Clustering of Peptide-Loaded MHC Class I Molecules for
Endoplasmic Reticulum Export Imaged by Fluorescence
Resonance Energy Transfer1
Tsvetelina Pentcheva and Michael Edidin2
Fluorescence resonance energy transfer between cyan fluorescent protein- and yellow fluorescent protein-tagged MHC class I
molecules reports on their spatial organization during assembly and export from the endoplasmic reticulum (ER). A fraction of
MHC class I molecules is clustered in the ER at steady state. Contrary to expectations from biochemical models, this fraction is
not bound to the TAP. Instead, it appears that MHC class I molecules cluster after peptide loading. This clustering points toward
a novel step involved in the selective export of peptide-loaded MHC class I molecules from the ER. Consistent with this model, we
detected clusters of wild-type HLA-A2 molecules and of mutant A2-T134K molecules that cannot bind TAP, but HLA-A2 did not
detectably cluster with A2-T134K at steady state. Lactacystin treatment disrupted the HLA-A2 clusters, but had no effect on the
A2-T134K clusters. However, when cells were fed peptides with high affinity for HLA-A2, mixed clusters containing both HLA-A2
and A2-T134K were detected. The Journal of Immunology, 2001, 166: 6625–6632.
heavy chain, noncovalently associated with soluble ?2-micro-
globulin and a peptide of 8–10 aa, usually generated in the cytosol
by proteasomes. Biochemical studies have established a model for
the assembly of human MHC (HLA) class I molecules in the en-
doplasmic reticulum (ER)3(reviewed in Refs. 1 and 2). The model
describes the associations of nascent class I molecules with ER-
resident chaperones and with TAP, which enable them to acquire
antigenic peptides and enter the secretory pathway. However, it
provides little information about the spatial organization of the
nascent proteins within the ER or about the way in which they exit
the ER. The 4:1 stoichiometry of MHC to TAP1/TAP2 het-
erodimer (3) suggests that class I molecules could be clustered at
the TAP complex. Alternatively, fully folded, peptide-loaded
MHC class I molecules could be clustered after their dissociation
from TAP as part of a mechanism for ER export, analogous to that
for soluble and GPI-anchored proteins (4, 5).
Studying MHC class I spatial organization in the ER requires
imaging on a scale beyond the resolution limit of the light micro-
scope. Recently, we and others (6–10) have developed a quanti-
tative technique, fluorescence resonance energy transfer (FRET)
microscopy, which can detect clusters of proteins carrying appro-
ajor histocompatibility complex class I molecules
present intracellular peptides to CD8?CTLs. Each
molecule comprises a polymorphic transmembrane
priate fluorophores. In FRET, nonradiative energy transfer occurs
between two fluorophores, an energy donor and an energy accep-
tor. Because the efficiency of FRET decays as the sixth power of
the donor-to-acceptor distance, the maximum separation allowing
detectable FRET for typical donor-acceptor pairs is ?100 Å.
Hence, significant energy transfer reports molecular proximity.
Because FRET causes the quenching of donor fluorescence (ener-
gy is transferred to the acceptor instead of being emitted as a pho-
ton), it can be measured by imaging the increase in donor fluores-
cence after acceptor photobleaching (6, 7).
A recent analysis of the theory for FRET on membranes has
pointed the way to distinguishing between FRET due to clustering
of donors and acceptors and FRET due to high concentrations of
donors and acceptors randomly distributed in the membrane (7). If
all donor and acceptor fluorophores are clustered, FRET is inde-
pendent of acceptor concentration, whereas if they are randomly
distributed, FRET increases with increasing acceptor surface den-
sity, and is independent of the ratio of donor-to-acceptor fluoro-
phores. If a labeled population is a mixture of clustered and ran-
domly distributed molecules, FRET increases with increasing
acceptor concentration, but for a given acceptor surface density, it
also depends on the ratio of donor-to-acceptor fluorophores (Fig.
Because TAP was expected to mediate some class I cluster-
ing (3), we measured FRET between HLA-A2 or HLA-A2-
T134K, a mutant that does not associate with TAP (11, 12).
Wild-type and mutant HLA-A2 molecules were tagged with
either cyan fluorescent protein (CFP), or with yellow fluores-
cent protein (YFP), at their C termini (13). As a positive control
for FRET, we tagged HLA-A2 with both CFP and YFP, sepa-
rated by a 25-aa linker (14).
At steady state, we could detect clusters of HLA-A2 molecules
and clusters of A2-T134K molecules. However, there was no ev-
idence for clusters containing both HLA-A2 and A2-T134K. Sur-
prisingly, no HLA-A2 clusters could be detected by FRET after
cells were treated with the proteasome inhibitor lactacystin (15),
indicating that, if multiple molecules are simultaneously bound to
TAP, the distance between them is beyond the FRET limit. In
Department of Biology, Johns Hopkins University, Baltimore, MD 21218
Received for publication November 13, 2000. Accepted for publication March
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
1This work was supported by National Institutes of Health Grant AI14584 (to M.E.).
2Address correspondence and reprint requests to Dr. Michael Edidin, Department of
Biology, Johns Hopkins University, Baltimore, MD 21218. E-mail address:
3Abbreviations used in this paper: ER, endoplasmic reticulum; CFP, cyan fluores-
cent protein; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate;
COPII, coat protein complex; D, diffusion coefficient; Endo H, endoglycosidase H;
FRAP, fluorescence recovery after photobleaching; FRET, fluorescence resonance
energy transfer; GFP, green fluorescent protein; ROI, region of interest; YFP, yellow
Copyright © 2001 by The American Association of Immunologists0022-1767/01/$02.00
contrast, addition of exogenous peptides that reached the ER in-
dependently of TAP (16, 17) resulted in some coclustering of wild-
type and mutant HLA-A2 molecules. These data strongly suggest
that HLA-A2 molecules cluster after peptide loading, perhaps as
part of the process of their export from the ER.
Materials and Methods
Cells, Abs, and reagents
HeLa Tet-On cells (Clontech, Palo Alto, CA) were maintained in DMEM
(Mediatech, Herndon, VA), supplemented with 10% tetracycline-free FBS
(Clontech), 2 mM L-glutamine (Life Technologies, Gaithersburg, MD), 100
?g/ml G418 (Life Technologies), and antibiotic/antimycotic (100 U/ml peni-
cillin G (sodium salt), 100 ?g/ml streptomycin sulfate, and 25 ?g/ml ampho-
using LipofectAMINE (Life Technologies) with the pBI constructs described
below mixed in a 20:1 ratio with vector pTK-Hyg (Clontech). They were
selected with 200 ?g/ml hygromycin B (Roche Molecular Biochemicals, In-
dianapolis, IN), sorted for high expression by flow cytometry, and cloned by
limiting dilution. Positive clones were maintained in HeLa Tet-On medium,
supplemented with 200 ?g/ml hygromycin B. Expression of the transfected
48 h before each experiment. For FRET experiments, 105cells were plated
onto a sterile coverslip in 2 ml of medium. The second construct was trans-
fected transiently 24 h later using FuGENE (Roche Molecular Biochemicals),
and the cells were imaged after 48 h.
The B-lymphoblast cell line LCL-721.45.1 (18, 19) was maintained in
RPMI 1640 medium (Mediatech) containing 15% heat-inactivated FBS
(Intergen, Purchase, NY). Cell line HMy2.C1R expressing HLA-A2-
T134K, a gift from Dr. J. Frelinger (University of North Carolina, Chapel
Hill, NC) (11), was maintained in RPMI 1640, supplemented with 10%
heat-inactivated FBS and 300 ?g/ml G418.
The mAb BB7.2 (ATCC HB-84) (20) was purified from hybridoma
supernatants using GammaBindPlus Sepharose (Pharmacia Biotech, Pis-
cataway, NJ). It recognizes a determinant in the ?2 domain and is specific
for conformed HLA-A2 molecules. Cy3-conjugated F(ab?)2goat anti-
mouse IgG was purchased from Jackson ImmunoResearch Laboratories
(West Grove, PA). Anti-TAP1 antiserum was purchased from StressGen
Biotechnologies (Victoria, BC, Canada).
All oligonucleotides were synthesized and purified by Integrated DNA
Technologies (Coralville, IA).
The Tax peptide (LLFGYPVYV) is derived from the human T cell
lymphotropic virus HTLV-1 (aa 11–19) and binds with high affinity to
HLA-A2 (21, 22). It was synthesized and purified to 97% purity by New
England Peptide (Fitchberg, MA). It was dissolved in 5% DMSO to make
a 2.5 mM stock and used at 250 ?M.
YFP was generated from pEGFP-N3 (Clontech) by site-directed mutagen-
esis, which introduced the following amino acid substitutions: L64F,
T65G, V68L, S72A, and T203Y. CFP was generated from pECFP (Clon-
tech) by site-directed mutagenesis, which introduced the N212K substitu-
tion. The final PCR products were ligated into vector pGEM-T (Promega,
Madison, WI) and sequenced. The EGFP was excised from the pEGFP-N3
with BamHI and BsrG I (New England Biolabs, Beverly, MA) and re-
placed with the PCR-generated products, cut with the same two enzymes,
thus creating pYFP-N3 or pCFP-N3.
The cDNA coding for HLA-A2 and HLA-A2-T134K were obtained by
RT-PCR from LCL-721.45.1 and HMy2.C1R.T134K cells, respectively.
RNA was purified from 107cells using TRIzol (Life Technologies). It was
converted to cDNA with random hexamer primers from the Advantage
RT-PCR kit (Clontech). The two cDNA were amplified with specific prim-
ers and sequenced. They were excised with SalI (New England Biolabs)
and BamHI and cloned into pYFP-N3 and pCFP-N3, digested with the
same two enzymes. The constructs were excised out of the N3 vectors with
SalI and XbaI (New England Biolabs) and cloned into pBI (Clontech), cut
with the same two enzymes.
To generate the untagged HLA-A2 and HLA-A2-T134K constructs
(containing a STOP codon), their cDNA were PCR amplified and se-
quenced. They were excised with SalI and NheI and ligated into pBI, cut
with SalI and XbaI.
To generate the positive control for energy transfer, YFP and CFP were
physically linked to the C terminus of HLA-A2. The linker SSMTG-
GQQMGGDLYDDDDGDPPAGS (based on Ref. 14) was created by PCR.
The YFP STOP codon was deleted by PCR, and the linker was fused at the
YFP C terminus. The product was sequenced, digested with BglII (New
England Biolabs) and BamHI, and introduced into pA2-CFP-N3, cut with
BamHI, and treated with calf intestinal phosphatase (New England Bio-
labs). The entire construct was moved into pBI, as described above for the
Cells were harvested in PBS containing trypsin, chicken serum, collage-
nase, and EDTA (Worthington Biochemical, Lakewood, NJ). They were
washed once with 1% BSA in PBS and either analyzed directly or stained
with 40 ?g/ml BB7.2 mAb for 1 h at 4°C, then washed in PBS, and
incubated with 5 ?g/ml Cy3-conjugated F(ab?)2goat anti-mouse IgG for
30 min at 4°C. After washing, cells were resuspended in PBS-1% BSA and
analyzed on an EPICS 752 flow cytometer (Coulter, Miami, FL).
Pulse chase and immunoprecipitation
Cells were incubated in cysteine- and methionine-free medium for 30 min,
then labeled with 260 ?Ci/ml Tran35S-label (ICN Biochemicals, Costa
Mesa, CA) for 20 min. The cells were washed in PBS and chased in com-
plete medium, supplemented with 2 mM cysteine and 2 mM methionine for
the indicated time intervals. They were washed with cold PBS and lysed in
buffer containing 0.5% Triton X-100. Postnuclear supernatants were pre-
cleared overnight with protein A-Sepharose (Sigma), then incubated with
25 ?g/ml BB7.2 mAb, and the HLA-Ab complexes were recovered with
protein A-Sepharose. They were washed in buffer containing 0.1% Triton
X-100 and eluted. The eluates were digested overnight with endoglycosi-
dase H (Endo H; Roche Molecular Biochemicals) in buffer G5 (New En-
gland Biolabs). The samples were analyzed by 10% SDS-PAGE and
To detect MHC class I interactions with TAP, cells were radiolabeled
and washed as above, then lysed in buffer containing 1% 3-[(3-cholami-
dopropyl)dimethylammonio]-1-propanesulfonate (CHAPS; Sigma). Pre-
cleared lysates were incubated with anti-TAP1 antiserum at 1/100 dilution.
The Ag-Ab complexes were recovered with protein A-Sepharose, washed
for different fluorophore distributions. a, If all donor and acceptor fluoro-
phores are clustered, FRET is independent of acceptor concentration. b, If
all fluorophores are randomly distributed, FRET increases with increasing
acceptor concentration and is independent of the ratio of donor:acceptor. c,
If a fluorophore population is a mixture of clustered and randomly distrib-
uted molecules, FRET increases with increasing acceptor concentration,
but, for a given acceptor surface density, it also increases with higher ratios
of acceptor:donor fluorophores.
Theoretical dependence of FRET on acceptor concentration
6626 CLUSTERING OF PEPTIDE-LOADED MHC CLASS I REVEALED BY FRET
in buffer containing 0.1% CHAPS, and eluted. The eluates were treated
with Endo H and analyzed by 10% SDS-PAGE and autoradiography, as
above. Bands were quantified using Scion Image (Scion, Frederick, MD).
For experiments in the presence of peptides or lactacystin, cells were in-
cubated for 1.5 h in medium supplemented with 250 ?M Tax peptide or
100 ?M lactacystin (Kamiya Biomedical, Seattle, WA) before their lysis.
Fluorescence recovery after photobleaching (FRAP)
Cells were grown on coverslips for 2 days prior to the experiments. Cov-
erslips were washed twice in HBSS (Life Technologies), supplemented
with 1% FBS and 10 mM HEPES (pH 7.3), mounted on slides in the same
solution, and sealed with nail polish. The peptide and lactacystin treatments
were the same as in the TAP coimmunoprecipitation experiments.
Lateral diffusion of A2-YFP or A2-T134K-YFP was measured as
described previously (17). Data were collected with custom software.
From each curve, the percentage of recovery of fluorescence (the mo-
bile fraction) and the half-time for recovery were obtained. The diffu-
sion coefficient (D) was calculated from the half-time value, assuming
Fluorescence microscopy and FRET measurements
Cells on coverslips were washed in PBS and fixed in 4% paraformaldehyde
in PBS for 30 min at room temperature. They were washed five times in
PBS, and then incubated for 10 min in equilibration buffer from the Slow-
Fade Light antifade kit (Molecular Probes, Eugene, OR). They were
mounted on slides in antifade from the same kit and sealed with nail polish.
The peptide and lactacystin treatments were the same as in the TAP co-
immunoprecipitation experiments and the FRAP measurements.
Cells were imaged on a Zeiss Axiovert 135 TV microscope (Zeiss,
Thornwood, NY) using a 1.4 NA ?100 Zeiss Plan-apochromat objective.
Fluorescence was excited with a 75-W arc lamp. CFP and YFP were de-
tected with XF114 (excitation 440DF20, dichroic 455DRLP, emission
480DF30) and XF30 (excitation 510DF23, dichroic 540DRLP, emission
OG550) filter sets, respectively (Omega Optical, Brattleboro, VT). Digital
images were collected with a 12-bit Series 300 cooled CCD (Roper Sci-
entific, Trenton, NJ), operated by the IC300 digital imaging system (Ino-
vision, Raleigh, NC).
FRET was measured as the percentage increase in CFP fluorescence
after the bleaching of YFP (adapted for CFP and YFP from Ref. 7). Four
images were acquired in a FRET experiment: 1) an image of the CFP
fluorescence (in the presence of YFP), 2) an image of the YFP fluores-
cence, 3) an image of the YFP fluorescence after 30-s-long continuous
excitation leading to its destruction, and 4) an image of the CFP fluores-
cence after the YFP bleach. Data were collected from more than five fields
per coverslip, and the results from more than four independent experiments
were pooled together, because they were comparable. To control for CFP
bleaching and noise at low CFP levels, cells expressing only CFP-tagged
molecules were taken through all four steps of a FRET experiment.
The two CFP images were registered using ISee (Inovision) to account
for any x-y drift of the slide during the bleach. Custom software output the
mean values for CFP and YFP fluorescence before and after the bleach,
after a dark current correction, from five 5 ? 5-pixel (340 nm ? 340 nm)
regions of interest (ROI) per cell. The ROI were placed on the nu-
clear envelope. For each ROI, FRET was calculated as (CFPpostbleach?
CFPprebleach)/CFPpostbleach, and the ratio of acceptor to donor was calculated
as YFPprebleach/CFPpostbleach. To convert the fluorescence ratios to approx-
imate protein ratios, cells were transiently transfected with chimeric mol-
ecules in known cDNA ratios. Because the noise of the experiment (? 5%
FRET) increased at very low CFP levels, data with CFP below 150
arbitrary units were excluded from the graphs.
Our data could not be fit by theoretical curves (23, 24). This is because
the dependence of FRET on YFP surface density is not linear, and because
a number of the variables, needed for a fit, are unknown. These variables
include the fraction of clustered molecules (which in turn may depend on
YFP concentration), and the FRET efficiency within a cluster, which may
not be the same for all clusters. Furthermore, in a mixture of clustered and
randomly distributed molecules, FRET may occur both within and between
clusters. No single theoretical model accounts for all these parameters, and
any fit that tries to take them into account has too many free parameters to
yield any useful information. However, the relative extent of clustering can
be evaluated in terms of the dependence of FRET on donor-to-acceptor
ratio for a given surface concentration of acceptor, and can be expressed as
the difference in the mean FRET, over a range of YFP concentration, for
different YFP:CFP ratios. This was done for YFP surface density in the
range 1500–2500 fluorescence units, because in this range all plots of
FRET vs YFP concentration reached a plateau for all YFP:CFP ratios.
Tagging with CFP or YFP did not perturb the folding and surface
expression of the class I molecules. Both the chimeric and the
untagged proteins were detected on the surface of transfected cells
with mAb BB7.2 specific for native HLA-A2. The plasma mem-
brane levels of the mutant A2-T134K were lower than the levels of
the wild-type molecules, as expected (12) (data not shown).
with the intracellular processing or with the interactions of the class I
molecules with the TAP complex. Cells expressing A2-YFP (a), untagged
HLA-A2 (b), A2-T134K-CFP (c), or untagged A2-T134K (d) were meta-
bolically labeled with a mixture of [35S]methionine and cysteine and
chased in nonradioactive medium for the indicated intervals. The cells were
lysed in 0.5% Triton X-100 and immunoprecipitated with mAb BB7.2. The
samples were digested with Endo H. Acquisition of resistance to Endo H
indicates that the sugar moieties of the labeled molecules have been pro-
cessed by enzymes residing in the medial Golgi. R and S refer to the Endo
H-resistant and the Endo H-sensitive forms of the proteins, respectively. e,
Cells expressing A2-YFP or A2-T134K-CFP were metabolically labeled as
before and lysed in 1% CHAPS. They were immunoprecipitated with either
anti-TAP1 antiserum or BB7.2 mAb and treated with Endo H. R and S
refer to the Endo H-resistant and the Endo H-sensitive forms of the pro-
teins, respectively. f, Cells expressing A2-YFP were treated with 100 ?M
lactacystin or 250 ?M Tax peptide for 1.5 h, metabolically labeled, and
lysed in 1% CHAPS. Equal amount of lysates was immunoprecipitated
with anti-TAP1 antiserum or BB7.2 mAb (as a positive control). Untrans-
fected cells were used as a negative control. Mix refers to the immuno-
precipitate from a 1:1 lysate mixture of radiolabeled untreated cell and
nonradioactive cells treated with 250 ?M Tax peptide for 1.5 h as a control
for peptide-induced A2-YFP dissociation from TAP in vitro.
Tagging with YFP or CFP does not significantly interfere
6627The Journal of Immunology
Tagging with CFP or YFP had some effect on the intracellular
processing of the HLA molecules, without affecting their associ-
ations with the TAP complex. In pulse-chase experiments, the in-
tracellular processing rates of both A2-YFP and A2-T134K-CFP
were lower than those of their untagged equivalents (Fig. 2, a–d),
which is consistent with previous data (25). However, the associ-
ations of the tagged molecules with the TAP complex were the
same as those of their untagged counterparts. A2-YFP coimmu-
noprecipitated with TAP, whereas A2-T134K-CFP did not coim-
munoprecipitate with TAP, consistent with results in other cells
(Fig. 2e) (11, 12). To confirm that inhibiting the proteasome results
in a larger number of A2-YFP molecules associating with TAP,
whereas adding exogenous peptides has the opposite effect, we
coimmunoprecipitated A2-YFP and TAP after treatment with 100
?M lactacystin for 1.5 h or 250 ?M Tax peptide for 1.5 h (Fig. 2f).
Densitometric analysis showed that lactacystin increased the num-
ber of A2-YFP molecules coimmunoprecipitating with TAP by a
factor of ?1.2, whereas the addition of Tax peptides decreased that
number by a factor of ?1.2. To ensure that peptide-induced dis-
sociation did not occur in vitro, we compared the number of A2-
YFP molecules coimmunoprecipitating with TAP from lysate of
radiolabeled untreated cells and from the same lysate mixed in a
1:1 ratio with lysate of nonradioactive cells incubated with 250
?M Tax peptide for 1.5 h (Fig. 2f, “mix”). Densitometric analysis
showed that the same number of molecules coimmunoprecipitated
with TAP under both conditions.
The lateral diffusion of green fluorescent protein (GFP)-tagged
MHC class I molecules measured by FRAP reports their associa-
tion with TAP in the ER (17). The data are summarized in Table
I. At steady state, the diffusion coefficient (D) represents the av-
erage of two populations of MHC class I: TAP bound (D ?1 ?
10?9cm2s?1) and TAP free (D ?4 ? 10?9cm2s?1) (17). At steady
state, A2-YFP diffusion was intermediate to that previously mea-
sured for free or TAP-bound molecules, D ?2.3 ? 10?9
cm2s?1? 2 ? 10?10cm2s?1, indicating that some fraction of the
population was associated with TAP. D of A2-T134K-YFP was
higher, D ?3.5 ? 10?9cm2s?1? 3 ? 10?10cm2s?1, consistent
with the biochemical evidence that A2-T134K does not bind TAP
at all (11, 12) (Fig. 2e). Lactacystin treatment lowered the average
A2-YFP diffusion to D ?1.7 ? 10?9cm2s?1? 3 ? 10?10
cm2s?1, indicating an increase in the fraction of A2-YFP bound to
TAP, but it had no effect on the lateral diffusion of A2-T134K-
YFP. Exogenously added peptides did not significantly affect the
diffusion of either A2-YFP or A2-T134K-YFP; however, when
added after lactacystin treatment, they restored A2-YFP diffusion
to its original value, D ?2.5 ? 10?9cm2s?1? 4 ? 10?10
cm2s?1. This suggests that the peptides were bound by empty,
TAP-associated HLA-A2 molecules that were then released from
TAP. The fraction of mobile A2-YFP molecules was slightly, but
significantly, lower than that of A2-T134K-YFP (70 ? 4% vs
80 ? 3%) and was unaffected by the treatments.
FRET was measured in terms of the dequenching of donor
(CFP) fluorescence after acceptor (YFP) bleaching (Fig. 3). Af-
ter acquiring images of the initial donor (CFPpre, Fig. 3a) and
acceptor (YFPpre, Fig. 3b) fluorescence, the acceptor is de-
stroyed by continuous excitation (YFPpost, Fig. 3d). The in-
crease in donor fluorescence (CFPpost, Fig. 3c) after acceptor
bleaching is proportional to the amount of FRET occurring
before the destruction of the acceptor: FRET (%) ? 100 ?
Our positive control for FRET was YFP and CFP connected
with a 25-aa linker and fused to the C terminus of HLA-A2. In
cells expressing this construct, FRET was insensitive to YFP sur-
face density, indicating that all donor and acceptor fluorophores
were clustered. Neither lactacystin nor the addition of Tax peptides
altered this distribution of FRET values (Fig. 4, compare with the
theoretical predictions in Fig. 1a).
FRET between labeled HLA-A2 molecules depended on accep-
tor surface density, but for a given surface density of acceptor,
FRET increased with increasing acceptor:donor ratios, indicating
that a fraction of the molecules was clustered at steady state (Fig.
5a, compare with the theoretical predictions in Fig. 1c). The dif-
ference between the mean FRET over a range of YFP concentra-
tions (1500 and 2500 arbitrary units) for YFP:CFP ratios of 1:1
and 2:1 was statistically significant (Fig. 5d). Because we assumed
that the clusters reflected multiple HLA-A2 molecules bound si-
multaneously to TAP, it was surprising to observe that some A2-
T134K molecules were also clustered. FRET between labeled A2-
T134K molecules depended on acceptor surface density, but for a
given acceptor concentration, FRET increased with increasing ac-
ceptor:donor ratios (Fig. 5, b and d). In contrast, FRET between
A2-T134K-CFP and A2-YFP was insensitive to the acceptor:do-
nor ratio, indicating that these molecules were randomly distrib-
uted relative to one another (Fig. 5, c and d, also compare with the
theoretical predictions in Fig. 1b).
After lactacystin treatment, which increased the fraction of
TAP-bound HLA-A2 molecules (Fig. 2f), FRET among
HLA-A2 molecules no longer depended on the donor-to-accep-
tor ratio (Fig. 6, a and d), whereas FRET among A2-T134K
molecules did (Fig. 6, b and d). Thus, cutting off peptide supply
reduced the fraction of clustered HLA-A2 molecules, without
affecting A2-T134K clustering, confirming that the clusters we
observed at steady state were not mediated by TAP. Lactacystin
had no effect on FRET in the mixture of HLA-A2 and A2-
T134K (Fig. 6, c and d).
When Tax peptides were added to otherwise untreated cells, a
fraction of the molecules in the mixture of HLA-A2 and A2-
T134K was now clustered, because for a given surface concentra-
tion of acceptor, FRET between these molecules depended on the
donor-to-acceptor ratio (Fig. 7, c and d). Peptide addition had no
effect on FRET among HLA-A2 or among A2-T134K molecules
Table I. Diffusion coefficients (D), mobile fraction (R), and 95% confidence limits (C.L.) for HLA-A2 and
D (?10?9cm2s?1) ? 95% C.L.
R (%) ? 95% C.L.
A2-YFP2.3 ? 0.2
1.7 ? 0.3
2.5 ? 0.3
2.5 ? 0.4
3.5 ? 0.3
3.7 ? 0.3
3.5 ? 0.3
3.7 ? 0.3
70 ? 4
73 ? 4
71 ? 3
72 ? 4
80 ? 3
83 ? 3
82 ? 2
81 ? 2
Lactacystin ? Tax peptide
Lactacystin ? Tax peptide
6628CLUSTERING OF PEPTIDE-LOADED MHC CLASS I REVEALED BY FRET
Tagging with CFP or YFP did not significantly interfere with the
folding, the intracellular processing, and the associations of the class
I molecules with TAP (Fig. 2) (25). This allowed us to use FRET
between CFP- and YFP-tagged class I molecules to monitor molec-
ular clustering during their assembly in and export from the ER.
Changes in the fraction of TAP-associated HLA-A2 molecules
were measured biochemically and in terms of changes in lateral
diffusion in the ER membrane using FRAP. D measured for A2-
YFP in HeLa cells was comparable with that reported for A2-GFP
in Mel JuSo cells (26). However, whereas adding peptide in-
creased D of the mouse MHC class I molecule, H2Ld-GFP, in L
cells (17), it had no detectable effect on D of HLA-A2 in HeLa
cells. This probably reflects the relatively low affinity, 6 ? 107
M?1(22, 27), of Tax peptide for HLA-A2 compared with the
30-fold higher affinity of mouse cytomegalovirus peptide amino
acid sequence YPHFMPTNL (17) peptide for H2Ld. Exogenous
peptides reached the ER of HeLa cells, because they decreased the
number of A2-YFP molecules coimmunoprecipitating with TAP
(Fig. 2f), and also reversed the effects of lactacystin on D. How-
ever, the fraction of TAP-bound HLA-A2 displaced by Tax was
much smaller than reported for mouse cytomegalovirus peptide
amino acid sequence YPHFMPTNL (17) peptide and H2Ld(17).
D for A2-T134K was higher than D for wild-type HLA-A2 un-
der all conditions. This is consistent with the inability of A2-
T134K to bind to TAP; however, because each of these constructs
was expressed in a stable, clonal, cell line, it may be that the
maximum value for D represents clonal variation in some aspect of
the ER affecting lateral diffusion in its membrane. Clonal variation
may also account for the slightly higher mobile fraction of A2-
T134K-YFP molecules relative to A2-YFP.
FRET was measured by imaging the increase in CFP fluores-
cence after YFP bleaching (Fig. 3). An advantage of this method
is that the experiments are internally controlled: the parameters
needed to calculate FRET (donor fluorescence in the presence and
the absence of the acceptor) can be obtained from the same cells,
without having to correct for absolute donor and acceptor concen-
trations. Furthermore, in contrast to quantitative measurements of
acceptor-sensitized emission due to FRET, this method does not
require the experimental determination of spectral correlation fac-
intensity, and the color scheme is shown in the bar on the lower right. a, Donor fluorescence before bleaching the acceptor (CFPpre); b, acceptor fluorescence
before photobleaching (YFPpre); c, donor fluorescence after bleaching the acceptor (CFPpost); and d, acceptor fluorescence after the bleach (YFPpost).
FRET was measured by imaging the increase in CFP fluorescence after photobleaching YFP. Images are pseudocolored for fluorescence
cells expressing CFP and YFP physically linked to the cytoplasmic tail of
HLA-A2. Each point represents the calculated FRET and the mean value
of YFP fluorescence, from a 5 ? 5-pixel square placed on the nuclear
membrane of the cells. In untreated cells, FRET was independent of YFP
concentration (?). This distribution did not change upon treatment with
100 ?M lactacystin for 1.5 h ( ) or 250 ?M Tax peptide for 1.5 h (F).
Dependence of FRET on acceptor (YFP) concentration in
6629The Journal of Immunology
As expected, the positive control, in which CFP and YFP were
physically linked to the C terminus of HLA-A2, reported that all
fluorescent molecules were clustered (Fig. 4). This set an upper
limit for our observations on FRET between mixtures of CFP- and
YFP-labeled molecules. At steady state, these mixtures showed
that some HLA-A2 molecules were clustered (Fig. 5, a and d) and
some A2-T134K molecules were clustered (Fig. 5, b and d), but no
clustering could be detected in the FRET profile of a mixture of
HLA-A2 and A2-T134K (Fig. 5, c and d).
Given the 4:1 stoichiometry of MHC class I association with
TAP, we expected that lactacystin treatment would increase the
fraction of clustered HLA-A2 molecules, because it increased the
fraction bound to TAP (Fig. 2f). Surprisingly, FRET did not detect
any clustered HLA-A2 molecules after lactacystin treatment (Fig.
6, a and d), although the A2-T134K clusters were not perturbed by
the proteasome inhibitor (Fig. 6, b and d), consistent with the in-
ability of A2-T134K to bind TAP (Fig. 2e) (11, 12). Because
HLA-A2 can bind peptides derived from proteasome-independent
signal sequences, it is possible that some of the clustered A2-
T134K molecules are still peptide loaded. To be out of FRET
range, the fluorophores of the multiple HLA-A2 molecules bound
to TAP must be separated by ?80 Å, because R0, the distance for
50% FRET between CFP and YFP, is 60 Å. This is consistent with
our earlier calculation that the TAP complex is large, ?600–1000
Å in diameter (17). It is also consistent with our recent observation
that FRET between mouse TAP1d-CFP and H2Ld-YFP was less
than 10% even after lactacystin treatment (data not shown).
FRET among HLA-A2 molecules or among A2-T134K mole-
cules did not change after the addition of Tax peptides (Fig. 7, a,
b, and d). It depended both on acceptor surface density and, for a
given acceptor density, on the donor-to-acceptor ratio, indicating
that each labeled population was a mixture of clustered and ran-
domly distributed molecules. However, after peptide addition,
FRET between A2-YFP and A2-T134K-CFP depended both on
tration for different acceptor:donor ratios: YFP:CFP,
?1:2 (?); YFP:CFP, ?1:1 ( ); YFP:CFP, ?2:1 (F).
a, In cells expressing A2-YFP and A2-CFP, FRET
depended both on YFP concentration and the ratio of
YFP:CFP. b, In cells expressing A2-T134K-YFP and
A2-T134K-CFP, FRET depended both on YFP con-
centration and the ratio of YFP:CFP. c, In cells ex-
pressing A2-YFP and A2-T134K-CFP, FRET de-
pended on YFP concentration, but was independent of
the ratio of YFP:CFP. d, The mean FRET (%) for YFP
concentration in the range 1500–2500 fluorescence
units for YFP:CFP ratios 1:1 and 2:1, as well as the
98% confidence limits (bars), were plotted for the mix-
tures of A2-YFP and A2-CFP (o), A2-T134K-YFP
and A2-T134K-CFP (p), and A2-YFP and A2-T134K-
Dependence of FRET on YFP concen-
tration after treatment with 100 ?M lactacystin for
1.5 h for different acceptor:donor ratios: YFP:CFP,
?1:2 (?); YFP:CFP, ?1:1 ( ); YFP:CFP, ?2:1 (F).
a, In cells expressing A2-YFP and A2-CFP, FRET
depended on YFP concentration but was independent
of the YFP:CFP ratio. b, In cells expressing A2-
T134K-YFP and A2-T134K-CFP, FRET depended
both on YFP concentration and the ratio of YFP:CFP.
c, In cells expressing A2-YFP and A2-T134K-CFP,
FRET depended on YFP concentration, but was in-
dependent of the YFP:CFP ratio. d, The mean FRET
(%) for YFP concentration in the range 1500–2500
fluorescence units for YFP:CFP ratios 1:1 and 2:1, as
well as the 98% confidence limits (bars), were plotted
for the mixtures of A2-YFP and A2-CFP (o), A2-
T134K-YFP and A2-T134K-CFP (p), and A2-YFP
and A2-T134K-CFP (s).
Dependence of FRET on YFP concen-
6630 CLUSTERING OF PEPTIDE-LOADED MHC CLASS I REVEALED BY FRET
acceptor surface density and, for a given acceptor density, on the
donor-to-acceptor ratio, indicating that some of these molecules
were also clustered (Fig. 7, c and d), even though they are loaded
with peptide by different mechanisms and at different sites in the
ER (see below).
Taken together, the data indicate that MHC class I molecules are
clustered after peptide loading, either in proximity to TAP (HLA-
A2) or elsewhere in the ER (A2-T134K). This strongly suggests
that MHC class I molecules are clustered for export out of the ER.
MHC class I exit from the ER may occur by either nonselective
bulk flow (28) or by specific receptor-mediated export (29). The
specific clustering of peptide-loaded MHC that we observed sup-
ports the second ER export mechanism. It is generally believed
that transmembrane proteins are selectively recruited into ER exit
sites by interactions of their cytoplasmic tails with the coat protein
complex (COPII) coat (30, 31). HLA-A2 lacks either of the puta-
tive signals for ER export, the di-phenylalanine (FF), or the di-
acidic (D?E) motifs (32–36). Conceivably, it may contain an-
other, yet unidentified, ER exit signal. However, because COPII
proteins are concentrated in distinct punctate structures, known as
ER exit sites (37, 38), and we randomly sampled uniformly fluo-
rescent regions of the nuclear envelope, if COPII is responsible for
the clusters that we observed, it must form them before the proteins
reach the ER exit sites.
Another possibility is that the clusters are created by class I
interactions with cargo receptors for transmembrane proteins. Both
of the known cargo receptors, ERGIC-53 and p24, are multimeric
transmembrane complexes (for reviews, see Refs. 39 and 40).
ERGIC-53 exists as dimers and hexamers (39), whereas p24 com-
plexes are heterotetramers (41, 42). If the exit of MHC class I
molecules involves interactions with specific cargo receptors, and
if these have the properties of known cargo receptors, then
HLA-A2 molecules should be clustered for export.
The observation of clusters containing only HLA-A2 molecules
or only A2-T134K molecules and the failure to observe steady
state clustering in the mixture of HLA-A2 and A2-T134K proba-
bly reflects differences in their spatial distribution for export, i.e.,
their physical segregation in distinct ER subdomains. It is possible
that HLA-A2 molecules are sequestered immediately after their
TAP dissociation by factors selecting cargo for the vesicles leaving
the ER and transported along the secretory pathway. In contrast,
A2-T134K does not bind to TAP and leaves the ER loaded with
peptides of suboptimal affinity (43). It is likely that the peptide-
loaded A2-T134K molecules are also capable of interacting with
the cargo-selecting factors; however, they will not bind them in
proximity to TAP, but elsewhere in the ER. In this scenario, in the
mixed population, there are separate clusters containing HLA-A2
or A2-T134K molecules, but few, if any, clusters containing both
proteins. The addition of high affinity peptides that reach the ER
independently of TAP may enable both wild-type and mutant mol-
ecules to bind peptide and complete their folding away from TAP.
Thus, they would have an equal chance of binding to the factors
selecting cargo for export.
Clustering of peptide-loaded MHC class I molecules immedi-
ately after their dissociation from TAP, as our data indicate, may
enhance host response against viral infections. It is likely that in
cases in which viruses appropriate the host protein synthesis ma-
chinery for the dedicated production of their own proteins, the
clusters would contain class I molecules loaded almost exclusively
with a few dominant viral peptides. Assuming that the clusters
persist throughout the secretory pathway until their delivery at the
plasma membrane, they might simultaneously engage multiple
TCRs, and as a result, may constitute better targets than single
molecules. The question whether the cargo receptors responsible
for their formation, assuming they exist, would be dedicated to
MHC class I export or would be shared with other transmembrane
proteins, as well as their actual identification, is a matter of future
We thank Dr. J. Frelinger for providing the HMy2.C1R.T134K cells;
T. Wei and A. Nechkin of the Integrated Imaging Center, Department of
Biology, for expert technical support; and Dr. D. Marguet, Dr. A. Ken-
worthy, and E. Spiliotis for technical advice.
1. Pamer, E., and P. Cresswell. 1998. Mechanisms of MHC class I-restricted antigen
processing. Annu. Rev. Immunol. 16:323.
2. Cresswell, P., N. Bangia, T. Dick, and G. Diedrich. 1999. The nature of the MHC
class I peptide loading complex. Immunol. Rev. 172:21.
3. Ortmann, B., J. Copeman, P. J. Lehner, B. Sadasivan, J. A. Herberg,
A. G. Grandea, S. R. Riddell, R. Tampe, T. Spies, J. Trowsdale, and P. Cresswell.
1997. A critical role for tapasin in the assembly and function of multimeric MHC
class I-TAP complexes. Science 277:1306.
tion after addition of 250 ?M Tax peptide for 1.5 h for
different acceptor:donor ratios: YFP:CFP, ?1:2 (?); YFP:
CFP, ?1:1 ( ); YFP:CFP, ?2:1 (F). In cells expressing
A2-YFP and A2-CFP (a), A2-T134K-YFP and A2-
T134K-CFP (b), or A2-YFP and A2-T134K-CFP (c),
FRET depended both on YFP concentration and the ratio
of YFP:CFP. d, The mean FRET (%) for YFP concentra-
tion in the range 1500–2500 fluorescence units for YFP:
CFP ratios 1:1 and 2:1, as well as the 98% confidence
limits (bars), were plotted for the mixtures of A2-YFP and
A2-CFP (o), A2-T134K-YFP and A2-T134K-CFP (p),
and A2-YFP and A2-T134K-CFP (s).
Dependence of FRET on YFP concentra-
6631 The Journal of Immunology
4. Appenzeller, C., H. Andersson, F. Kappeler, and H. P. Hauri. 1999. The lectin Download full-text
ERGIC-53 is a cargo transport receptor for glycoproteins. Nat. Cell Biol. 1:330.
5. Muniz, M., C. Nuoffer, H. P. Hauri, and H. Riezman. 2000. The Emp24 complex
recruits a specific cargo molecule into endoplasmic reticulum-derived vesicles.
J. Cell Biol. 148:925.
6. Bastiaens, P. I., I. V. Majoul, P. J. Verveer, H. D. Soling, and T. M. Jovin. 1996.
Imaging the intracellular trafficking and state of the AB5 quaternary structure of
cholera toxin. EMBO J. 15:4246.
7. Kenworthy, A. K., and M. Edidin. 1998. Distribution of a glycosylphosphatidy-
linositol-anchored protein at the apical surface of MDCK cells examined at a
resolution of ?100 Å using imaging fluorescence resonance energy transfer.
[Published erratum appears in 1998 J. Cell Biol. 142:881.] J. Cell Biol. 142:69.
8. Nagy, P., G. Vamosi, A. Bodnar, S. J. Lockett, and J. Szollosi. 1998. Intensity-
based energy transfer measurements in digital imaging microscopy. Eur. Biophys.
9. Ng, T., A. Squire, G. Hansra, F. Bornancin, C. Prevostel, A. Hanby, W. Harris,
D. Barnes, S. Schmidt, H. Mellor, et al. 1999. Imaging protein kinase C? acti-
vation in cells. Science 283:2085.
10. Pollok, B. A., and R. Heim. 1999. Using GFP in FRET-based applications.
Trends Cell Biol. 9:57.
11. Peace-Brewer, A. L., L. G. Tussey, M. Matsui, G. Li, D. G. Quinn, and
J. A. Frelinger. 1996. A point mutation in HLA-A*0201 results in failure to bind
the TAP complex and to present virus-derived peptides to CTL. Immunity 4:505.
12. Lewis, J. W., A. Neisig, J. Neefjes, and T. Elliott. 1996. Point mutations in the
?2 domain of HLA-A2.1 define a functionally relevant interaction with TAP.
Curr. Biol. 6:873.
13. Miyawaki, A., J. Llopis, R. Heim, J. M. McCaffery, J. A. Adams, M. Ikura, and
R. Y. Tsien. 1997. Fluorescent indicators for Ca2?based on green fluorescent
proteins and calmodulin. Nature 388:882.
14. Heim, R., and R. Y. Tsien. 1996. Engineering green fluorescent protein for im-
proved brightness, longer wavelengths and fluorescence resonance energy trans-
fer. Curr. Biol. 6:178.
15. Fenteany, G., R. F. Standaert, W. S. Lane, S. Choi, E. J. Corey, and
S. L. Schreiber. 1995. Inhibition of proteasome activities and subunit-specific
amino-terminal threonine modification by lactacystin. Science 268:726.
16. Day, P. M., J. W. Yewdell, A. Porgador, R. N. Germain, and J. R. Bennink. 1997.
Direct delivery of exogenous MHC class I molecule-binding oligopeptides to the
endoplasmic reticulum of viable cells. Proc. Natl. Acad. Sci. USA 94:8064.
17. Marguet, D., E. T. Spiliotis, T. Pentcheva, M. Lebowitz, J. Schneck, and
M. Edidin. 1999. Lateral diffusion of GFP-tagged H2Ld molecules and of GFP-
TAP1 reports on the assembly and retention of these molecules in the endoplas-
mic reticulum. Immunity 11:231.
18. Kavathas, P., F. H. Bach, and R. DeMars. 1980. ? Ray-induced loss of expression
of HLA and glyoxalase I alleles in lymphoblastoid cells. Proc. Natl. Acad. Sci.
19. Kittur, D., Y. Shimizu, R. DeMars, and M. Edidin. 1987. Insulin binding to
human B lymphoblasts is a function of HLA haplotype. Proc. Natl. Acad. Sci.
20. Parham, P., and F. M. Brodsky. 1981. Partial purification and some properties of
BB7.2: a cytotoxic monoclonal antibody with specificity for HLA-A2 and a vari-
ant of HLA-A28. Hum. Immunol. 3:277.
21. Kannagi, M., H. Shida, H. Igarashi, K. Kuruma, H. Murai, Y. Aono,
I. Maruyama, M. Osame, T. Hattori, H. Inoko, et al. 1992. Target epitope in the
Tax protein of human T-cell leukemia virus type I recognized by class I major
histocompatibility complex-restricted cytotoxic T cells. J. Virol. 66:2928.
22. Lauvau, G., K. Kakimi, G. Niedermann, M. Ostankovitch, P. Yotnda, H. Firat,
F. V. Chisari, and P. M. van Endert. 1999. Human transporters associated with
antigen processing (TAPs) select epitope precursor peptides for processing in the
endoplasmic reticulum and presentation to T cells. J. Exp. Med. 190:1227.
23. Dewey, T. G., and G. G. Hammes. 1980. Calculation on fluorescence resonance
energy transfer on surfaces. Biophys. J. 32:1023.
24. Wolber, P. K., and B. S. Hudson. 1979. An analytic solution to the Forster energy
transfer problem in two dimensions. Biophys. J. 28:197.
25. Gromme, M., F. G. Uytdehaag, H. Janssen, J. Calafat, R. S. van Binnendijk,
M. J. Kenter, A. Tulp, D. Verwoerd, and J. Neefjes. 1999. Recycling MHC class
I molecules and endosomal peptide loading. Proc. Natl. Acad. Sci. USA 96:
26. Reits, E. A., J. C. Vos, M. Gromme, and J. Neefjes. 2000. The major substrates
for TAP in vivo are derived from newly synthesized proteins. Nature 404:774.
27. Sette, A., J. Sidney, M. F. del Guercio, S. Southwood, J. Ruppert, C. Dahlberg,
H. M. Grey, and R. T. Kubo. 1994. Peptide binding to the most frequent HLA-A
class I alleles measured by quantitative molecular binding assays. Mol. Immunol.
28. Wieland, F. T., M. L. Gleason, T. A. Serafini, and J. E. Rothman. 1987. The rate
of bulk flow from the endoplasmic reticulum to the cell surface. Cell 50:289.
29. Kuehn, M. J., and R. Schekman. 1997. COPII and secretory cargo capture into
transport vesicles. Curr. Opin. Cell Biol. 9:477.
30. Springer, S., and R. Schekman. 1998. Nucleation of COPII vesicular coat com-
plex by endoplasmic reticulum to Golgi vesicle SNAREs. Science 281:698.
31. Kuehn, M. J., J. M. Herrmann, and R. Schekman. 1998. COPII-cargo interactions
direct protein sorting into ER-derived transport vesicles. Nature 391:187.
32. Kappeler, F., D. R. Klopfenstein, M. Foguet, J. P. Paccaud, and H. P. Hauri.
1997. The recycling of ERGIC-53 in the early secretory pathway: ERGIC-53
carries a cytosolic endoplasmic reticulum-exit determinant interacting with
COPII. J. Biol. Chem. 272:31801.
33. Dominguez, M., K. Dejgaard, J. Fullekrug, S. Dahan, A. Fazel, J. P. Paccaud,
D. Y. Thomas, J. J. Bergeron, and T. Nilsson. 1998. gp25L/emp24/p24 protein
family members of the cis-Golgi network bind both COP I and II coatomer.
J. Cell Biol. 140:751.
34. Nishimura, N., and W. E. Balch. 1997. A di-acidic signal required for selective
export from the endoplasmic reticulum. Science 277:556.
35. Nishimura, N., S. Bannykh, S. Slabough, J. Matteson, Y. Altschuler, K. Hahn,
and W. E. Balch. 1999. A di-acidic (DXE) code directs concentration of cargo
during export from the endoplasmic reticulum. J. Biol. Chem. 274:15937.
36. Sevier, C. S., O. A. Weisz, M. Davis, and C. E. Machamer. 2000. Efficient export
of the vesicular stomatitis virus G protein from the endoplasmic reticulum re-
quires a signal in the cytoplasmic tail that includes both tyrosine-based and di-
acidic motifs. Mol. Biol. Cell 11:13.
37. Scales, S. J., R. Pepperkok, and T. E. Kreis. 1997. Visualization of ER-to-Golgi
transport in living cells reveals a sequential mode of action for COPII and COPI.
38. Presley, J. F., N. B. Cole, T. A. Schroer, K. Hirschberg, K. J. Zaal, and
J. Lippincott-Schwartz. 1997. ER-to-Golgi transport visualized in living cells.
39. Hauri, H. P., F. Kappeler, H. Andersson, and C. Appenzeller. 2000. ERGIC-53
and traffic in the secretory pathway. J. Cell Sci. 113:587.
40. Kaiser, C. 2000. Thinking about p24 proteins and how transport vesicles select
their cargo. Proc. Natl. Acad. Sci. USA 97:3783.
41. Marzioch, M., D. C. Henthorn, J. M. Herrmann, R. Wilson, D. Y. Thomas,
J. J. Bergeron, R. C. Solari, and A. Rowley. 1999. Erp1p and Erp2p, partners for
Emp24p and Erv25p in a yeast p24 complex. Mol. Biol. Cell 10:1923.
42. Fullekrug, J., T. Suganuma, B. L. Tang, W. Hong, B. Storrie, and T. Nilsson.
1999. Localization and recycling of gp27 (hp24?3): complex formation with
other p24 family members. Mol. Biol. Cell 10:1939.
43. Lewis, J. W., and T. Elliott. 1998. Evidence for successive peptide binding and
quality control stages during MHC class I assembly. Curr. Biol. 8:717.
6632 CLUSTERING OF PEPTIDE-LOADED MHC CLASS I REVEALED BY FRET