| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Departments of
*
Immunology and
Molecular Biology, and
The Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, CA 92037
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-chain with I-Ad but has a peculiar ß-chain. As with
most ß-chain alleles linked to diabetes susceptibility,
I-Ag7 contains a nonaspartic residue at position ß57. We
have produced large amounts of empty I-Ag7 molecules using
a fly expression system to characterize its biochemical properties and
peptide binding by phage-displayed peptide libraries. The
identification of a specific binding peptide derived from glutamic acid
decarboxylase (GAD65) has allowed us to crystallize and obtain the
three-dimensional structure of I-Ag7. Structural
information was critical in evaluating the binding studies.
I-Ag7, like I-Ad, appears to be very
promiscuous in terms of peptide binding. Their binding motifs are
degenerate and contain small and/or small hydrophobic residues at P4
and P6 of the peptide, a motif frequently found in most globular
proteins. The degree of promiscuity is increased for I-Ag7
over I-Ad as a consequence of a larger P9 pocket that can
specifically accommodate negatively charged residues, as well as
possibly residues with bulky side chains. So, although I-Ad
and I-Ag7 are structurally closely related, stable
molecules and good peptide binders, they differ functionally in their
ability to bind significantly different peptide repertoires that are
heavily influenced by the presence or the absence of a negatively
charged residue at position 57 of the ß-chain. These characteristics
link I-Ag7 with autoimmune diseases, such as
insulin-dependent diabetes mellitus. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A striking feature of class II molecules associated with IDDM is the
presence of a serine, alanine, or valine residue at position 57 of the
ß-chain instead of a conserved aspartic acid in IDDM-resistant
haplotypes (3). In the case of
I-Ag7, the class II heterodimer is made of an
I-Ad -chain paired with the unique
I-Ag7 ß-chain that differs from
I-Ad ß at 17 positions, all clustered in the
first variable domain. Only two of these differences are unique across
all I-A haplotypes at position ß56 and ß57. At ß57, a serine
residue replaces an aspartic acid, and, at ß56, a histidine replaces
a proline (6). Introduction of various alleles, such as
Ak/Akß or
Adß alone into the NOD background protects
these animals from or reduces diabetes (7, 8).
Furthermore, replacement of the normal I-Ag7
serine ß57 by an aspartic acid reduces diabetes in transgenic NOD
mice (9), and replacement of histidine ß56 by a proline
has an even deeper impact by preventing both diabetes and insulitis
(10). Thus, positions ß56 and ß57 play a critical role
in the context of I-Ag7 and are most likely
associated with the altered binding and selectivity of peptides
compared with other MHC class II molecules.
In terms of peptide binding to I-Ag7 molecules,
the number of studies so far is somewhat limited. Elution of peptides
from B cell-purified I-Ag7 molecules has not
uncovered any striking features compared with other class II molecules
(11). However, a few peptides had a negatively charged
residue in their C terminus (11) but the series of
sequenced peptides was too limited to draw general conclusions. Peptide
binding to I-Ag7 has also been studied by
alanine-scanning mutagenesis of known binding peptides, as well as by
binding of substituted polyalanine peptides (12, 13).
These studies did not come to a consensus regarding a requirement for a
negatively charged residue in the C terminus. From a structural point
of view, the advantage of such a residue can be easily explained. The
exchange of aspartate ß57 by a serine removes the salt bridge with
arginine 76, as seen, for example, in I-Ad
(14), and exposes a positive charge in the P9 pocket that
would favor the pairing with a negative charge provided by the peptide
(15).
To date, only a few biochemical studies on I-Ag7 have been conducted. In one study, the half-life of I-Ag7 molecules at 37°C was shorter than that of I-Ad and I-Ab molecules (16). This characteristic was associated with the SDS instability of the molecule (12, 16, 17, 18), and was interpreted as being a sign of poor binding capability, because SDS stability has been associated with both peptide binding (19, 20, 21) and MHC class II half-life (22). However, SDS stability is not a reliable marker of peptide occupancy. Indeed, some MHC class II molecules, such as I-Ad, are very unstable with or without peptide, whereas others, such as HLA-DR1 or I-Ab, are very stable, even when occupied with low affinity peptides. On the contrary, a normal peptide binding capacity and a normal cell surface half-life of I-Ag7 molecules have also been reported (12, 13), prompting us to re-evaluate these issues but with a different approach. The final and most important reason to reassess the function of I-Ag7 molecules is the existence of the Biozzi AB/H mouse, which expresses only I-Ag7 as its MHC class II (23) but does not develop IDDM or any spontaneous autoimmune features, yet mounts above-normal immune responses against most Ags. This feature constitutes the strongest in vivo argument to rule out any major misfolding and malfunction of the I-Ag7 molecule itself in terms of peptide binding.
To carry out this study, recombinant empty I-Ag7 molecules were produced in Drosophila melanogaster cells. Purified empty molecules were used to evaluate peptide binding to a large panel of peptides, to screen random cDNA fragment phage display libraries for the identification of I-Ag7 epitopes within OVA and glutamic acid decarboxylase 65 (GAD65) proteins, and to screen a random 12-mer phage-displayed peptide library in the search for an I-Ag7 binding motif. The results were used to construct and express tethered peptide I-Ag7 molecules for crystallographic studies.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Phage vector pJCM13-88 containing a second synthetic copy of gene VIII has been described elsewhere (24). The random 12-mer peptide M13 phage display library JCM13-88-X12 was constructed in our laboratory in collaboration with Dr. C. Glass (R.W. Johnson Pharmaceutical Research Institute, La Jolla, CA). Plasmid pBSK-GAD65 containing the murine GAD65 coding sequence previously cloned from a C57BL/6 brain cDNA library was provided by Dr. J. Fehling (Basel Institute, Basel, Switzerland). The OVA coding sequence (25) subcloned into pCMU4 was obtained from Dr. K. Frueh (R.W. Johnson Pharmaceutical Research Institute). Bacterial strains Escherichia coli XL1-Blue and E. coli XLOLR were obtained from Stratagene (La Jolla, CA).
Construction, expression, and purification of I-Ag7 molecules
The wild-type cDNA for the I-Ag7 ß-chain
was modified to generate two constructs to express empty and
peptide-filled molecules, as previously reported for
I-Ad (26, 27). For the expression of
empty molecules, the cDNA was interrupted after the codon for the last
tryptophan before the transmembrane segment and extended by a linker
(SSADL), a thrombin site (LVPRGS), a basic zipper, and a hexa-histidine
tail. To tether peptides to the ß-chain, a hybrid cDNA was
constructed by substituting the natural signal sequence of
I-Ag7 by the HLA-A2 signal sequence followed by
the first four residues of the 1 domain of HLA-A2, a multiple
cloning site, a glycine-serine linker, and the ß1 domain of
I-Ag7. Oligonucleotides corresponding to the
peptide sequence were cloned between the SacII and
BglII sites of the polylinker. The coding sequence of the
extracellular domain of the human invariant chain (Ii) was cloned in
frame with the HLA-A2 signal sequence followed by six histidine codons
(26). The I-Ad -chain was also
modified by the addition of a BirA biotinylation sequence following the
acidic zipper and proceeding the histidine tag (28). All
constructs were subcloned into the metallothionein promoter-based fly
expression vector pRMHa3 (29). Modified ß-chains were
transfected along with the soluble I-Ad acid
zipper -chain or soluble I-Ad acid zipper
biotin -chain (26, 27) to produce empty and
peptide-filled I-Ag7 molecules.
I-Ag7/Ii complexes were expressed by transfecting
cDNAs for the - and ß-chain without histidine coding sequences and
the human Ii construct. The various cDNA combinations were
cotransfected into D. melanogaster SC2 cells by
calcium phosphate precipitation. A neomycin-resistance gene was added
to the transfection at a 1:200 ratio to allow the selection of stable
cell lines in the presence of 0.5 mg/ml of G418. Expression was tested
at 4 wk after a 3-day induction with 0.7 mM CuSO4
and immunoprecipitation with nickel-agarose beads (Qiagen, Chatsworth,
CA). Large-scale preparations were made using serum-free media
(Insect-Xpress; BioWhittaker, Walkersville, MD) in roller bottles.
Supernatants were concentrated and buffer exchanged against PBS by
tangential flow. Recombinant proteins were purified at 4°C by batch
binding to nickel-agarose beads and elution with a step gradient of
imidazole. Empty molecules were further purified by gel filtration over
a Superdex 200 column. Peptide-filled molecules were purified to
homogeneity by ion exchange chromatography over a MonoQ10/10 column at
pH 8.4 (Pharmacia, Piscataway, NJ). I-Ag7 Ii
complexes were produced in Schneider’s media, 10% FCS, in
triple-layered flasks. Metal induction was conducted in the presence of
5 mM pepstatin A and 5 mM leupeptin. Following the initial
nickel-agarose step, purification was conducted by a succession of
anion exchange (MonoQ10/10), hydrophobic interaction chromatography
(Phenylsuperose 10), and gel filtration (Superdex 200). Purified
proteins were kept in PBS, pH 7.4, containing 0.02%
NaN3 at 4°C.
Biochemical analysis, SDS-PAGE, and isoelectric focusing
Purifications were monitored by SDS-PAGE followed by Coomassie Brilliant Blue R 250 staining. Two-dimensional gel electrophoresis was performed as described (26) using 3.5–10.0 nonequilibrium pH gradient electrophoresis (NEPHGE) in the first dimension and 12% SDS-PAGE in the second dimension followed by Coomassie Brilliant Blue staining.
Vector constructs and random cDNA fragment libraries
Control phage constructs pJCM13-88-OVA and pJCM13-88-ROI coding for the OVA peptide 323-339 (ISQAVHAAHAEINEAGR) and the synthetic peptide ROI (VHAAHAVHAAHAVHA), respectively, were cloned as fusions to the N terminus of gene VIII by a GGGGS linker sequence. Vector pORFES IV was digested with NheI/HindIII and the stuffer replaced by the following annealed oligonucleotides: 5'-CTAGCGTAGCTCAGGCCGGTGGCGGTATATCTCAAGCTGTCCATGCAGCACATGCAGAAATCAATGAAGCAGGCAGAGGTGGCGGTGGCTCCCA-3', 5'-AGCTTGGGAGCCACCGCCACCTCTGCCTGCTTCATTGATTTCTGCAT GTGCTGCATGGACAGCTTGAGATATACCGCCACCGGCCTGAGCTAC G-3' (pORFESIVOVA); 5'-CTAGCGTAGCTCAGGCCGGTGGCGGTGTTC ACGCTGCTCACGCAGTTCACGCTGCACACGCTGTTCACGCAGGTGG CGGTGGCTCCCA-3' and 5'-AGCTTGGGAGCCACCGCCACCTGCGTGA ACAGCGTGTGCAGCGTGAACTGCGTGAGCAGCGTGAACACCGCCAC CGGCCTGAGCTACG-3' (pORFESIV-ROI). The ligated vectors were digested with XbaI/HindIII, and the inserts were directionally cloned into JCM13-88.
The construction of cDNA fragment phage display libraries will be
described in details elsewhere (T. Stratmann, L. Teyton, and A. S.
Kang, manuscript in preparation). Briefly, plasmids pCMU4 and
pBSk-GAD65 were treated with BamHI and
SacI/HindIII to release the GAD65 and OVA coding
sequence, respectively. Random DNA fragmentation was established by
digesting the inserts with DNase I and, subsequently, with mung bean
nuclease to obtain blunt ends. The DNA fragments were ligated into
pORFESIV that had been cleaved with NaeI and transformed
into E. coli XLOLR cells, resulting in two primary libraries
of 6 x 105 (OVA) and 2.6 x
106 (GAD65) independent clones. The fragments
restoring the reading frame of the selection marker ß-lactamase were
rescued by treatment of the recovered plasmids with
XbaI/HindIII and separated by agarose gel
electrophoresis. Inserts of 
160–210 bp of length (corresponding to
40–90 bp of random DNA fragment length) were isolated and cloned into
pJCM13-88 to express peptide fragments at the N terminus of the
synthetic copies of gene VIII of bacteriophage M13 upon
transformation into E. coli XL1-Blue cells. Bacteriophages
were propagated overnight at 30°C and prepared by standard methods.
The sizes of the final libraries were 1.7 x
106 independent clones for the OVA library,
JCM13-88-rOVA, and 7 x 106 clones for the
GAD65 library, JCM13-88-rGAD65.
Phage inhibition assay and selection of M13 random 12-mer peptide and cDNA fragment display libraries
To assess specific binding of phages JCM13-88-OVA and JCM13-88-ROI to empty I-Ag7molecules, phages were incubated with 20 ng of recombinant I-Ag7 for 2 h at 37°C in 100 µl of 50 mM NaAc, pH 5.0, and 0.05% Tween 20 containing 0.02% NaN3 in a micro tube in duplicates in the presence or absence of the ROI peptide at various concentrations. After neutralization with 1 M Tris, samples were transferred to Reacti-Bind metal chelate microtiter plates (100 µl per well; Pierce, Rockford, IL) that had been blocked with 0.5% BSA in PBS containing 0.02% NaN3, and incubated at room temperature for 2 h under gentle agitation. Each well was washed 10 times with 300 µl of PBS containing 0.05% Tween 20 (PBST), and binding phages were eluted with 100 µl of 100 mM glycine, pH 2.2, for 10 min at room temperature. After neutralization with 1 M Tris, the phage titer was determined. Selection of the libraries JCM13-88-X12, JCM13-88-rOVA, and JCM13-88-rGAD65 was as described above, with the difference that 200 ng instead of 20 ng of I-Ag7 was used and that no free peptide was added. Eluted phages were propagated in E. coli XL1-Blue cells, and the selection was repeated three or four times. After selection, DNA from randomly chosen individual clones was prepared and analyzed by sequencing.
Peptide competition assay to MHC class II molecules
A modified competition assay (ELISA) was used to determine
binding of peptides to I-Ag7 and
I-Ad molecules using biotinylated
OVA323–339, or in some cases, biotinylated ROI
peptide and unlabeled peptides (13, 30). Microtiter plates
(96-well, Maxisorb; Nunc, Naperville, IL) were coated with anti-I-A
ß-chain-specific mAb M5.114 (100 µl/well, 5 µg/ml in 100 mM
NaCO3, pH 9.6; American Type Culture Collection,
Manassas, VA) overnight at 4°C. Plates were washed five times with
PBST, blocked overnight with 200 µl of PBS containing 5% BSA at
37°C, and washed. In a second set of plates, 200 ng of empty
I-Ag7 or I-Ad was incubated
in 100 µl binding buffer (6.7 mM citric phosphate, pH 10.4, 0.15 M
NaCl, 2% Nonidet P-40, 2 mM EDTA) containing 300 nM of biotinylated
OVA323–339 peptide with varying concentrations
of test peptide at 37°C overnight. The samples were transferred to
the coated plates and incubated for 4 h at room temperature. The
plates were washed 10 times with PBST for >1 h, followed by five
additional washes with PBS for >30 min. Plates were incubated with 100
µl/well of PBS containing streptavidin-conjugated alkaline
phosphatase (0.7 µg/ml) for 1 h at room temperature and washed
as before. Biotinylated molecules were colorimetrically detected by
overnight incubation with para-nitrophenyl phosphate at
4°C, and the absorbance at 
405 nm was
determined using a 96-well plate reader. Although we observed
differences between batches of I-Ag7 and
I-Ad molecules in terms of maximal binding
capacity and half-lives of the preparations, their
IC50 values and the relative binding capacity of
I-Ag7 vs I-Ad were
reproducible.
Biotinylation and FACS analysis
MHC class II molecules with the biotinylation sequence were purified, as described above, and biotinylated with the BirA enzyme according to the manufacturer instructions (Avidity, Denver, Co). Biotinylated molecules were tetramerized with PE-labeled streptavidin (BioSource International, Camarillo, CA) and used for FACS. Staining was performed at 4°C for 2 h with 10 µM of labeled tetramers. Analysis was performed on a FACScaliber instrument (Becton Dickinson, Mountain View, CA). Data were analyzed using CellQuest software (Becton Dickinson). GAD65524–538-specific hybridomas were provided by Dr. P. Reich (Anergen, Redwood City, CA).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We have previously reported the successful production of large
amounts of I-Ad molecules in D.
melanogaster SC2 cells and their crystallization with
covalently linked single peptides (14, 26, 27). Soluble,
secreted, and empty I-Ag7 molecules were
expressed in a similar manner. Their transmembrane and cytoplasmic
domains were deleted and replaced by leucine zippers to enhance
heterodimeric chain pairing (14, 27). Levels of expression
in SC2 cells were comparable to I-Ad expression
levels, and oscillated between 0.2 and 0.5 mg/L of final purified
product. No aggregation of the empty molecules was noted at any stage
of purification. The purification of these molecules was conducted by a
combination of nickel-agarose affinity chromatography and gel
filtration at neutral pH (Fig. 1
). The
- and ß-chains were secreted at equimolar ratios (Fig. 1A). To test the stability of these dimers, leucine zippers
were removed by thrombin digestion at 37°C for 4 h and the
digest was analyzed by gel filtration and SDS-PAGE after undigested
molecules were removed using nickel-agarose beads. Uncut and cut
material were incubated at 37°C for 60 min before gel filtration.
Both samples had the same retention time (15 ml, 55 kDa) on the
Superdex 200 profile (Fig. 1B) indicating that, with or
without zippers, these molecules had no tendency to dissociate into
monomeric chains or to aggregate, suggesting that the loss of the
ß57/76 salt bridge was not critical for the stability of the
I-Ag7 heterodimer. Unlike Ab affinity
chromatography (26), this purification procedure does not
use extreme pH and has a major impact in term of assessing the binding
of peptides to purified empty I-Ag7 or
I-Ad molecules. Whereas Ab affinity-purified
empty MHC molecules were really empty (26), molecules
purified at neutral pH were found to be associated with a large
collection of peptides that could be stripped at acidic pH and analyzed
by mass spectrometry (Fig. 1C). This same phenomenon has
been described for empty MHC class I molecules expressed in the fly
system (31). Thus, it would be proper to call these
molecules "functionally empty", instead of "empty", because
they are associated with low affinity peptides. These peptides could
easily be displaced by higher affinity peptides in a binding assay
(Fig. 2). Direct binding of peptides and
specificity of peptide binding (tested by competition) were shown for
both I-Ad and I-Ag7 with
the ROI peptide (VHAAHAVHAAHAVHA) (Fig. 2, A and
B).
|
|
Correct folding of the recombinant I-Ag7
molecule in the fly expression system was further demonstrated by its
ability to associate with the Ii. A soluble form of human Ii with an
N-terminal hexa-histidine tag was cotransfected with
I-Ag7 and ß constructs retaining zippers
but no histidine tag. The cell line was cloned and the single cell
clone 134, which expresses equivalent amounts of the three chains, was
used for further studies. Complexes of the three chains, -ß-Ii,
were isolated by nickel-agarose chromatography (binding only Ii)
followed by gel filtration, hydrophobic interaction chromatography, and
ion exchange chromatography. The presence of all three chains was
demonstrated by SDS-PAGE and two-dimensional gel analysis of the final
pooled fractions (Fig. 1
A). Again, the behavior of
I-Ag7/Ii complexes was almost identical with the
behavior of I-Ad/Ii complexes (26).
Peptides could be loaded on these complexes at acidic pH in the absence
of accessory molecules (data not shown, 26). These results
confirmed the observation by Peterson and Sant (17) that
I-Ag7 associates normally with the Ii, excluding
possible gross defects of intercellular trafficking.
Promiscuous peptide motif for I-Ad and I-Ag7 molecules
A broad number of MHC class I and class II molecules binding motifs have been identified by direct sequencing of natural peptides and/or binding of random peptides and sequencing (32). For HLA-DQ and I-A molecules, the same strategy has not proven very fruitful, and most of these haplotypes have no clearly defined motif to date. In the case of I-Ag7, the number of studies addressing the issue is very limited and they have yielded rather controversial results (11, 12, 13). To try to address this question, we followed a strategy previously used by Hammer and colleagues to identify the HLA-DR1 class II molecule binding motif based on the selection of random peptide M13 phage display libraries with purified MHC molecules (33).
A random 12-mer peptide phage library was selected by four consecutive
rounds of panning against empty I-Ag7 molecules.
An increase of the absolute amount of eluted phages, combined with an
increase of binding over background up to 2700-fold after the third
round of selection indicated the enrichment of binding clones from the
original library (Fig. 3A).
One hundred and two clones of the third and fourth round of selection
and 107 clones from the naive library were further analyzed by direct
sequencing (Table I). In the selected
library, four different clones were identified multiple times. A
binding motif was sought by manual alignment and by using CLUSTAL W 1.7
(Human Center, Baylor College of Medicine, Houston, TX; Ref.
34), as well as PILEUP software (program of the Genetics
Computer Group, Madison, WI). When the total numbers of all 20 aa from
the enriched library (averaged to 100 clones) were compared with the
total numbers of the same amount of sequences from the naive library
(Table II), only minor enrichments were
found in the selected library for cysteine (65 vs 50), glutamate (41 vs
31), and phenylalanine (75 vs 51). However, these residues were
randomly distributed along the sequence, and no motif was apparent.
|
|
|
). These results suggested specific
selection for I-Ag7 binding during panning. With
respect to negatively charged residues that could compensate for the
absence of the aspartic acid ß57 in I-Ag7,
aspartic and glutamic acids were found more frequently within the last
four residues of the I-Ag7-selected library
(30.4%, Table I) than in the I-Ad (12.3%, Table IV) or naive (22.4%, Table II)
libraries.
|
|
B). A strong enrichment of binding clones was
observed after one and two rounds of panning. A total of 57 clones were
sequenced and analyzed for a possible binding motif (Table IV). Two
hydrophobic amino acids were enriched in the selected library, leucine
(151 vs 117) and phenylalanine (72 vs 51), whereas lysine was also
increased (44 vs 28) and serine diminished (116 vs 142). Again, no
apparent motif could be determined, but the prevalence of hydrophobic
residues satisfies the optimal requirement for such residues at the P4
and P6 positions in binding peptide to I-Ad, as
deduced from the crystal structure of two I-Ad
peptide complexes (14).
We used the three-dimensional structure of I-Ag7
complexed with GAD207–220 to rationalize these
binding studies. Like I-Ad,
I-Ag7 does not use large side chains to anchor in
its pockets for peptide binding. Instead, the binding energy is
provided mainly by interaction of the peptide backbone with the
peptide-binding groove. From the structural data, we were able to
suggest the following binding motif: P1, degenerate; P4 optimal, P, L,
V, I, G, A, and C; P4 nonoptimal, M, S, T, N, Q, and H; P6 optimal, L,
V, I, P, S, T, G, A, C, and N; P6 nonoptimal, D; P9 optimal, D, E, S,
G, and A; P9 nonoptimal, M, L, I, V, P, C, N, H, Q, and T
(15). Using that motif, we could align the vast majority
of peptides known to bind to I-Ag7. Peptides from
the unselected, as well as the selected I-Ag7 and
I-Ad peptide libraries, respectively, were then
aligned based on that motif. In the unselected (naive) library, 36% of
the sequences could satisfactorily be fitted using the optimal
alignment for I-Ag7 (P4, P6, and P9 using optimal
side chains). This number rose to 74% if one nonoptimal residue was
allowed only at one position. After selection against
I-Ag7, these numbers changed for round 3 to 27
and 65% (optimal and nonoptimal fit peptides, respectively) and to 32
and 64% for round 4. Two clones, which were identified multiple times
in round 3, were eliminated under these strict rules for
motif-containing clones (Table I). However, it is likely that these
sequences bind to I-Ag7; this hypothesis will be
tested later using synthetic peptides. For I-Ad,
54 and 55% could be aligned in the unselected and the selected library
using the optimal and nonoptimal motif, respectively. In summary,
essentially no enhancement of sequences containing this binding motif
was observed after selection. This result can be easily explained by
the high numbers of potential binding clones in the unselected library.
For comparison, <1% of clones of the naive library contained a motif
for I-Eb or I-Ak (33, 35). Selection of the same library for binding to these
particular MHC class II molecules should lead to a clear enrichment of
motif-containing clones.
Isolation of OVA and GAD65 peptides able to associate to I-Ag7 molecules
The number of peptides known to bind I-Ag7
molecules is limited. The number of known self-peptides is even smaller
and, for most of them, their affinity is unknown. Because functional
and structural studies of MHC molecules require the identification of
the appropriate peptide, we decided to design a general method to
identify I-Ag7 binding peptides. Several groups,
including ours, have previously shown that Ab-specific epitopes can be
retrieved from random cDNA fragment phage display libraries (36, 37, 38). Generally, these libraries consist of 15–30 aa peptide
fragments that are generated by random cDNA fragmentation and expressed
as fusions to one of the surface proteins of bacteriophage M13. A
similar strategy was used to isolate peptides that bound to empty MHC
class II molecules. The approach was tested by constructing two phages
displaying peptides known to bind I-Ag7 and
I-Ad: OVA323–339 (phage
JCM13-88-OVA) and ROI (phage JCM13-88-ROI) (13, 39). These
peptides were fused to the N terminus of protein VIII and a four
glycine-one serine linker. Specificity of binding was demonstrated by
incubating JCM13-88-OVA and JCM13-88-ROI phages with empty
I-Ag7 molecules with increasing concentrations of
free ROI peptide. Binding of phages was dose-dependently decreased by
increasing doses of ROI but not by a
non-I-Ag7-specific control peptide (data not
shown). Based on this result, two cDNA fragment libraries were
constructed from the OVA and murine GAD65 cDNAs called JCM13-88-rOVA
and JCM13-88-rGAD65, respectively. GAD65 was chosen because it has been
identified as one of the major autoantigens in IDDM in both humans and
mice (40, 41, 42). OVA was used as a control. Both
libraries were subjected to four successive rounds of selection (Fig. 3, C and D). After the fourth round, binding of
the JCM13-88-rOVA library had doubled as compared with the first round
of selection, and binding was 650 times more than background. To
identify binding peptides, phage clones were randomly chosen from the
third and fourth rounds of selection and sequenced (Table V). Of 60 clones analyzed, 13 clones were
coded for OVA peptides. Alignment of these sequences revealed two major
consensus sequences: GISSAESLKISQA, OVA314–326
and SVSEE FRADHPFL, OVA353–365. The
previously described I-Ag7 and
I-Ad epitope OVA323–339
sequence was found in only one clone. Both consensus sequences could be
aligned multiple times using our predicted I-Ag7
motif. The affinity of OVA323–339 and
OVA353–365 synthetic peptides was measured
directly for empty I-Ag7 molecules (Table III).
Interestingly, OVA353–365 did not bind to
I-Ad but, at 25 nM, was one of the best
I-Ag7 binders that we identified in this study
compared with 0.1 µM for OVA323–339. Thus,
these data suggest that I-Ag7 and
I-Ad can bind both similar and haplotype-specific
epitopes from the same protein.
|
). The most
represented sequence was YEIAPVFVLLE, GAD65 residues 207-217, which
appeared in 10 clones (59%) and which mapped within a sequence
previously identified as the major GAD65 epitope in vivo (43, 44, 45). The second consensus sequence,
GAD65105–119, had not been described as being a
functional epitope, but its binding to I-Ag7
molecules was confirmed in a binding assay (Table III). The second
major functional epitope identified by Chao and McDevitt as well as
Zechel et al. (43, 44, 45), GAD65 residues 221–235,
partially overlapped with clone TSS116. Therefore, it is noticeable
that, in vitro, we were able to identify two of the three major
functional determinants for GAD65.
|
). From the compilation
of these data, three main features emerged: 1) none of the peptides had
very high affinity (<1 nM) for I-Ad or
I-Ag7 molecules, 2) at the cutoff of 10 µM
(medium affinity), I-Ag7 bound slightly more
peptides (16/20) than I-Ad molecules (9/20), and
3) specificity was high with only 9 of 20 peptides binding to both
haplotypes (<10 µM). Also, as recently reported (46),
the binding of the full-length class II-associated invariant chain
peptide (81-104, human sequence) to I-Ag7 was
found to be of relatively low affinity (1.3 µM) compared with its
binding to I-Ad (0.18 µM). Expression and crystallization of a series of I-Ag7 peptide complexes
Because one of our goals was to carry out functional and
structural studies of I-Ag7 molecules, a series
of peptide sequences were cloned in frame into the
HLA-A2-I-Ag7 ß-chain hybrid construct (see
Materials and Methods) and transfected with the
I-Ad -chain into SC2 cells. The functionality
of the recombinant product was tested for its ability to stimulate T
cell hybridomas in a plate assay (26 , and data not shown)
and to stain T cells after biotinylation and tetramerization with
PE-labeled streptavidin (Fig. 4). All
molecules were purified by a succession of nickel-agarose
chromatography, ion exchange, and gel filtration (isoelectric point
values ranged from 5.5 to 7.5). The two-dimensional analysis of some of
these molecules is shown in Fig. 5.
Protein yields ranged from 0.2 to 2 mg/L and correlated very closely
with peptide affinities. The crystallization of four complexes
was successful: OVA323–339, ROI,
GAD207–220, and
GAD221–235. However, the
I-Ag7 OVA323–339 and
I-Ag7 ROI complex crystals were needles that
could not be grown to sufficient size for crystallographic analysis.
But, data quality crystals could be obtained, after removal of the
leucine zipper with thrombin digestion, for I-Ag7
GAD207–220 and GAD221–235
complexes. Diffraction data to 2.6 and 3.0Å were obtained for
I-Ag7 GAD207–220 and
GAD221–235, respectively, upon cryocooling of
the crystals, and these structures have been determined by x-ray
crystallography (15).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To address some of the most fundamental questions about the biology of
this molecule, we expressed I-Ag7 in a soluble
recombinant form in fly cells. The previous expression of its close
relative I-Ad in the same system (14, 26, 27) allowed us to compare the two molecules in terms of
stability, aggregation, and peptide binding. From the few biochemical
studies that have been published on I-Ag7, one of
the main impressions was that the molecule was unstable, and a poor
peptide binder (12, 16, 17, 18). The loss of the salt bridge
between ßAsp57 and
Arg86, seen in other MHC class II molecules,
could be the structural basis for poor chain pairing, instability, and
poor peptide binding of I-Ag7. Contrary to our
expectations, the biochemical and biophysical comparison between the
two I-Ad and I-Ag7
haplotypes revealed very few differences. Expression levels of both
molecules were almost identical, and they exhibited no tendency to
aggregate. In addition, the I-Ag7 dimer was shown
to be stable at 37°C, even after removal of the leucine zipper. More
importantly, both molecules bound peptides equally well. The first
evidence of peptide binding was provided by the elution of peptides
from the "empty" molecules. Most of these peptides are low affinity
binders that can be readily exchanged in a binding assay and that were
stripped from the groove when molecules were Ab affinity purified
(26). Peptide binding was lost in a few days for low
pH-eluted molecules, whereas histidine-tagged molecules were stable for
several weeks. The occupancy of empty MHC molecules made in the fly
expression system by low affinity peptides has already been noted for
the class I molecules H-2Kb (31) and
H-2Ld (48). The screen of a random
peptide library with "empty" I-Ad and
I-Ag7 molecules confirmed the degeneracy of
binding motif for both molecules as no particular anchor residues
became apparent in the screen. Then, it became obvious from the crystal
structures of I-Ag7 and
I-Ad molecules, respectively, that both molecules
could bind peptide without using large anchor residues, unlike most
other MHC class I and class II molecules (14, 15, 32).
However, both structures, in combination with previously reported
peptide binding studies and in vivo data as well as our own peptide
binding studies, allowed a rational approach to determine a binding
motif for both molecules. The major energy for peptide binding appears
to be primarily provided by the interaction of the peptide binding
groove with the peptide backbone; the accommodation of side chains into
the P1, P4, P6, and P9 pockets was based on the combination of size and
hydrophobicity of each pocket. The peptide P1 position appears
degenerate; at P4, small hydrophobic residues, and, at P6, small
hydrophobic or hydrophilic side chains can be accommodated. Finally, P9
has a preference for either negatively charged or small neutral
residues, in case of I-Ag7, and for small neutral
residues in case of I-Ad, respectively. Alignment
of our random peptide library, before and after selection against
I-Ag7 and I-Ad, revealed no
increase in the number of clones containing a motif. The combination of
high persistence of the motif in the unselected library with the
necessity to analyze randomly picked clones after selection explains
these results. In contrast, an enrichment of motif-displaying clones
would be expected in case of selection against, e.g.,
I-Ek or I-Ak because <1%
of clones of the unselected library contains a motif for either
molecule (32, 35).
The second most important point of our study was the identification of
new I-Ag7 epitopes using phage display of cDNA
fragments. The advantage of this approach is that it is fast and doable
for any cDNA or cDNA library. The main disadvantage is that it is based
solely on MHC binding; therefore, binding peptides that will not be
available in vivo by degradation in Ag processing cells will still be
selected. As a consequence, the identification of functional epitopes
using this method requires the immunization of animals with the whole
protein and the test for reactivity of this peptide. In the case of
GAD65, the immunization of NOD mice with the protein and the screen for
peptide reactivity had already been conducted by two groups (43, 44, 45). It is interesting to note that the major sequence that
we selected (207–217) was also the dominant functional epitope
(43, 44, 45). In the GAD65 sequence, a surprising 126
peptides can be identified that meet the requirement of the
I-Ag7 motif (Table VII). In comparison, no sequence can be
aligned with the strict motif of either I-Ek or
I-Ak.
|
In summary, recombinant I-Ad and I-Ag7 molecules appeared to be very similar in term of stability, solubility, and peptide binding. Binding motifs of both molecules are degenerate, allowing them to bind a wide selection of peptides. Based on the binding of a large collection of unselected peptides, I-Ag7 molecules appeared more promiscuous than I-Ad molecules. The same conclusion was reached recently by Carrasco-Martin and colleagues using a different approach (49). This promiscuity is certainly associated with a lower average affinity of peptides for I-Ag7 molecules, and will translate in vivo by shorter half-lives at the cell surface for I-Ag7 peptide complexes (22). This "functional instability" could lead to inoperative thymic selection and high frequency of autoreactive T cells on a diabetic genetic background (50). The determination of the crystal structure of I-Ag7 complexed to self-peptides will help us design new experiments to test these hypotheses.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Current address: Department of Medicine 0613-C, University of California at San Diego, 9500 Gilman Drive, La Jolla, CA 92083-0613.
3 Current address: Abgenix Inc., 7601 Dumbarton Circle, Fremont, CA 94555.
4 Address correspondence and reprint requests to Dr. Luc Teyton, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037.
5 Abbreviations used in this paper: IDDM, insulin-dependent diabetes mellitus; NOD, nonobese diabetic; GAD, glutamic acid decarboxylase; NEPHGE, nonequilibrium pH gradient electrophoresis; Ii, invariant chain; PBST, PBS containing 0.05% Tween 20.
Received for publication April 5, 2000. Accepted for publication June 26, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-chain. Nature 345:727.[Medline]
ß heterodimers in the absence of antigenic peptide. Cell 68:465.[Medline]
This article has been cited by other articles:
|
|
 |
|
  R. Arakaki, A. Nagaoka, N. Ishimaru, A. Yamada, S. Yoshida, and Y. Hayashi Role of Plasmacytoid Dendritic Cells for Aberrant Class II Expression in Exocrine Glands from Estrogen-Deficient Mice of Healthy Background Am. J. Pathol., May 1, 2009; 174(5): 1715 - 1724. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Ishimaru, R. Arakaki, S. Yoshida, A. Yamada, S. Noji, and Y. Hayashi Expression of the retinoblastoma protein RbAp48 in exocrine glands leads to Sjogren's syndrome-like autoimmune exocrinopathy J. Exp. Med., November 24, 2008; 205(12): 2915 - 2927. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Tian, M. J. I. Ansari, J. Paez-Cortez, J. Bagley, J. Godwin, M. Donnarumma, M. H. Sayegh, and J. Iacomini Induction of Robust Diabetes Resistance and Prevention of Recurrent Type 1 Diabetes Following Islet Transplantation by Gene Therapy J. Immunol., November 15, 2007; 179(10): 6762 - 6769. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Gardiner, K. A. Richards, A. J. Sant, and L. S. Arneson Conformation of MHC class II I-Ag7 is sensitive to the P9 anchor amino acid in bound peptide Int. Immunol., September 1, 2007; 19(9): 1103 - 1113. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. G. Levisetti, A. Suri, S. J. Petzold, and E. R. Unanue The Insulin-Specific T Cells of Nonobese Diabetic Mice Recognize a Weak MHC-Binding Segment in More Than One Form J. Immunol., May 15, 2007; 178(10): 6051 - 6057. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J. Mellanby, C. H. Koonce, A. Monti, J. M. Phillips, A. Cooke, and E. K. Bikoff Loss of Invariant Chain Protects Nonobese Diabetic Mice against Type 1 Diabetes J. Immunol., December 1, 2006; 177(11): 7588 - 7598. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. R. Bill, D. G. Mack, M. T. Falta, L. A. Maier, A. K. Sullivan, F. G. Joslin, A. K. Martin, B. M. Freed, B. L. Kotzin, and A. P. Fontenot Beryllium Presentation to CD4+ T Cells Is Dependent on a Single Amino Acid Residue of the MHC Class II {beta}-Chain J. Immunol., November 15, 2005; 175(10): 7029 - 7037. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. S. Smith, T. Patterson, and M. E. Pauza Transgenic Ly-49A Inhibits Antigen-Driven T Cell Activation and Delays Diabetes J. Immunol., April 1, 2005; 174(7): 3897 - 3905. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. J. Carven, S. Chitta, I. Hilgert, M. M. Rushe, R. F. Baggio, M. Palmer, J. E. Arenas, J. L. Strominger, V. Horejsi, L. Santambrogio, et al. Monoclonal Antibodies Specific for the Empty Conformation of HLA-DR1 Reveal Aspects of the Conformational Change Associated with Peptide Binding J. Biol. Chem., April 16, 2004; 279(16): 16561 - 16570. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Zavala-Ruiz, E. J. Sundberg, J. D. Stone, D. B. DeOliveira, I. C. Chan, J. Svendsen, R. A. Mariuzza, and L. J. Stern Exploration of the P6/P7 Region of the Peptide-binding Site of the Human Class II Major Histocompatability Complex Protein HLA-DR1 J. Biol. Chem., November 7, 2003; 278(45): 44904 - 44912. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-H. Jang, N. P. Seth, and K. W. Wucherpfennig Ex Vivo Analysis of Thymic CD4 T Cells in Nonobese Diabetic Mice with Tetramers Generated from I-Ag7/Class II-Associated Invariant Chain Peptide Precursors J. Immunol., October 15, 2003; 171(8): 4175 - 4186. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Jaeckel, L. Klein, N. Martin-Orozco, and H. von Boehmer Normal Incidence of Diabetes in NOD Mice Tolerant to Glutamic Acid Decarboxylase J. Exp. Med., June 16, 2003; 197(12): 1635 - 1644. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Cantu III, K. Benlagha, P. B. Savage, A. Bendelac, and L. Teyton The Paradox of Immune Molecular Recognition of {alpha}-Galactosylceramide: Low Affinity, Low Specificity for CD1d, High Affinity for {alpha}{beta} TCRs J. Immunol., May 1, 2003; 170(9): 4673 - 4682. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Suri, J. J. Walters, O. Kanagawa, M. L. Gross, and E. R. Unanue Specificity of peptide selection by antigen-presenting cells homozygous or heterozygous for expression of class II MHC molecules: The lack of competition PNAS, April 29, 2003; 100(9): 5330 - 5335. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. E. Starwalt, E. L. Masteller, J. A. Bluestone, and D. M. Kranz Directed evolution of a single-chain class II MHC product by yeast display Protein Eng. Des. Sel., February 1, 2003; 16(2): 147 - 156. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Sidney, M.-F. del Guercio, S. Southwood, and A. Sette The HLA Molecules DQA1*0501/B1*0201 and DQA1*0301/B1*0302 Share an Extensive Overlap in Peptide Binding Specificity J. Immunol., November 1, 2002; 169(9): 5098 - 5108. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. V. Tarbell, M. Lee, E. Ranheim, C. C. Chao, M. Sanna, S.-K. Kim, P. Dickie, L. Teyton, M. Davis, and H. McDevitt CD4+ T Cells from Glutamic Acid Decarboxylase (GAD)65-specific T Cell Receptor Transgenic Mice Are Not Diabetogenic and Can Delay Diabetes Transfer J. Exp. Med., August 19, 2002; 196(4): 481 - 492. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Wen, F. S. Wong, R. Sherwin, and C. Mora Human DQ8 Can Substitute for Murine I-Ag7 in the Selection of Diabetogenic T Cells Restricted to I-Ag71 J. Immunol., April 1, 2002; 168(7): 3635 - 3640. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Suri, I. Vidavsky, K. van der Drift, O. Kanagawa, M. L. Gross, and E. R. Unanue In APCs, the Autologous Peptides Selected by the Diabetogenic I-Ag7 Molecule Are Unique and Determined by the Amino Acid Changes in the P9 Pocket J. Immunol., February 1, 2002; 168(3): 1235 - 1243. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Monneaux, H. Dumortier, G. Steiner, J.-P. Briand, and S. Muller Murine models of systemic lupus erythematosus: B and T cell responses to spliceosomal ribonucleoproteins in MRL/Faslpr and (NZB NZW)F1 lupus mice Int. Immunol., September 1, 2001; 13(9): 1155 - 1163. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Kelemen, D. R. Wegmann, and J. C. Hutton T-Cell Epitope Analysis on the Autoantigen Phogrin (IA-2{beta}) in the Nonobese Diabetic Mouse Diabetes, August 1, 2001; 50(8): 1729 - 1734. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
