Cytoskeletal Control of CD36
Diffusion Promotes Its Receptor
and Signaling Function
Khuloud Jaqaman,1,2,6Hirotaka Kuwata,3,6Nicolas Touret,4Richard Collins,3William S. Trimble,3Gaudenz Danuser,2,5,*
and Sergio Grinstein3,*
1Department of Systems Biology, Harvard Medical School, Boston, MA 02115, USA
2Department of Cell Biology, The Scripps Research Institute, La Jolla, CA 92037, USA
3Cell Biology Program, The Hospital for Sick Children, Toronto, ON M5G 1X8, Canada
4Department of Biochemistry, University of Alberta, Edmonton, AB T6G 2E1, Canada
5Department of Cell Biology, Harvard Medical School, Boston, MA 02115, USA
6These authors contributed equally to this work
*Correspondence: firstname.lastname@example.org (G.D.), email@example.com (S.G.)
The mechanisms that govern receptor coalescence
into functional clusters—often a critical step in their
stimulation by ligand—are poorly understood. We
used single-molecule tracking to investigate the
dynamics of CD36, a clustering-responsive receptor
that mediates oxidized LDL uptake by macrophages.
We found that CD36 motion in the membrane was
spatially structured by the cortical cytoskeleton. A
subpopulation of receptors diffused within linear
confinement regions whose unique geometry simul-
taneously facilitated freedom of movement along
one axis while increasing the effective receptor den-
sity. Co-confinement within troughs enhanced the
probability of collisions between unligated recep-
tors and promoted their clustering. Cytoskeleton
perturbations that inhibited diffusion in linear con-
finement regions reduced receptor clustering in the
absence of ligand and, following ligand addition,
suppressed CD36-mediated signaling and internali-
zation. These observations demonstrate a role for
the cytoskeleton in controlling signal transduction
by structuring receptor diffusion within membrane
regions that increase their collision frequency.
Receptor clustering and organization into membrane microdo-
mains is an essential feature of transmembrane signal transduc-
tion (Cebecauer et al., 2010; Groves and Kuriyan, 2010; Scott
and Pawson, 2009). While receptors were initially thought to
cluster upon binding multivalent ligands (Heldin, 1995), there is
increasing evidence that receptors can also exist in pre-formed
clusters that get reorganized and activated upon ligand binding
(Cambi et al., 2006; Chung et al., 2010; Iino et al., 2001; Livnah
et al., 1999; Sako et al., 2000; Schamel et al., 2005; Varma and
Mayor, 1998). Membrane microdomains enriched in cholesterol
and sphingolipids (Foster etal., 2003; Friedrichson and Kurzcha-
actions between transmembrane proteins and the cytoskeleton
(Andrews et al., 2008; Bouzigues et al., 2007; de Keijzer et al.,
2011; Goswami et al., 2008; Kaizuka et al., 2009; Plowman
et al., 2005; Serge et al., 2003; Suzuki et al., 2007), and interac-
tions between proteins within the membrane (Douglass and
Vale, 2005; Espenel et al., 2008) have been implicated in regu-
their roles in controlling the signaling competence of receptors.
CD36 is a clustering-responsive class B scavenger receptor
expressed on the surface of platelets, endothelial cells, and
macrophages (Febbraio et al., 2001). In macrophages, it binds
to multivalent ligands such as oxidized low-density lipoprotein
(oxLDL), apoptotic cells, and malaria-infected erythrocytes,
implicating it in a wide range of processes, from lipid metabolism
to innate immunity to tissue remodeling (Endemann et al., 1993;
McGilvray et al., 2000; Savill, 1997). Biochemical studies
suggest that CD36 clustering at the cell surface upon engage-
ment of multivalent ligands triggers signal transduction and
receptor-ligand complex internalization (Daviet et al., 1997;
McGilvray et al., 2000). However, it is not known whether unli-
gated CD36 exists as monomers or as clusters that facilitate
the cellular response to ligand, and what factors contribute to
CD36 clustering. To address these questions, we combined
ical approaches to study the dynamics, clustering, and signaling
of CD36 in primary human macrophages.
Single-Molecule Imaging of CD36 on the Surface
of Primary Human Macrophages
To image single receptors, we immunolabeled CD36 with a
primary anti-CD36 Fab fragment followed by a secondary Fab
Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc. 593
fragment conjugated largely (>85%) with a single Cy3 fluoro-
phore, and imaged the dorsal surface of macrophages using
wide-field epifluorescence microscopy. The resulting images
consisted of diffraction-limited spots (Figure 1A), the subpixel
positions and peak intensities of which were determined by
fitting mixtures of Gaussian kernels (Jaqaman et al., 2008) (see
Figure S1A available online).
To assess whether the spots corresponded to single fluoro-
phores, we imaged fixed cells using a range of primary Fab
fragment dilutions at a fixed concentration of secondary Fab
fragment. Modal analysis of the particle intensity histograms
(Yang et al., 2007) revealed multiple intensity modes with
conserved mean intensities across all dilutions (Figure 1B).
Moreover, individual particles photobleached in a stepwise
fashion, with a step size similar to the mean of the first mode in
the intensity histograms (Figure 1A). Thus, the first mode of the
intensity histograms most likely corresponded to a single Cy3
fluorophore, demonstrating our ability to detect single mole-
cules. Of note, notall secondary Fab fragments were conjugated
to exactly one Cy3 and, at the labeling densities used, not every
CD36 molecule on the surface was labeled. For these reasons,
the following analysis does not assume that one fluorophore
represents one CD36 molecule.
CD36 Exhibits Multiple Motion Types on the Surface
To measure the dynamics of CD36 in live cells, we chose an
intermediate labeling density (Figure 1B, right) that balanced
the conflicting requirements of tracking single receptors while
at the same time capturing interactions between them. Movies
were collected with a frame rate of 10 Hz for 10 s (Movie S1,
left), over which period photobleaching was negligible (Figures
S1B and S1C). Receptor trajectories were reconstructed using
a multiple-particle tracking algorithm (Jaqaman et al., 2008)
designed to follow individual particles in densely populated
fields and to explicitly capture their merging and splitting with
other particles (Movie S1, right, and Movie S2).
In unstimulated macrophages CD36 exhibited several trajec-
tory types (Figure 2A), which we classified using two measures:
Thefirstcharacterized trajectoriesaslinear orisotropicbasedon
the scatter of receptor positions regardless of the underlying
mobility (Huet et al., 2006; Jaqaman et al., 2008). The second
Figure 1. Single-Molecule Imaging of CD36 in Primary Human Macrophages
(A) Single-molecule image (left) and particle fluorescence step photobleaching (right). Particles 1 and 2 are composed of one and four Cy3 molecules,
respectively. Scale bar, 1 mm. Insets, representative images of Particle 2.
(B) Fluorescence intensity histograms of detected CD36 particles at different primary Fab fragment dilutions in fixed (left and middle) and live (right) cells,
decomposed into individual intensity modes (orange lines, individual modes; red lines, sum of all modes). Arrows show mode centers; numbers indicate the
number of Cy3s each mode represents. Insets: Fraction of detected particles in each mode, and mode centers (in parentheses).
See also Figure S1.
594 Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc.
identified the mobility by a moment scaling spectrum (MSS)
analysis of receptor displacements (Ewers et al., 2005; Ferrari
et al., 2001) (Figure S2A). The combination of these two
measures revealed that 27% ± 1% of receptors had linear
trajectories, 18% ± 1% had isotropic trajectories generated by
unconfined diffusion (referred to as isotropic-unconfined), and
55% ± 1% had isotropic trajectories generated by confined
diffusion (referred to as isotropic confined) (Figure 2B).
The linear trajectories of CD36 radiated from the perinuclear
region (Figures 2A and 2C and Figure S2B). Since unligated
CD36 is thought to reside exclusively at the cell surface (Collins
et al., 2009), this unexpected linear movement raised the possi-
bility that binding to Fab fragments triggered CD36 internaliza-
tion and displacement along microtubules (MTs). However,
following an acute acid wash that preserved cell integrity, only
15% of the fluorescent particles remained (Figures S2C–S2H).
This fraction was smaller than the fraction of linearly moving
receptors, and none of the remaining label exhibited linear
motion (Figure S2F). These results indicated that the majority
of the imaged receptors were surface-bound, including those
CD36 Motion along Linear Trajectories Promotes
The imaged receptors, even though unligated, underwent
merging and splitting events (Figures 2D–2F and Movie S2).
These events could be apparent fusions reflecting incidental
colocalization of receptors within distances closer than the
resolution limit (?300 nm), or they could be genuine reversible
clustering events formed, for example, through direct interac-
tions between receptors, indirect interactions via other mole-
cules, and/or receptor co-confinement within membrane nano-
domains. We used several measures to distinguish genuine
clustering from incidental colocalization. First, we compared
the measured distribution of fusion times (Figure 2G) to the
distribution expected were merging and splitting events solely
due to incidental colocalization (Kasai et al., 2011). Specifically,
we simulated noninteracting receptors that moved on the cell
of apparent fusion times caused by resolution limitations. With
this distribution, if the probability of observing a simulated
apparent fusion time R X s was pðXÞ, then we defined the
confidence that an experimental fusion lasting for X s repre-
sented a true clustering event as 1 ? pðXÞ (Figure 2G). We found
that CD36 fusion events could not be accounted for solely by
incidental encounters: 60% of the experimentally observed
fusions lasted longer than 1 s, the 90% confidence threshold.
Second, since a protein’s diffusion speed in the membrane is
linked to its dimension (Gambin et al., 2006), we investigated
whether fused receptors moved slower than before merging or
after splitting. We found that 65% of receptors indeed exhibited
slower speeds while fused (Figures S2I and S2J). In addition,
we found a significant negative cross-correlation between
particle intensity and mobility (Figure S2K). These results implied
that at least 60%–65% of the observed merging and splitting
events reflected genuine reversible clustering events, while
the rest were most likely apparent mergers due to resolution
Even though clustering events were rare overall (Figure 2F),
they depended on the type of receptor motion. First, there was
a gradient in particle intensities: linearly moving particles had
the highest intensity and isotropic-confined particles the lowest
(Figure 2H), implying that the chance of a detected particle
consisting of multiple CD36 molecules was highest for linearly
moving particles, and lowest for isotropic-confined particles.
Second, we observed a gradient in the probability of merging
and splitting: again linearly moving receptors had the highest
probability and isotropic-confined receptors the lowest (Fig-
ure 2I). These observations collectively indicated that the
linear movement of CD36 favored metastable clustering in the
absence of ligand.
CD36 Linear Motion Is Diffusion within Linear
MSS analysis of the diffusion of linearly moving receptors rarely
classified them as super-diffusive (only ?7%). To further dissect
the linear motion characteristics, we determined the orientation
axis of each linear trajectory, defined as the axis of largest posi-
tional variation within a trajectory (Figure S2B), and extracted
two parameters (Figure 3A): (1) the component of the frame-to-
frame displacement parallel to the orientation axis, and (2) the
run time, i.e., the number of steps taken in one direction before
switching to the opposite direction. We found no difference
between motion away from and toward the perinuclear region
(Figure 3B). Also, the distribution of run times resembled that
of a 1D random walk, where the probability of taking n consecu-
tive steps in one direction is 2?n(Berg, 1993). These results
suggested that the linear motion of unligated CD36 was not
motor driven but rather diffusive.
To test this hypothesis further, we collected movies with
higher sampling frequencies (33, 62.5, and 125 Hz) and com-
we labeled CD36 with quantum dots (Qdots) instead of Cy3
because of their brighter and more photostable signal. Qdot
blinking was compensated for by the gap-closing feature of
the multiple-particle tracking algorithm (Jaqaman et al., 2008).
This multiscale analysis yielded several results: First, the mean
parallel component of the frame-to-frame displacements scaled
with the square root of time, as expected for diffusion (Figure 3C;
see also Figure S3A). In contrast, motors would scale linearly
with time. Second, as expected for a timescale invariant process
such as diffusion (Sethna, 2006), the distributions of run times
expressed in frames did not vary with frame rate (Figure 3D). In
contrast, motor-driven motion should have a characteristic
processivity, in which case frame rate changes would alter the
run time distribution expressed in frames. Third, in agreement
with the isotropic nature of diffusion, the average receptor
displacement component perpendicular to the orientation axis
(Figure 3A) was very similar to the average parallel displacement
component at 125 Hz sampling (Figure 3C; at lower sampling
rates, confinement in the perpendicular direction prevented the
full square-root of-timescaling of the perpendicular component).
Fourth, receptors visited all positions within their linear con-
finement regions, albeit with a bias toward the edges, providing
further evidence for motion isotropy (Figure 3E and Figures S3B–
S3D). Combined, these observations provided strong evidence
Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc. 595
Figure 2. CD36 Moves along Linear Trajectories that Enhance Receptor Clustering
(A) CD36 trajectories in a human macrophage from a 10 Hz/10 s movie (Movie S1). Scale bar, 3 mm. Red, linear trajectories; cyan, isotropic-unconfined
trajectories; blue, isotropic-confined trajectories; purple, isotropic-undetermined trajectories, i.e., trajectories that are isotropic but too short for MSS analysis
(lasting less than 20 frames). Arrow points to perinuclear region.
(B) Probability of CD36 undergoing the different motion types (from 5455 trajectories in 19 unperturbed cells). Error bars, SEM from 200 bootstrap samples
(C) Mean deviation, in degrees, of the orientation of linear trajectories from a perfect radial arrangement about the perinuclear region. Error bars, SEM from
19 unperturbed cells for CD36 and 20 simulations. **p value < 10?4.
596 Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc.
that the linear motion of CD36 on the surface of macrophages
resulted from diffusion within linear confinement regions.
The Three Different Motion Types Have Common
Differences or similarities between the motion characteristics
of the linear, isotropic-unconfined and isotropic-confined re-
ceptors could give further insight into the regulation of CD36
motion in the membrane. Thus, we first compared the confine-
ment width of receptors in linear trajectories to the confinement
dimension of isotropic-confined receptors. We approximated
linear confinement regions by rectangles and isotropic confine-
ment regions by squares. Interestingly, the confinement width
of linearly moving receptors and the confinement dimension of
isotropic-confined receptors were similar, both with a median
of 190 nm (Figures 3E and 3F).
Next, we compared the diffusion coefficient between the three
motion categories (Figure 3G). To accommodate the anisotropic
geometry of linear trajectories, which caused apparent differ-
ences between movements parallel and perpendicular to the
orientation axis (Figure 3C), we also calculated for linearly
moving receptors their 1D diffusion coefficients parallel and
perpendicular to the orientation axis (Figure 3H) (Long and Vu,
2010). The 1D parallel diffusion coefficient was ?0.1 mm2/s,
similar to what was previously measured for receptors diffusing
freely in the plane of the membrane (Serge et al., 2003). Impor-
tantly, the 1D perpendicular diffusion coefficient was similar to
the diffusion coefficient of isotropic-confined trajectories, indi-
cating that both motion types were generated by one diffusive
movement that was confined within either linear regions or small
ries had an apparent diffusion coefficient <0.1 mm2/s also sug-
gested that these trajectories did not undergo truly free diffusion
10 Hz (Saxton and Jacobson, 1997).
CD36 Linear Motion Does Not Depend on Rafts
CD36 has been reported to localize in the cholesterol-enriched
microdomains known as rafts (Dorahy et al., 1996; Zeng et al.,
2003). Thus we investigated whether rafts played a role in
organizing CD36 motion in the membrane. First we tracked raft
dynamics using the raft marker cholera toxin subunit B (CTB)
conjugated to Alexa555 (Brown and London, 1998) (Figure S4A).
We found that most rafts exhibited isotropic, primarily confined
diffusion, although a small fraction exhibited radially arranged
linear motion (Figures 4A–4C). Of note, radially arranged linear
motion was not a general feature of macrophage membrane
components; Fcg receptors, for example, did not exhibit any
(Figures S4B–S4D). Next, we tracked rafts or CD36 after treating
the macrophages with methyl-b-cyclodextrin (MbCD; 10 mM for
30 min) which depleted ?50% of cholesterol from the cells
(Figures S4E and S4F). While MbCD treatment disrupted raft
motion as previously reported (Kilsdonk et al., 1995; Ohtani
et al., 1989), it had no effect on CD36 motion (Figures 4D and
4E). These results indicated that while some raft-associated
molecules could undergo linear motion, the association of
CD36with cholesterol-enriched microdomains wasnotessential
for it to move linearly.
CD36 Linear Motion Depends on the Cortical
The actin cytoskeleton has been previously implicated in regu-
lating membrane protein dynamics (Andrews et al., 2008; Chung
et al., 2010; Goswami et al., 2008; Kaizuka et al., 2009; Plowman
et al., 2005; Suzuki et al., 2007). Therefore, we investigated
whether CD36 motion depended on the actin cytoskeleton.
Indeed, macrophage treatment with latrunculin B (10 mM for
20 min) to depolymerize F-actin markedly reduced the fraction
of linearly moving CD36 (Figures 5A and 5B). These relatively
short incubation periods sufficiently preserved the actin cortex
to maintain stable cell-substrate adhesion for single-molecule
imaging, yet receptor motion was disrupted. Macrophage
treatment with blebbistatin (10 mM for 10 min), a specific inhibitor
of myosin II, also decreased the fraction of linearly moving
receptors (Figures 5Aand 5C).Thus, the motion of CD36 in linear
confinement regions depended on the integrity and flow of the
cortical actomyosin meshwork.
CD36 Linear Motion Depends on MTs
The dependence of CD36 linear motion on the cortical actomy-
osin meshwork raised the question of what could underlie the
formation of the linear structures in the path of CD36. The radial
arrangement of the linear trajectories around the nucleus sug-
gested that MTs, closely apposed to the cell cortex in macro-
phages (Figure S5A), could play a role. Previous studies have
implicated MTs in regulating receptor dynamics (Bouzigues
et al., 2007; de Keijzer et al., 2011; Serge et al., 2003), although
generally resulting in directed movement and not ‘‘1D diffusion’’
as observed for CD36.
Using two-color imaging of Qdot-labeled CD36 and Cy3-
immunolabeled MTs in fixed cells, we found a significant
fraction of receptors colocalizing with MTs, 27% ± 1% (Figures
6A and 6B and Figures S5B–S5D), in remarkable agreement
with the fraction of CD36 diffusing within linear confinement
magenta and green, trajectories after splitting. Scale bar, 500 nm.
(E) Stills of the receptors whose trajectories are shown in (D). Arrow color coding same as in (D). The two receptors are last detected at time points 7 and 8.8 s.
Scale bar, 2 mm.
(F) Overall probability of labeled CD36 undergoing merging and splitting events. Error bar, SEM from 200 bootstrap samples.
(orange, right y axis), for linearly moving receptors. Similar results were obtained for Brownian receptors (not shown). This analysis was not possible for confined
(H) Intensity distributions of particles per motion type. Inset: box plots showing second quartile, median and third quartile. *Median comparison p value < 0.05.
(I) Conditional probability of CD36 merging and splitting per motion type. Error bars, SEM from 200 bootstrap samples. **p < 10?4.
See also Figure S2.
Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc. 597
Figure 3. CD36 Linear Motion Is Diffusion within Linear Confinement Regions
(A) Illustration of run time and frame-to-frame displacement decomposition into parallel and perpendicular components.
(B) Distribution of frame-to-frame displacement parallel components and run times toward and away from the perinuclear region.
(C) Mean magnitude of frame-to-frame displacement parallel (filled circles) and perpendicular (open squares) components at different sampling rates. Error bars,
SEM from >1500 data points per sampling rate. Upper x axis, sampling rates; lower x axis, corresponding time between frames. Dashed line, linear scaling of
parallel displacement with time; solid line, square root scaling with time. Red symbols, Cy3 data; blue symbols, Qdot data.
598 Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc.
regions (Figure 2B). Live-cell imaging of Qdot-labeled CD36
in macrophages transduced with baculovirus to express
tubulin-GFP revealed that linear CD36 trajectories primarily
colocalized with MTs (Figure 6C), and were more abundant in
areas where MTs were more organized (right versus left side
of cell in Figure 6D). Consistent with these observations,
macrophage treatment with nocodazole (50 mM for 30 min)
to depolymerize MTs significantly decreased the fraction of
receptors undergoing linear motion (Figures 6E and 6F). There-
fore, in addition to the cortical actomyosin meshwork, MTs
(D) Distributions of run times, in frames, at the indicated sampling rates.
isotropic-confined trajectory. Inset box size, 200 3 200 nm. Histogram (right, in cyan): distribution of receptor positions across the width of the linear trajectory.
(F) Distribution of confinement widths for linear and isotropic-confined trajectories. Inset: box plots showing the second quartile, median and third quartile. ns,
median comparison p value > 0.05.
(G) Distribution of 2D diffusion coefficients for different motion types.
(H) Box plots showing the second quartile, median and third quartile of the 2D diffusion coefficients shown in (G) and of the 1D diffusion coefficients parallel
and perpendicular to the orientation axis of linear trajectories. The 1D parallel diffusion coefficient is shown in magenta and green stripes (colors used in A) to
indicate that it includes movement both toward and away from the perinuclear region. *, ns: median comparison p < 0.05 and p > 0.05, respectively.
See also Figure S3.
Figure 4. CD36 Linear Motion Does Not Depend on Rafts
(A) Alexa555-conjugated CTB trajectories in a primary human macrophage from a 10 Hz/10 s movie. Trajectory color coding and arrow as in Figure 2A. Scale
bar, 3 mm.
(B) Probability of CTB undergoing the different motion types (from 14,805 trajectories in 23 unperturbed cells). Error bars, SEM from 200 bootstrap samples.
(C) Mean deviation, in degrees, of the orientation of linear trajectories from a perfect radial arrangement about the perinuclear region. Error bars, SEM from
23 unperturbed cells for CTB and 20 simulations. **p value < 10?4.
(D and E)Relative change in CTB (D) and CD36 (E) motion typeprobabilities after treatment with MbCD in comparison to unperturbed cells (from 6719trajectories
in 22 MbCD-treated cells for CTB and 1603 trajectories in ten MbCD-treated cells for CD36). Error bars, SEM from 200 bootstrap samples. *, **, ns: changes
associated with p < 0.05, p < 10?4, and p > 0.1, respectively.
See also Figure S4.
Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc. 599
also played a role in mediating the radially arranged linear
motion of CD36.
Motion Changes upon Cytoskeleton Perturbation
Reduce CD36 Clustering
The propensity of unligated CD36 to form metastable clusters
depended on its mobility (Figures 2H and 2I). Thus, we sus-
would also perturb its clustering. To investigate this, we com-
pared clustering in the different conditions using two measures:
(1) modal analysis of the particle intensity histograms (similar to
Figure 1B) and (2) the probability of receptor merging and
splitting (similar to Figure 2F). At face value, all drug treatments
reduced receptor clustering, with nocodazole showing the
weakest reduction (Figures 7A and 7B). However, in addition to
altering receptor motion, these drug treatments reduced
receptor density on the cell surface (Figure 7C). To separate
the effects of motion perturbation and density reduction, we
repeated the clustering comparisons between conditions using
a subset of cells (called ‘‘density-normalized subset’’) that had
comparable receptor densities (Figures 7D and 7E). In this
subset, latrunculin and blebbistatin treatments reduced clus-
tering to a similar extent as they did in all cells, implying that
in reducing receptor clustering. The effects of nocodazole on the
density-normalized subset were not significant, implying that
with this drug the reduction in receptor density was more likely
Figure 5. CD36 Linear Motion Depends on the
Cortical Actomyosin Meshwork
(A) Relative change in CD36 motion type probabilities
after treatment with latrunculin or blebbistatin in compar-
ison to unperturbed cells (from 4886 trajectories in 20
latrunculin-treated cells and 2588 trajectories in 12 bleb-
bistatin-treated cells). Error bars, SEM from 200 bootstrap
samples. *, **, ns: changes associated with p < 0.05,
p < 10?4and p > 0.1, respectively.
(Band C) CD36 trajectoriesafter treatment withlatrunculin
B(B) or blebbistatin (C) from 10Hz/10 smovies. Trajectory
color coding and arrow as in Figure 2A. Scale bar, 3 mm.
See also Figure S7.
the cause of decreased receptor clustering.
Overall, this analysis provided evidence that
clustering of unligated CD36 was regulated by
geometric constraints mediated by cortical
Cytoskeleton Perturbation Inhibits CD36
Function and Signaling
Biochemical evidence suggests that CD36
clustering is essential for its signaling and
internalization upon engagement to multivalent
ligands (Daviet et al., 1997; McGilvray et al.,
2000). The metastable clustering of unligated
CD36 might prime the cell and facilitate its
response when exposed to ligand that, in turn,
could stabilize the clusters and/or increase their
size, leading to receptor activation. Our observation that CD36
diffusion within linear confinement regions promoted unengaged
receptor clustering thus led us to hypothesize that the cytoskel-
eton-mediated organization of receptor diffusion in the mem-
brane might enhance CD36 responsiveness to ligand.
To test this, we monitored the response of macrophages to
oxLDL, a physiologically important ligand that binds to macro-
ization of oxLDL and the activation of c-Jun N-terminal kinase
(JNK), a well-established effector of CD36 (Kennedy et al.,
2011; Rahaman et al., 2006).
In unperturbed macrophages, DiI-labeled oxLDL bound
rapidly to the surface. Within 5 min of DiI-oxLDL addition, a frac-
tion of the receptor-ligand complexes, 25% ± 2%, moved along
linear trajectories as described for CD36 (Figures 7F and 7G).
This behavior was observed before anysignificant internalization
occurred, as verified by acid-stripping the cells, which removed
most of the bound oxLDL and eliminated most of the linearly
moving complexes (Figure 7G). After 20 min of oxLDL addition,
?60% of the oxLDL was internalized and could no longer be dis-
placed from the cells by an acid wash (data not shown). Binding
and internalization of oxLDL were associated with JNK activa-
tion, as assessed using antibodies that specifically recognized
the phosphorylated form of its substrate cJun (phospho-cJun;
Figure 7H and Figure S6B).
To assess the effect of perturbing CD36 motion and clustering
on its ability to signal and internalize oxLDL, we pretreated the
600 Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc.
cells with latrunculin, blebbistatin or nocodazole before adding
oxLDL. Pretreatment with all three agents reduced ligand
internalization (Figure 7I), in all cases to a larger extent than
what would be expected from the reduction in receptor density
alone (Table S1). This reduction did not result from wholesale
inhibition of endocytosis, as the same drug treatments did not
significantly alter transferrin internalization (Figure 7J). Pretreat-
ment with blebbistatin or nocodazole also suppressed c-Jun
phosphorylation (Figure 7H and Figure S6B). The effect of
latrunculin on CD36-mediated JNK activation could not be
evaluated; as described in other systems (Subbaramaiah et al.,
2000; Yujiri et al., 1999), actin-perturbation itself markedly acti-
vated JNK, precluding subsequent stimulation via CD36 (Fig-
ure 7H; Figure S6B). The results of these experiments combined
thus supported the hypothesis that the cytoskeleton-mediated
organization of CD36 diffusion was essential for its proper sig-
naling and ability to internalize ligands.
CD36 Diffusion in the Membrane Is Spatially Organized
by the Cortical Cytoskeleton
Our study reveals that the diffusion of CD36 in the membrane
of human macrophages is regulated by interactions between
CD36 and the cytoskeleton. Indeed, Triton extraction experi-
ments provide evidence that in macrophages CD36 interacts
with F-actin (Figure S7), although most likely transiently and indi-
rectly, perhaps via integrins (Thorne et al., 2000). It is tempting to
speculate thatsomelipidmicrodomains exhibitradiallyarranged
linear motion like CD36 because of similar interactions with
F-actin (Harder et al., 1997; Viola and Gupta, 2007).
While details of the molecular mechanism by which the cyto-
skeleton controls CD36 diffusion in the membrane remain to
be determined, our current data suggest two models: in regions
without MTs, the submembranous actin meshwork is isotropic;
thus receptors diffuse isotropically and, due to CD36-actin inter-
actions, would get slowed down (Saxton and Jacobson, 1997)
and often confined. On the other hand, where MTs are apposed
to the membrane (Manneville et al., 2003), they might disrupt the
integrity of the submembranous actomyosin meshwork by
chemical and/or mechanical interactions (Rodriguez et al.,
2003), generating actin-delimited channels along which CD36
would move relatively unobstructed (model 1 in Figure S5E).
Alternatively, the submembranous actin meshwork is isotropic
everywhere, including regions with MTs, however MTs might
locally detach the actin meshwork from the plasmalemma,
generating linear patches of bare membrane where receptor
diffusion is unimpeded by actin (model 2 in Figure S5E). The re-
markable conservation of confinement width between isotropic
compartments and linear channels favors the first model. In
either case, the reversible interactions between CD36 and
F-actin would lead to the observed bias of CD36 localization
toward the channel edges (Figure 3E).
The compartmentalization of CD36 diffusion in the membrane
is reminiscent of the membrane matrix corrals proposed by
Kusumi et al. (2005a, 2005b). However, there are two main
differences between CD36 compartments and those described
previously. First, for CD36 we observe not only isotropic
compartments but also linear channels. Second, the confine-
ment of CD36 seems to be more long-lived than the previously
observed corrals (at least 10 s versus 1 ms timescale [Kusumi
et al., 2005a]). Of note, cortical actin turnover is also on the order
of tens of seconds (McGrath et al., 1998; Ponti et al., 2005).
Therefore, we propose that the linear and isotropic compart-
ments described here are salient features of the cortical archi-
tecture in macrophages, controlling receptor diffusion over
long periods and thus having major implications for the steady-
state of CD36-mediated signal transduction.
The Spatial Organization of CD36 Diffusion
in the Membrane Enhances Signal Transduction
A critical implication of the compartmentalization of CD36
diffusion is its impact on receptor interactions. In particular,
our data show that CD36 diffusion in linear channels promotes
receptor encounters and clustering, which can be attributed to
the unique geometry of linear channels: when compared with
the small regions of isotropic confinement, the comparatively
long linear channels accommodate more receptors and offer
them greater freedom of movement parallel to the orientation
axis. Conversely, when compared to free diffusion, linear chan-
nels restrict movement perpendicular to the orientation axis,
thereby increasing the effective local density by approximately
While the exact molecular nature of the CD36 clusters remains
to be determined, our study provides evidence that the meta-
stable clusters of unligated CD36 prime the cell to respond
cytoskeleton organization and signaling could be probed only
by global disruption of actin and MT dynamics. Thus we cannot
formally exclude that effects besides reduced receptor clus-
tering contribute to the documented shifts in signaling down-
working via different molecular mechanisms had similar effects
on CD36 function provides compelling evidence that reducing
unligated CD36 clustering—common among all the pertur-
bants—is a major contributor to the inhibition of CD36 function
upon disruption of cytoskeleton organization.
In conclusion, by dictating the spatial organization of receptor
motion, cortical cytoskeletal structures appear to play a critical
role in CD36 signal transduction, where the outside-in activation
by inside-out feedback regulating receptor clustering. We spec-
ulate that this reciprocal interaction may be a general mecha-
nism for enhancing or silencing signals at the level of the plasma
Macrophage Isolation and Culture
Human blood samples from healthy volunteers were collected with heparin.
Peripheral blood mononuclear cells were isolated by density-gradient cen-
trifugation using Ficoll-Paque Plus (Amersham). Cells were resuspended
(107cells/ml) in RPMI-1640 with L-glutamine containing 10% heat-inactivated
fetal calf serum (FCS; from Wisent) and seeded onto 18 mm glass coverslips
(Fisher Scientific) at 5 3 105cells/coverslip. After 1 hr at 37?C, nonadherent
cells were removed by multiple washes with Hank’s buffered saline solution
(HBSS). Adherent cells were incubated in RPMI-1640 with 10% FCS and
Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc. 601
Figure 6. CD36 Linear Motion Depends on MTs
(A) Fixed-cell image overlaying Qdot-labeled CD36 (cyan) and Cy3-immunolabeled MTs (red). Scale bar, 3 mm.
(B) Fraction of CD36 particles colocalizing with MTsin experimentaldata, and simulated fraction ofcolocalization for arandom distributionof thesame numberof
particles. Experimental fraction calculated from three images. Error bars, SEM from three experiments and associated simulations. **p value < 10?4.
(C and D) Live-cell images overlaying CD36 trajectories (CD36 tracked using Qdot labeling) and tubulin-GFP, from 10 Hz/14 s movies. Scale bars, 10 mm. CD36
trajectories are color-coded based on motion type, as in Figure 2A. Insets in (C): Zoom-in on two areas highlighting the colocalization of CD36 trajectories
first and last time points in each trajectory. Scale bars in insets, 1 mm.
602 Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc.
100 U/ml penicillin, 100 mg/ml streptomycin and 10 mg/ml polymyxin B
(Invitrogen) for 7–14 days.
Monoclonal antibodiestohuman CD36 (clone 131.1; mouse IgG1)were thegift
of Dr. N. Tandon (Otsuka AmericaPharmaceutical, Inc., Rockville, MD). Mono-
valent Fab fragments were prepared using the ImmunoPure Fab Preparation
Kit (Pierce). To minimize nonspecific binding, cells were blocked with 4%
donkey serum for 10 min, then incubated with anti-CD36 Fab fragments at
1:2000–1:3000 dilution for 10 min. After washing with HBSS, cells were
incubated with either (1) Cy3-conjugated donkey anti-mouse Fab fragments
(Jackson ImmunoResearch Laboratories) at a 1:3000 dilution for 10 min; (2)
Qdot 655-goat F(ab’)2anti-mouse IgG conjugates at 1:2000–1:3000 dilution
for 10–15 min (to prevent crosslinking, unoccupied antibody-binding sites on
the Qdots were blocked with nonimmune mouse IgG antibody [10 mg/ml]); or
(3) biotinylated secondary Fab (1:1000 dilution) followed by streptavadin-655
Qdots (1:10,000 dilution); medium with excess free biotin was then added to
block sites on avidin, thereby preventing crosslinking. All labeling protocols
led to comparable results.
All of the preceding steps were performed at 4?C to minimize lateral mobility
and clustering. Cells were then warmed to 37?C before filming.
To visualize cell-associated oxLDL, cells were labeled with DiI-oxLDL (1:1000)
for 5 min and washed with HBSS. To visualize rafts, cells were labeled with
0.5 mg/ml AlexaFluor 555-conjugated CTB for 10 min on ice.
Live-cell imaging was performed using a Zeiss Axiovert 200 epifluorescence
microscope equipped with a 1003 oil-immersion objective (NA 1.45), a
custom-made 2.53 lens and either a Cy3 filter set or a 32012 cube from
Chroma Technology for Qdots. Illumination was provided by an Exfo X-Cite
120 light source, and a Hamamatsu 9100-13 deep-cooled EM-CCD camera
was used for recording. Image acquisition was controlled by Volocity
(Perkin-Elmer). Images were acquired continuously at 10, 33, 62.5, and 125
frames per second for 10–20 s.
To strip anti-CD36 antibodies, oxLDL, transferrin, or CTB bound extracellu-
larly, cells were incubated with 200 mM acetic acid and 150 mM NaCl
(pH 2.8) for 5 min at 37?C. Acid stripping was terminated by transferring the
cells to prewarmed medium RPMI 1640 buffered with HEPES.
Transferrin and oxLDL Uptake Assays
Primary human macrophages (7–10 days old) were serum-starved in HEPES-
buffered RPMI (Wisent Inc.) for 1 hr, treated with or without latrunculin B,
blebbistatin or nocodazole followed by addition of either 25 mg/ml transferrin
Alexa Fluor-555 conjugate (Molecular Probes) for 30 min at 37?C or 50 mg/ml
DiI-oxLDL for 20 min at 37?C. Cells were then acid-washed to remove extra-
cellular adherent ligand and fixed in 4% paraformaldehyde. The amount of
cell-associated ligand remaining (i.e., internalized) was quantified by acquiring
epifluorescence images that were analyzed using ImageJ software.
To quantify JNK activation, the cells were serum-starved for 3 hr and treated
with the indicated cytoskeleton perturbants and/or oxLDL. The cells were
then fixed, permeabilized, and immunostained with phospho-cJun antibodies
(1:200) followed by Cy3-labeled secondary and counterstained with DAPI.
Images were acquired by epifluorescence and quantified using ImageJ.
All experiments were performed in triplicate, with at least 50 cells analyzed
To visualize microtubules in fixed cells, cells were treated with methanol for
2 min at ?20?C, rinsed, and incubated with monoclonal Cy3-conjugated
anti-btubulinantibody (TUB2.1). Tovisualize microtubulesinlivecells, macro-
phages (z30% confluent) were incubated with 400 CellLight Tubulin-GFP
BacMam 2.0 baculovirus particles per cell for 16 hr at 37?C and visualized
by epifluorescence with the same system used for single-particle tracking.
The imaged molecules were detected and tracked as described in (Jaqaman
et al., 2008). In brief, particle subpixel positions and intensities—which were
subresolution features even for receptor clusters—were estimated by (1) de-
tecting significant local intensity maxima and (2) fitting Gaussian kernels
approximating the two-dimensional point spread function of the microscope.
Importantly, our algorithm fitted multiple Gaussians simultaneously (i.e.,
Gaussian mixture-models) for particles with overlapping signals, enhancing
the accuracy and resolution of the detection.
The detected particles were tracked using a two step particle tracking
tories by closing gaps and capturing merging and splitting events. In the first
between particlepairs.Thetracks generatedinthisstepstarted andended not
of temporary particle disappearance, merging, and splitting. Thus, in the
second step, the algorithm generated complete particle trajectories by linking
the track segments from the first step in three ways: end-to-start, to close
gaps resulting from temporary particle disappearance; end-to-middle, to
capture particle merging events; and middle-to-start, to capture particle split-
ting events. In this step, all track segments were allowed to compete with
each other, resulting in a spatially and temporally optimal global assignment
of track pairs, adding stability to the tracker under high particle density condi-
tions. The cost functions employed to weigh competing particle and track
segment assignments were based on distance and intensity, as well as on
motion models that aided tracking by allowing particle position propagation
via the Kalman filter. All movies were analyzed using the same tracking param-
eters. Tracks lasting at least five frames were retained for trajectory analysis.
classified trajectory shape based on the degree of anisotropy of the scatter of
particle positions along a trajectory (Huet et al., 2006; Jaqaman et al., 2008).
The second extracted diffusion types using an MSS analysis of particle
displacements (Ewers et al., 2005; Ferrari et al., 2001). The MSS analysis
was applied to full 2D displacements and to 1D displacements in linear trajec-
tories. Particle diffusion coefficients were calculated from the MSS analysis.
The confinement dimension for confined and linear trajectories was derived
via eigenvalue decomposition of the variance-covariance matrix of particle
positions along each trajectory.
The radial arrangement of linear trajectoriesaround the perinuclear region of
acell wasquantified by calculating for each lineartrajectory the angle between
its orientation axis and the ray connecting the ‘‘cell center’’ to the trajectory
center, and averaging the angle over all linear trajectories in the cell.
Please see Extended Experimental Procedures for a full description of these
Please see Extended Experimental Procedures for reagents, immunolabeling
controls, measuring cholesterol content and quantifying cytoskeleton-associ-
ated CD36; also for a detailed description of apparent fusion time simulations
and receptor density estimation.
(E) CD36 trajectories in a nocodazole-treated cell from a 10 Hz/10 s movie. Trajectory color coding and arrow as in Figure 2A. Scale bar, 3 mm.
(F) Relative change in CD36 motion type probabilities after treatment with nocodazole in comparison to unperturbed cells (from 2849 trajectories in 16 noco-
dazole-treated cells). Error bars, SEM from 200 bootstrap samples. **, ns: changes associated with p < 10?4and p > 0.1, respectively.
See also Figure S5.
Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc. 603
604 Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc.
Supplemental Information includes Extended Experimental Procedures,
seven figures,four movies, and one table and can be found with this article on-
line at doi:10.1016/j.cell.2011.06.049.
WorkintheGrinstein labwassupportedbytheHeartand StrokeFoundationof
Ontario and by Canadian Institutes of Health Research Grant MOP-102474.
Work in the Danuser lab was supported by NIH grant U01 GM67230. K.J.
is an investigator in the Center for Cell Decision Processes (NIH P50
GM068762). HK was supported in part by the Uehara Memorial Foundation,
the Mochida Memorial Foundation for Medical and Pharmaceutical Research
and the Kanae Foundation for the Promotion of Medical Science. N.T. is sup-
ported by an Alberta Heritage Foundation for Medical Research Scholarship.
S.G. is the current holder of the Pitblado Chair in Cell Biology.
Received: September 1, 2010
Revised: April 28, 2011
Accepted: June 21, 2011
Published: August 18, 2011
Andrews, N.L., Lidke, K.A., Pfeiffer, J.R., Burns, A.R., Wilson, B.S., Oliver,
antigen-induced receptor immobilization. Nat. Cell Biol. 10, 955–963.
Berg, H.C. (1993). Random Walks in Biology (Princeton, NJ: Princeton Univer-
Bouzigues, C., Morel, M., Triller, A., and Dahan, M. (2007). Asymmetric redis-
tribution of GABA receptors during GABA gradient sensing by nerve growth
cones analyzed by single quantum dot imaging. Proc. Natl. Acad. Sci. USA
Brown, D.A., and London, E. (1998). Functions of lipid rafts in biological
membranes. Annu. Rev. Cell Dev. Biol. 14, 111–136.
Cambi, A., Joosten, B., Koopman, M., de Lange, F., Beeren, I., Torensma, R.,
Fransen, J.A., Garcia-Parajo ´, M., van Leeuwen, F.N., and Figdor, C.G. (2006).
Organization of the integrin LFA-1 in nanoclusters regulates its activity. Mol.
Biol. Cell 17, 4270–4281.
Cebecauer, M., Spitaler, M., Serge ´, A., and Magee, A.I. (2010). Signalling
complexes and clusters: functional advantages and methodological hurdles.
J. Cell Sci. 123, 309–320.
Chung, I., Akita, R., Vandlen, R., Toomre, D., Schlessinger, J., and Mellman, I.
(2010). Spatial control of EGF receptor activation by reversible dimerization on
living cells. Nature 464, 783–787.
Collins, R.F., Touret, N., Kuwata, H., Tandon, N.N., Grinstein, S., and Trimble,
W.S. (2009). Uptake of oxidized low density lipoprotein by CD36 occurs by an
actin-dependent pathway distinct from macropinocytosis. J. Biol. Chem. 284,
Daviet, L., Malvoisin, E., Wild, T.F., and McGregor, J.L. (1997). Thrombospon-
din induces dimerization of membrane-bound, but not soluble CD36. Thromb.
Haemost. 78, 897–901.
de Keijzer, S., Galloway, J., Harms, G.S., Devreotes, P.N., and Iglesias, P.A.
(2011). Disrupting microtubule network immobilizes amoeboid chemotactic
receptor in the plasma membrane. Biochim. Biophys. Acta 1808, 1701–1708.
Dorahy, D.J., Lincz, L.F., Meldrum, C.J., and Burns, G.F. (1996). Biochemical
isolation of a membrane microdomain from resting platelets highly enriched in
the plasma membrane glycoprotein CD36. Biochem. J. 319, 67–72.
Douglass, A.D., and Vale, R.D. (2005). Single-molecule microscopy reveals
plasma membrane microdomains created by protein-protein networks that
exclude or trap signaling molecules in T cells. Cell 121, 937–950.
Endemann, G., Stanton, L.W., Madden, K.S., Bryant, C.M., White, R.T., and
Protter, A.A. (1993). CD36 is a receptor for oxidized low density lipoprotein.
J. Biol. Chem. 268, 11811–11816.
Espenel, C., Margeat, E.,Dosset, P.,Arduise, C.,Le Grimellec,C., Royer, C.A.,
Boucheix, C., Rubinstein, E., and Milhiet, P.E. (2008). Single-molecule analysis
of CD9 dynamics and partitioning reveals multiple modes of interaction in the
tetraspanin web. J. Cell Biol. 182, 765–776.
Ewers, H., Smith, A.E., Sbalzarini, I.F., Lilie, H., Koumoutsakos, P., and Helen-
ius, A. (2005).Single-particle tracking of murine polyoma virus-like particles on
live cells and artificial membranes. Proc. Natl. Acad. Sci. USA 102, 15110–
Febbraio, M., Hajjar, D.P., and Silverstein, R.L. (2001). CD36: a class B scav-
enger receptor involved in angiogenesis, atherosclerosis, inflammation, and
lipid metabolism. J. Clin. Invest. 108, 785–791.
Ferrari, R., Manfroi, A.J., and Young, W.R. (2001). Strongly and weakly self-
similar diffusion. Physica D 154, 111–137.
Foster, L.J., De Hoog, C.L., and Mann, M. (2003). Unbiased quantitative
proteomics of lipid rafts reveals high specificity for signaling factors. Proc.
Natl. Acad. Sci. USA 100, 5813–5818.
Friedrichson, T., and Kurzchalia, T.V. (1998). Microdomains of GPI-anchored
proteins in living cells revealed by crosslinking. Nature 394, 802–805.
Gambin, Y.,Lopez-Esparza,R.,Reffay, M.,Sierecki, E.,Gov,N.S., Genest,M.,
Hodges, R.S., and Urbach, W. (2006). Lateral mobility of proteins in liquid
membranes revisited. Proc. Natl. Acad. Sci. USA 103, 2098–2102.
Goswami, D., Gowrishankar, K., Bilgrami, S., Ghosh, S., Raghupathy, R.,
Chadda, R., Vishwakarma, R., Rao, M., and Mayor, S. (2008). Nanoclusters
Figure 7. CD36 Linear Motion Enhances CD36 Function
(A and B) Effect of cytoskeleton perturbations on CD36 clustering as measured by modal analysis of the intensity histogram (A) and the relative change in the
probability of labeled CD36 undergoing merging and splitting events (B). Error bars in (B), SEM from 200 bootstrap samples. **, ns: changes associated with
p < 10?4and p > 0.1, respectively.
(C) CD36 density on cell surface, measured by immunofluorescence at saturating antibody concentrations. Data from three experiments with 6–12 data points
each. Box plots show median (central mark), 25th and 75th percentiles (lower and upper box edges), extent of inlier data points (whiskers) and outliers (blue
crosses). *p values < 0.05 in both the Wilcoxon rank-sum test comparing medians and the Kolmogorov-Smirnov test comparing distributions.
(D and E) As in (A) and (B) but calculated from a density-normalized subset of cells. See (A) and (B) for details.
(F)DiI-labeled oxLDL trajectories 5minafteradditiontoamacrophage,froma10Hz/5 smoviebefore acidwashanda10Hz/5smovie afteracidwash.Trajectory
color coding and arrows as in Figure 2A. Scale bar, 3 mm.
(G) Probability of oxLDL undergoing linear motion before and after acid washing (from 994 (before) and 484 (after) trajectories in six cells). Error bars, SEM from
200 bootstrap samples. **p < 10?4.
(H) Ratio of phospho-cJun intensity in nucleus to cytoplasm measured by immunofluorescence. Higher ratios indicate more JNK activation. Data from three
(I) oxLDL uptake. Data from three experiments with 5–30 data points each. Box plots and *, as in (C).
(J) Transferrin uptake. Data points from three to five experiments with 5–33 data points each. Box plots and ns, as in (H).
See also Figure S6 and Table S1.
Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc. 605
of GPI-anchored proteins are formed by cortical actin-driven activity. Cell 135,
Groves,J.T.,and Kuriyan, J.(2010). Molecular mechanisms insignal transduc-
tion at the membrane. Nat. Struct. Mol. Biol. 17, 659–665.
Harder,T., Kellner,R., Parton, R.G., and Gruenberg, J.(1997). Specificrelease
of membrane-bound annexin II and cortical cytoskeletal elements by seques-
tration of membrane cholesterol. Mol. Biol. Cell 8, 533–545.
Heldin, C.H. (1995). Dimerization of cell surface receptors in signal transduc-
tion. Cell 80, 213–223.
Hess, S.T., Gould, T.J., Gudheti, M.V., Maas, S.A., Mills, K.D., and Zimmer-
berg, J. (2007). Dynamic clustered distribution of hemagglutinin resolved at
40 nm in living cell membranes discriminates between raft theories. Proc.
Natl. Acad. Sci. USA 104, 17370–17375.
Huet,S.,Karatekin, E.,Tran, V.S.,Fanget,I., Cribier, S.,and Henry, J.P. (2006).
Analysis of transient behavior in complex trajectories: application to secretory
vesicle dynamics. Biophys. J. 91, 3542–3559.
Iino, R., Koyama, I., and Kusumi, A. (2001). Single molecule imaging of green
fluorescent proteins in living cells: E-cadherin forms oligomers on the free cell
surface. Biophys. J. 80, 2667–2677.
Jaqaman, K., Loerke, D., Mettlen, M., Kuwata, H., Grinstein, S., Schmid, S.L.,
and Danuser, G. (2008). Robust single-particle tracking in live-cell time-lapse
sequences. Nat. Methods 5, 695–702.
Kaizuka, Y., Douglass, A.D., Vardhana, S., Dustin, M.L., and Vale, R.D. (2009).
The coreceptor CD2 uses plasma membrane microdomains to transduce
signals in T cells. J. Cell Biol. 185, 521–534.
Kasai, R.S., Suzuki, K.G.N., Prossnitz, E.R., Koyama-Honda, I., Nakada, C., Fu-
jiwara, T.K., and Kusumi, A. (2011). Full characterization of GPCR monomer-
Kennedy, D.J., Kuchibhotla, S., Westfall, K.M., Silverstein, R.L., Morton, R.E.,
and Febbraio, M. (2011). A CD36-dependent pathway enhances macrophage
and adipose tissue inflammation and impairs insulin signalling. Cardiovasc.
Res. 89, 604–613.
Kilsdonk, E.P.C., Yancey, P.G., Stoudt, G.W., Bangerter, F.W., Johnson, W.J.,
Phillips, M.C., and Rothblat, G.H. (1995). Cellular cholesterol efflux mediated
by cyclodextrins. J. Biol. Chem. 270, 17250–17256.
Kusumi, A., Ike, H., Nakada, C., Murase, K., and Fujiwara, T. (2005a). Single-
molecule tracking of membrane molecules: plasma membrane compartmen-
talization and dynamic assembly of raft-philic signaling molecules. Semin.
Immunol. 17, 3–21.
Kusumi, A., Nakada, C., Ritchie, K., Murase, K., Suzuki, K., Murakoshi, H.,
Kasai, R.S., Kondo, J., and Fujiwara, T. (2005b). Paradigm shift of the plasma
membrane concept from the two-dimensional continuum fluid to the parti-
tioned fluid: high-speed single-molecule tracking of membrane molecules.
Annu. Rev. Biophys. Biomol. Struct. 34, 351–378.
Livnah, O., Stura, E.A., Middleton, S.A., Johnson, D.L., Jolliffe, L.K., and
Wilson, I.A. (1999). Crystallographic evidence for preformed dimers of erythro-
poietin receptor before ligand activation. Science 283, 987–990.
Long, B.R., and Vu, T.Q. (2010). Spatial structure and diffusive dynamics from
single-particle trajectories using spline analysis. Biophys. J. 98, 1712–1721.
Manneville, J.B., Etienne-Manneville, S., Skehel, P., Carter, T., Ogden, D., and
Ferenczi, M. (2003). Interaction of the actin cytoskeleton with microtubules
regulates secretory organelle movement near the plasma membrane in human
endothelial cells. J. Cell Sci. 116, 3927–3938.
McGilvray, I.D., Serghides, L., Kapus, A., Rotstein, O.D., and Kain, K.C. (2000).
parasitized erythrocytes: a role for CD36 in malarial clearance. Blood 96,
McGrath, J.L., Tardy, Y., Dewey, C.F., Jr., Meister, J.J., and Hartwig, J.H.
(1998). Simultaneous measurements of actin filament turnover, filament frac-
tion, and monomer diffusion in endothelial cells. Biophys. J. 75, 2070–2078.
Ohtani, Y., Irie, T., Uekama, K., Fukunaga, K., and Pitha, J. (1989). Differential
effects of alpha-cyclodextrins, beta-cyclodextrins and gamma-cyclodextrins
on human erythrocytes. Eur. J. Biochem. 186, 17–22.
Pike, L.J. (2003). Lipid rafts:bringing order to chaos. J.Lipid Res. 44,655–667.
Plowman, S.J., Muncke, C., Parton, R.G., and Hancock, J.F. (2005). H-ras,
K-ras, and inner plasma membrane raft proteins operate in nanoclusters
with differential dependence on the actin cytoskeleton. Proc. Natl. Acad.
Sci. USA 102, 15500–15505.
Ponti, A., Matov, A., Adams, M., Gupton, S., Waterman-Storer, C.M., and
Danuser, G. (2005). Periodic patterns of actin turnover in lamellipodia and
lamellae of migrating epithelial cells analyzed by quantitative Fluorescent
Speckle Microscopy. Biophys. J. 89, 3456–3469.
Rahaman, S.O., Lennon, D.J., Febbraio, M., Podrez, E.A., Hazen, S.L., and
Silverstein, R.L. (2006). A CD36-dependent signaling cascade is necessary
for macrophage foam cell formation. Cell Metab. 4, 211–221.
Rodriguez, O.C., Schaefer, A.W., Mandato, C.A., Forscher, P., Bement, W.M.,
and Waterman-Storer, C.M. (2003). Conserved microtubule-actin interactions
in cell movement and morphogenesis. Nat. Cell Biol. 5, 599–609.
Sako, Y., Minoghchi, S., and Yanagida, T. (2000). Single-molecule imaging of
EGFR signalling on the surface of living cells. Nat. Cell Biol. 2, 168–172.
Savill, J. (1997). Recognition and phagocytosis of cells undergoing apoptosis.
Br. Med. Bull. 53, 491–508.
membrane dynamics. Annu. Rev. Biophys. Biomol. Struct. 26, 373–399.
Schamel, W.W.A.,Arechaga,I.,Risuen ˜o,R.M.,vanSanten,H.M.,Cabezas,P.,
Risco, C., Valpuesta, J.M., and Alarco ´n, B. (2005). Coexistence of multivalent
and monovalent TCRs explains high sensitivity and wide range of response.
J. Exp. Med. 202, 493–503.
Scott, J.D., and Pawson, T. (2009). Cell signaling in space and time: where
proteins come together and when they’re apart. Science 326, 1220–1224.
Serge, A., Fourgeaud, L., Hemar, A., and Choquet, D. (2003). Active surface
transport of metabotropic glutamate receptors through binding to microtu-
bules and actin flow. J. Cell Sci. 116, 5015–5022.
Sethna, J.P. (2006). Statistical mechanics: entropy, order parameters and
complexity (New York: Oxford University Press).
Sharma, P., Varma, R., Sarasij, R.C., Ira, Gousset, K., Krishnamoorthy, G.,
Rao, M., and Mayor, S. (2004). Nanoscale organization of multiple GPI-
anchored proteins in living cell membranes. Cell 116, 577–589.
Subbaramaiah, K., Hart, J.C., Norton, L., and Dannenberg, A.J. (2000). Micro-
tubule-interfering agents stimulate the transcription of cyclooxygenase-2.
Evidence for involvement of ERK1/2 AND p38 mitogen-activated protein
kinase pathways. J. Biol. Chem. 275, 14838–14845.
Suzuki, K.G.N., Fujiwara, T.K., Sanematsu, F.,Iino, R.,Edidin, M.,and Kusumi,
A. (2007). GPI-anchored receptor clusters transiently recruit Lyn and G alpha
for temporary cluster immobilization and Lyn activation: single-molecule
tracking study 1. J. Cell Biol. 177, 717–730.
Thorne, R.F., Marshall, J.F., Shafren, D.R., Gibson, P.G., Hart, I.R., and Burns,
G.F. (2000). The integrins alpha3beta1 and alpha6beta1 physically and
functionally associate with CD36 in human melanoma cells. Requirement for
the extracellular domain OF CD36. J. Biol. Chem. 275, 35264–35275.
Varma, R., and Mayor, S. (1998). GPI-anchored proteins are organized in
submicron domains at the cell surface. Nature 394, 798–801.
Viola, A., and Gupta, N. (2007). Tether and trap: regulation of membrane-raft
dynamics by actin-binding proteins. Nat. Rev. Immunol. 7, 889–896.
(2007). Architectural dynamics of the meiotic spindle revealed by single-fluo-
rophore imaging. Nat. Cell Biol. 9, 1233–1242.
Yujiri, T., Fanger, G.R., Garrington, T.P., Schlesinger, T.K., Gibson, S., and
Johnson, G.L. (1999). MEK kinase 1 (MEKK1) transduces c-Jun NH2-terminal
kinase activation in response to changes in the microtubule cytoskeleton. J.
Biol. Chem. 274, 12605–12610.
Zeng, Y.C., Tao, N.B., Chung, K.N., Heuser, J.E., and Lublin, D.M. (2003).
Endocytosis of oxidized low density lipoprotein through scavenger receptor
CD36 utilizes a lipid raft pathway that does not require caveolin-1. J. Biol.
Chem. 278, 45931–45936.
606 Cell 146, 593–606, August 19, 2011 ª2011 Elsevier Inc.