| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,
,
*
Department of Molecular and Cellular Biology and Howard Hughes Medical Institute, and
Department of Chemistry and Biological Chemistry, Harvard University, Cambridge, MA 02138;
Laboratoire de Chimie Biophysique, Institut le Bel, Université Louis Pasteur, Strasbourg, France; and
Ludwig Institute for Cancer Research, Lausanne Branch, University of Lausanne, Epalinges, Switzerland
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Melan-A/melanoma Ag recognized by T cells-1 (Melan-A/MART-1)4 is one of the most frequently recognized tumor Ags in melanoma patients expressing HLA-A2 *0201 (HLA-A2) (5, 6). Initially, two parental peptides were identified that mimic the protein Ag, the nonapeptide Melan-A/MART-127–35 AAGIGILTV and the decapeptide Melan-A/MART-126–35 EAAGIGILTV, which contains an additional residue at the amino terminus (6). Discovery of those peptides led to the synthesis of a variety of modified Melan-A/MART-1 peptides that are now used in phase I vaccination trials in patients with high-risk stage III-IV malignant melanoma (7).
All CTL clones obtained from patients against the Melan-A/MART-1 gene product recognize both wild-type nonamer and decamer peptides bound to HLA-A2. Indeed, target cells sensitized with saturating concentrations of either peptide are lysed with equal efficiency by all CTL clones. However, peptide titrations reveal that the wild-type decamer is generally recognized more efficiently than the natural nonamer peptide (12 of 18 CTL analyzed) (8). Although the natural decamer binds to HLA-A2 three to five times more efficiently than the natural nonamer, this difference does not fully account for the higher efficiency of decamer Ag recognition. A smaller, second category of CTL clones recognizes the two natural peptides with similar efficiency (4 of 18 CTL analyzed). A rare, third category of CTL recognizes the nonamer more efficiently than the decamer peptide (2 of 18 CTL clones analyzed) (6, 8). The responses of these different categories of CTL suggest that the HLA-A2-bound natural nonamer and decamer exhibit some surfaces that are similar and some that are different to CTL.
Both Melan-A/MART-1 parental peptides display an intermediate binding
affinity for the HLA-A2 molecule, because they have Ala rather than a
large anchor residue at peptide position two (P2) (Table I
). Replacement of P2 by residues such as
Leu or Thr leads to detectable increases in HLA-A2 binding, as measured
in functional assays (9). Although the complexes with
either parental peptide have very short
t1/2 of <1 h, the introduction of the
better P2 anchor residues prolongs the stability of both complexes to
t1/2 of >6 h. Curiously, however,
while the modified decamer led to an improvement in the efficiency of
Ag recognition by most CTL clones tested, the modified nonamer led to a
strong reduction in Ag recognition, in the same population of CTL
clones, even at very high peptide concentrations. This loss of CTL
reactivity suggests that the modified nonamer peptide binds to HLA-A2
differently than the natural nonamer and differently than the natural
and modified decamer (10, 11).
|
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Both peptides, ELA and ALG, were synthesized and purified using HPLC, as described previously (9).
Folding and purification of HLA-A2 complexes
Soluble HLA-A2 was refolded in vitro in the presence of peptide
after expression in bacteria, as described previously
(12). In brief, the HLA-A2 heavy chain (truncated at
Glu275) and
2-microglobulin (2m),
isolated from Escherichia coli inclusion bodies, were
refolded by dilution to final concentrations of 1 and 2 µM,
respectively, in a solution containing 10 µg/ml of the desired
peptide. The solution was stirred at 4°C for 3–4 days.
HLA-A2/peptide complexes were purified by gel filtration chromatography
(HiPrep 26/10; Pharmacia Biotech, Uppsala, Sweden), followed by ion
exchange chromatography (Resource Q; Pharmacia Biotech) and then a
second gel filtration step (Superdex 75; Pharmacia Biotech). The final
buffer solution was 25 mM MES, 0.1% NaN3, pH
6.5. Satisfactory yields (i.e., 4 mg of purified complex/L refolding
buffer) were obtained with both peptides.
Crystallization
Crystals of both HLA-A2/peptide complexes were prepared by
microseeding under identical conditions, as previously described for
crystals of the HLA-A2/HepB complex (13). Briefly, the
seed crystals were prepared by the hanging-drop method at 20°C, by
suspending 1 µl of 10 mg ml-1 protein mixed
with 1 µl in 25 mM MES buffer, pH 6.5, over a reservoir of 16%
polyethylene glycol 6000 in 25 mM MES and 0.10%
NaN3 at 6.5 pH. Microcrystals were prepared by
crushing these initial crystals with a glass rod and serially diluting
them in fresh well solution buffer. Large crystals with platelike
morphology were then grown in hanging drops composed of protein and
1/1000 dilution of microcrystals and would reach the maximum size
(HLA-A2/ELA: 150 µm x 150 µm x 
40 µm, and
HLA-A2/ALG: 150 µm x 150 µm x 80 µm) within
2–3 days.
Data collection and processing
Crystals of both complexes were transferred into cryoprotectant
buffer containing 80% mother liquor and 20% glycerol, and cooled
directly in liquid nitrogen. Diffraction data for HLA-A2/ALG
were collected on beamline 14-BM-C at Advanced Photon Source (Argone,
IL) using a Quantum 4 charge-coupled device (Area Detector Systems
Corporation, Poway, CA). Data were processed with DENZO and SCALEPACK
(14). The HLA-A2/ELA data set was collected on
the beamline A1 in Cornell High Energy Synchrotron Source (Ithaca, NY)
using a Quantum 4 charge-coupled device detector. Data were processed
using MOSFILM (15). Data statistics are shown in Table II.
|
The structure of HLA-A2/ALG was determined by molecular replacement in MOLREP (16) using free HLA-A2/Tax as a search model (17). Cross-rotation functions identified a single complex in the asymmetric unit. The structure of HLA-A2/ELA was determined by rigid body refinement in CNS (18), with the refined structure of HLA-A2/ALG as a starting model.
Both complexes were refined using CNS version 1.0. Model building was
performed with O (19) using 2Fo-Fc and simulated annealing
composite omit-maps. The final refinement statistics for both
structures are presented in Table III.
|
Metal ion identification
Strong 13.5
electron density was observed in the composite
omit-map of the HLA-A2/ELA complex and was assigned to a
metal ion. The distances between the metal ion and the coordinating
atoms are 2.01, 2.07, 2.06, and 2.06 Å. Based on these distances and
the preferences of metals for the tetrahedral coordination
(21), it is likely that the metal binding site is occupied
by either Mg2+, Zn2+, or
Cu2+ (minimum metal-ligand oxygen distances 2.07,
1.91, and 1.82 Å, respectively; Ref. 21). B factor
refinement with all three metals eliminated Mg2+
as a candidate. The B factor for Mg2+ refined to
below 1 Å2, whereas the B factors for
Zn2+ and Cu2+ both refined
to 22.4 Å2, in agreement with the B factors of
the atoms coordinating the metal, which are in the 19–25
Å2 range.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, A
and B). The structures were refined to a reasonable
agreement between observed and calculated structure factors, as well as
good stereochemistry and geometry (Table III
). The 1.8 Å resolution of
HLA-A2/ELA complex allowed incorporation of additional water
molecules (Table III), but the high resolution and quality of both
structures permitted a detailed interpretation of the peptide
binding.
|
The HLA-A2 structures in the HLA-A2/ELA and
HLA-A2/ALG complexes are very similar to the highest
resolution, 1.8 Å MHC class I (MHCI)/peptide structure in the Protein
Data Bank, HLA-A2/Tax peptide (1DUZ; Ref. 17), with
root-mean-square difference (rmsd) of 0.79 Å2
and 0.86 Å2 for the heavy chain and 0.55
Å2 and 0.63 Å2 for
2m, respectively. The few differences include
disorder of Gln226 and
Asp227 in the 
3 domains of the
melanoma-peptide complexes. In addition, the HLA-A2/ELA
complex contains a crystallographic metal binding site (Fig. 1C) that has not been observed in any other MHCI/peptide
crystal structure. The tetrahedral metal is coordinated by
His192 from the heavy chain of HLA-A2 and
Asp98 from the 2m of the
same molecule, as well as by His151 and
Glu154 from the heavy chain of a symmetry-related
molecule. Although no metal ions were added to the solutions during the
purification or crystallization procedures, it is possible that even
trace amounts of contaminants would be enough to become incorporated.
Metal binding sites can stabilize the oligomerization of proteins in
solution, and therefore it would be expected that the presence of metal
might have stabilized the crystal packing and increased the diffraction
limit of the HLA-A2/ELA crystals (full data set is 1.8 Å,
but some weak reflections were observed at 1.5 Å). Although the
crystals of the two complexes are isomorphous, the metal is not present
in the HLA-A2/ALG complex, and one of the metal ligands,
Glu154, is in a different conformation, rotated
away from the position of the metal in the
other crystal (see discussion of
Gln155 in Results).
Modified melanoma-peptide structures
The Melan-A/MART-1 nonamer and decamer peptides bind to HLA-A2 in
modes similar to that of five viral peptides, including nonamers and a
decamer previously observed bound to HLA-A2 (22). To
facilitate comparison, we divide the peptides into three regions (Fig. 2A): an N-terminal region,
which includes three N-terminal residues P1-P3; a C-terminal region,
which includes four C-terminal residues (P6-P9); and the central region
consisting of a three-residue bulge in the decamer peptide (P4, P*, P5)
and a two-residue link in a nonamer peptide (P4-P5). Arrows in Fig. 2A show the directions of the peptide side chains (up, out
of the groove; down, into the groove; right, toward the 2 helix; or
left, toward the 1 helix).
|
, B and C). All the atoms that are
common to both peptides (all atoms of the ALG peptide form a
subset of those in the ELA peptide) are in the same position
in both structures. The C atoms can be superimposed with rmsd of
0.03 Å2, and common atoms (excluding the
carbonyl group of P3) can be superimposed with the rmsd of 0.07
Å2. The N-terminal motif contains an anchor
residue at position P2 (Leu in both peptides), which occupies the B
pocket of HLA-A2. This confirms that the increased binding of
Ala
Leu-substituted peptides can be attributed to the
introduction of the P2 anchor. The "additional" atoms of the
ELA peptide, which are not present in the ALG
peptide, include the carboxylate group of P1 Glu, which forms a
secondary anchor for the peptide and points toward the surface of the
complex, and the C carbon of Ala P3, which points down into the
groove. In the observed conformation, the side chain of
Glu26/P1 could participate in the interaction
with TCR and hence affect the immunogenicity of the decapeptide.
The central region (GIG(ELA) vs
IG(ALG)) is drastically different in
the two complexes, which might explain the differences in CTL
recognition of the peptides. When compared with the two-residue link in
the nonamer, the first and the last residues of the decamer bulge are
shifted in opposite directions, creating an opening for the extra
residue, P*, and a zigzag when viewed from above the binding site (at
P* in Fig. 2C). The first residue of the bulge,
Gly29(ELA)/P4, is shifted
toward the N terminus and up and to the left toward the 1
helix, by 2 Å, while the last residue of the bulge,
Gly31(ELA)/P5, is shifted
toward the C terminus and down and to the right toward the 2
helix, by 2.5 Å. In the side view (Fig. 2B), the central
residue, Ile30(ELA)/P*, is
shifted up with the C atom position 6.75 Å up from the peptide
axis (line connecting C atoms of the first and last residue of the
peptide). The P4 and P5 residues are elevated by 5.7 Å, while the
rest of the peptide is displaced from the axis by between 0 and 2.7 Å.
In the ALG peptide link, the Ile30/P4
is displaced upward by 4.8 Å and Gly31/P5 by 6.2
Å. The large nonpolar side chain of the Ile30 of
the decamer peptide not only protrudes more extensively from the
molecular surface, but its conformation is also very different from
that of the nonamer (dotted red lines in Fig. 2, B and
C). This centrally located difference could account for the
differences in the T cell recognition of these peptides.
The C-terminal regions of both peptides have identical sequences (ILTV)
and similar conformations. Because this is the only region of the
peptide that is invariant, its structure might be required for
recognition of HLA-A2/melanoma complexes by all three categories of the
CTL clones. There are only two minor differences between the two
complexes that we can observe. When the structures of the peptides are
superimposed, the C atoms of Ile32/P6 anchor
are positioned 0.8 Å apart and their side chains are rotated in the
opposite direction (Fig. 3B). In addition, the
surface-exposed, hydrophobic side chain of
Leu33/P7 has two alternate conformations in the
decamer, but only one conformation in the nonamer. Nevertheless, the
general direction of Leu33/P7 side chain in both
peptides is identical, and in fact the orientation of that side chain
in the nonamer corresponds to the average of the two conformations
observed in the decamer (see Leu33 in Figs. 1, A and B and 2, B and
C).
|
In all MHCI/peptide structures, solvent molecules appear to
stabilize the interaction of the melanoma peptides in five locations
(corresponding water molecules in HLA-A2/ELA and
HLA-A2/ALG complexes are color coded similarly in Fig. 3, A
and B): at the peptide N terminus and C terminus (yellow),
below the peptide near P7 (green), between the peptide and the 1
helix (red) and between peptide and the 2 helix (light blue). The
hydrogen-bonding networks involving water molecules at both the N- and
C-terminal regions of the peptides are very similar in all MHCI/peptide
complexes (23, 24). The water molecules underneath the
peptide are likely to be similar in the subset of HLA-A2/peptide
complexes, where residues Arg97 and
Tyr116 of the HLA-A2 are oriented toward the
peptide C terminus (see discussion in Ref. 24). The
interactions with bound water molecules on the sides of the peptides
(Fig. 3, C, D, and E) appear to be
distinctive.
Between the peptide and the 2 helix, the hydrogen-bonding pattern
differs between the two complexes investigated in this study. The
decapeptide carbonyl group from Ile30/P* and the
side chain of Ala28/P3 are positioned too close
to the 2 helix to allow water entry into this volume. In the
nonamer, there is an opening between the peptide chain and 2 helix,
which contains three water molecules (Fig. 3E). The water
molecules interact with the peptide main chain and with two residues
from the adjacent -helix, His114 and
Gln155 (blue spheres in Fig. 3E). To
accommodate this interaction, Gln155 is rotated
in the nonapeptide complex toward the water cluster (toward the
N-terminal region of ALG peptide, blue in Fig. 3E). Gln155 in the
HLA-A2/ELA complex and other MHCI/peptide complexes
(22) in absence of the TCR is exposed on the surface and
points in the direction of the central region of the peptide (yellow in
Fig. 3E). The conformation of Gln155
is altered in the complexes of HLA/Tax with the human A6 and B7 TCRs,
in which Gln155 points upward toward the TCR
(1AO7, Ref. 25 ; 1BD2, Ref. 26).
Although there are five water molecules involved in a
hydrogen-bonding network on the 1 domain side of both peptides, the
hydrogen-bonding patterns are different in the two complexes (Fig. 3, C and D). In the nonapeptide complex, the waters
fill the gap between the peptide and the 1 helix and create a
tightly packed network that connects the peptide backbone to
Thr73 and Arg65. Three
water molecules (34, 51, and 95) interact directly with the peptide
backbone, water 189 connects waters, and water 206 bridges water 34 to
Arg65 (Fig. 3C). In the decapeptide
complex, water molecules also fill the gap between the peptide and 1
domain helix, but they are packed more loosely. Two water molecules
interact directly with peptide backbone, water 225 interacts with
Thr73, and water 279 with
Arg65 (Fig. 3D). However, there is no
water molecule that bridges them together and, in general, the bonds
between water molecules and peptide atoms are longer than in the
nonapetide complex. This results in higher B factors for all these
water molecules. This difference in water structure might influence TCR
binding in this region because the positions of the water molecules,
the adjacent protein atoms, or both may be more easily altered upon TCR
binding in the decamer peptide complex.
Potential peptide flexibility
The effect of deformability on TCR binding has been considered previously in a poorly binding HER-2/neu epitope GP2 bound to HLA-A2. Based on the x-ray structure (the rigidity of the peptide in the binding groove can be judged by the quality of the corresponding electron density map), the weak interaction of the GP2 peptide with HLA-A2 has been attributed to increased flexibility in the center of the peptide (3).
Electron density maps for Melan-A/MART-1-modified peptides,
calculated without the peptide atoms or any atoms within 5 Å of the
peptide to avoid model bias, indicate that the central residues of the
nonapeptide are less well defined than those of the decapeptide (Fig. 1). In the nonapeptide, the electron density for two residues from the
central region, Ile30/P4 and
Gly31/P5 (Fig. 1B), is weak (contoured
at 1.4, although present at 0.7.). In the decapeptide, by
contrast, only the distal side chain atom C
of
Ile30/P* and the side chain atoms of
Leu33/P7 are missing at the same contour levels.
When the electron density map is contoured at 0.7, electron density
is observed for the C atom of Ile30/P*, and
two alternate conformations of Leu33 are visible
(Fig. 1A). The observation that the central region of
decamer peptide may be somewhat more ordered than the nonamer is
consistent with the peptide-to-MHC contacts observed in the structures.
In the decamer peptide, in addition to the P6 anchor, the central
region of the peptide is stabilized by two bound water molecules (W1
and W2 in Fig. 2B, orange in Fig. 3A) bridging
the Glu26/P1 side chain and the
Gly29/P3 and additional constraints created by
the zigzagging of the peptide. Sparse water packing on the 1 domain
side of the decamer peptide, by reducing constraints, might permit the
peptide to fold into a more energetically stable conformation.
Relative to the major conformation differences between the modified Melan-A/MART-1 nonamer and decamer described in previous sections, the minor differences in peptide flexibility may be less likely to play a prominent role in T cell recognition.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Improving ligands for generating tumor-reactive CTL
Modifying the natural decamer sequence of the Melan-A/MART-126–35 peptide by changing the P2 anchor from Ala to Leu stabilized the peptide/HLA-A2 complex, increased the efficiency of T cell recognition by a number of clones, and improved the in vitro generation of tumor-reactive CTLs by stimulation of PBMC from HLA-A2 melanoma patients (9, 27). Substitution of P1 Glu to Ala increases HLA affinity, reactivity by tumor-infiltrated lymph node 6- to 60-fold, and reactivity by Melan-A-specific CTL clones by 4- to 1000-fold (10). These improved properties suggest that modified peptides might be better candidates for peptide-based vaccine trials and/or for improving the generation of tumor-reactive CTL ex vivo for adoptive transfer therapy (9, 27). Improvements in peptide affinity and HLA-A2 complex stability of the gp100 154–162 epitope have been reported with a substitution at a nonanchor peptide position, P8 (28).
In some cases, even very subtle peptide modifications can have profound
effects on T cell recognition and signaling, switching from agonists to
antagonists (reviewed in Ref. 29 ; for structural examples,
see Ref. 30), indicating that modifying peptides can have
unpredictable consequences that need to be tested experimentally.
Nevertheless, the array of water molecules around both the modified
peptides investigated in this study (Fig. 3) suggests atomic positions
in which other atoms might be added to the peptide or MHC molecule to
replace waters and provide novel ligands of potential therapeutic
value.
T cell reactivity data and the observed binding modes of the modified peptides suggest how the wild-type peptides bind
T cell reactivity data suggest that the binding mode in the HLA-A2
complex of the wild-type decamer is very likely to be the same as the
zigzag binding mode observed in the modified decamer, but the wild-type
nonamer is more likely to bind with a zigzag like the modified decamer
than stretched out like the modified nonamer observed in this study.
The observation that decamer peptides modified with P2 substitutions,
which are more suitable anchors for HLA-A2 than the wild-type P2 Ala,
are universally recognized by the same T cells as the wild-type
peptide, but at lower peptide concentrations (9),
indicates that the wild-type and modified decamers share the same
zigzag binding mode. Increasing the peptide affinity for HLA-A2 by
substituting a more suitable P2 anchor apparently increases the number
of HLA-A2 molecules loaded with the decamer, which then present
essentially the same antigenic surface to T cells specific for the
Melan-A/MART-1 protein. The observation that P1 substitutions that
replace the P1 Glu with Ala increase HLA affinity, reactivity by
tumor-infiltrated lymph node 6- to 60-fold, and reactivity by
Melan-A-specific CTL clones by 4- to 1000-fold (10)
suggests that the same antigenic surface is being presented by
wild-type and these P1-substituted peptide/HLA-A2 complexes, and that
the P1 glutamic acid side chain is either not optimal for HLA-A2
binding or constrains the peptide conformation (see Fig. 2B)
away from a conformation that TCR prefer.
Decamers with the large aromatic residues, Tyr or Phe, substituted at P1 are more T cell reactive than wild-type peptide (8), possibly indicating better interaction with HLA-A2 than P1 Glu, but they are selectively recognized by only some CTL clones, suggesting that these large P1 side chains may interact with the TCR as well.
CTL reactivity data indicate that the substitution of Leu, a good P2 anchor residue, for P2 Ala in the nonapeptide significantly increased its binding to HLA-A2, as expected, but strongly reduced its reactivity with Melan-A-specific T cells (9, 11). This loss of reactivity suggests that the wild-type nonamer binds differently than the P2 Leu-substituted nonamer, observed in this study (10, 11). A possibility consistent with the cross-reactivity of all the Melan-A-specific T cells with the wild-type nonamer and decamer peptides is that the wild-type nonamer binds predominately in the zigzag mode like the modified decamer observed in this study, but with the first pocket, P1, empty, rather than like the modified nonamer that binds in the stretched-out mode typical on nonamers with strong P2 anchor residues. This possibility has been suggested before, based on the cell reactivity (9, 31). The observation that nonamers with P1 substitutions to Leu or Met, which are good anchors for the P2 pocket, have increased CTL reactivity (9, 31) is consistent with this possibility, as those modifications are expected to cause P1 to bind in the P2 pocket, making those modified nonamers bind in the zigzag decamer mode. Although there are no examples of crystal structures, as yet, of nonamer peptides that bind with the P1 pocket empty and in a zigzag mode characteristic of decamers, an octamer of the human T cell leukemia virus-1 Tax peptide has been observed bound to HLA-A2, leaving the P1 pocket empty (17). In that Tax-8/HLA-A2 complex, a number of water molecules are observed in place of a peptide residue, linking the peptide to HLA-A2 and forming hydrogen bonds with the HLA-A2 residues that normally form hydrogen bonds with the main chain polar atoms of the first peptide residue (17).
Because it seems likely that both wild-type nonamer and decamer Melan-A/Mart-1 peptides bind predominately in a similar conformation, like that of the modified decamer observed in this study, the structure of the modified decamer bound to HLA-A2 can be used as a model for further studies involving mutational analysis of this peptide family.
The fine specificity of T cell responses to Melan-A/MART-1 peptides
All T cells tested recognize both the wild-type nonamer and
decamer Melan-A/MART-1 peptides, as indicated by the lysis of target
cells sensitized with high concentrations of either peptide. But fine
specificities are revealed in peptide titration studies: most clones
are more reactive with the decamer than the nonamer, although a few
clones react to both equally well, and a rare clone favors the nonamer
(6, 8, 10). These T cell reactivity data suggest that the
antigenic surface presented by the wild-type nonamer and decamer
peptides and HLA-A2 has some common features contacted by the clones
that recognize both peptides equally well and at least one structural
difference contacted by clones that discriminate in favor of either the
decamer or nonamer. If, as suggested by the data discussed above, the
wild-type nonamer binds in the zigzag mode like the decamer, most of
the center and C-terminal sections of the peptide/HLA-A2 complex should
have the same structure in the nonamer and decamer complexes as that
observed in this study in the modified decamer complex (Figs. 2 and 3A) providing the basis for cross-reactivity. The x-ray
crystal structure of a TCR/MHC-peptide complex of an alloreaction has
been described in which the TCR makes no contacts to the N-terminal
half of a bound peptide, indicating that such focused interactions can
exist and can initiate strong T cell signals (32). Another
example of CTL cross-reactivity between different lengths of peptides
from one antigenic epitope has been documented in the case of
H2-Kd-restricted CTL directed against the
Plasmodium berghei peptide nonamer 252–260, in
which most clones also recognized the octamer 253–260 efficiently
(33).
The structures of the wild-type nonamer and decamer Melan-A/MART-1
peptide/HLA-A2 complexes would be expected to diverge toward the N
terminus of the peptides. Clones that distinguish the nonamer
from the decamer may make direct contacts near the P1 pocket, which
would be filled with a peptide amino acid in the decamer complexes, but
empty or more likely water filled in the wild-type nonamer complex,
offering a structural basis for distinguishing the complexes. Because
the two waters, W1 and W2 (red, Fig. 2B), that form a
hydrogen-bonding network connecting P1 Glu26 to
P4 Gly29 in the modified decamer complex would be
absent in a wild-type nonamer complex missing P1 Glu, the peptide
conformations in the P4-P5 may differ in structure, as the nonamer
complex would lack the stabilization of this hydrogen-bonding network.
Structural differences resulting from the absence of the P1 Glu residue
and its stabilizing interactions via waters to P4 could also affect
residues on the surface of HLA-A2, for example at
Gln155 and Glu154 (Fig. 3E), creating a different surface for TCR recognition. The
absence of a good P2 anchor residue in the wild-type nonamer might also
have structural consequences in peptide and MHC residues in its
vicinity that could be distinguished by T cells, although such
structural differences are difficult to predict.
It is possible that some wild-type nonamers bind in the stretched-out mode observed in this study for the modified nonamer, but if that were the case, some T cell clones might have been expected to recognize the modified nonamer even better as a result of its higher affinity to HLA-A2 than the wild-type nonamer, but no clones of that type have been observed.
Alanine-scanning mutagenesis of the decamer peptide has established that both the P7 Ile32 and P10 Val35 anchors, critical for stabilizing the C-terminal region of the bound peptides, are crucial for recognition by all CTL clones, regardless of their fine specificity (10). Neither of the two glycines in either peptide have the unusual dihedral angles permitted for glycines only, yet alanine substitutions at these glycine positions decrease peptide affinity to HLA-A2, as deduced from competition experiments (10). Because both residues have been previously shown to be crucial for peptide binding, it is possible that the presence of Gly at positions P4 and P6 is required for the formation of the zigzag binding mode of these peptides, as previously suggested (10).
A complete structural description of the fine specificity of T cell reactivity for the immunodominant nonamer and decamer peptides from the melanoma tumor-associated protein Melan-A/MART-1 would require crystallization of TCR complexes with these peptide/HLA-A2 complexes. The instability of both wild-type peptide complexes with HLA-A2 has to date precluded their purification. We have determined high resolution crystal structures of the complexes of HLA-A2 molecules with two modified Melan-A/MART-1 peptides, with increased HLA-A2 affinity. Our conclusions can be summarized as follows: 1) Differences in the structures of the modified peptides, including in clusters of bound water molecules, readily explain the differences in their recognition by T cells: the modified nonamer binds stretched out, the modified decamer in a zigzag mode. 2) T cell reactivity and structural data suggest that the wild-type nonamer binds predominately like the modified decamer, and that the fine specificity of T cell reactivity of the wild-type peptides results from the similarities between the nonamer and decamer binding in the middle and C terminus of the peptide, while the differences in reactivity are caused by differences in the peptide and MHC surfaces near the N-terminal end of the peptides. 3) Modifications in the decamer peptide to increase its affinity for HLA-A2 and possibly to form better Ags for therapy might include chemical modifications to place atoms in the location of many bound waters observed clustered in five locations around the peptide. 4) A metal, probably zinc, found at a crystal lattice contact in the decamer complex, may have stabilized that crystal, resulting in observable x-ray diffraction to a Bragg spacing beyond 1.8 Å.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 P.S. and O.M. contributed equally to this work.
3 Address correspondence and reprint requests to Dr. Don C. Wiley, Department of Molecular and Cellular Biology and Howard Hughes Medical Institute, Harvard University, Cambridge, MA 02138. E-mail address: dcwadmin{at}crystal.harvard.edu
4 Abbreviations used in this paper: Melan-A/MART-1, Melan-A/Melanoma Ag recognized by T cells-1; ALG, modified nonapeptide ALGIGILTV; 2m, 2-microglobulin; ELA, modified decapeptide ELAGIGILTV; MHCI, MHC class I; rmsd, root-mean-square difference.
Received for publication May 7, 2001. Accepted for publication July 9, 2001.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Derre, M. Ferber, C. Touvrey, E. Devevre, V. Zoete, A. Leimgruber, P. Romero, O. Michielin, F. Levy, and D. E. Speiser A Novel Population of Human Melanoma-Specific CD8 T Cells Recognizes Melan-AMART-1 Immunodominant Nonapeptide but Not the Corresponding Decapeptide J. Immunol., December 1, 2007; 179(11): 7635 - 7645. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Yagi, H. Hashizume, T. Horibe, Y. Yoshinari, M. Hata, A. Ohshima, T. Ito, M. Takigawa, A. Shibaki, H. Shimizu, et al. Induction of therapeutically relevant cytotoxic T lymphocytes in humans by percutaneous Peptide immunization. Cancer Res., October 15, 2006; 66(20): 10136 - 10144. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Martinez-Hackert, N. Anikeeva, S. A. Kalams, B. D. Walker, W. A. Hendrickson, and Y. Sykulev Structural Basis for Degenerate Recognition of Natural HIV Peptide Variants by Cytotoxic Lymphocytes J. Biol. Chem., July 21, 2006; 281(29): 20205 - 20212. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Colombetti, T. Fagerberg, P. Baumgartner, L. Chapatte, D. E. Speiser, N. Rufer, O. Michielin, and F. Levy Impact of Orthologous Melan-A Peptide Immunizations on the Anti-Self Melan-A/HLA-A2 T Cell Cross-Reactivity. J. Immunol., June 1, 2006; 176(11): 6560 - 6567. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Thilo, P. Berglund, S. E. Applequist, J. W. Yewdell, H.-G. Ljunggren, and A. Achour Dissection of the Interaction of the Human Cytomegalovirus-derived US2 Protein with Major Histocompatibility Complex Class I Molecules: PROMINENT ROLE OF A SINGLE ARGININE RESIDUE IN HUMAN LEUKOCYTE ANTIGEN-A2 J. Biol. Chem., March 31, 2006; 281(13): 8950 - 8957. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. B. E. Stewart-Jones, G. Gillespie, I. M. Overton, R. Kaul, P. Roche, A. J. McMichael, S. Rowland-Jones, and E. Y. Jones Structures of Three HIV-1 HLA-B*5703-Peptide Complexes and Identification of Related HLAs Potentially Associated with Long-Term Nonprogression J. Immunol., August 15, 2005; 175(4): 2459 - 2468. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Madurga, I. Belda, X. Llora, and E. Giralt Design of enhanced agonists through the use of a new virtual screening method: Application to peptides that bind class I major histocompatibility complex (MHC) molecules Protein Sci., August 1, 2005; 14(8): 2069 - 2079. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Y. Borbulevych, T. K. Baxter, Z. Yu, N. P. Restifo, and B. M. Baker Increased Immunogenicity of an Anchor-Modified Tumor-Associated Antigen Is Due to the Enhanced Stability of the Peptide/MHC Complex: Implications for Vaccine Design J. Immunol., April 15, 2005; 174(8): 4812 - 4820. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. K. Lee, G. Stewart-Jones, T. Dong, K. Harlos, K. Di Gleria, L. Dorrell, D. C. Douek, P. A. van der Merwe, E. Y. Jones, and A. J. McMichael T Cell Cross-Reactivity and Conformational Changes during TCR Engagement J. Exp. Med., December 6, 2004; 200(11): 1455 - 1466. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. K. Baxter, S. J. Gagnon, R. L. Davis-Harrison, J. C. Beck, A.-K. Binz, R. V. Turner, W. E. Biddison, and B. M. Baker Strategic Mutations in the Class I Major Histocompatibility Complex HLA-A2 Independently Affect Both Peptide Binding and T Cell Receptor Recognition J. Biol. Chem., July 9, 2004; 279(28): 29175 - 29184. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. I. Webb, M. A. Dunstone, W. Chen, M.-I. Aguilar, Q. Chen, H. Jackson, L. Chang, L. Kjer-Nielsen, T. Beddoe, J. McCluskey, et al. Functional and Structural Characteristics of NY-ESO-1-related HLA A2-restricted Epitopes and the Design of a Novel Immunogenic Analogue J. Biol. Chem., May 28, 2004; 279(22): 23438 - 23446. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. G. Rudolph, L. Q. Shen, S. A. Lamontagne, J. G. Luz, J. R. Delaney, Q. Ge, B. K. Cho, D. Palliser, C. A. McKinley, J. Chen, et al. A Peptide That Antagonizes TCR-Mediated Reactions with Both Syngeneic and Allogeneic Agonists: Functional and Structural Aspects J. Immunol., March 1, 2004; 172(5): 2994 - 3002. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Guillaume, D. F. Legler, N. Boucheron, M.-A. Doucey, J.-C. Cerottini, and I. F. Luescher Soluble Major Histocompatibility Complex-Peptide Octamers with Impaired CD8 Binding Selectively Induce Fas-dependent Apoptosis J. Biol. Chem., February 7, 2003; 278(7): 4500 - 4509. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 |
