| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, January 2002, p. 274-284, Vol. 82, No. 1
and
*Department Cell and Developmental Biology, University of North
Carolina at Chapel Hill, North Carolina 27599 USA; Kusumi
Membrane Organizer Project, Exploratory Research for Advanced
Technology Organization (ERATO), Japan Science and Technology
Corporation, Nagoya 460-0012, Japan; Department of
Biological Science, Nagoya University, Nagoya 464-8602, Japan; and
§Lineberger Comprehensive Cancer Center, University of
North Carolina at Chapel Hill, North Carolina 27599 USA
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
We examined the physical and chemical characteristics of transient confinement zones (TCZs) that are detected in single particle trajectories of molecules moving within the membrane of C3H 10T1/2 murine fibroblasts and their relationship to "rafts." We studied the lateral movement of different membrane molecules thought to partition to varying degrees into or out of the putative lipid domains known as rafts. We found that lipid analogs spend significantly less time in TCZs compared with Thy-1, a glycosylphosphatidylinositol-anchored protein, and GM1, a glycosphingolipid. For Thy-1, we found that zone abundance was markedly reduced by cholesterol extraction, suggesting that a major source of the observed temporary confinement is related to the presence of raft domains. More detailed analysis of particle trajectories reveals that zones can be revisited even tens of seconds after the original escape and that diffusion within the zones is reduced by a factor of ~2, consistent with the zone being a cholesterol-rich liquid-ordered phase. Surprisingly, transient confinement was not strongly temperature dependent. Overall, our data demonstrate that there are raft-related domains present in certain regions of the plasma membrane of C3H cells, which can persist for tens of seconds.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
The notion of lipid rafts (Simons and Ikonen,
1997
, 2000) stems from the observation that a significant fraction of
plasma membrane lipids are detergent resistant (Brown and Rose, 1992). The raft concept has proved useful in organizing structural and functional knowledge concerning detergent-resistant membranes (DRM)
(Brown and London, 1998a,b). Rafts, defined largely by their insolubility in various detergents, are enriched in cholesterol and
sphingomyelin, glycosphingolipids (GSL), GPI-anchored proteins, and a
variety of signaling molecules, potentially imparting a high degree of
functional significance to these microdomains. At this juncture, there
are a number of outstanding issues about rafts that preclude critical
acceptance of the concept. Among them is exactly what is the in vivo
correlate of the biochemically isolated DRM or rafts. Various
sophisticated light microscopic techniques have produced a number of
estimates of raft size ranging from 0 to 700 nm in dimension (for
reviews, see Jacobson and Dietrich, 1999; Anderson and Jacobson,
manuscript submitted).
One of these techniques is single particle tracking (SPT), which allows
the study of particle movements and therefore also the molecules
attached to the particles with submicron spatial resolution on the
surfaces of living cells (Kusumi and Sako, 1996; Saxton and Jacobson,
1997). Transient confinement zones (TCZ) have been detected by SPT
(Simson et al., 1995, 1998; Sheets et al., 1997). These zones, ~200
to 300 nm in diameter, in which the particle is trapped for ~5 to
10 s, were shown to preferentially trap particles attached to
GPI-anchored proteins and the GSL, GM1; furthermore, the abundance and
size of the zones was diminished when the biosynthesis of GSLs was
depressed by inhibitors (Sheets et al., 1997). These results suggested
that the zones bear some relationship to rafts.
In this paper, we examine further the physical and chemical
characteristics of the TCZs in relation to "rafts." We found that zone abundance was markedly reduced by cholesterol extraction, that
zones could be revisited after escape, that zone abundance was
surprisingly temperature independent, and that diffusion within the
zones was reduced but only by a factor of ~2. This indicates that
trapping is not simply due to particles encountering a much more
viscous region but suggests that some type of barrier is involved. This
permanent barrier could be due to a phase boundary in the lipid matrix
and/or the involvement of the membrane skeleton fence (Kusumi and Sako,
1996).
| |
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Materials
Forty-nanometer colloidal gold functionalized with antibiotin
antibodies was purchased from British Biocell International (Cardiff, UK),
1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-fluorescein (FL-DOPE) from Avanti Polar Lipids (Birmingham, AL),
1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-fluorescein (FL-DPPE), and biotinylated rabbit polyclonal antifluorescein antibodies were purchased from Molecular Probes (Eugene, OR). Biotinylated monoclonal anti-Thy-1.2 rabbit was obtained from PharMingen (San Diego, Ca). Biotin-conjugated cholera toxin B subunit
(biotin-CTB) and the cholesterol-sequestering agent
methyl-
-cyclodextrin (MCD) were obtained from Sigma (St. Louis, MO).
Cells
C3H 10T1/2 murine fibroblasts (American Type Culture Collection, Rockville, MD) were maintained in Basal Medium Eagle with Earle's salts and L-glutamine (BME) that has been supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 µg/ml streptomycin. Two to 4 days before an SPT experiment, fibroblasts were plated onto sterile 22 mm × 22 mm coverslips (#1.5, glass) that were placed into 35-mm petri dishes at an appropriate cell density that yielded single cells for SPT measurements.
Cells were depleted of cholesterol by treatment with 5 mM MCD in
unsupplemented BME at 37°C for 1 h. Cholesterol levels (Heider and Boyett, 1978) of C3H cells were reduced by ~60% (±10%) after a
30-min incubation with MCD without detectable effects on cell morphology or cell activity, as judged by membrane ruffling.
Before each SPT experiment, cells on coverslips were washed three times with HAMS F12 nutrition mixture (Life Technologies/Gibco-BRL, Carlsbad, CA) supplemented with 25 mM HEPES (HH) and either mounted with spacers on a microscope slide for measurements at 22°C (room temperature) and 37°C (air curtain) or on a temperature-controlled chamber holder made from stainless steel to conduct measurements at 10 or 15°C (±1°C). Before placing the chambers on the microscope and adjusting to the desired temperature, preparation and incubation steps (see below) were conducted by injection of adequate amounts of reaction solutions into the chambers, each holding ~100 µl of solution. After the final preparation step, chambers were sealed with wax and mounted on the microscope for imaging.
Gold conjugation to cells
Gold was bound to Thy-1.2, a GPI-anchored protein, GM1, a GSL, and the two fluorescent phospholipid analogs, FL-DPPE and FL-DOPE. C3H cells expressed significant amounts of GM1, whereas a detectable level of Thy-1.2 expression occurred only in ~20% of the cells in a preparation, as judged by staining with fluorescein-CTB and anti-Thy-1.2 immunofluorescence, respectively. Synthetic phospholipid analogs were introduced in the cell membranes as follows. Fifteen microliters of 1 mg/ml lipid analog in 50:50 ethanol/chloroform (v/v) solution was placed in a borosilicate disposable culture tube and exposed to a gentle stream of Argon gas until most of the solvent was evaporated and the remaining material at the bottom of the tube was of an oily consistency. Two milliliters of warm HH (~50°C) was injected into the tube while vigorously vortexing. The pellet should dissolve completely without forming flakes. The stock solution was kept for up to 3 days at 4°C. After being mounted in sample chambers, cells were incubated with the lipid solutions at room temperature for 30 to 40 s and were then flushed with 600 µl of HH to remove unbound lipids. Fluorescence imaging revealed a uniform membrane stain, and spot fluorescence recovery after photobleaching (FRAP) measurements gave lateral mobility (DFL-DOPE = 0.33 µm2/s; DFL-DPPE = 0.41 µm2/s at 37°C) and fractional recovery values (fFL-DOPE = 0.95 ± 4% and fFL-DPPE = 0.92 ± 3% at 37°C), indicating correct reconstitution of the two lipid analogs into the plasma membrane. The phenotype of cells labeled under this protocol does not differ from that of untreated cells either in morphology or activity. Moreover, if cells are labeled with FL-DOPE and trajectories for GM1 measured, they do not differ significantly from GM1 trajectories measured when no FL-DOPE is added.
We used biotinylated Thy-1.2, biotin-CTB, and biotinylated antifluorescein to bind colloidal antibiotin gold to Thy-1.2, GM1, and fluorescent lipid analogs, respectively (Fig. 1). Gold was dissolved in HAMS F12 nutrition mixture (Life Technologies/Gibco-BRL) supplemented with 25 mM HEPES and 15% serum (HHS), sedimented by centrifugation for 15 min at 11,000 × g at 4°C and redissolved in HHS. Cells mounted in the chamber were incubated for 5 min at room temperature with 100-µl-injected biotinylated ligands, which were dissolved in HH to a concentration of 10 µg/ml, and the cells were then washed with 600 µl of HH. The gold concentration of the injection solution was adjusted to yield ~1.5 µg/ml biotin antibody bound to gold as determined from the manufacturer's specification for that lot number. We avoided a final washing step to remove unbound gold, and the sample chamber was immediately sealed with wax and mounted on the microscope; particle trajectories were recorded for the following 60 min. Even after 30 min, significant amounts of gold in the bulk did not bind to the cell, whereas the number of particles at the cell lamella (5-15 per 30 × 22-µm field of observation) remained constant. In control experiments, where cells were not incubated with biotinylated ligands, gold showed little binding to the plasma membrane (<1 gold particles per field of view), which demonstrated the specificity of the binding. Thus, specific cell surface binding appears quite stable over tens of minutes with little exchange with gold in the bulk solution.
|
Gold conjugation to planar model membranes
Glass supported POPC lipid bilayers were formed by vesicle
fusion as described previously (Lee et al., 1991). To conjugate antibiotin gold to the lipid bilayer, we added small amounts of FL-DPPE
to the lipid sample (~1 mol %), and used biotinylated antifluorescein antibody as a bridging reagent as described in the
section for gold conjugation to cells. Instead of cell medium, the
preparations for model system were conducted with 25 mM Hepes buffer at
pH 7.3.
Time lapse gold imaging for SPT
Computer-enhanced video microscopy, described earlier (Lee et
al., 1991), was used to image colloidal gold bound to different ligands
in the plasma membrane of C3H cells. Briefly, the cell lamella with
bound gold was imaged in brightfield mode and recorded with a video
camera (Newvicon Hamamatsu). After online background subtraction and
contrast enhancement with an image processing unit (Argus 20, Hamamatsu, Japan), video frames were recorded in time lapse mode (500 frames with 5 frames/s) on the hard disk of a computer (O2,
Silicon Graphics, Mountain View, CA). Recorded movies were analyzed by
the commercial software package ISEE (Inovision Corp., Durham, NC) that
identifies relative changes of gold particle positions on the cell
lamella with a precision of ±20 nm (this figure is not a standard
deviation but rather characterizes the complete distribution of
positional fluctuations that are measured for an immobile particle).
All trajectories were visually inspected to ensure correct tracking of
gold particles. Erroneous or false particle trajectories were excluded
from analysis.
High-speed SPT
For image recordings with temporal resolution of 2 ms,
instrumentation described elsewhere was used (Tomishige, 1998).
Briefly, a digital high-speed video camera with a C-MOS sensor
(model FASTCAM-Ultima; Photron, Tokyo, Japan) was used to acquire,
digitize, and store images at a rate of 500 frames/s for up to
10 s. Digitized images were continuously transferred to a standard
video out port, which allowed recording of video sequences on tape at
the standard video rate (30 frames/sec). This camera allowed
simultaneous monitoring at the video rate, and thus fast recordings
could be started at any arbitrary moment during the video recording as
defined by the microscopist.
Computer simulation
Particle trajectories were generated with a custom written
program designed to study the effect of detection uncertainties when
diffusion occurs within confinement zones. Random walks were generated
by a Monte Carlo technique calculating particle jumps with random jump
angle (random number generator was used from commercial programming
environment; LabWindows, National Instruments, Austin, TX) and fixed
jump length 
To simulate a confinement in two dimensions, we assumed that particles were entrapped in circular domains of radius rc. The diffusion coefficient within the domain Din was assumed to be constant and was selected to be roughly consistent with experiments; however, the transfer out of the domain into the surrounding domain was restricted. For every basic simulation step that led a particle across the domain rim, an evenly distributed random number between 0 and 1 was generated and compared with a predefined escape probability levels (Pescape) from the interval 0 to 1. In case the generated random number fell below the critical probability level, the particle path was recalculated, assuming an inelastic reflection at the domain border, which conserved the overall jump length. Otherwise, the simulated position was accepted and used as the starting point for the next simulation step. To simulate an infinitely high barrier, Pescape was set to 1, which ensured entrapment of the particle in the domain.
Detection uncertainties in an experiment were caused by measurement
errors and blurring effects of the recording cameras. To account for
this detection noise, we displaced every trajectory point
(xi,yi)
of a simulated trajectory by a random distance
(dxi, dyi), which was generated by assuming
noise after a Gaussian distribution, characterized by the standard
deviation 
x and y in
the x and y directions, respectively. When not
otherwise stated, simulations with noise were conducted with setting
x = y = 10 nm, a
characteristic value for the instrumentation used as assessed from
particle trajectories from immobile gold particles.
Analysis of trajectories
In experiments with cells, we selected gold particles that moved
long distances over the lamella. Immobile gold, which comprised up to
20% of the bound gold found on the cell lamella, was visually rejected
in the measurement. To implement an objective criteria for long
distance lateral mobility we calculated the parameter MSD90
(Kusumi et al., 1993), which assesses the long distance displacement
(as measured for MSD) on basis of analyzed short-term particle mobility D13 (see below). We rejected
trajectories with MSD90 < 0.3 to exclude strongly
confined and stationary trajectories and trajectories with
MSD90 > 1.9 to avoid trajectories dominated by
directed motion. Note that the procedure of trajectory selection was
applied to screen simulated and experimental data and affected only a
small fraction (~15% for experimental and ~7% for simulated trajectories) of the data. For experimental or computer-generated particle trajectories, the mean squared displacement
MSDdt = 
x
+ y of particle positions (x,
y) was calculated for different time intervals
dt. According to MSD = 4 × D × dt for a randomly diffusing particle in
two dimensions, we obtained D by fitting the MSDs
at 1dt, 2dt, and 3dt to a straight line, in which dt is the basic time interval in a recording.
Detection noise adds a constant offset to the MSD, and
therefore it is not advisable to include the origin
(MSD = 0, dt = 0) into the line fit
(see Appendix).
Transient confinement
TCZs occurring within long-term trajectories (100 s, 5 frames/s) were detected by an algorithm described previously and characterized with respect to size and dwell time (Simson et al., 1995). To obtain a measure for the tendency of a molecule to reside in
TCZs, we introduced the parameter-termed relative transient confinement time (RCT), which is defined as the ratio of detected transient confinement duration to the total recording time of the trajectories analyzed. In a few cases (~2%), the algorithm did not properly distinguish between spatial separated TCZs that were obvious upon visual inspection of the particle trajectory. These confinement zones
were detected as one and therefore resulted in an unreasonably large
domain radius; such zones were excluded from analysis.
The uncertainty in RCT was computed as a standard error of the mean.
This presumes that the recorded trajectories randomly sample the cell
surface in an ergodic fashion, and the RCT is a measure of the
partition coefficient if confinement areas are referenced to
unrestricted diffusion areas. Thus, by averaging the amount of time
spent in TCZs for each trajectory, we will arrive at an estimate of the
degree of transient confinement that would tend to be a true mean as
more and more trajectories are collected.
Revisting TCZs
The distance between two successive detected TCZs was defined as the distance between their centers of mass calculated from their associated particle positions. By definition, a particle revisited the same TCZ when the distance between the two visited TCZs was smaller than the average of the radii attributed to the TCZs (for definition of the zone radius see Simson et al. (1995)). To characterize the time
between visits to successive TCZs we used the time that had elapsed
between the frame when the particle left the first zone and the frame
when it entered the second zone (which could on occasion be the same
TCZ [i.e., revisiting]).
High time resolution SPT
To identify TCZs in high-speed recordings (500-Hz frame rate), we applied TCZ analysis (Simson et al., 1995) on the simultaneously conducted video (30 Hz) rate recordings. This allowed us to distinguish and identify zones and periods of confined and free diffusion. The
frame rate allowed the acquisition of data sets with typically several
100 particle positions while a particle resided in a TCZ. For the time
sequences of transient confinement,
MSDTCZ versus dt plots were
calculated, which typically showed a saturation behavior, while the
MSDfree versus dt plots of the
remaining portions of the trajectory were linear. Displacements in the
millisecond time regime are of the order of the detection noise, and
this must be accounted for (see Appendix). Therefore
MSDfree versus dt plots were
fitted to a straight line for time intervals covering frame 1 to 3, i.e., dt is 2 to 6 ms. The slope of the line provided the
particle mobility Dfree in the
millisecond time domain and the intersection with ordinate equaled the
MSDnoise-offset, which was subtracted from
MSDTCZ to obtain noise corrected
MSDTCZc versus dt plots. This
correction allowed inclusion of MSDTCZc(dt = 0) = 0 as the starting point at time dt = 0. The initial slope of the MSD versus
dt plot, and, in consequence, the diffusion coefficient within the confinement zone, was determined as
DTCZ = f × MSDTCZc(dt = 2 ms)/(4 × dt(dt = 2 ms)) in which
f is a correction factor determined from Monte Carlo
simulations (see Appendix). We tested this procedure on simulated
particle trajectories and found that this protocol reproduced the
diffusion values injected into the simulation within a few percent. The
domain radius rc was obtained from the
saturation level of MSDTCZc in the limit of
long-interval times (Appendix). The ratio
DTCZ/Dfree
provided a measure of the differences of the membrane mobility between
the TCZ and surrounding membrane.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Labeling various raft components for SPT
To address the differential partitioning of several representative membrane molecules into putative lipid microdomains, SPT was applied to follow the movements of Thy-1, a GPI-anchored protein, GM1, a representative GSL, as well as fully saturated FL-DPPE and unsaturated FL-DOPE fluorescent lipid analogs on the surface of fibroblasts. As depicted in Fig. 1, 40 nm colloidal gold conjugated with antibiotin antibody was attached to these molecules using a biotinylated ligand as a bridging component. Thus, the same antibiotin conjugated gold could be used to tag the investigated molecules with similar spacing between the colloidal gold and the cell membrane.
Relative confinement for putative raft components
We analyzed particle movement of the different membrane probes for
100-s observation times. We applied an algorithm previously described
(Simson et al., 1995) to detect TCZ within the long-term trajectories.
To compare the tendency of the different probe molecule to undergo
transient confinement, we introduce the RCT, which is defined as the
ratio of the time a particle spends in TCZs to the total recording time
of the trajectory. Fig. 2 shows the average RCTs calculated from sets of particle trajectories that were
recorded for the different molecules investigated. The RCT values
indicate Thy-1 and GM1, both putative raft occupants, spend similar and
the largest amounts of time in the TCZs. Even lipid analogs experience
smaller but significant transient confinement when moving laterally in
the plasma membrane of C3H fibroblasts with the fully saturated analog
confined more than the unsaturated analog. By contrast, trajectories
obtained from particles attached to FL-DPPE embedded in a planar
supported model membrane system (see Materials and Methods) revealed
negligible transient confinement (Fig. 2). As a negative control, the
RCT analyzed from random walk particle trajectories generated by
computer assuming lateral diffusion in a homogeneous membrane is given
(right panel, Fig. 2). The negligible RCT obtained confirms that the
algorithm with the parameter settings used is robust against the
statistical variations inherent in Brownian diffusion.
|
Chemical and physical characteristics of TCZs
Cholesterol extraction reduces TCZ abundance
Earlier, we showed that GSL depletion by metabolic inhibition of GSL biosynthesis resulted in a considerable reduction in TCZ abundance and size (Sheets et al., 1997). We used MCD as a cholesterol
extraction reagent and found that ~60% extraction of cell
cholesterol results in a major reduction in TCZ abundance at both 37 and 22°C, as inferred from the decrement in RCT (Fig. 3, top left panel). Note that in the
unextracted controls, mean domain diameters are ~200 nm and dwell
times are ~5 s (Fig. 3, top right and bottom left panels,
respectively). A significant reduction of the size and dwell time of
the remaining TCZs (Fig. 3, top right and bottom left panels,
respectively) was also observed. The lower right panel shows, that as
expected, the frequency of visiting TCZs drops drastically after
cholesterol extraction. These findings are consistent with the
hypothesis that TCZs reflect lipid rafts that are dependent on
cholesterol for their integrity (Simons and Ikonen, 1997, 2000).
|
Diffusion within TCZs is reduced
To carefully examine diffusion of Thy-1 within TCZs, we used a high-speed camera to record bright field images every 2 ms for a period of 10 s. The high-speed recording was randomly activated while the particle position was recorded at the video rate (see Materials and Methods). The video rate (30 Hz) recording allowed application of the TCZ analysis revealing time periods and regions of temporary confinement. By chance, these events may be overlaid with the high-speed recording, as illustrated in Fig. 4 a (with a high magnification inset). In Fig. 4 b, corresponding plots depicting the displacements in x-direction as a function of time (Fig. 4, bottom panel) and the calculated likelihood profile (Fig. 4, top panel) are given. From such measurements, MSD versus time plots could be constructed from the high-time resolution trajectories both inside and outside of the TCZs (Fig. 4 c). Outside the TCZ, the linear MSD versus time plot indicates the free diffusive behavior with the diffusion coefficient, Dout, calculated from the slope of the line between points 1 and 3. The MSD plot calculated for periods, when the particle resided in the TCZ saturated quickly, indicating confined diffusion. We applied a more sophisticated analysis described in the Materials and Methods and Appendix sections to deduce Din and size for the domains as compiled in Table 1. The measured values for Dout and Din vary significantly for different particles; however, the relative reduction in diffusion, Din/Dout, indicates a similar modest reduction of the particle mobility by a factor of ~2 (Table 1). This reduction is in line with what one would expect for a cholesterol-rich liquid ordered phase (Wu et al., 1977;
Almeida et al., 1992), such as is postulated for rafts (Simons and
Ikonen, 1997). This moderate reduction indicates that the observed
particle confinement is not due to the particle entering a zone of much
greater viscosity but rather that the particle experiences an
significant activation barrier, which it has to overcome to escape from
the confinement zone.
|
|
TCZs can be revisited
We also studied the temporal and spatial distribution of TCZs for Thy-1 and GM1. Interestingly, we find that particles revisit TCZs (for illustration, see particle trajectory in Fig. 5 a). For Thy-1 and GM1 we find that 12 and 26% of the TCZs for these components, respectively, are revisited. The distribution of times between successive visits to the same zone is depicted in the histograms in Fig. 5 b and indicates for Thy-1 that revisiting of the same zone can occur even up to 80 s after the first visit. This implies that at least certain zones were stable for 80 s or longer.
|
Interzone distances are broadly distributed
Fig. 5, c and d gives the interzone distance distribution and the distribution of times between visiting successive TCZs for GM1 and Thy-1, respectively. We were particularly interested if molecules undergoing transient confinement reflect the hop diffusion of a molecule from one zone to the adjacent confinement region. We find that a particle can travel several microns before it again encounters a TCZ. This results in a wide variation from nearest neighbor zones (<0.5-µm separation) to trajectories that have only one or no TCZ; in fact, these later categories are most prevalent as shown earlier for a GPI-anchored neural cell adhesion molecule isoform (Simson et al., 1998). By contrast, strict conformance to the membrane skeleton
fence model would predict that a particle attached to a membrane
protein most often hops to a neighboring corral (Kusumi and Sako,
1996).
Temperature dependency of TCZs
Fig. 6 shows the temperature dependence of RCT, zone diameter, and duration of confinement for GM1. Surprisingly, within the considerable uncertainty of the measurement, there was not significant temperature dependence. Similar results were obtained for Thy-1, FL-DPPE and FL-DOPE (data not shown).
|
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
In this study we find several lines of evidence that TCZs detected
in particle trajectories reflect the presence of putative lipid
microdomains. First, the abundance of TCZs strongly depends on the
presence of cholesterol, consistent with the hypothesis that the
integrity of rafts depends on appropriate levels of cholesterol (Simons
and Ikonen, 1997, 2000). Second, the measured reduction of particle
mobility after entering a TCZ is modest, in line with the
interpretation that rafts reflect the existence of lipid domains in a
liquid ordered, cholesterol-enriched phase. Third, we find that the
abundance of TCZs depends on the membrane molecule investigated. Thy-1,
a GPI-anchored protein was found to be significantly enriched in the
DRM fractions in neurons (Madore et al., 1999) and in lymphoid cells
(Hoessli and Rungger-Brandle, 1985; Draberova and Draber, 1993) and, on
this basis and based on corollary electron microscopy (Madore
et al., 1999), was suggested to be present in membrane rafts. The GSLs,
including GM1, are also thought to partition favorably into rafts
(Brown and London, 1998a,b; Stauffer and Meyer, 1997; Dietrich et al.,
2001). Consistent with this, we find a significantly higher tendency
for these probe molecules to reside in TCZs compared with the
phospholipid analogs. The lowest value for RCT is measured for FL-DOPE,
which would be thought least likely to enter rafts due to its
unsaturated acyl chains, making partitioning into the more ordered raft
phase unfavorable.
Current limitations
Labeling
Although we devised a scheme (Fig. 1) in which the same colloidal gold can be coupled to various target molecules via a biotinylated ligand as a bridging component, in fact, the absolute valency of the observed complexes cannot be determined. This limitation may produce a range of partitioning into various membrane domains including rafts, due to a possible light degree of cross-linking.TCZ analysis
The algorithm to detect TCZs was optimized to pick up anomalies in the video rate trajectories in which sometimes transient confinement is obvious by inspection (Simson et al., 1995). However, hop diffusion on
a very fine scale (Suzuki et al., 2001) is generally not detected. This
is because the time (approximately microseconds) and distance scales
(approximately tens of nanometers) needed to visualize the
membrane skeleton fence in full detail require much higher temporal
resolution than are needed to visualize TCZs, which are normally
separated by hundreds of nanometers and in which the particle resides
for ~5 s.
Models for TCZs and their relationship to lipid rafts
Several alternative models could account for the data (Jacobson
and Dietrich, 1999). Of course, at this stage of development of the
raft field, these models do not comprise an exhaustive list. First,
zones could simply be contiguous cholesterol-dependent liquid ordered
regions ~200 nm in dimension. This model would readily account for
the loss of TCZs upon cholesterol extraction and the slightly slower
diffusion within the TCZ. As demonstrated in model membrane systems
(Dietrich et al., 2001), fluid-fluid phase coexistence can provide
quite measurable differential partitioning with a high rate of
molecular exchange between phases. Thus, the boundary between the
liquid crystalline and the liquid-ordered phase would not, in itself,
provide the duration of confinement required. An effective
"barrier" would be indicated if the partition coefficients for
GPI-anchored proteins and GSL were very favorable. However, with few
exceptions (Schutz et al., 2000), fluorescent stains of putative raft
components, in an uncross-linked state, show little unambiguous
evidence of localized concentrations of such components. A cytoskeletal
barrier could be provided by the membrane apposed cytoskeleton as
transmembrane proteins linked to the underlying cytoskeleton would
provide a "picket fence" that is able to restrict the diffusion of
GPI-anchored proteins residing in the outer monolayer of the plasma
membrane (Sheets et al., 1997). Cytoskeletal components
including actin (Rodgers and Zavzavadjian, 2001; Holowka et al., 2000),
annexin (Olieferenko et al., 1999), and filamin (Stahlhut and Deurs,
2000) have been found associated with rafts and/or caveolae. The
phenomenon that zones can be revisited by the same particle implies a
stable location most likely originating in membrane-cytoskeletal
interactions. However, preliminary attempts to date to perturb this
putative interaction with drugs have not met with success. In regard to a cytoskeletal basis for confinement, the lack of a temperature dependence is something of a conundrum. If confinement is not due to
viscous trapping, why is not a significant activation energy seen for
escape (dwell time) from the TCZ?
In Monte Carlo simulations that assume a domain radius independent of temperature, we found that escape probabilities in the range of p ~ 0.2 (this corresponds to an activation energy of 1.6 kBT at 37°C) reproduce the weak temperature dependence of transient confinement. However, to obtain this result, diffusion had to be ~10 times slower within the TCZ compared to outside of the TCZ, whereas the measured Dout/Din ~ 2. At this point, one might speculate that the model, using the measured Dout/Din, produces erroneous results because it assumes normal (not anomalous) diffusion within the TCZs.
Alternatively, rafts could be considerably smaller than the measured
zone diameters, and the entire raft might diffuse into a TCZ, become
trapped for ~5s, and then escape. Such entities might be of the size
suggested by Varma and Mayor (1998) and Pralle et al. (2000). Or they
might be even smaller
the lipid shells postulated by R. Anderson and
K. Jacobson (manuscript submitted) to surround individual GPIAPssuch
that a gold particle might bind one or several shells. In this sort of
model, cholesterol depletion would dismantle the elementary raft or
shell and thereby prevent extensive transient confinement. Diffusion
would be slower within the TCZ because of an elementary raft or shell
concentration with the zone-producing obstructed diffusion.
| |
APPENDIX |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Effect of detection noise on MSD plots
For high-speed recordings conducted in this study, particle
displacements between two time frames are of the same magnitude of the
spatial detection noise. Assuming a particle with D = 7 × 10
10 cm2/s, one
obtains a mean jump distance of ~17 nm for a 2-ms time interval. Fig.
7 a shows
MSD-dt plots calculated from simulated trajectories assuming a random walk (open squares) and random walk with
superposed Gaussian distributed noise (filled circles). We generated
MSD-dt plots from simulated trajectories assuming unrestricted diffusion (straight lines) and diffusion restricted to a
circular domain (saturating curves) characterized by an infinitely deep
potential well (Pescape = 0).
|
For an immobile particle the MSD is zero; however, it
differs from zero if the spatial detection noise is included, which results in apparent displacements and shifts the MSD-plot at
all time frames by a constant value
MSDnoise. For our instrumentation, detection uncertainties of immobile particles are well described with a
normal distribution probability. Then,
|
x and
y are the standard deviations in x
and y direction, respectively. Thus, from knowledge of the
spatial noise level, one obtains the corrected
MSDc-dt plot by subtraction of MSDnoise for all time frames other than
dt = 0, in which
MSDc(dt = 0) 
0.
Extraction of Din for particles entrapped in a circular domain
For a circular domain geometry, the
MSDTCZ-plot saturates to the value
MSDTCZ(dt 
) = rc2 (Saxton and
Jacobson, 1997). For short time periods, when the particle does not
experience the barrier at the domain rim, the diffusion is free and the
initial slope of the plot gives the diffusion coefficient of the
particle within the domain with 4 × Din = d(MSDTCZc)/d(dt)|dt=0.
It was important to inquire whether the initial slope is reasonably
well represented by the slope as obtained form the first two points of
a noise corrected MSDc-dt plot.
Therefore, we generated trajectories, simulating particles stably
entrapped (Pescape = 0) in a circular
domain for different particle mobilities and domain diameters and
assuming a particle detection rate of 500 Hz. Fig. 7 b shows
a selection of MSD-dt plots after correction of
the noise related MSD shift with typical experimental
parameter settings. The MSD-dt plots are given in dimensionless units, ([MSDscale] = [MSD/r
(MSDmaster)/(dtmaster)|dt=0 = 4 and MSDmaster|dt
= 1. The first data point at dtmaster(1) in
the MSDc plot then provides an estimate for
the mobility of the particle: Destimate = MSDmaster(1)/(4 × dtmaster(1)). Simulations were conducted
with different temporal resolutions to obtain factors by which
D measured from increasing time intervals is underestimated for typical experimental settings (inset in Fig. 7 b). We
found that the slope of the first two time points systematically
underestimated the "true" D by ~10% for parameter
settings that are typical for our experiments. However, this was small
compared with the mobility variation experimentally measured for
different TCZs.
| |
ACKNOWLEDGMENTS |
|---|
We thank Thierry Horner for working out the cholesterol depletion assay and Alex Valm for assistance conducting SPT measurements.
This work was supported by the National Institutes of Health (GM41402, to K.J.) and the Deutsche Forschungsgemeinschaft Scholarship (Di 691/1-1, to C.D.), and the ERATO Kusumi Membrane Organizer project (to A.K.).
| |
FOOTNOTES |
|---|
Received for publication 21 June 2001 and in final form 15 September 2001.
Address reprint requests to Dr. Ken Jacobson, University of North Carolina, Department Cell and Developmental Biology, CB#7090, 108 Taylor Hall, Chapel Hill, NC 27599-7090. Tel.: 919-966-3855; Fax: 919-966-1856; E-mail: frap{at}med.unc.edu.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Biophys J, January 2002, p. 274-284, Vol. 82, No. 1
© 2002 by the Biophysical Society 0006-3495/02/01/274/11 $2.00
This article has been cited by other articles:
 |
|
 |
|
  V. Nechyporuk-Zloy, P. Dieterich, H. Oberleithner, C. Stock, and A. Schwab Dynamics of single potassium channel proteins in the plasma membrane of migrating cells Am J Physiol Cell Physiol, April 1, 2008; 294(4): C1096 - C1102. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M. Davey, K. M. Krise, E. D. Sheets, and A. A. Heikal Molecular Perspective of Antigen-mediated Mast Cell Signaling J. Biol. Chem., March 14, 2008; 283(11): 7117 - 7127. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Crane and A. S. Verkman Long-Range Nonanomalous Diffusion of Quantum Dot-Labeled Aquaporin-1 Water Channels in the Cell Plasma Membrane Biophys. J., January 15, 2008; 94(2): 702 - 713. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Jin, P. M. Haggie, and A. S. Verkman Single-Particle Tracking of Membrane Protein Diffusion in a Potential: Simulation, Detection, and Application to Confined Diffusion of CFTR Cl- Channels Biophys. J., August 1, 2007; 93(3): 1079 - 1088. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Ghosh and M. J. Wirth Parsing the Motion of Single Molecules: A Novel Algorithm for Deconvoluting the Dynamics of Individual Receptors at the Cell Surface Sci. Signal., May 29, 2007; 2007(388): pe28 - pe28. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. A. Siddiqui, K. A. Harvey, G. P. Zaloga, and W. Stillwell Modulation of Lipid Rafts by {Omega}-3 Fatty Acids in Inflammation and Cancer: Implications for Use of Lipids During Nutrition Support Nutr Clin Pract, February 1, 2007; 22(1): 74 - 88. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Wenger, F. Conchonaud, J. Dintinger, L. Wawrezinieck, T. W. Ebbesen, H. Rigneault, D. Marguet, and P.-F. Lenne Diffusion Analysis within Single Nanometric Apertures Reveals the Ultrafine Cell Membrane Organization Biophys. J., February 1, 2007; 92(3): 913 - 919. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Bouzigues and M. Dahan Transient Directed Motions of GABAA Receptors in Growth Cones Detected by a Speed Correlation Index Biophys. J., January 15, 2007; 92(2): 654 - 660. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Davey, R. P. Walvick, Y. Liu, A. A. Heikal, and E. D. Sheets Membrane Order and Molecular Dynamics Associated with IgE Receptor Cross-Linking in Mast Cells Biophys. J., January 1, 2007; 92(1): 343 - 355. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. B. Nicholson and C. P. Stanners Specific inhibition of GPI-anchored protein function by homing and self-association of specific GPI anchors J. Cell Biol., November 20, 2006; 175(4): 647 - 659. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Chen, W. R. Thelin, B. Yang, S. L. Milgram, and K. Jacobson Transient anchorage of cross-linked glycosyl-phosphatidylinositol-anchored proteins depends on cholesterol, Src family kinases, caveolin, and phosphoinositides J. Cell Biol., October 9, 2006; 175(1): 169 - 178. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. G. Ayuyan and F. S. Cohen Lipid Peroxides Promote Large Rafts: Effects of Excitation of Probes in Fluorescence Microscopy and Electrochemical Reactions during Vesicle Formation Biophys. J., September 15, 2006; 91(6): 2172 - 2183. [Abstract] [Full Text] [PDF]
|
|
|
|
|
) or including Gaussian distributed spatial noise
(
). Linear
MSD-dt plots result from simulations
assuming unrestricted diffusion whereas saturation in
MSD-dt plots occurs when particle motion
is restricted to a circular domain of 50-nm radius and infinitely high
barrier at the boundary (Pescape = 0).
For these plots, the detection rate was 500 Hz, and the particle
diffusion constant was Din = 5 cm × 10