Cluster size regulates protein sorting in the
Nin ˜a C. Hartmana, Jeffrey A. Nyeb, and Jay T. Grovesa,c,d,1
Departments ofaChemistry andbChemical Engineering,cHoward Hughes Medical Institute, anddPhysical Biosciences and Materials Sciences Divisions,
Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720
Edited by Douglas T. Fearon, University of Cambridge, Cambridge, United Kingdom, and approved June 18, 2009 (received for review March 12, 2009)
During antigen recognition by T cells, signaling molecules on the T
cell engage ligands on the antigen-presenting cell and organize
into spatially distinctive patterns. These are collectively known as
the immunological synapse (IS). Causal relationships between
large-scale spatial organization and signal transduction have pre-
viously been established. Although it is known that receptor
mechanisms by which different proteins become spatially sorted
remain unclear. These sorting processes contribute a facet of signal
regulation; thus their elucidation is important for ultimately un-
derstanding signal transduction through the T cell receptor. Here
we investigate protein cluster size as a sorting mechanism using
the hybrid live T cell?supported membrane system. The clustering
state of the co-stimulatory molecule lymphocyte function-
associated antigen-1 (LFA-1) is modulated, either by direct anti-
body crosslinking or by crosslinking its intercellular adhesion
molecule-1 ligand on the supported bilayer. In a mature IS, native
LFA-1 generally localizes into a peripheral ring surrounding a
central T cell receptor cluster. Higher degrees of LFA-1 clustering,
induced by either method, result in progressively more central
localization, with the most clustered species fully relocated to the
central zone. These results demonstrate that cluster size directly
influences protein spatial positioning in the T cell IS. We discuss a
sorting mechanism, based on frictional coupling to the actin
cytoskeleton, that is consistent with these observations and is, in
principle, extendable to all cell surface proteins in the synapse.
actin ? mechanobiology ? membrane ? transport ? receptor
from molecular dimensions to the size of the cell itself. One
particularly dramatic example is the T cell immunological syn-
and an antigen-presenting cell (APC), within which a variety of
receptor and adhesion proteins engage their cognate ligands on
the apposed cell surface (1, 2). Before contact, no large-scale
organization is present on either cell surface. However, within
minutes of the initial encounter between an APC displaying
appropriate antigen and a complementary T cell, a highly
coordinated transport process is initiated that ultimately sorts
dozens of membrane proteins on both cell surfaces into a series
of concentric rings. This spatial pattern of proteins is not unique
to T cells; similar structures have also been observed between
natural killer cells and B cells and their target cells, as well as
between certain immune cells and neurons (3, 4). Recent work
has demonstrated the importance of protein spatial distribution
to both effector functions and as a signal regulatory mechanism
(5–7). However, the mechanism of IS formation is unclear.
In the case of the T cell-B cell interaction, the rapid spatial
sorting of cell membrane proteins into multiple specific regions
within the intercellular junction is driven from within the T cell
(8). Substitution of the B cell with a synthetic supported lipid
bilayer (SLB) displaying key proteins produces minimal differ-
ences in the antigen specificity or protein spatial organization
(2). The transport mechanism is postulated to be based on actin
espite their dynamic liquid nature, cell membranes exhibit
distinctive spatial organization on length scales ranging
T cell (8–10). Disruption of this actin flow using cytoskeletal
inhibitors blocks transport (8, 10). Associations of cell-surface
proteins with actin are thought to exist via adaptor proteins, such
as talin, which is known to couple lymphocyte function-
associated antigen-1 (LFA-1) to the actin cytoskeleton (11).
Similarly, ezrin has been reported to be an adaptor protein for
the T cell receptor (TCR) (12). Although simply coupling to
actin flow may be sufficient for protein transport, additional
regulation is required to achieve differential sorting of multiple
proteins within the IS.
Differential protein sorting is markedly illustrated by com-
parison of TCR and LFA-1, which become segregated into the
central supramolecular activation cluster (cSMAC) and the
surrounding peripheral supramolecular activation cluster
motion on the cell surface reveals that these 2 proteins move in
the same direction under the influence of centripetal F-actin
flow (9, 10, 13). The mechanism directing them to different
destinations, microns apart, remains mysterious. Before trans-
port, TCR engagement of the peptide-major histocompatibility
complex (pMHC) on the apposed membrane leads to the
formation of TCR clusters that contain approximately 100 TCR
molecules, which are subsequently transported to the cSMAC (9,
14). LFA-1 does not form such large-scale clusters upon binding
its ligand, intercellular adhesion molecule-1 (ICAM-1) (11, 15).
Thus, we speculate that differences in cluster size may contribute
to protein sorting.
In the present study, we manipulate the clustered state of
LFA-1 in primary murine T cells to directly investigate the effect
of protein cluster size on spatial sorting. Cluster size is increased
by crosslinking LFA-1 with a non-blocking bivalent monoclonal
antibody (Bi-X LFA-1). Crosslinking the primary antibody with
a secondary antibody to form a tetravalent crosslinker against
LFA-1 leads to a more clustered state (Tetra-X LFA-1). Thus,
2 degrees of clustering beyond the native state, which may
already be clustered to a small degree, are accessible. Compared
with non-crosslinked LFA-1, Bi-X LFA-1 is transported further
inward toward the inner zone of the pSMAC, immediately
surrounding the cSMAC. Crosslinking with a tetravalent agent
leads to localization of LFA-1 to the cSMAC along with TCR.
Use of bivalent and tetravalent crosslinkers against ICAM-1
induces similar alterations in the spatial position of LFA-
1:ICAM-1 complexes. Thus, increasing the clustering state of
LFA-1, either directly by using anti-LFA-1 antibodies or indi-
rectly by crosslinking its ICAM-1 ligand, biases its distribution
toward the synapse center. Moreover, spatial sorting of the
Author contributions: N.C.H., J.A.N., and J.T.G. designed research; N.C.H. and J.A.N. per-
formed research; N.C.H., J.A.N., and J.T.G. analyzed data; and N.C.H., J.A.N., and J.T.G.
wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
August 4, 2009 ?
vol. 106 ?
no. 31 ?
differentially crosslinked LFA-1 species from each other within
a single synapse is also observed. These results illustrate that
differences in protein cluster size are sufficient to direct discrete
sorting of proteins into different regions of the IS. We discuss
how a frictional mechanism for coupling of proteins to centrip-
synapse, predicts this.
ICAM-1 Distribution Reflects Ligated LFA-1 Pattern on the T Cell. The
formation of LFA-1:ICAM-1 complexes is necessary for the
recruitment of active LFA-1 and ligated ICAM-1 to the pSMAC
(7). During cell migration and upon TCR activation, LFA-1
binds ICAM-1 and induces its accumulation on the apposed
surface in contact with the T cell membrane (11). This interac-
tion depletes the local concentration of unbound ICAM-1
into the contact area by the law of mass action (16). The total
ICAM-1 distribution thus reflects the pattern of active, ligated
LFA-1 on the T cell surface. To confirm that the ICAM-1
pattern provides a real-time means of tracking ligated LFA-1 on
the T cell, we monitor its reversibility of binding to crawling T
cells. We use an ICAM-1-YFP fusion protein linked to nickel-
chelating lipids in the SLB by a stable decahistidine tag (17).
Although the commonly used H155 antibody for LFA-1 also
allows for direct tracking, it indiscriminately binds both active
(ICAM-1 bound) and inactive (non-ICAM-1 bound) forms of
LFA-1 (18). To quantitatively characterize ICAM-1 fluores-
cence as a function of its interactions with LFA-1 on T cells, we
deposit the SLB on substrates pre-patterned with grid arrays of
metal lines (3 ?m ? 3 ?m corrals) that act as barriers to lateral
mobility. Membrane lipids and proteins diffuse normally within
each corral (6, 19). A constant number of freely mobile ICAM-1
(20–22). Restriction of ICAM-1 and pMHC lateral motion
within the SLB by the metal grid lines indirectly disrupts the
transport of bound cognate receptors on the T cell (Fig. 1A).
Receptor-ligand pairs do not directly interact with the metal
Initially, ICAM-1 accumulates beneath the crawling T cell and
the area of ICAM-1 accumulation decreases in size as the cell
migrates away from the corral. Without T cell contact (t ? 110
sec; Fig. 1C), ICAM-1 rapidly returns to a homogeneous distri-
bution by passive diffusion. The total integrated fluorescence,
relative to off-grid ICAM-1, remains constant, confirming that
fluorescence intensity linearly maps ICAM-1 concentration. No
measurable quenching is observed, nor does ICAM-1 desorb
from the membrane over the course of these experiments (see
Fig. S1 for ensemble data from a population of cells). These
results confirm that ICAM-1 spatial distribution quantitatively
reflects the distribution of LFA-1:ICAM-1 complexes.
LFA-1 Is Transported Inward During IS Formation. In the native IS,
LFA-1:ICAM-1 complexes form a ring at the pSMAC and
TCR:pMHC micro-clusters congregate into a central cluster at
the cSMAC (Fig. 2A). Synapse assembly on substrates displaying
diffusion barriers leads to altered protein patterns in the final IS,
spatial mutation reveal aspects of the mechanism that drives
synapse assembly. Here, the inward radial transport of both
TCR:pMHC and LFA-1:ICAM-1 complexes is made apparent
by TCR and ICAM-1 accumulation in areas along the grid lines
that are closest to the synaptic center (Fig. 2B). Similar evidence
with H155 antibody (Fig. S2). These observations confirm that
during synapse formation in primary T cells by processes that
Jurkat cells (10). Disruption of pattern formation by addition of
the actin polymerization blocker latrunculin A confirms that
inward transport is mediated by actin centripetal flow (Fig. S3).
TCR-pMHC Micro-Clusters Exclude and Displace LFA-1:ICAM-1 Com-
plexes. Simultaneous imaging of TCR and ICAM-1 reveals
exclusion of LFA-1:ICAM-1 complexes by TCR micro-clusters.
Diagram of a T cell exposed to a planar bilayer on a patterned substrate.
diffusion barriers. (B) A cell migrates over a supported bilayer containing
ICAM-1 and pMHC corralled by metal lines into 3 ?m squares. As the cell
moves, ICAM-1 distribution within the corrals changes as a result of LFA-1
binding. When the cell is no longer in contact with the grid square, ICAM-1 is
released and diffuses back to homogeneity. (C) Magnified images of the
3-?m ? 3-?m square indicated in B. Total integrated fluorescence within the
corral indicated, normalized to t ? 0, is given beneath each image, showing
that clustering does not significantly affect total ICAM-1 fluorescence.
turbed IS, LFA-1:ICAM-1 complexes form a ring, as visualized with ICAM-YFP,
and TCR labeled with H57 ?TCR-fab(Alexa Fluor 568) forms a central cluster
(observed in 75% of T cells, n ? 81). (B) Upon T cell interaction with an SLB
constrained by metal grid lines, LFA-1:ICAM-1 complexes and TCR accumulate
in areas closest to the synaptic center (observed in 80% of T cells, n ? 41).
Throughout the interface, TCR displaces LFA-1:ICAM-1 complexes to occupy
the most central position available. Images were taken using epifluorescence
LFA-1 is transported inward during IS formation. (A) In an unper-
www.pnas.org?cgi?doi?10.1073?pnas.0902621106Hartman et al.
Upon initial cell contact with the bilayer, ICAM-1 begins to
accumulate and TCR assembles into micro-clusters at the inter-
face (Fig. 3 A–C, column 1; observed in 90% of cells, n ? 44).
After 5 min of cell interaction with the bilayer, 67% of the cells
(n ? 122) contain voids (?300 nm in diameter) within the
ICAM-1 ring that colocalize with the TCR:pMHC micro-
clusters (Fig. 3 A–C, column 2; and magnified images in Fig. 3D).
The ICAM-1 density within these voids is comparable to bulk
density, indicating that unligated ICAM-1 is free to diffuse
throughout the interface while LFA-1:ICAM-1 complexes are
specifically excluded. Nascent TCR:pMHC micro-clusters con-
tinue to colocalize with the voids in the LFA-1:ICAM-1 ring
after 15 min (Fig. 3 A–C, column 3) of cell interaction with the
SLB (observed in 77% of cells, n ? 196).
LFA-1 and TCR do not translocate together, although both
are transported inward in an actin-dependent manner. More-
over, the larger TCR:pMHC micro-clusters efficiently displace
LFA-1:ICAM-1 complexes within the relatively static pSMAC
(Movie S1 and Fig. S4). In addition to this dynamic displace-
ment, we observe a static exclusion and spatial sorting between
these proteins within peripheral regions of the synapse when
substrate barriers block further inward transport (Fig. 3E). Note
how TCR:pMHC micro-clusters out-compete LFA-1:ICAM-1
complexes for the innermost positions within each corralled
zone (observed in 66% of cells, n ? 101). We also show a similar
observation for differentially clustered, ligated LFA-1 (Fig. 5G).
An important corollary of this observation is that segregation
between the pSMAC and cSMAC is not driven exclusively by an
internal difference in cellular structure between these 2 regions.
From these observations, we conclude that an active differential
sorting mechanism capable of distinguishing between TCR:p-
MHC micro-clusters and LFA-1:ICAM-1 complexes exists
throughout the synaptic interface.
Cluster Size Determines LFA-1 Spatial Sorting in the IS. We manip-
ulated the cluster size of LFA-1 to determine its effect on LFA-1
transport and radial distribution. LFA-1 distribution at the
fragments (?LFA-fab), which are monovalent and lack the fc
portion (Fig. S5A) (5, 18). Crosslinking LFA-1 with the H155
bivalent antibody (?LFA-mAb) increases its cluster size (Bi-X
LFA-1). The ?LFA-mAb may be crosslinked itself by a second-
ary antibody (?mAb), which specifically binds to the fcportion.
T cell incubation with this tetravalent crosslinker further in-
creases the degree of LFA-1 clustering (Tetra-X LFA-1).
Cluster size-based protein sorting is observed by simulta-
neously labeling LFA-1 with fabfragments (No-X LFA-1) and
either the bivalent or tetravalent crosslinker. Before interaction
with the SLB, live T cells were concurrently incubated with
fluorescently labeled H155 ?LFA-mAb and ?LFA-fab. No-X
LFA-1 displayed the native broad annular pattern at the
pSMAC. Bi-X LFA-1 also sorted into the pSMAC. However, the
Bi-X LFA-1 clusters were transported further inward, leading to
an enriched ring pattern in the inner zone of the pSMAC (Fig.
micro-clusters are recruited (Fig. 4C). Labeling with the differ-
ent crosslinkers in the absence of LFA-1 pre-clustering did not
result in the observed differences in LFA-1 spatial patterns (Fig.
S6). The differential sorting can be quantified for a population
of cells by generating averaged protein density plots of the
normalized and azimuthally integrated intensities obtained for
the various forms of LFA-1 as a function of the normalized cell
radii (Fig. 4D Inset). These averaged radial profiles (n ? 53 cells
each) of the Bi-X LFA-1 (Fig. 4D) and Tetra-X LFA-1 (Fig. 4E),
compared with the radial profile of No-X LFA-1, reveal the
different characteristic spatial sorting as a function of cluster
size. The mean normalized radii at which peak intensities (i.e.,
clearly different with the more inward positions occupied by the
more highly crosslinked species (P ? 0.001 for both sets, Student
t test). Slight differences between the mean radial positions at
peak intensities for the No-X LFA-1 may occur as a result of
No-X LFA-1 exclusion from the inner pSMAC zone by the Bi-X
The overall synapse morphologies of ICAM-1 (pSMAC) and
TCR (cSMAC) are largely unaffected by the induced clustering
of a subpopulation of LFA-1 in these experiments (Fig. 4F). The
unaltered distribution of ICAM-1 at the pSMAC for all degrees
of crosslinking applied (Fig. S5 B and C) indicates that a
sufficient population of non-crosslinked and unlabeled ligated
LFA-1 is present to preserve the broad pSMAC pattern. Direct
the cSMAC given that ICAM-1 accumulation, which maps only
ligated LFA-1, does not reflect the distributions of crosslinked
LFA-1 at the IS.
Crosslinked ICAM-1 Increases Inward Transport of Active LFA-1. To
selectively study crosslinking effects on LFA-1 ligated to
ICAM-1, we used a bivalent crosslinking antibody against the
YFP domain (?YFP-mAb) of the ICAM-1-YFP fusion protein.
Crosslinked ICAM-1 (Bi-X ICAM-1) is formed by incubation of
the bilayer containing ICAM-YFP and pMHC with ?YFP-mAb
Epifluorescence images of ICAM-YFP (A) and T cells labeled with labeled
?TCR-fab(Alexa Fluor 568) (B) that were fixed 2 min (column 1), 5 min (column
2), and 15 min (column 3) after contact with SLBs containing ICAM-1 and
pMHC. Upon formation, TCR micro-clusters exclude and displace ICAM-1 as
they translocate to the cSMAC. The ICAM-1 density within the excluded areas
is comparable to the bulk density. (C) Composite image of ICAM-1 from A and
TCR from B. (D) Magnified images of the area indicated in C of the T cell fixed
after 5 min. (E) Fluorescent images taken with TIRF illumination of ICAM-YFP
and TCR on live T cells exposed to a bilayer constrained by diffusion barriers
(parallel lines with 2 ?m spacing). Within the pSMAC, TCR:pMHC micro-
clusters occupy the most central positions and exclude LFA-1:ICAM-1
TCR micro-clusters exclude and displace LFA-1:ICAM-1 complexes.
Hartman et al. PNAS ?
August 4, 2009 ?
vol. 106 ?
no. 31 ?
before T cell addition (Fig. 5A). The different spatial distribu-
tions of LFA-1:ICAM-1 complexes, as a function of ICAM-1
crosslinking, are evident in the quantitative fluorescence images
presented in Fig. 5. These images are calibrated to reveal
absolute protein densities using a method based on supported
bilayer standards (23). Imaging the fluorescently labeled ?YFP-
mAb responsible for crosslinking the ICAM-YFP reveals that it
is primarily concentrated in the cSMAC (Fig. 5C Right). Line
scans of protein densities along the dashed lines on Fig. 5 B and
C are also shown in Fig. 5D. An ICAM-1:?YFP-mAb ratio
greater than 2:1 requires that additional ICAM-1 be indirectly
ICAM-YFP:?YFP-mAb ratio observed in the cSMAC (Fig. 5D
Inset) provides evidence for native LFA-1 organization into
preexisting clusters of no more than a few molecules (11, 15).
These observations indicate that LFA-1:ICAM-1 complexes,
which have been crosslinked through ICAM-1, preferentially
was labeled in live cells (B and C) and imaged with TIRF microscopy. (A)
Diagram of the degrees of LFA-1 clustering. (B) T cells were labeled with H155
?LFA-fabfragment (No-X LFA-1) and ?LFA-mAb (Bi-X LFA-1). (C) H155 ?LFA-
mAb was pre-clustered by crosslinking with an antibody (?mAb) and T cells
were labeled as in B to give No-X LFA-1 and Tetra-X LFA-1. Averaged radial
profile plots (n ? 53 cells each) of intensities as a function of the normalized
circle at a given radius and is normalized by the number of pixels in the circle.
representative of 3 independent experiments. (F) For T cells incubated with
the tetravalent crosslinker, the ICAM-1 ring and TCR central cluster is main-
tained. Thus, ICAM-1 and TCR synapse morphology is unaffected by crosslink-
ing a subpopulation of cell-surface LFA-1.
Increased clustering of LFA-1 changes its spatial localization. LFA-1
ICAM-YFP crosslinked with ?YFP-mAb (Bi-X ICAM-1). (B) Quantitative protein
(Alexa Fluor 568) in a synapse with Bi-X ICAM-1. Images were taken by
ICAM-YFP density (left axis) and ?YFP-mAb density (right axis). (Inset) Plot of
?YFP-mAb/ICAM-YFP using line scan values for areas within the cell. (E) Bi-X
ICAM-1 and non-crosslinked ICAM-1 (No-X ICAM-1) are simultaneously pre-
and were obtained as in Fig. 4 (P ? 0.001, Student t test). Data are represen-
tative of 5 independent experiments. (G) Bi-X ICAM-1 and No-X ICAM-1
sorting are also shown for when the bilayer is corralled by 5-?m ? 5-?m grid
squares. As seen with TCR and ICAM-1, Bi-X ICAM-1 occupies the most radially
inward positions along the grid lines within the pSMAC, displacing No-X
Crosslinking ICAM-1 alters synapse morphology. (A) Diagram of
www.pnas.org?cgi?doi?10.1073?pnas.0902621106Hartman et al.
segregate closer to the cSMAC, whereas non-crosslinked LFA-
a tetravalent crosslinker against ICAM-1 (Tetra-X ICAM-1)
results in complete transport of ICAM-1 to the cSMAC (Fig.
S7). Cytosolic calcium levels, used to measure cell signaling, do
not differ between cells interacting with Tetra-X ICAM-1 and
cells forming native patterns (Fig. S8). Addition of latrunculin A
to T cells interacting with these differently crosslinked ICAM-1
species result in synaptic patterns that are not well resolved (Fig.
S3 B and C).
Protein sorting between crosslinked and non-crosslinked
ICAM-1 can be directly observed when both populations simul-
taneously interact with a T cell. Bilayers containing pMHC,
fluorescently labeled ICAM-1 (Alexa Fluor 647) lacking the
YFP moiety (No-X ICAM-1), and ICAM-YFP (Bi-X ICAM-1)
are incubated with ?YFP-mAb before T cell addition. Upon T
cell activation, the Bi-X ICAM-1 can be seen to segregate from
the No-X ICAM-1 (Fig. 5G). The more centrally biased distri-
bution of LFA-1:Bi-X ICAM-1 relative to the LFA-1:No-X
ICAM-1 is also evident in the cell population-averaged radial
profiles (Fig. 5F, n ? 63, mean ? SE). No-X ICAM-1 fluores-
cence intensities in the cSMAC are comparable to bulk intensity
values, indicating free diffusion of unligated ICAM-1 through-
out the synaptic junction. When the same bilayer system is
constrained into patterned grid arrays of diffusion barriers,
LFA-1:Bi-X ICAM-1 complexes clearly occupy the most radially
inward positions along the barriers and exclude LFA-1:No-X
ICAM-1 complexes (Fig. 5H). Crosslinking alters the clustering
state of ligated LFA-1, and the cell transports crosslinked
LFA-1:ICAM-1 complexes further inward than non-crosslinked
Upon T cell activation, LFA-1 and TCR engage their ligands and
become sorted into the pSMAC and cSMAC by radial transport
within the T cell-APC junction. Direct or indirect external
crosslinking of LFA-1 on T cells preceding IS formation alters
its spatial sorting. Specifically, we have observed a graduated
response in which a bivalent crosslinker redirects LFA-
1:ICAM-1 complexes to the innermost radii of the pSMAC
whereas a tetravalent crosslinker causes them to localize in the
cSMAC. These results indicate that the crosslinking state of
LFA-1 can determine its final position within the IS.
We propose a mechanistic model whereby the sorting of
proteins results from differential strengths of coupling to the
moving actin cytoskeletal network. Linking of cell surface
proteins to actin (e.g., via talin, ezrin, or other less direct
methods) may be described as frictional coupling (24, 25). The
result is that a protein, or cluster of proteins, on the cell surface
experiences a driving force in the direction of actin cytoskeletal
flow. If individual bonds form and break rapidly, it is not
necessary that the proteins be driven at the same speed, or even
in the same direction, as the actin flow. Indeed, it has been
previously reported that TCR clusters can be driven at angles to
the preferred flow when they encounter physical barriers (13). If
we speculate that the strength of coupling to the cytoskeleton
exhibits a nonlinear scaling with protein density (e.g., there is a
degree of cooperative binding), the result is a cluster size-
mediated sorting of the type observed in the experiments
Under such a mechanism, all proteins in the membrane will
experience a driving force determined by the composite of their
specific coupling chemistry to the actin network and their local
clustering state. The transport progresses until a quasi-
equilibrium radial distribution is reached in which the inward
driving force of each protein cluster species is balanced by local
competition from surrounding clusters. This is analogous to
sedimentation equilibrium, in which denser species out-compete
less dense species for the down-field positions in a force field
(e.g., centrifugal, gravitational, electric) (26, 27). In the case of
the T cell IS, the driving force is the actin cytoskeletal flow and
the coupling force per unit area experienced by the protein
clusters determines their relative positions. TCR clusters clearly
have a higher coupling force density than native LFA-1:ICAM-1
complexes. This conclusion is partly based on our observations
of the ability of TCR clusters to physically displace LFA-
1:ICAM-1 en route to the cSMAC (Movie S1 and Fig. S4).
Additionally, TCR trapped in the pSMAC region on patterned
substrates displace LFA-1 from the innermost radial positions of
each corral (see Fig. 2B and Fig. 3E). For the same reasons, TCR
clusters, along with other cSMAC localizing proteins, such as
CD28:B7–1 (28) (Fig. S9), could prevent the more weakly
coupled LFA-1 from entering the cSMAC (Fig. 6A). External
crosslinking of LFA-1 into larger clusters increases the coupling
force density, enabling LFA-1 clusters to effectively compete
with TCR for cSMAC territory (Fig. 6 B and C). The mechanism
we suggest here is consistent with all the observations discussed.
However, other mechanisms have also been proposed.
Size exclusion and membrane bending effects have long been
considered as possible contributors to protein sorting within
the IS (29–32). It has been suggested that the larger ectodomains
of LFA-1:ICAM-1 complexes (?42 nm) prevent their colocal-
ization with TCR:pMHC complexes (?15 nm) at the cSMAC.
TCR transport are coupled to actin centripetal flow and the strength of actin
attachment scales with the protein cluster size. (A) Micro-clusters of TCR and
relatively non-clustered LFA-1 molecules are transported inward and TCR
forms a central cluster in the cSMAC surrounded by a ring of LFA-1 in the
pSMAC. (B) Crosslinking of LFA-1 with a bivalent crosslinker increases its
coupling to actin, resulting in LFA-1 accumulation in the inner zone of the
clusters, leading to comparable coupling strengths of LFA-1 and TCR to actin.
Both TCR and LFA-1 occupy the cSMAC. (D) Fluorescence images of the
small-scale segregation between TCR:pMHC (red) and LFA-1:ICAM-1 (green)
darker areas indicate closer membrane interactions. (E) Magnified fluores-
cence and RICM images of the area indicated in D. TCR:pMHC complexes
colocalize with darker regions in the RICM and LFA-1:ICAM-1 complexes
colocalize with the lighter regions.
Hartman et al.PNAS ?
August 4, 2009 ?
vol. 106 ?
no. 31 ?