results in the loss of bilayer integrity responsible for permeation. To confirm this inference, we
253
compared the permeability of GPMVs produced in a standard “open” dish to a sealed chamber designed
254
to prevent bulk fluid flow around the cells (Fig. 5B). In contrast to the “open” configuration where GPMVs
255
became permeable upon inducing movement of bulk solution, GPMVs produced in the sealed chamber
256
remained uniformly impermeable regardless of mechanical perturbation of the plate (Fig. 5D and S5B).
257
When such sealed chambers were inverted, detached GPMVs settled to the bottom coverslip allowing
258
them to be visualized separately from cells (Fig. 6A). Importantly, these GPMVs were almost exclusively
259
sealed (Fig. 6B), offering an effective method to visualize sealed GPMVs.
260 261
DISCUSSION
262 263
Here, we characterize the surprising observation that GPMVs, a widely used plasma membrane model,
264
are passively permeable to macromolecular solutes. We show that GPMVs are permeable to solutes as
265
large as 40 kDa and that they remain permeable after their isolation, regardless of the specific procedure
266
used to isolate them. Leakiness has been previously reported in other cell blebs (34). We observe a clear
267
solute size-dependence, with the majority of vesicles being permeable to solutes below 10 kDa, and
268
decreasing fractions being permeable to larger solutes up to a maximal cutoff of ~60 kDa. These
269
observations combined suggest stable, somewhat variably sized, hydrophilic pores as the most likely
270
potential mechanism underlying GPMV permeation.
271 272
Such pores are generally not observed in pure lipid model membranes, as confirmed here in GUVs (Fig.
273
1B-C), prompting the question of how they are formed and why they are stable in GPMVs. Our
274
observations suggest a plausible model wherein shear stress induced by bulk fluid flow mechanically
275
ruptures the GPMV away from the cell plasma membrane. This rupture may induce a membrane pore that
276
can be stabilized by amphiphilic proteins in the GPMV lumen or unfolding of GPMV membrane proteins.
277
Consistent with this inference, detachment of vesicles in the absence of mechanical perturbation yields
278
GPMVs that are almost completely sealed (Fig. 5D and Fig. 6).
279 280
How do some vesicles detach from the PM without forming a pore? One possibility involves protein
281
machinery whose specific function is to mediate defect-free fission of membranes with the relevant
282
topology, namely the endosomal sorting complexes required for transport (ESCRT) complex (35). We
283
speculate that in some cells undergoing blebbing, ESCRT components assemble at the GPMV neck and
284
catalyze scission, in analogy with the formation of intraluminal vesicles in late endosomes, microvesicles
285
at the PM, or viruses budding from the PM (36). If this machinery is preempted by mechanical rupture,
286
then a stable pore may form. This hypothesis could be tested directly by imaging for the presence of core
287
ESCRT components at the neck of the GPMV bud (35). Alternatively, abrogation of ESCRT machinery
288
should prevent formation of sealed GPMVs entirely.
289 290
These findings are of particular relevance to studies using GPMVs for drug encapsulation (27, 28),
291
therapeutic targeting (26, 27), and studies of membrane permeation (26, 29–32). Another important
292
consideration is whether observations of protein binding to GPMVs may be biased by some
293
probes/interactors leaking into the vesicle lumen. Our suggestion for possible approaches to avoid
294
leakage in GPMVs may be of value therein.
295 296
Author Contributions
297
AS and IL designed the studies and wrote the paper. AS performed almost all experimental work, KRL
298
performed control experiments for revision.
299 300
Acknowledgements
301
All fluorescence microscopy was performed at the Center for Advanced Microscopy, Department of
302
Integrative Biology & Pharmacology at McGovern Medical School, UTHealth. We gratefully acknowledge
303
the members of the Levental lab for extensive discussions and support on this manuscript. Funding for this
304
work was provided by the NIH/National Institute of General Medical Sciences (GM114282, GM124072,
305
GM120351, GM134949), the Volkswagen Foundation (grant 93091), and the Human Frontiers Science
306
Program (RGP0059/2019). Authors have no competing interests.
307 308
Figure Legends
309 310 311 312 313 314 315 316 317 318
Figure 1. Large solutes permeate GPMVs. (A) Isolated GPMVs exposed to FITC-modified 3 kDa dextran (FITCdextran) and Alexa-Fluor-modified ATP analog (ATP-647) exhibit accumulation of probes in the intravesicular space of some vesicles. (B) In contrast, GUVs are not permeable to the same probes. This effect is quantified in (C) where points represent relative intensities of individual vesicles in a given preparation. A robust population of permeable GPMVs (defined by a relative intensity > 0.7; see Methods) is observed, while almost all GUVs remain sealed. (D) Permeable GPMVs fill with both probes proportionally and in all standard preparations two populations emerge: sealed and permeable vesicles. (E) While the percentage of permeable vesicles varies considerably in repeat preparations, on average around 65% of isolated GPMVs are permeable. Here each data point represents the percentage of permeable vesicles in a given preparation of GPMVs.
319 320 321 322 323 324 325
Figure 2. Large solutes permeate GPMVs rapidly. (A) Time-lapse imaging of GPMVs exposed to hydrophilic probes revealed that vesicles that fill with solutes in approximately one minute. (B) While FITC-dextran signal quickly levels out after this time, ATP-647 signal continues to accumulate, likely through interactions with ATP-binding proteins within vesicles. Dotted lines indicate time to vesicle permeation (reaching a relative vesicle intensity of 0.7) as calculated via one phase exponential association kinetics. Time to filling is 55 s for FITC-dextran and 49 s for ATP-647. Scale bar is 5 ߤ m.
326 327 328 329 330 331 332 333 334
Figure 3. Large solutes permeate GPMVs via stable pores. When exposed to different molecular weight fluorescent dextrans, GPMV permeation showed a clear size dependence. (A) Vesicles that were permeable to 3 kDa dextran nearly completely excluded 60 kDa dextran. (B) Lower molecular weight dextrans permeated GPMVs with decreasing frequency up to 60 kDa, suggesting existence of pores of heterogeneous sizes. Each data point represents a population of GPMVs from the same preparation incubated with differently sized dextrans. (C) Vesicles that were prepared in the presence of FITC-dextran only were exposed to ATP-647 30 mins after their isolation. Vesicles that were permeable to FITC-dextran at isolation were also permeable to ATP-647 after 30 mins suggesting pores remain stable for extended periods. Scale bars are 10 ߤm.
335 336 337 338 339 340 341 342 343 344 345 346 347
Figure 4. GPMVs become permeable upon isolation independent of temperature cycling or pipetting stress. (A) Time-lapse of cells during GPMV production indicate GPMVs are not permeable immediately upon their budding or during their growth. Both blebbing (x) and detached (arrow) vesicles remained impermeable to both fluorescent probes. (B) GPMVs isolated from cells were compared to those that were left undisturbed alongside the cells they formed from over the course of 2.5 hours. While undisturbed GPMVs remained impermeable, the isolated samples had a distribution of permeable and impermeable vesicles that remained consistent throughout time. (C) Keeping o o temperature constant between formation and isolation by isolating GPMVs at 37 C as opposed to 22 C did not affect the permeation induced by isolation, suggesting that temperature changes during GPMV handling were not responsible for permeation. (D) Changing shear strain by pipetting with different pipette tip bores did not affect the population of permeable vesicles suggesting pipetting is not responsible for eliciting permeation. Asterisk indicates pipette tip was cut to increase bore. All results (B-D) repeated in the presence of ATP-647 as demonstrated in the supplement. Scale bar is 10 ߤ m.
348 349 350 351 352 353
Figure 5. GPMVs become permeable during shear-induced detachment from cells. (A) When a plate of cells forming GPMVs is jostled, shear stress from movement of the bulk solution is induced and a previously sealed population of GPMVs becomes permeable. (B) GPMVs forming in this manner (“open” configuration) become permeable due to shear flow, likely by rupture from cells. In contrast, GPMV formation can be induced in a sealed chamber (“chambered” configuration), which eliminates shear forces from movement of the bulk solution. Chambered GPMVs remain sealed despite jostling of the plate. (C) GPMVs forming in the “open” configuration became
354 355 356 357
permeable upon generating shear from movement of the bulk solution. (D) In contrast, GPMVs forming in the chambered configuration remain impermeable regardless of bulk fluid movement. In (C-D) data points represent individual vesicles in a preparation of GPMVs. Scale bar is 10 ߤ m.
358 359 360 361 362
Figure 6. GPMVs formed in shear-protected conditions remain sealed. (A) GPMVs formed in a shear-protected chamber can be separated from cells by inverting the chamber. Because of their density, detached GPMVs sink towards the bottom coverslip allowing them to be visualized in isolation from the cells. (B) Such GPMVs remain sealed. Scale bar is 10 ߤ m.
363 364 365
REFERENCES
366
1.
Baumgart, T., A.T. Hammond, P. Sengupta, S.T. Hess, D.A. Holowka, B.A. Baird, and W.W. Webb.
367
2007. Large-scale fluid/fluid phase separation of proteins and lipids in giant plasma membrane
368
vesicles. Proc. Natl. Acad. Sci. U. S. A. 104: 3165–3170.
369
2.
Sezgin, E., H.-J. Kaiser, T. Baumgart, P. Schwille, K. Simons, and I. Levental. 2012. Elucidating
370
membrane structure and protein behavior using giant plasma membrane vesicles. Nat. Protoc. 7:
371
1042–1051.
372
3.
373 374
Levental, K.R., and I. Levental. 2015. Giant plasma membrane vesicles: models for understanding membrane organization. Curr. Top. Membr. 75: 25–57.
4.
Sengupta, P., A. Hammond, D. Holowka, and B. Baird. 2008. Structural determinants for partitioning
375
of lipids and proteins between coexisting fluid phases in giant plasma membrane vesicles. Biochim.
376
Biophys. Acta. 1778: 20–32.
377
5.
Kaiser, H.-J., H. -J. Kaiser, D. Lingwood, I. Levental, J.L. Sampaio, L. Kalvodova, L. Rajendran, and
378
K. Simons. 2009. Order of lipid phases in model and plasma membranes. Proceedings of the National
379
Academy of Sciences. 106: 16645–16650.
380
6.
381 382
Veatch, S.L., P. Cicuta, P. Sengupta, A. Honerkamp-Smith, D. Holowka, and B. Baird. 2008. Critical fluctuations in plasma membrane vesicles. ACS Chem. Biol. 3: 287–293.
7.
Moreno-Pescador, G., C.D. Florentsen, H. Østbye, S.L. Sønder, T.L. Boye, E.L. Veje, A.K. Sonne, S.
383
Semsey, J. Nylandsted, R. Daniels, and P.M. Bendix. 2019. Curvature- and Phase-Induced Protein
384
Sorting Quantified in Transfected Cell-Derived Giant Vesicles. ACS Nano. 13: 6689–6701.
385
8.
Schneider, F., D. Waithe, M.P. Clausen, S. Galiani, T. Koller, G. Ozhan, C. Eggeling, and E. Sezgin.
386
2017. Diffusion of lipids and GPI-anchored proteins in actin-free plasma membrane vesicles
387
measured by STED-FCS. Mol. Biol. Cell. 28: 1507–1518.
388
9.
Worch, R., Z. Petrášek, P. Schwille, and T. Weidemann. 2017. Diffusion of Single-Pass
389
Transmembrane Receptors: From the Plasma Membrane into Giant Liposomes. J. Membr. Biol. 250:
390
393–406.
391
10. Lewis, J.D., A.L. Caldara, S.E. Zimmer, S.N. Stahley, A. Seybold, N.L. Strong, A.S. Frangakis, I.
392
Levental, J.K. Wahl 3rd, A.L. Mattheyses, T. Sasaki, K. Nakabayashi, K. Hata, Y. Matsubara, A.
393
Ishida-Yamamoto, M. Amagai, A. Kubo, and A.P. Kowalczyk. 2019. The desmosome is a mesoscale
394
lipid raft-like membrane domain. Mol. Biol. Cell. 30: 1390–1405.
395
11. Wei, X., H. Song, L. Yin, M.G. Rizzo, R. Sidhu, D.F. Covey, D.S. Ory, and C.F. Semenkovich. 2016.
396
Fatty acid synthesis configures the plasma membrane for inflammation in diabetes. Nature. 539: 294–
397
298.
398
12. Garten, M., L.D. Mosgaard, T. Bornschlögl, S. Dieudonné, P. Bassereau, and G.E.S. Toombes. 2017.
399
Whole-GUV patch-clamping. Proceedings of the National Academy of Sciences. 114: 328–333.
400
13. Sedgwick, A., M. Olivia Balmert, and C. D’Souza-Schorey. 2018. The formation of giant plasma
401
membrane vesicles enable new insights into the regulation of cholesterol efflux. Experimental Cell
402
Research. 365: 194–207.
403 404
14. Lin, X., A.A. Gorfe, and I. Levental. 2018. Protein Partitioning into Ordered Membrane Domains: Insights from Simulations. Biophys. J. 114: 1936–1944.
405
15. Diaz-Rohrer, B.B., K.R. Levental, K. Simons, and I. Levental. 2014. Membrane raft association is a
406
determinant of plasma membrane localization. Proc. Natl. Acad. Sci. U. S. A. 111: 8500–8505.
407
16. Sezgin, E., I. Levental, M. Grzybek, G. Schwarzmann, V. Mueller, A. Honigmann, V.N. Belov, C.
408
Eggeling, U. Coskun, K. Simons, and P. Schwille. 2012. Partitioning, diffusion, and ligand binding of
409
raft lipid analogs in model and cellular plasma membranes. Biochim. Biophys. Acta. 1818: 1777–
410
1784.
411
17. Lorent, J.H., B. Diaz-Rohrer, X. Lin, K. Spring, A.A. Gorfe, K.R. Levental, and I. Levental. 2017.
412
Structural determinants and functional consequences of protein affinity for membrane rafts. Nature
413
Communications. 8.
414
18. Zhou, Y., K.N. Maxwell, E. Sezgin, M. Lu, H. Liang, J.F. Hancock, E.J. Dial, L.M. Lichtenberger, and I.
415
Levental. 2013. Bile acids modulate signaling by functional perturbation of plasma membrane
416
domains. J. Biol. Chem. 288: 35660–35670.
417 418 419
19. Gray, E., J. Karslake, B.B. Machta, and S.L. Veatch. 2013. Liquid general anesthetics lower critical temperatures in plasma membrane vesicles. Biophys. J. 105: 2751–2759. 20. Gray, E.M., G. Díaz-Vázquez, and S.L. Veatch. 2015. Growth Conditions and Cell Cycle Phase
420
Modulate Phase Transition Temperatures in RBL-2H3 Derived Plasma Membrane Vesicles. PLoS
421
One. 10: e0137741.
422
21. Johnson, S.A., B.M. Stinson, M.S. Go, L.M. Carmona, J.I. Reminick, X. Fang, and T. Baumgart. 2010.
423
Temperature-dependent phase behavior and protein partitioning in giant plasma membrane vesicles.
424
Biochim. Biophys. Acta. 1798: 1427–1435.
425
22. Levental, K.R., M.A. Surma, A.D. Skinkle, J.H. Lorent, Y. Zhou, C. Klose, J.T. Chang, J.F. Hancock,
426
and I. Levental. 2017. ω-3 polyunsaturated fatty acids direct differentiation of the membrane
427
phenotype in mesenchymal stem cells to potentiate osteogenesis. Sci Adv. 3: eaao1193.
428
23. Levental, K.R., J.H. Lorent, X. Lin, A.D. Skinkle, M.A. Surma, E.A. Stockenbojer, A.A. Gorfe, and I.
429
Levental. 2016. Polyunsaturated Lipids Regulate Membrane Domain Stability by Tuning Membrane
430
Order. Biophys. J. 110: 1800–1810.
431
24. Levental, K.R., E. Malmberg, Y.-Y. Fan, R. Chapkin, R. Ernst, and I. Levental. Homeostatic
432
remodeling of mammalian membranes in response to dietary lipids is essential for cellular fitness.
433
25. Yang, S.-T., A.J.B. Kreutzberger, V. Kiessling, B.K. Ganser-Pornillos, J.M. White, and L.K. Tamm.
434
2017. HIV virions sense plasma membrane heterogeneity for cell entry. Sci Adv. 3: e1700338.
435
26. Manni, M.M., J. Sot, and F.M. Goñi. 2015. Interaction of Clostridium perfringens epsilon-toxin with
436
biological and model membranes: A putative protein receptor in cells. Biochim. Biophys. Acta. 1848:
437
797–804.
438
27. Patel, J.M., V.F. Vartabedian, E.N. Bozeman, B.E. Caoyonan, S. Srivatsan, C.D. Pack, P. Dey, M.J.
439
D’Souza, L. Yang, and P. Selvaraj. 2016. Plasma membrane vesicles decorated with glycolipid-
440
anchored antigens and adjuvants via protein transfer as an antigen delivery platform for inhibition of
441
tumor growth. Biomaterials. 74: 231–244.
442 443 444
28. Gadok, A.K., D.J. Busch, S. Ferrati, B. Li, H.D.C. Smyth, and J.C. Stachowiak. 2016. Connectosomes for Direct Molecular Delivery to the Cellular Cytoplasm. J. Am. Chem. Soc. 138: 12833–12840. 29. Pae, J., P. Säälik, L. Liivamägi, D. Lubenets, P. Arukuusk, Ü. Langel, and M. Pooga. 2014.
445
Translocation of cell-penetrating peptides across the plasma membrane is controlled by cholesterol
446
and microenvironment created by membranous proteins. J. Control. Release. 192: 103–113.
447
30. Säälik, P., A. Niinep, J. Pae, M. Hansen, D. Lubenets, Ü. Langel, and M. Pooga. 2011. Penetration
448
without cells: membrane translocation of cell-penetrating peptides in the model giant plasma
449
membrane vesicles. J. Control. Release. 153: 117–125.
450
31. Dubavik, A., E. Sezgin, V. Lesnyak, N. Gaponik, P. Schwille, and A. Eychmüller. 2012. Penetration of
451 452 453 454 455 456
amphiphilic quantum dots through model and cellular plasma membranes. ACS Nano. 6: 2150–2156. 32. Faust, J.E., P.-Y. Yang, and H.W. Huang. 2017. Action of Antimicrobial Peptides on Bacterial and Lipid Membranes: A Direct Comparison. Biophys. J. 112: 1663–1672. 33. Dimitrov, D.S., and M.I. Angelova. 1988. Lipid swelling and liposome formation mediated by electric fields. Bioelectrochemistry and Bioenergetics. 19: 323–336. 34. Sarabipour, S., R.B. Chan, B. Zhou, G. Di Paolo, and K. Hristova. 2015. Analytical characterization of
457
plasma membrane-derived vesicles produced via osmotic and chemical vesiculation. Biochimica et
458
Biophysica Acta (BBA) - Biomembranes. 1848: 1591–1598.
459
35. Jimenez, A.J., P. Maiuri, J. Lafaurie-Janvore, S. Divoux, M. Piel, and F. Perez. 2014. ESCRT
460
Machinery Is Required for Plasma Membrane Repair. Science. 343: 1247136–1247136.
461 462 463
36. McDonald, B., and J. Martin-Serrano. 2009. No strings attached: the ESCRT machinery in viral budding and cytokinesis. J. Cell Sci. 122: 2167–2177.
A
FITC-dextran
ATP-647
Merge
Isolated GPMVs in dye solution
DiI
10 𝝁m
5 𝝁m
B
FITC-dextran
Merge
ATP-647
5 𝝁m 5 𝝁m
GUVs in dye solution
DiI
ATP-647 rel. vesicle intensity
rel. vesicle intensity
1.5 1.0 0.5 0.0 GUVs GPMVs FITC-dextran ATP-647
1.2
Isolated GPMVs permeable
1.0 FITC-dextran 0.8 ATP-647 0.6 0.4
sealed
0.4 0.6 0.8 1.0 1.2 100 FITC-dextran 80 vesicle intensity rel.
E % permeable vesicles
D
2.0
esicles
C
100 80 60 40 20 0
GPMV preparations FITC-dextran FITC-dextran ATP-647 ATP-647
DiI
ATP-647
0s
30 s
60 s
150 s
B FITC-dextran ATP-647
1.5 rel. vesicle intensity
FITC-dextran
A
1.0
0.5
0.0 0
50
100 150 time (s)
200
250
60 kDa dextran
B
Merge
10 𝝁m
10 𝝁m
% permeable vesicles
3 kDa dextran
Isolated GPMVs
A
60 40 20 0 1
C
DiI
FITC-dextran
ATP-647
After 30 mins ATP-647 introduced
10 100 dextran size (kDa)
Merge
Time from induction of blebbing (min)
A 20
60
40
GPMVs forming in dye solution
30
0.5
1.0 0.5
1.5
isolated 2.0 undisturbed 1.5
1.0
1.0
2.0
0.5
0.5
D FITC-dextran rel. vesicle intensity
1.0
C
FITC-dextran rel. vesicle intensity
1.5
2.0 isolated undisturbed 1.5
FITC-Dextran rel. vesicle intensity
2.0
FITC-dextran rel. vesicle intensity
FITC-dextran rel. vesicle intensity
B
X 2.0 isolated isolated undisturbed undisturbed 1.5 1.0 0.5
0.0 0.0 0.0 0.0 *p1000 50 60 90 120 150 180 50 60 90 120p200 150 180 50 60 90 12037150 180 o o 22 time from induction of blebbing (min) time from induction of blebbing (min) time from induction of blebbing (min) o pipette tip bore temperature ( C)
0.0
A
FITC-dextran
ATP-647
Bulk shear induced
Undisturbed
DiI
After inducing bulk shear
B “Open”
FITC-dextran rel. vesicle intensity
1.5
“Open”
D 1.5 FITC-dextran rel. vesicle intensity
C
“Chambered”
1.0 0.5 0.0 before shear
after shear
“Chambered”
1.0 0.5 0.0 before shear
after shear
Merge
A
B DiI
FITC-dextran
ATP-647
Merge
S0006-3495(20)30068-0
DOI:
https://doi.org/10.1016/j.bpj.2019.12.040
Reference:
BPJ 10274
To appear in:
Biophysical Journal
Received Date: 9 August 2019 Accepted Date: 23 December 2019
Please cite this article as: Skinkle A, Levental K, Levental I, Cell-derived plasma membrane vesicles are permeable to hydrophilic macromolecules, Biophysical Journal (2020), doi: https://doi.org/10.1016/ j.bpj.2019.12.040. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Biophysical Society.
1
TITLE: Cell-derived plasma membrane vesicles are permeable to hydrophilic macromolecules
2
AUTHORS: AD Skinkle1,2, KR Levental1, I Levental1,*
3
AFFILIATIONS:
4
1
5
Pharmacology, Houston, Texas 77030
6
2
7
*
University of Texas Health Science Center at Houston, Department of Integrative Biology and
University of North Carolina at Chapel Hill, Biological and Biomedical Sciences Program
corresponding author; [email protected]
8 9
ABSTRACT: Giant Plasma Membrane Vesicles (GPMVs) are a widely used experimental platform for
10
biochemical and biophysical analysis of isolated mammalian plasma membranes (PM). A core advantage
11
of these vesicles is that they maintain the native lipid and protein diversity of the plasma membrane while
12
affording the experimental flexibility of synthetic giant vesicles. In addition to fundamental investigations of
13
PM structure and composition, GPMVs have been used to evaluate the binding of proteins and small
14
molecules to cell-derived membranes, and the permeation of drug-like molecules through them. An
15
important assumption of such experiments is that GPMVs are sealed; i.e. that permeation occurs by
16
diffusion through the hydrophobic core rather than through hydrophilic pores. Here we demonstrate that
17
this assumption is often incorrect. We find that most GPMVs isolated using standard preparations are
18
passively permeable to various hydrophilic solutes as large as 40 kDa, in contrast to synthetic Giant
19
Unilamellar Vesicles (GUVs). We attribute this leakiness to stable, relatively large, and heterogeneous
20
pores formed by rupture of vesicles from cells. Finally, we identify preparation conditions that minimize
21
poration and allow evaluation of sealed GPMVs. These unexpected observations of GPMV poration
22
important for interpreting experiments utilizing GPMVs as plasma membrane models, particularly for drug
23
permeation and membrane asymmetry.
24 25
STATEMENT OF SIGNIFICANCE: A central assumption in using Giant Plasma Membrane Vesicles to
26
study membrane penetration and interactions is that these vesicles maintain the permeability barrier of the
27
native membrane from which they originate. Using large fluorescently-labeled hydrophilic probes, we
28
demonstrate that this assumption is often incorrect and conclude that macromolecular solutes permeate
29
GPMVs through stable pores formed during shear-induced rupture of vesicles from cells. Using these
30
insights into the mechanisms of poration, we demonstrate an approach to isolate sealed GPMVs.
31 32
INTRODUCTION
33 34
Model membranes are among the most widely used tools to investigate the physicochemical behaviors of
35
biological lipids, interactions between lipids and proteins, and the properties of lipid assemblies. While
36
synthetic lipid membranes effectively recapitulate certain aspects of biological membranes, it is not
37
experimentally feasible to construct model systems with the lipid diversity and protein density of living
38
cells. In this context, an important development in membrane biophysics has been the discovery (1) and
39
characterization of Giant Plasma Membrane Vesicles (GPMVs), which are isolated from live mammalian
40
cells as intact PM vesicles (2). These GPMVs fill a unique niche, maintaining the approximate lipid and
41
protein content of living membranes while offering the experimental tractability of synthetic models (3).
42
This combination has allowed investigation of PM composition (4), physical properties (5–7), dynamics (8,
43
9), and potentially signaling and transport (8, 10–13) in isolation from the complexity of living cells.
44 45
Because of these advantages, and despite several important caveats distinguishing isolated vesicles from
46
living membranes (3), GPMVs have become an important and widely used tool for membrane biophysics.
47
Perhaps their most notable feature is their spontaneous separation into coexisting liquid-ordered and
48
disordered membrane phases, allowing investigations into the properties and compositions of ordered
49
membrane domains, as models for lipid rafts (14–21). These vesicles have also been used in combination
50
with mass spectrometric lipidomics to characterize the detailed and comprehensive lipid profiles of PMs
51
and their remodeling during various cellular processes (22–24). Recently, GPMVs have also been
52
employed to study mechanisms of viral binding and fusion to the PM (25), as well as toxin binding (26).
53 54
In addition to their considerable utility for basic research, GPMVs have recently gained attention for
55
possible translational applications, particularly for their potential as drug delivery vehicles (27, 28) or for
56
studies of membrane permeability and poration (29–31). An important premise for these and other similar
57
studies is that GPMVs are sealed to extravesicular compounds. This assumption stems from analogy to
58
synthetic Giant Unilamellar Vesicles (GUVs), which are unambiguously sealed to permeation of
59
hydrophilic solutes (26, 32). Here, we report the surprising finding that GPMVs are permeable to large,
60
hydrophilic macromolecules. Probes as large as 40 kDa rapidly accumulate inside GPMVs, apparently
61
through stable aqueous pores. We investigate the mechanism for this surprising permeability and suggest
62
that stable pores are formed during shear-induced rupture of GPMVs from cells. Based on these
63
observations, we develop a technique to limit pore formation and image sealed GPMVs.
64 65
MATERIALS AND METHODS
66 67
Materials: The following lipids were purchased from Avanti Polar Lipids (Alabaster, AL) and used without
68
further purification: dipalmitoyl phosphatidylcholine (DPPC), dioleoyl PC (DOPC) and cholesterol. Fast DiI
69
(Life Technologies, Carlsbad, CA), fluorescein- and TMR-labeled dextrans (Sigma, St. Louis, MO),
70
AlexaFluor-647-ATP (Fisher Scientific, Hampton, NH). Other chemicals, buffers, and media were
71
purchased from Sigma.
72 73
Cell culture: Rat basophilic leukemia (RBL) cells were maintained in medium containing 60% modified
74
Eagle’s medium (MEM), 30% RPMI, 10% fetal calf serum, 2 mM glutamine, 100 units/mL penicillin, and
75
100 µg/mL streptomycin at 37oC in humidified 5% CO2.
76
77
GPMV isolation and imaging: GPMVs were prepared as previously described (2). Briefly, cells were
78
washed in isolation buffer (10 mM HEPES, 150 mM NaCl, 2 mM CaCl2, pH 7.4) and then incubated with
79
isolation buffer supplemented with 25 mM paraformaldehyde (PFA) and 2 mM dithiothreitol (DTT) (or 2
80
mM N-ethylmaleimide (NEM) as indicated) for 1 h at 37oC (unless otherwise indicated). Before isolation,
81
cell membranes were labeled with 5 µg/mL of the membrane marker FAST DiI for 10 min on ice. After 1 h
82
incubation, the GPMV-rich cellular supernatant was transferred to a microcentrifuge tube, then
83
concentrated GPMVs were collected after 30 min sedimentation on ice. All imaging was done by confocal
84
microscopy on a Nikon A1R.
85 86
Quantification of GPMV leakiness: GPMVs were either prepared with (or incubated after isolation with)
87
250 nM AlexaFluor-647-ATP and/or 250 ng/mL fluorescein-labeled 3 kDa dextran. Confocal images were
88
then taken with the fluorescent molecules still in solution. In these images, the lumens of “sealed” vesicles
89
appear dark relative to the background fluorescence, whereas leaky vesicles have bright fluorescent
90
lumens (see Fig. 1). These images were analyzed using a custom-built algorithm in the open-source
91
software Cell Profiler. GPMVs were identified and masked as circular objects, then the median
92
fluorescence intensity (to eliminate edge-effects) of the vesicle lumen was calculated for each vesicle. The
93
fluorescence intensity of the bulk/background was calculated as the median intensity of all pixels outside
94
GPMVs. The “relative vesicle intensity” was calculated as the ratio of the intensity of the vesicle lumen to
95
the bulk intensity.
96 97
Although most “leaky” vesicles had an internal fluorescence that was approximately equal to bulk
98
fluorescence, some were slightly dimmer while others slightly brighter than the bulk fluorescence. Thus, a
99
“leaky” vesicle was defined as one whose intensity was at least 70% of bulk fluorescence (relative vesicle
100
intensity >0.7). Although this value is somewhat arbitrary, it was chosen because it generally segregated
101
the two populations of GPMVs that emerged in our analyses (e.g. Fig. 1D). Sealed vesicles also had non-
102
zero “relative vesicle intensities”, which would appear to suggest that some permeation still occurred.
103
Although we cannot absolutely rule out some leakage into apparently “dark” vesicles, it is much more likely
104
that the non-zero lumenal fluorescence comes from analyzing images with a very bright fluorescent
105
background. Even in confocal mode, some out-of-focus signal is detectable from above and below the
106
vesicle, and the magnitude of this signal depends on the size of the vesicle. Thus, there are no truly “dark”
107
pixels in our images, which prohibits accurate background subtraction. An example of this effect (non-zero
108
normalized intensity in putatively sealed vesicles) is shown in Fig. S1. Because of the large contrast
109
between the “sealed vesicle” lumens and the bright bulk fluorescence, and because most putatively sealed
110
vesicles had similar lumenal intensities, we believe they allow minimal permeation of the fluorescent
111
solutes.
112 113
Preparation of Giant Unilamellar Vesicles: GUVs were prepared by electroformation, as previously
114
described (33). Briefly, a lipid mixture of dipalmitoyl phosphatidylcholine (DPPC), dioleoyl PC (DOPC) and
115
cholesterol was prepared in chloroform with a molar ratio of 5:3:2 (lipids purchased from Avanti). The lipid
116
mixture was applied to Pt-wires in a custom electroformation chamber and solvent was evaporated in a
117
vacuum chamber. The electroformation chamber was filled with sucrose solution (0.1 M) and heated to
118
53°C. An alternating sinusoidal current was applied across the cell unit with 2.5 V, 10 Hz at 53 °C fo r 2
119
hours. GUVs were then tranferred to a microcentrifuge tube and maintained at room temperature, before
120
imaging and analysis as described above for GPMVs. We observed no effect of liquid-liquid phase
121
separation in GUVs or GPMVs on their leakage (Fig. S6).
122 123
RESULTS AND DISCUSSION
124 125
Isolated GPMVs are permeable to hydrophilic macromolecules
126 127
GPMVs are widely used models of cellular plasma membranes and as such have been presumed to retain
128
the barrier function of the native PM. This assumption has been reinforced by GPMVs’ superficial
129
resemblance to synthetic GUVs, which generally do not have large hydrophilic pores. Surprisingly, we
130
observed that cell-detached GPMVs exhibited considerable heterogeneity in their permeability to
131
extravesicular solutes in the form of fluorescently-labeled hydrophilic probes. When GPMVs were
132
prepared in the presence of an Alexa647-modified ATP analog (2 kDa) or FITC-modified dextran (3 kDa),
133
a major fraction of isolated vesicles showed accumulation of these probes in the intravesicular space (Fig.
134
1A), a phenomenon hereafter referred to as permeation.
135 136
While fluorescent dextran was generally slightly less concentrated in the lumen of permeable GPMVs
137
compared to the bulk extravesicular solution, fluorescent ATP was often clearly enriched (Fig. 1A inset).
138
We presume this enrichment reflects the interaction with intravesicular proteins. In striking contrast,
139
synthetic giant unilamellar vesicles (GUVs) prepared by electroformation were almost uniformly
140
impermeable to the same solutes, as exemplified in Fig. 1B and quantified in Fig. 1C. Within any given
141
isolated GPMV population, there was considerable variation in the accumulation of probes inside the
142
vesicle, ranging from impermeable to slightly brighter than the bulk fluorescence of the extravesicular
143
solution (Fig. 1D). Notably, all permeable GPMVs were similarly permeable to both fluorescent solutes,
144
while GPMVs that were impermeable to dextran were also impermeable to fluorescent ATP (Fig. 1D).
145
These observations allowed us to separate the GPMVs in a given preparation into clearly distinct sealed
146 147
and permeable populations (Fig. 1D).
148
The fraction of permeable GPMVs was somewhat variable between repeated preparations (the reasons
149
underlying this variability are discussed below); however, on average, the majority of vesicles were
150
permeable to both fluorescent solutes (Fig. 1E). GPMVs were similarly permeable to FITC-modified
151
dextran in the absence of fluorescent ATP (Fig. S2) revealing that permeation is not driven by the ATP
152
analogue. These observations were not dependent on the chemicals chosen for GPMV production, as
153
GPMVs prepared with N-ethylmaleimide (NEM) were similarly permeable to those prepared with
154
PFA+DTT (Fig. S3)
155 156
Time-lapse imaging revealed that permeation occurred abruptly, and after the formation of an initially
157
sealed cell-attached vesicle. Fig. 2A and Video S1 show the temporal evolution of one such vesicle, which
158
is sealed at an arbitrary starting point (approximately 30 mins after induction of vesiculation), but then
159
rapidly fills with both macromolecular solutes within approximately one minute of the permeation-inducing
160
event (the putative nature of the permeation-inducing event is discussed below). This permeation was not
161
associated with any change in the gross morphology of the vesicle or its membrane, as shown by the lipid
162
dye DiI (Fig. 2; bottom row). The kinetics of filling can be fit to a first-order rate constant, with both solutes
163
reaching the ‘permeability threshold’ (rel. vesicle intensity of 0.7) in <1 min. This rapid accumulation is
164
inconsistent with the permeation of large hydrophilic probes through an intact membrane hydrophobic
165
core, suggesting alternative permeation pathways. To test the mechanisms underlying these surprising
166
observations, we evaluated the size, temperature, time, and mechanical perturbation dependence of
167
GPMV permeability.
168 169
Large solutes permeate GPMVs via stable pores
170 171
To determine whether permeation through the GPMV membrane was occurring via well-defined pores, we
172
tested permeation to fluorescent dextrans of increasing molecular weight. We observed a clear solute size
173
dependence, with larger substrates permeating less efficiently. This dependence is demonstrated by
174
incubating a vesicle preparation with two differently sized dextran molecules labeled with two different
175
fluorophores. As shown in Fig. 3A, some vesicles are completely sealed (dark in both channels), while a
176
large fraction are completely permeable to 3 kDa dextran but not 60 kDa dextran (red only). This
177
dependence of permeation on size of the solute was observed up to ~40 kDa, with decreasing permeation
178
for larger substrates (Fig. 3B). However, the size-dependence was not as sharp as might be expected for
179
very consistent, homogeneous pores. In that case, we would expect a sharp cut-off at a particular
180
molecular weight, with all vesicles being permeable to smaller substrates and completely sealed to larger
181
ones. Instead, we observed a gradual decrease in the fraction of permeable vesicles to larger and larger
182
substrates, suggesting that a distribution of pore sizes exists in any given population of vesicles.
183 184
Finally, the possibility of stable pores was supported by observations that permeable GPMVs retained
185
their permeability well after isolation. GPMVs that were isolated in the presence of FITC-dextran alone
186
were incubated for 30 mins at 37oC, then exposed to fluorescent ATP. After this incubation, vesicles that
187
were permeable to one substrate were also permeable to the other, whereas vesicles that remained
188
impermeable to dextran after 30 mins were also sealed to the second substrate (Fig. 3C). These
189
observations suggest that stable pores (some of which can accommodate substrates up to 40 kDa) are
190
responsible for accumulation of solutes within GPMVs.
191
Cell-attached GPMVs are not permeable during their formation
192 193
To elucidate the conditions that caused stable pores in GPMVs, we first examined whether GPMVs were
194
permeable immediately upon their generation or became permeable due to handling during/after isolation.
195
GPMV formation was imaged from its initiation, allowing observation of vesicles as they began to bud,
196
inflated, then ultimately detached from cells. In these observations, both budding vesicles and almost all
197
detached vesicles remained sealed to the small molecule probes (Fig. 4A), revealing that cell-attached
198
GPMVs are generally not permeable but rather become leaky at some point after their generation or
199
isolation.
200 201
To test the possibility that GPMVs become permeable over time regardless of handling, we observed
202
GPMVs up to 2.5 hrs after isolation. We also compared GPMVs that were isolated from cells to those that
203
were left undisturbed in the original plate in which they formed. Isolated GPMVs were characteristically
204
heterogeneous in their accumulation of the extravesicular probes, with the majority being leaky to both
205
solutes (Fig. 4B and S4A). Notably, the general distribution of permeable vesicles remained consistent
206
throughout the three-hour time course, suggesting that the permeation-inducing event is not time-
207
dependent. In striking contrast, GPMVs that were not isolated from cells, but rather imaged without
208
manipulation in the same plate in which they formed, remained sealed throughout the time course, up to 3
209
h after induction of blebbing (Fig. 4B, S4A). These results suggested that the permeation-inducing event
210
occurs during the process of isolating GPMVs from the cells.
211 212
Neither temperature changes, phase separation, nor shear from pipetting contribute to GPMV
213
permeability
214 215
Given that GPMVs remain sealed until isolation from cells, we hypothesized that either temperature
216
changes or the mechanical stress of pipetting during isolation was responsible for eliciting vesicle
217
permeation. To test the effect of temperature changes, GPMVs were prepared at 37˚C (as usual) then
218
observed either at 37˚C or 23˚C. Both cases showed similar and characteristic distributions of permeable
219
vesicles, while samples that were left undisturbed remained sealed regardless of temperature changes
220
(Fig. 4C, S4B). Similarly, heating of the samples after isolation was not sufficient to “heal” pre-formed
221
pores. Thus, temperature changes occurring during isolation are insufficient to elicit permeation.
222 223
The permeability of GPMVs was also unaffected by their phase separation into Lo and Ld phases. We
224
observed both sealed and “leaky” vesicles in microscopically homogeneous and phase-separated
225
populations, with no notable difference in leakiness between them. GUVs remained sealed to
226
macromolecules regardless of whether they were mixed or phase-separated (Fig S6).
227
228
To test the possibility that the mechanical stress of pipetting causes lipid bilayer disruptions large enough
229
to allow permeation of solutes into the vesicle, GPMVs were isolated using pipette tips of varying bores
230
and pipetted >20 times (Fig 4D, S4C). In all conditions, GPMVs displayed a characteristic distribution of
231
permeable vesicles, suggesting that pipette shear during handling had little effect on GPMV permeability.
232
Additionally, further pipetting after removal from the plate did not affect permeability (Fig S7). We also
233
minimal mechanical stress conditions, by transferring the vesicles one time using 1000 µL pipette tips from
234
which the ends were cut to minimize shear stress on the vesicles. Vesicles isolated in this fashion were
235
similar to those pipetted many times using small bore pipettes (Fig 4D), suggesting that the mechanical
236
perturbations during transferring were not inducing permeation.
237 238
GPMVs become permeable upon mechanical rupture from cells
239
During the course of these investigations, we observed that plates of GPMV-forming cells that were
240
mechanically jostled during preparation contained many fewer cell-attached GPMVs and many more
241
permeable GPMVs. These observations suggested that shear from flow of the bulk solution surrounding
242
GPMVs may be inducing permeability. To test this hypothesis directly, GPMV production was imaged
243
while cells were immobilized on the microscope stage. As previously observed, unperturbed cells
244
produced exclusively sealed GPMVs, most of which remained attached to the cells. Immediately after
245
observation, the same plate of cells was jostled to induce flow of the bulk solution. After this perturbation,
246
the previously sealed population of GPMVs now contained a significant subset of permeable, free-floating
247
vesicles (Fig. 5A).
248 249
Given that extensive pipetting of isolated GPMVs did not induce permeation but bulk fluid flow around
250
blebbing cells did, it is unlikely that shear forces from fluid flow damage intact isolated GPMVs. Rather, it
251
seems more likely that shear flow results in mechanical rupture of GPMVs from cells, and it is this rupture
252
event that results in the loss of bilayer integrity responsible for permeation. To confirm this inference, we
253
compared the permeability of GPMVs produced in a standard “open” dish to a sealed chamber designed
254
to prevent bulk fluid flow around the cells (Fig. 5B). In contrast to the “open” configuration where GPMVs
255
became permeable upon inducing movement of bulk solution, GPMVs produced in the sealed chamber
256
remained uniformly impermeable regardless of mechanical perturbation of the plate (Fig. 5D and S5B).
257
When such sealed chambers were inverted, detached GPMVs settled to the bottom coverslip allowing
258
them to be visualized separately from cells (Fig. 6A). Importantly, these GPMVs were almost exclusively
259
sealed (Fig. 6B), offering an effective method to visualize sealed GPMVs.
260 261
DISCUSSION
262 263
Here, we characterize the surprising observation that GPMVs, a widely used plasma membrane model,
264
are passively permeable to macromolecular solutes. We show that GPMVs are permeable to solutes as
265
large as 40 kDa and that they remain permeable after their isolation, regardless of the specific procedure
266
used to isolate them. Leakiness has been previously reported in other cell blebs (34). We observe a clear
267
solute size-dependence, with the majority of vesicles being permeable to solutes below 10 kDa, and
268
decreasing fractions being permeable to larger solutes up to a maximal cutoff of ~60 kDa. These
269
observations combined suggest stable, somewhat variably sized, hydrophilic pores as the most likely
270
potential mechanism underlying GPMV permeation.
271 272
Such pores are generally not observed in pure lipid model membranes, as confirmed here in GUVs (Fig.
273
1B-C), prompting the question of how they are formed and why they are stable in GPMVs. Our
274
observations suggest a plausible model wherein shear stress induced by bulk fluid flow mechanically
275
ruptures the GPMV away from the cell plasma membrane. This rupture may induce a membrane pore that
276
can be stabilized by amphiphilic proteins in the GPMV lumen or unfolding of GPMV membrane proteins.
277
Consistent with this inference, detachment of vesicles in the absence of mechanical perturbation yields
278
GPMVs that are almost completely sealed (Fig. 5D and Fig. 6).
279 280
How do some vesicles detach from the PM without forming a pore? One possibility involves protein
281
machinery whose specific function is to mediate defect-free fission of membranes with the relevant
282
topology, namely the endosomal sorting complexes required for transport (ESCRT) complex (35). We
283
speculate that in some cells undergoing blebbing, ESCRT components assemble at the GPMV neck and
284
catalyze scission, in analogy with the formation of intraluminal vesicles in late endosomes, microvesicles
285
at the PM, or viruses budding from the PM (36). If this machinery is preempted by mechanical rupture,
286
then a stable pore may form. This hypothesis could be tested directly by imaging for the presence of core
287
ESCRT components at the neck of the GPMV bud (35). Alternatively, abrogation of ESCRT machinery
288
should prevent formation of sealed GPMVs entirely.
289 290
These findings are of particular relevance to studies using GPMVs for drug encapsulation (27, 28),
291
therapeutic targeting (26, 27), and studies of membrane permeation (26, 29–32). Another important
292
consideration is whether observations of protein binding to GPMVs may be biased by some
293
probes/interactors leaking into the vesicle lumen. Our suggestion for possible approaches to avoid
294
leakage in GPMVs may be of value therein.
295 296
Author Contributions
297
AS and IL designed the studies and wrote the paper. AS performed almost all experimental work, KRL
298
performed control experiments for revision.
299 300
Acknowledgements
301
All fluorescence microscopy was performed at the Center for Advanced Microscopy, Department of
302
Integrative Biology & Pharmacology at McGovern Medical School, UTHealth. We gratefully acknowledge
303
the members of the Levental lab for extensive discussions and support on this manuscript. Funding for this
304
work was provided by the NIH/National Institute of General Medical Sciences (GM114282, GM124072,
305
GM120351, GM134949), the Volkswagen Foundation (grant 93091), and the Human Frontiers Science
306
Program (RGP0059/2019). Authors have no competing interests.
307 308
Figure Legends
309 310 311 312 313 314 315 316 317 318
Figure 1. Large solutes permeate GPMVs. (A) Isolated GPMVs exposed to FITC-modified 3 kDa dextran (FITCdextran) and Alexa-Fluor-modified ATP analog (ATP-647) exhibit accumulation of probes in the intravesicular space of some vesicles. (B) In contrast, GUVs are not permeable to the same probes. This effect is quantified in (C) where points represent relative intensities of individual vesicles in a given preparation. A robust population of permeable GPMVs (defined by a relative intensity > 0.7; see Methods) is observed, while almost all GUVs remain sealed. (D) Permeable GPMVs fill with both probes proportionally and in all standard preparations two populations emerge: sealed and permeable vesicles. (E) While the percentage of permeable vesicles varies considerably in repeat preparations, on average around 65% of isolated GPMVs are permeable. Here each data point represents the percentage of permeable vesicles in a given preparation of GPMVs.
319 320 321 322 323 324 325
Figure 2. Large solutes permeate GPMVs rapidly. (A) Time-lapse imaging of GPMVs exposed to hydrophilic probes revealed that vesicles that fill with solutes in approximately one minute. (B) While FITC-dextran signal quickly levels out after this time, ATP-647 signal continues to accumulate, likely through interactions with ATP-binding proteins within vesicles. Dotted lines indicate time to vesicle permeation (reaching a relative vesicle intensity of 0.7) as calculated via one phase exponential association kinetics. Time to filling is 55 s for FITC-dextran and 49 s for ATP-647. Scale bar is 5 ߤ m.
326 327 328 329 330 331 332 333 334
Figure 3. Large solutes permeate GPMVs via stable pores. When exposed to different molecular weight fluorescent dextrans, GPMV permeation showed a clear size dependence. (A) Vesicles that were permeable to 3 kDa dextran nearly completely excluded 60 kDa dextran. (B) Lower molecular weight dextrans permeated GPMVs with decreasing frequency up to 60 kDa, suggesting existence of pores of heterogeneous sizes. Each data point represents a population of GPMVs from the same preparation incubated with differently sized dextrans. (C) Vesicles that were prepared in the presence of FITC-dextran only were exposed to ATP-647 30 mins after their isolation. Vesicles that were permeable to FITC-dextran at isolation were also permeable to ATP-647 after 30 mins suggesting pores remain stable for extended periods. Scale bars are 10 ߤm.
335 336 337 338 339 340 341 342 343 344 345 346 347
Figure 4. GPMVs become permeable upon isolation independent of temperature cycling or pipetting stress. (A) Time-lapse of cells during GPMV production indicate GPMVs are not permeable immediately upon their budding or during their growth. Both blebbing (x) and detached (arrow) vesicles remained impermeable to both fluorescent probes. (B) GPMVs isolated from cells were compared to those that were left undisturbed alongside the cells they formed from over the course of 2.5 hours. While undisturbed GPMVs remained impermeable, the isolated samples had a distribution of permeable and impermeable vesicles that remained consistent throughout time. (C) Keeping o o temperature constant between formation and isolation by isolating GPMVs at 37 C as opposed to 22 C did not affect the permeation induced by isolation, suggesting that temperature changes during GPMV handling were not responsible for permeation. (D) Changing shear strain by pipetting with different pipette tip bores did not affect the population of permeable vesicles suggesting pipetting is not responsible for eliciting permeation. Asterisk indicates pipette tip was cut to increase bore. All results (B-D) repeated in the presence of ATP-647 as demonstrated in the supplement. Scale bar is 10 ߤ m.
348 349 350 351 352 353
Figure 5. GPMVs become permeable during shear-induced detachment from cells. (A) When a plate of cells forming GPMVs is jostled, shear stress from movement of the bulk solution is induced and a previously sealed population of GPMVs becomes permeable. (B) GPMVs forming in this manner (“open” configuration) become permeable due to shear flow, likely by rupture from cells. In contrast, GPMV formation can be induced in a sealed chamber (“chambered” configuration), which eliminates shear forces from movement of the bulk solution. Chambered GPMVs remain sealed despite jostling of the plate. (C) GPMVs forming in the “open” configuration became
354 355 356 357
permeable upon generating shear from movement of the bulk solution. (D) In contrast, GPMVs forming in the chambered configuration remain impermeable regardless of bulk fluid movement. In (C-D) data points represent individual vesicles in a preparation of GPMVs. Scale bar is 10 ߤ m.
358 359 360 361 362
Figure 6. GPMVs formed in shear-protected conditions remain sealed. (A) GPMVs formed in a shear-protected chamber can be separated from cells by inverting the chamber. Because of their density, detached GPMVs sink towards the bottom coverslip allowing them to be visualized in isolation from the cells. (B) Such GPMVs remain sealed. Scale bar is 10 ߤ m.
363 364 365
REFERENCES
366
1.
Baumgart, T., A.T. Hammond, P. Sengupta, S.T. Hess, D.A. Holowka, B.A. Baird, and W.W. Webb.
367
2007. Large-scale fluid/fluid phase separation of proteins and lipids in giant plasma membrane
368
vesicles. Proc. Natl. Acad. Sci. U. S. A. 104: 3165–3170.
369
2.
Sezgin, E., H.-J. Kaiser, T. Baumgart, P. Schwille, K. Simons, and I. Levental. 2012. Elucidating
370
membrane structure and protein behavior using giant plasma membrane vesicles. Nat. Protoc. 7:
371
1042–1051.
372
3.
373 374
Levental, K.R., and I. Levental. 2015. Giant plasma membrane vesicles: models for understanding membrane organization. Curr. Top. Membr. 75: 25–57.
4.
Sengupta, P., A. Hammond, D. Holowka, and B. Baird. 2008. Structural determinants for partitioning
375
of lipids and proteins between coexisting fluid phases in giant plasma membrane vesicles. Biochim.
376
Biophys. Acta. 1778: 20–32.
377
5.
Kaiser, H.-J., H. -J. Kaiser, D. Lingwood, I. Levental, J.L. Sampaio, L. Kalvodova, L. Rajendran, and
378
K. Simons. 2009. Order of lipid phases in model and plasma membranes. Proceedings of the National
379
Academy of Sciences. 106: 16645–16650.
380
6.
381 382
Veatch, S.L., P. Cicuta, P. Sengupta, A. Honerkamp-Smith, D. Holowka, and B. Baird. 2008. Critical fluctuations in plasma membrane vesicles. ACS Chem. Biol. 3: 287–293.
7.
Moreno-Pescador, G., C.D. Florentsen, H. Østbye, S.L. Sønder, T.L. Boye, E.L. Veje, A.K. Sonne, S.
383
Semsey, J. Nylandsted, R. Daniels, and P.M. Bendix. 2019. Curvature- and Phase-Induced Protein
384
Sorting Quantified in Transfected Cell-Derived Giant Vesicles. ACS Nano. 13: 6689–6701.
385
8.
Schneider, F., D. Waithe, M.P. Clausen, S. Galiani, T. Koller, G. Ozhan, C. Eggeling, and E. Sezgin.
386
2017. Diffusion of lipids and GPI-anchored proteins in actin-free plasma membrane vesicles
387
measured by STED-FCS. Mol. Biol. Cell. 28: 1507–1518.
388
9.
Worch, R., Z. Petrášek, P. Schwille, and T. Weidemann. 2017. Diffusion of Single-Pass
389
Transmembrane Receptors: From the Plasma Membrane into Giant Liposomes. J. Membr. Biol. 250:
390
393–406.
391
10. Lewis, J.D., A.L. Caldara, S.E. Zimmer, S.N. Stahley, A. Seybold, N.L. Strong, A.S. Frangakis, I.
392
Levental, J.K. Wahl 3rd, A.L. Mattheyses, T. Sasaki, K. Nakabayashi, K. Hata, Y. Matsubara, A.
393
Ishida-Yamamoto, M. Amagai, A. Kubo, and A.P. Kowalczyk. 2019. The desmosome is a mesoscale
394
lipid raft-like membrane domain. Mol. Biol. Cell. 30: 1390–1405.
395
11. Wei, X., H. Song, L. Yin, M.G. Rizzo, R. Sidhu, D.F. Covey, D.S. Ory, and C.F. Semenkovich. 2016.
396
Fatty acid synthesis configures the plasma membrane for inflammation in diabetes. Nature. 539: 294–
397
298.
398
12. Garten, M., L.D. Mosgaard, T. Bornschlögl, S. Dieudonné, P. Bassereau, and G.E.S. Toombes. 2017.
399
Whole-GUV patch-clamping. Proceedings of the National Academy of Sciences. 114: 328–333.
400
13. Sedgwick, A., M. Olivia Balmert, and C. D’Souza-Schorey. 2018. The formation of giant plasma
401
membrane vesicles enable new insights into the regulation of cholesterol efflux. Experimental Cell
402
Research. 365: 194–207.
403 404
14. Lin, X., A.A. Gorfe, and I. Levental. 2018. Protein Partitioning into Ordered Membrane Domains: Insights from Simulations. Biophys. J. 114: 1936–1944.
405
15. Diaz-Rohrer, B.B., K.R. Levental, K. Simons, and I. Levental. 2014. Membrane raft association is a
406
determinant of plasma membrane localization. Proc. Natl. Acad. Sci. U. S. A. 111: 8500–8505.
407
16. Sezgin, E., I. Levental, M. Grzybek, G. Schwarzmann, V. Mueller, A. Honigmann, V.N. Belov, C.
408
Eggeling, U. Coskun, K. Simons, and P. Schwille. 2012. Partitioning, diffusion, and ligand binding of
409
raft lipid analogs in model and cellular plasma membranes. Biochim. Biophys. Acta. 1818: 1777–
410
1784.
411
17. Lorent, J.H., B. Diaz-Rohrer, X. Lin, K. Spring, A.A. Gorfe, K.R. Levental, and I. Levental. 2017.
412
Structural determinants and functional consequences of protein affinity for membrane rafts. Nature
413
Communications. 8.
414
18. Zhou, Y., K.N. Maxwell, E. Sezgin, M. Lu, H. Liang, J.F. Hancock, E.J. Dial, L.M. Lichtenberger, and I.
415
Levental. 2013. Bile acids modulate signaling by functional perturbation of plasma membrane
416
domains. J. Biol. Chem. 288: 35660–35670.
417 418 419
19. Gray, E., J. Karslake, B.B. Machta, and S.L. Veatch. 2013. Liquid general anesthetics lower critical temperatures in plasma membrane vesicles. Biophys. J. 105: 2751–2759. 20. Gray, E.M., G. Díaz-Vázquez, and S.L. Veatch. 2015. Growth Conditions and Cell Cycle Phase
420
Modulate Phase Transition Temperatures in RBL-2H3 Derived Plasma Membrane Vesicles. PLoS
421
One. 10: e0137741.
422
21. Johnson, S.A., B.M. Stinson, M.S. Go, L.M. Carmona, J.I. Reminick, X. Fang, and T. Baumgart. 2010.
423
Temperature-dependent phase behavior and protein partitioning in giant plasma membrane vesicles.
424
Biochim. Biophys. Acta. 1798: 1427–1435.
425
22. Levental, K.R., M.A. Surma, A.D. Skinkle, J.H. Lorent, Y. Zhou, C. Klose, J.T. Chang, J.F. Hancock,
426
and I. Levental. 2017. ω-3 polyunsaturated fatty acids direct differentiation of the membrane
427
phenotype in mesenchymal stem cells to potentiate osteogenesis. Sci Adv. 3: eaao1193.
428
23. Levental, K.R., J.H. Lorent, X. Lin, A.D. Skinkle, M.A. Surma, E.A. Stockenbojer, A.A. Gorfe, and I.
429
Levental. 2016. Polyunsaturated Lipids Regulate Membrane Domain Stability by Tuning Membrane
430
Order. Biophys. J. 110: 1800–1810.
431
24. Levental, K.R., E. Malmberg, Y.-Y. Fan, R. Chapkin, R. Ernst, and I. Levental. Homeostatic
432
remodeling of mammalian membranes in response to dietary lipids is essential for cellular fitness.
433
25. Yang, S.-T., A.J.B. Kreutzberger, V. Kiessling, B.K. Ganser-Pornillos, J.M. White, and L.K. Tamm.
434
2017. HIV virions sense plasma membrane heterogeneity for cell entry. Sci Adv. 3: e1700338.
435
26. Manni, M.M., J. Sot, and F.M. Goñi. 2015. Interaction of Clostridium perfringens epsilon-toxin with
436
biological and model membranes: A putative protein receptor in cells. Biochim. Biophys. Acta. 1848:
437
797–804.
438
27. Patel, J.M., V.F. Vartabedian, E.N. Bozeman, B.E. Caoyonan, S. Srivatsan, C.D. Pack, P. Dey, M.J.
439
D’Souza, L. Yang, and P. Selvaraj. 2016. Plasma membrane vesicles decorated with glycolipid-
440
anchored antigens and adjuvants via protein transfer as an antigen delivery platform for inhibition of
441
tumor growth. Biomaterials. 74: 231–244.
442 443 444
28. Gadok, A.K., D.J. Busch, S. Ferrati, B. Li, H.D.C. Smyth, and J.C. Stachowiak. 2016. Connectosomes for Direct Molecular Delivery to the Cellular Cytoplasm. J. Am. Chem. Soc. 138: 12833–12840. 29. Pae, J., P. Säälik, L. Liivamägi, D. Lubenets, P. Arukuusk, Ü. Langel, and M. Pooga. 2014.
445
Translocation of cell-penetrating peptides across the plasma membrane is controlled by cholesterol
446
and microenvironment created by membranous proteins. J. Control. Release. 192: 103–113.
447
30. Säälik, P., A. Niinep, J. Pae, M. Hansen, D. Lubenets, Ü. Langel, and M. Pooga. 2011. Penetration
448
without cells: membrane translocation of cell-penetrating peptides in the model giant plasma
449
membrane vesicles. J. Control. Release. 153: 117–125.
450
31. Dubavik, A., E. Sezgin, V. Lesnyak, N. Gaponik, P. Schwille, and A. Eychmüller. 2012. Penetration of
451 452 453 454 455 456
amphiphilic quantum dots through model and cellular plasma membranes. ACS Nano. 6: 2150–2156. 32. Faust, J.E., P.-Y. Yang, and H.W. Huang. 2017. Action of Antimicrobial Peptides on Bacterial and Lipid Membranes: A Direct Comparison. Biophys. J. 112: 1663–1672. 33. Dimitrov, D.S., and M.I. Angelova. 1988. Lipid swelling and liposome formation mediated by electric fields. Bioelectrochemistry and Bioenergetics. 19: 323–336. 34. Sarabipour, S., R.B. Chan, B. Zhou, G. Di Paolo, and K. Hristova. 2015. Analytical characterization of
457
plasma membrane-derived vesicles produced via osmotic and chemical vesiculation. Biochimica et
458
Biophysica Acta (BBA) - Biomembranes. 1848: 1591–1598.
459
35. Jimenez, A.J., P. Maiuri, J. Lafaurie-Janvore, S. Divoux, M. Piel, and F. Perez. 2014. ESCRT
460
Machinery Is Required for Plasma Membrane Repair. Science. 343: 1247136–1247136.
461 462 463
36. McDonald, B., and J. Martin-Serrano. 2009. No strings attached: the ESCRT machinery in viral budding and cytokinesis. J. Cell Sci. 122: 2167–2177.
A
FITC-dextran
ATP-647
Merge
Isolated GPMVs in dye solution
DiI
10 𝝁m
5 𝝁m
B
FITC-dextran
Merge
ATP-647
5 𝝁m 5 𝝁m
GUVs in dye solution
DiI
ATP-647 rel. vesicle intensity
rel. vesicle intensity
1.5 1.0 0.5 0.0 GUVs GPMVs FITC-dextran ATP-647
1.2
Isolated GPMVs permeable
1.0 FITC-dextran 0.8 ATP-647 0.6 0.4
sealed
0.4 0.6 0.8 1.0 1.2 100 FITC-dextran 80 vesicle intensity rel.
E % permeable vesicles
D
2.0
esicles
C
100 80 60 40 20 0
GPMV preparations FITC-dextran FITC-dextran ATP-647 ATP-647
DiI
ATP-647
0s
30 s
60 s
150 s
B FITC-dextran ATP-647
1.5 rel. vesicle intensity
FITC-dextran
A
1.0
0.5
0.0 0
50
100 150 time (s)
200
250
60 kDa dextran
B
Merge
10 𝝁m
10 𝝁m
% permeable vesicles
3 kDa dextran
Isolated GPMVs
A
60 40 20 0 1
C
DiI
FITC-dextran
ATP-647
After 30 mins ATP-647 introduced
10 100 dextran size (kDa)
Merge
Time from induction of blebbing (min)
A 20
60
40
GPMVs forming in dye solution
30
0.5
1.0 0.5
1.5
isolated 2.0 undisturbed 1.5
1.0
1.0
2.0
0.5
0.5
D FITC-dextran rel. vesicle intensity
1.0
C
FITC-dextran rel. vesicle intensity
1.5
2.0 isolated undisturbed 1.5
FITC-Dextran rel. vesicle intensity
2.0
FITC-dextran rel. vesicle intensity
FITC-dextran rel. vesicle intensity
B
X 2.0 isolated isolated undisturbed undisturbed 1.5 1.0 0.5
0.0 0.0 0.0 0.0 *p1000 50 60 90 120 150 180 50 60 90 120p200 150 180 50 60 90 12037150 180 o o 22 time from induction of blebbing (min) time from induction of blebbing (min) time from induction of blebbing (min) o pipette tip bore temperature ( C)
0.0
A
FITC-dextran
ATP-647
Bulk shear induced
Undisturbed
DiI
After inducing bulk shear
B “Open”
FITC-dextran rel. vesicle intensity
1.5
“Open”
D 1.5 FITC-dextran rel. vesicle intensity
C
“Chambered”
1.0 0.5 0.0 before shear
after shear
“Chambered”
1.0 0.5 0.0 before shear
after shear
Merge
A
B DiI
FITC-dextran
ATP-647
Merge