Probing molecular interactions within class II MHC Aq/ glycopeptide/T-cell receptor complexes associated with collagen-induced arthritis

Article abstract of DOI:10.1021/jm0705410

T cells obtained in a mouse model for rheumatoid arthritis are activated by a glycopeptide fragment from rat type II collagen (CII) bound to the class II major histocompatibility complex A^q molecule. We report a comparative model of A^q in complex with the glycopeptide CII260-267. This model was used in a structure-based design approach where the amide bond between Ala²⁶¹ and Gly²⁶² in the glycopeptide was selected for replacement with ψ[COCH₂], ψ[CH₂NH ₂⁺], and ψ[(E)-CH=CH] isosteres. Ala-Gly isostere building blocks were then synthesized and introduced in CII260-267 and CII259-273 glycopeptides. The modified glycopeptides were evaluated for binding to the A^q molecule, and the results were interpreted in view of the AVglycopeptide model. Moreover, recognition by a panel of T-cell hybridomas revealed high sensitivity for the backbone modifications. These studies contribute to the understanding of the interactions in the ternary A ^q/glycopeptide/T-cell receptor complexes that activate T cells in autoimmune arthritis and suggest possibilities for new vaccination approaches.

Full text of DOI:10.1021/jm0705410

J. Med. Chem. 2007, 50, 5627-5643
5627
Probing Molecular Interactions within Class II MHC Aq/Glycopeptide/T-Cell Receptor
Complexes Associated with Collagen-Induced Arthritis
Ida E. Andersson,† Balik Dzhambazov,‡ Rikard Holmdahl,‡ Anna Linusson,*,† and Jan Kihlberg*,†,§
Department of Chemistry, Umeå UniVersity, SE-901 87 Umeå, Sweden, Medical Inflammation Research, BMC I11, Lund UniVersity,
SE-221 84 Lund, Sweden, and AstraZeneca R&D Mo¨lndal, SE-431 83 Mo¨lndal, Sweden
ReceiVed May 9, 2007
T cells obtained in a mouse model for rheumatoid arthritis are activated by a glycopeptide fragment from
rat type II collagen (CII) bound to the class II major histocompatibility complex Aq molecule. We report a
comparative model of Aq in complex with the glycopeptide CII260-267. This model was used in a structure-
based design approach where the amide bond between Ala261 and Gly262 in the glycopeptide was selected
+
for replacement with ψ[COCH2], ψ[CH2NH2 ], and ψ[(E)-CHdCH] isosteres. Ala-Gly isostere building
blocks were then synthesized and introduced in CII260-267 and CII259-273 glycopeptides. The modified
glycopeptides were evaluated for binding to the Aq molecule, and the results were interpreted in view of the
Aq/glycopeptide model. Moreover, recognition by a panel of T-cell hybridomas revealed high sensitivity
for the backbone modifications. These studies contribute to the understanding of the interactions in the
ternary Aq/glycopeptide/T-cell receptor complexes that activate T cells in autoimmune arthritis and suggest
possibilities for new vaccination approaches.
Introduction
similar symptoms and histopathology of the affected joints as
seen in RA patients.12 In a similar way that RA is associated
with expression of DR4 and DR1, susceptibility to CIA in mice
is genetically linked to the class II MHC Aq molecule.13,14 It
has been found that T cells obtained from mice with CIA are
activated by the rat CII259-273 peptide sequence.15 Further
studies have revealed that Ile260 and Phe263 are critical residues
for binding to Aq, while Lys264 is a major T-cell receptor (TCR)
contact point.16,17 Importantly, a majority of the T-cell hybri-
domas specifically recognize Lys264 when post-translationally
hydroxylated and glycosylated with a â-D-galactopyranosyl
residue (GalHyl264).18-20 This recognition is specific for indi-
vidual hydroxyl groups on the carbohydrate moiety15,21 and, in
addition, also depends on interactions with the ꢀ-amino
group19,22,23 in the hydroxylysine side chain and the O-glycosidic
linkage.22,24 The Aq binding preferences for some of the amino
acid residues in CII260-267, which is the minimal epitope
required for binding to Aq, while still being able to elicit a T-cell
response,25 have recently been described by a quantitative
structure-activity relationship (QSAR) model.26
Autoimmune diseases are characterized by the immune
system responding to self-tissue antigens. In rheumatoid arthritis
(RAa), this specifically leads to chronic inflammation of the
peripheral cartilaginous joints, eventually resulting in degrada-
tion of joint cartilage, irreversible bone erosion, and severe joint
deformation.1 Susceptibility to RA has been strongly correlated
with the class II major histocompatibility complex (MHC) genes
HLA-DR4 and HLA-DR1, that code for the class II MHC
molecules DR4 and DR1, respectively.2-4 Class II MHC
molecules are membrane-bound glycoproteins, with a molecular
weight of approximately 60 kDa, that consist of two nonco-
valently associated chains, R and â.5 These glycoproteins are
found in antigen-presenting cells (APC), that is, macrophages,
B cells and dendritic cells, where they bind peptide antigens
obtained by degradation of extracellular proteins. The class II
MHC molecule in complex with the peptide antigen is subse-
quently transported to the cell surface of the APC where it is
presented for recognition by receptors on circulating CD4+
T
cells. Although the events initiating the immune response in
RA are unknown, the specific inflammatory attack on joint
cartilage indicates that T cells recognize a joint-specific self-
antigen. Both T cells6-8 and antibodies9-11 directed against type
II collagen (CII), which is the main protein in joint cartilage,
have been found in RA patients, indicating that it is an important
component of the disease pathology.
It has been shown that vaccination of mice with the CII259-
273 glycopeptide in complex with soluble Aq protein can protect
against CIA, and that those individuals that still develop the
disease experience a delayed onset and reduced severity.27
Importantly, treatment with the Aq/glycopeptide complex was
also successful in relieving symptoms in mice with an estab-
lished chronic relapsing disease. Although neonatal vaccination
with the CII259-273 glycopeptide alone can confer protection
against CIA,28 the therapeutic effect is not as pronounced
compared with the glycopeptide/Aq complex and much higher
doses of the glycopeptide are required. This is most likely effects
of proteolytic degradation and rapid clearance of the glycopep-
tide in vivo, thereby limiting its use as a therapeutic agent.
Development of peptidomimetics with improved proteolytic
resistance that are able to mimic the biological activity of the
native glycopeptide could circumvent some of these problems.
Such peptidomimetics with specific alterations could also be
used to study the interactions within the Aq/glycopeptide/TCR
Collagen-induced arthritis (CIA) is an animal model for RA,
where immunization of mice with CII derived from rat leads to
* To whom correspondence should be addressed. Phone: +46-90-
7866890 (A.L. and J.K.). Fax: +46-90-138885 (A.L. and J.K.). E-mail:
anna.linusson@chem.umu.se (A.L.); jan.kihlberg@chem.umu.se (J.K.).
† Umeå University.
‡ Lund University.
§ AstraZeneca R&D Mo¨lndal.
a Abbreviations: RA, rheumatoid arthritis; MHC, major histocompat-
ibility complex; APC, antigen-presenting cell; CII, type II collagen; CIA,
collagen-induced arthritis; TCR, T-cell receptor; GalHyl, â-D-galactopy-
ranosyl hydroxylysine; QSAR, quantitative structure-activity relationship;
MS, multiple sclerosis; PDB, protein data bank; RMSD, root-mean-square
deviation; CLIP, class II-associated invariant chain peptide; IL-2, interleukin-
2; RMSG, root-mean-square gradient.
10.1021/jm0705410 CCC: $37.00 © 2007 American Chemical Society
Published on Web 10/18/2007

5628 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
complexes to obtain knowledge of the autoimmune T-cell
response in CIA on a molecular level.
structures of the class II MHC molecule Ab and one of Ad. These
structures, determined with resolutions between 2.15 and 2.50
Å, were found suitable as modeling templates because Ab, Ad,
as well as Aq have an insertion of two amino acids, that is,
â65Pro and â66Glu, that are missing in the other I-A mol-
ecules.47 Backbone alignment of the crystal structures revealed
high conformational similarity with root-mean-square deviation
(RMSD) values for the R-carbons in the binding sites between
0.59 and 0.75 Å. Importantly, the RMSD values of the two
different proteins, Ab and Ad, were in the same range as for the
two crystal structures of the same protein (Ab), reflecting a high
preservation of the overall 3D structure of these proteins.
Additionally, the side chains displayed high conformational
similarity, where conserved amino acids were almost identically
positioned in the three crystal structures. Amino acids that
differed between Ab and Ad, but with side chains having related
molecular properties, also adopted strikingly similar conforma-
tions. Based on the high sequence identity and conformational
similarity of the three suitable templates, the crystal structure
with the highest resolution, that is, Ab in complex with a human
CLIP peptide (2.15 Å, PDB code: 1MUJ),47 was selected for
the comparative modeling.
Peptidomimetics that inhibit antigen presentation by the
human DR1 and DR4 proteins have recently been shown to
have a therapeutic effect in transgenic mouse models for both
RA and multiple sclerosis (MS).29 Moreover, peptide epitopes
with subtle modifications of TCR contact points, such as
substitution of one amino acid, can induce altered T-cell
activation states, including anergy where the T cells no longer
are responsive to the antigen.30 Such modified peptides have
been used to modulate the T-cell responses and prevent
experimental autoimmune encephalomyelitis in a mouse model
for MS.31 Design strategies for modified peptides and peptido-
mimetics that interact with human or murine class II MHC
proteins include introduction of D-amino acids,32 aza-amino
acids,33 and other non-natural amino acids.34 Amide bonds in
the peptide backbone have also been replaced with various
isosteres, such as retro-inverso,35 methyleneamine,36 (E)-alk-
ene,37 and ketomethylene38 isosteres. Furthermore, peptide
mimetics based on polypyrrolinone scaffolds,39 mimicking the
polyproline type II conformation of the native peptide ligands,
as well as nonpeptidic small organic compounds40 have been
reported to bind to class II MHC proteins. The Ile260-Ala261
amide bond of CII260-267 has previously been replaced by a
methylene ether isostere, which was found to decrease the
binding affinity to the Aq molecule while still allowing some
T-cell hybridomas to recognize the Aq/glycopeptide complex
weakly.41,42
In recent years, several computational tools have emerged
that can promote the discovery of new peptide-based vaccines.
These include online databases that use predictive algorithms
to perform B cell and T cell epitope mapping as well as
prediction of MHC binding.43-45 The design of peptide epitopes
with optimized immunogenicity can also be guided by informa-
tion from a crystal structure of the target class II MHC molecule.
Rational design can thus be combined with synthetic chemical
methods to develop modified peptides with altered MHC binding
and TCR signaling. Although the structure of Aq has not yet
been solved by X-ray crystallography, a comparative model of
Aq based on the human class II MHC DR1 molecule, which
has 63% sequence identity with Aq, has previously been
described.17 In recent years, however, X-ray crystallography has
provided several crystal structures of murine class II MHC
molecules that are more closely related to Aq.
In the initial comparative modeling, the side chain conforma-
tions of four amino acids in the binding site of Aq, that is,
R73Trp, R79Phe, â28Asn, and â37Trp, could not be unambigu-
ously determined, and two rotamers for each side chain were
therefore investigated, resulting in 16 models in total. Further-
more, two amino acids (â86Val and R52Thr) in the protein,
which also differed from the template, restricted the volume of
the P1 pocket as noted after initial attempts of docking the
CII260-267 glycopeptide and were therefore adjusted to make
the pocket as large as possible. Energy minimization was
successfully performed for 13 of the 16 models, and a final
comparative model of the Aq molecule was selected based on
the ability of the four investigated side chains (i.e., R73Trp,
R79Phe, â28Asn, and â37Trp) to participate in favorable
interactions. Evaluation of this model revealed that all torsion
angles were in allowed regions in the Ramachandran plot and
no outliers were detected among the dihedral angles, bond
angles, or bond lengths.
Modeling of the CII260-267 Glycopeptide. The modeling
studies were focused on the minimal epitope CII260-267, rather
than the longer CII259-273 glycopeptide, to facilitate the
computational operations by minimizing the number of rotatable
bonds in the ligand. The epitope CII260-267 was modeled into
the peptide-binding groove using two different strategies;
docking and comparative modeling. In the former, 106 different
low-energy conformations of the glycopeptide were rigidly
docked into the binding site of the Aq comparative model. The
binding pose ranked second by the scoring function was
consistent with previously experimentally determined binding
data.16,17 The latter approach, where the coordinates of the
backbone in the human CLIP peptide found in the Ab crystal
structure was used as template to model the backbone of the
glycopeptide, also resulted in conformations of the amino acid
side chains that fulfilled the evaluation criteria (see Experimental
Section). After energy minimization with the Aq molecule, the
two binding poses were found to be almost identical in the
backbone from Ile260 to GalHyl264 of the glycopeptide (Figure
1). However, differences were observed in the conformation of
the Gln267 residue at the C-terminal, which may indicate that
this part of the glycopeptide is more flexible in the binding site.
The conformation generated by the rigid docking was rotated
at the C-terminal, while the pose generated using the backbone
Here we report a new comparative model of the Aq molecule
in complex with CII260-267 that was developed to enable
studies of specific molecular interactions. The crystal structure
that was used as template had significantly higher sequence
identity with Aq and higher resolution, as compared to the
previous DR1 template. In a subsequent structure-based design,
the amide bond between Ala261 and Gly262 in the glycopeptide
ligand was selected for replacement with ψ[COCH2], ψ[CH2-
NH2+], and ψ[(E)-CHdCH] isosteres. The three Ala-Gly
isostere building blocks were synthesized and subsequently
introduced in CII260-267 and the longer glycopeptide CII259-
273 using Fmoc-based solid-phase peptide synthesis. The
backbone-modified glycopeptides were evaluated for binding
to the Aq molecule and recognition by T-cell hybridomas
obtained from mice with CIA.
Results and Discussion
Comparative Modeling of the Aq Molecule. Three crystal
structures in the Protein Data Bank (PDB)46 database were found
to have more than 90% sequence identity with Aq, that is, two

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5629
Figure 1. Superposition of two extracted binding poses of the CII260-
267 glycopeptide generated by docking (colored red) and comparative
modeling (colored gray); here shown after energy minimization with
the Aq comparative model.
coordinates of the CLIP peptide displayed an extended confor-
mation throughout the peptide backbone with the Gln267 side
chain pointing out of the peptide-binding groove. As for other
class II MHC proteins, the Aq molecule is capable of binding
longer peptides, which is consistent with an extended conforma-
tion of the peptide backbone. Consequently, the following
studies in this paper are focused on the model having CII260-
267 in its extended conformation.
Characterization of the Aq/CII260-267 Complex. As
expected, the overall backbone conformation of the modeled
Aq molecule in the energy-minimized Aq/CII260-267 complex
displayed high structural similarity with the template crystal
structure. Backbone alignment of the R-carbons gave RMSD
values of 1.06 Å and 0.88 Å for the whole protein and in the
binding site, respectively. Also, the conserved amino acids in
the binding site of Aq showed high conformational similarity
with the crystal structure homolog. The CII260-267 glycopep-
tide, which bound to Aq with an extended backbone conforma-
tion, was positioned deep in the peptide-binding groove, leaving
the side chain of GalHyl264 only partly extending out of the
groove to interact with a TCR (Figure 2). The position of the
flexible GalHyl264 side chain is somewhat uncertain and
structural interpretation of T-cell responses is further compli-
cated by the absence of a TCR in the model. Previous studies
have, however, shown that most T-cell hybridomas have a strong
dependence on the hydroxyl group at C-4 on the GalHyl264
moiety.15 In the Aq/glycopeptide model this hydroxyl group,
that is, HO-4, is the most exposed of all the galactose hydroxyl
groups for contacts with a TCR, providing overall support for
the extended orientation of GalHyl264. However, it should be
noted that a conformation where the GalHyl264 side chain instead
is flipped in the opposite direction could also lead to an exposed
HO-4.
A total of nine areas that bind peptide side chains, referred
to as P1-P9, with different properties could be distinguished
in the peptide-binding groove, thereby establishing the binding-
specificity of Aq. Residues Ile260 and Phe263 in the CII260-
267 glycopeptide were found to be extensively anchored in the
prominent P1 and P4 pockets, respectively, as proposed
earlier16,17 (Figure 3a). The well-defined, hydrophobic P1 pocket
was formed by residues R31Trp, R43Trp, R32Phe, â85Val,
R52Thr, and â82Asn of Aq. Similarly, as in the previous Aq
comparative model,17 the â84aGly residue is positioned in the
vicinity of the P1 pocket. The restricted size of this residue
probably contributes to making the P1 pocket large enough to
accommodate the bulky side chain of Ile260 in CII260-267.
Analogously, â13Gly, â28Asn, â26Ser, â15Cys, â79Cys,
Figure 2. (a) Secondary structure (shown as a ribbon diagram) of the
peptide-binding groove of the Aq comparative model with the CII260-
267 glycopeptide bound with an extended backbone conformation. (b)
The Aq/CII260-267 complex viewed along the peptide-binding groove
from the N- to the C-terminal of the glycopeptide and with HO-4 and
HO-6 of the GalHyl264 moiety indicated. The molecular surface of the
Aq molecule is colored blue for polar parts, brown for nonpolar parts,
and green for intermediate polarity.
â78Val, and R7Gly established the hydrophobic to medium
polarity of the P4 pocket. The amino acids in CII260-267 at
positions P3, P6, and P7 were found to be completely or
predominantly buried in the Aq binding groove. Gly262 was
positioned at P3 in a shallow, slightly polar binding area, while
the medium-sized, hydrophobic P6 accommodated Gly265. P7
was also found to be quite shallow, where the Glu266 side chain
in CII260-267 adopted a conformation where it was pointing
toward the helical domain rather than toward the â-sheet floor
of the binding groove, just as in the Ab/CLIP crystal structure.
The glycopeptide side chains Ala261, GalHyl264, and Gln267
positioned at P2, P5, and P8, respectively, were solvent exposed.
Conformational flexibility of the Gln267 residue in the C-
terminus of CII260-267 was however indicated in the modeling,
as previously stated. The properties of the P2, P3, and P6-P8
binding areas in the modeled Aq structure were consistent with
the preferred amino acids in each position suggested by the
recently published QSAR model for peptide binding to Aq.26
The medium-sized P9, which was not occupied by the glyco-
peptide in the model, was found to be of moderate polarity.
An extensive hydrogen-bonding network was found between
the CII260-267 glycopeptide and amino acids in Aq (Figure
3b). A total of 8 out of 13 hydrogen bonds (marked in Figure
3b) that were formed to the glycopeptide backbone were

5630 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
Figure 3. (a) View of the modeled Aq peptide-binding groove with the CII260-267 glycopeptide with an extended backbone conformation. The
nine binding areas are denoted P1-P9 (see text for details) and the molecular surface of the Aq molecule is colored blue for polar parts, brown for
nonpolar parts, and green for intermediate polarity. (b) The extensive network of hydrogen bonds and ionic interactions found between the CII260-
267 glycopeptide and the Aq molecule. The interactions (see Experimental Section for definition) are illustrated with dashed lines and are color
coded according to the hydrogen-acceptor distance; orange indicates a distance of 1.4-1.99 Å, while green indicates a distance of 2.0-2.7 Å.
Interactions that are conserved in the investigated Ab and Ad crystal structures are indicated by underlined amino acids in the Aq molecule. (c) The
three isosteres with varying geometry, charge, and hydrogen-bonding properties that were selected to replace the amide bond between Ala261 and
Gly262 in the glycopeptide backbone.
conserved in the investigated Ab and Ad crystal structures.47-49
Both amide groups on either side of the important GalHyl264
residue were involved in hydrogen bond interactions, thereby
contributing to the positioning of this important TCR contact
point in the binding site. Previous studies have shown that
replacement of Glu266 with alanine leaves the T-cell hybridoma
HCQ.3 nonresponsive.42 Somewhat surprisingly, the Glu266 side
chain was found to be almost completely buried in the binding
site in the Aq/CII260-267 model, although it was noted that it
formed a hydrogen bond to HO-2 of GalHyl264. Interestingly,
replacement of HO-2 with a hydrogen atom also led to that the
T-cell response was lost for HCQ.3.15 A possible explanation
could thus be that the TCR of HCQ.3 is not in direct contact
with the Glu266 side chain, but that this residue instead stabilizes
the position of the GalHyl264 side chain so that an epitope with
the proper contact points is presented to the TCR.
Design of Backbone-Modified Glycopeptides. The sequence
from Ile260 to GalHyl264 of CII260-267 contains the most
essential features, that is, the anchoring residues and the major
TCR contacts required for binding to Aq and for eliciting a T-cell
response. Hence, it was attractive to probe the molecular
interactions in this part of the ternary Aq/glycopeptide/TCR
complexes. Based on the structural information provided by the
Aq/CII260-267 model, the Ala261-Gly262 amide bond was
found to be interesting to modify because its carbonyl group
formed two hydrogen bonds to residues in Aq, while the amino
group did not participate in any interactions. Isosteric replace-
ment of this amide bond was thus aimed at exploring the
interactions within the Aq/glycopeptide/TCR complexes as well
as to validate the Aq/glycopeptide model. Isosteres with different
hydrogen-bonding properties selected for this study were the
ketomethylene (ψ[COCH2]), methyleneamine (ψ[CH2NH2+]),
and (E)-alkene (ψ[(E)-CHdCH]) isosteres (Figure 3c). Intro-
duction of the amide bond isosteres in the glycopeptide
backbone required synthesis of building blocks 1-3 (Figure 4)
for subsequent use in Fmoc-based solid-phase glycopeptide
synthesis.
Synthesis of Alaψ[COCH2]Gly Isostere 1. The synthetic
sequence toward the ketomethylene isostere 1 relied on a
strategy previously described by De´ziel et al.50 A key step was
the Horner-Wadsworth-Emmons reaction (Scheme 1) between
tert-butyl glyoxylate51 and â-ketophosphonate 5, prepared from
methyl ester 4 according to a literature procedure.50 To minimize
racemization, the reaction was initially performed with 1 equiv
of N-methylmorpholine in dichloromethane at 0 °C.50 Unfor-
tunately, this rendered alkene 6 in unacceptably low yields

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5631
Scheme 3. Synthesis of the Alaψ[CH2NH2+]Gly Isosterea
Figure 4. Ala-Gly isostere building blocks protected for solid-phase
synthesis of backbone-modified type II collagen glycopeptides.
Scheme 1. Synthesis of the Alaψ[COCH2]Gly Isosterea
a Reagents and conditions: (a) cf. refs 55 and 56; (b) Dess-Martin
periodinane, CH2Cl2/DMSO 1:1, rt; (c) HCl‚H-Gly-OtBu, NaBH(OAc)3,
CH2Cl2, MeOH, rt, 43% from 12; (d) TFA, CH2Cl2, rt; (e) Boc2O, DIPEA,
CH2Cl2, rt, 89% from 14.
>100:1 and it was thus concluded that no significant degree of
epimerization had occurred (see Supporting Information).
Synthesis of Alaψ[CH2N(Boc)]Gly Isostere 2. The synthetic
route toward the orthogonally protected methyleneamine isostere
2 involved reductive amination as a key step (Scheme 3).
Alcohol 1255 was first prepared from N-Fmoc-protected L-
alanine (11) via reduction of the corresponding mixed iso-butyl
anhydride using sodium borohydride according to a published
one-pot procedure.56 After workup, 12 was directly oxidized to
aldehyde 13 by treatment with Dess-Martin periodinane.57 This
oxidation was performed in dichloromethane/dimethyl sulfoxide
(1:1) due to solubility problems. To minimize racemization of
the stereocenter, 13 was kept cold during workup and was used
directly without further purification.58,59 Thus, reductive alky-
lation60 of glycine tert-butyl ester with 13 using sodium
triacetoxyborohydride as reducing agent afforded secondary
amine 14 in a modest yield (43% from 12). Deprotection of
the carboxylic acid to give 15 required stirring overnight in
trifluoroacetic acid/dichloromethane (1:3), and this was followed
by Boc-protection of the secondary amine to afford carboxylic
acid 2 (89% from 14).
a Reagents and conditions: (a) cf. ref 50; (b) tert-butyl glyoxylate,51 TEA,
LiCl, MeCN, 0 °C, 97% as an isomeric mixture of E/Z 3:1; (c) H2, Pd/C,
EtOAc, rt, 81%; (d) TFA, CH2Cl2, rt; (e) FmocOSu, NaHCO3, acetone,
H2O, rt, 88% from 7.
Scheme 2. Synthesis of Mosher Amidesa
Synthesis of Alaψ[(E)-CHdCH]Gly Isostere 3. Wiktelius
et al. have developed a synthetic route to the Pheψ[(E)-CHd
CH]Phe isostere relying on a Wittig-type reaction with excellent
E-selectivity for the Phe-Phe derivative.61 This methodology
was applied to the synthesis of Alaψ[(E)-CHdCH]Gly isostere
3, which allowed investigation of the selectivity in the Wittig
reaction for a derivative with less bulky side chains.
a Reagents and conditions: (a) (i) TFA, CH2Cl2, rt; (ii) TMSCl, MeOH,
rt; (iii) (S)-MTPA-Cl, DIPEA, CH2Cl2, rt; (b) DBU, THF, 80 °C.
(<22%), although exclusively as the E-isomer. By instead using
1 equiv of triethylamine and lithium chloride52 in acetonitrile
at 0 °C an excellent yield of 6 (97%) was obtained as a
diastereomeric mixture (E/Z 3:1) according to 1H NMR analy-
sis.51 Separation of the two diastereomers was not required
because the alkene functionality was subsequently reduced by
hydrogenation over palladium on charcoal to afford the corre-
sponding saturated isostere 7 in 81% yield.53 Simultaneous
cleavage of the Boc-group and the tert-butyl ester followed by
Fmoc-protection then afforded the desired building block 1 (88%
from 7).
Phosphonium salt 19 was prepared in three steps from
L-alaninol in 62% overall yield (Scheme 4). Unfortunately, small
impurities of triphenylphosphine oxide (∼3.5%) could not be
removed in the purification of 19. In the Wittig reaction, 19
was first treated with 2 equiv of n-butyllithium to generate the
ylide and to lithiate the amide.61 Subsequent addition of tert-
butyldiphenylsilyl-protected â-hydroxipropionic aldehyde62-64
at -78 °C was followed by slowly raising the temperature to 0
°C to allow equilibration of the alkene to the E-isomer. This
resulted in formation of 20 in 80-89% yield, but disappoint-
ingly, the E/Z ratios varied between 3:2 and 2:3. No effect of
the counterion on the selectivity was observed when the bromide
corresponding to 19 was employed instead. Because the alkene
isomers could not be separated by flash chromatography, the
mixture was treated with tetrabutylammonium fluoride to afford
the corresponding alcohols. According to TLC, this deprotection
appeared to be quantitative, however, repeated purification by
flash chromatography to separate (E)-21 and (Z)-21 resulted in
only modest yields (29% and 30%, respectively, when starting
from a mixture of 20 with E/Z 1:1). In the following step,
It had previously been reported that the Horner-Wadsworth-
Emmons reaction between benzyl glyoxylate and 5 under similar
conditions (i.e., 2 equiv of triethylamine in acetonitrile at room
temperature) was accompanied by racemization of the stereo-
center.50 To investigate how the modified reaction conditions
used in the synthesis of 6 had influenced the degree of
racemization, Mosher amide54 9 was synthesized from 7
(Scheme 2). A reference 1:1 diastereomeric mixture of Mosher
amides 10 was prepared analogously from racemic 7. Integration
of the signals from the diastereomeric methyl groups in the 1H
NMR spectrum of 9 established that the diastereoselectivity was

5632 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
Scheme 4. Synthesis of Alaψ[(E)-CHdCH]Gly and
Alaψ[(Z)-CHdCH]Gly Isosteresa
(Z)-21 in 75% yield over three steps according to the same
procedure as described for 3.
Synthesis of Backbone-Modified Glycopeptides. The im-
munological studies were initially focused on the minimal
epitope CII260-267 (27), and building blocks 1 and 2 were
therefore introduced in its backbone to give 28 and 29,
respectively (Scheme 5). However, use of the longer glycopep-
tide CII259-273 could also allow studies of the human class
II MHC protein DR4 because its epitope is shifted three amino
acids toward the C-terminus as compared to peptides bound by
the murine Aq molecule.65 Thus, all isostere building blocks,
that is, 1-3 and 26, were incorporated instead of residues
Ala261-Gly262 in the CII259-273 glycopeptide 30 to give 31-
34. The modified glycopeptides were synthesized on solid phase
under standard conditions19 according to the Fmoc protocol.
Glycopeptides 31-34 were synthesized with charged termini,
while the truncated glycopeptides 28 and 29 were synthesized
as C-terminal amides and with acetylated N-terminals to avoid
positioning glycopeptides with charged termini in the binding
site of Aq. An orthogonally protected galactosylated hydroxy-
lysine derivative, which was prepared according to a published
procedure66 but with some modifications (see Supporting
Information), was used in the syntheses.
An initial attempt to synthesize glycopeptide 28, containing
the ketomethylene isostere, relied on coupling of Fmoc-protected
isoleucine at the N-terminal followed by Fmoc deprotection and
acetylation. However, this rendered a complex mixture of
products after cleavage from the solid phase, with the major
product being the ring-closed analogue 35, as indicated by mass
spectrometry and NMR spectroscopy (Figure 5). Formation of
35 could be explained by nucleophilic attack of the R-amino
group in isoleucine on the ketone of the isostere, resulting in
formation of an imine38 followed by oxidation. This side reaction
was efficiently avoided by coupling of Ac-Ile-OH at the
N-terminus in the synthesis of 28, whereas Boc-Gly-Ile-OH was
coupled in the synthesis of 31. Gratifyingly, this strategy would
also minimize epimerization of the stereocenter in the isostere
during solid-phase synthesis by avoiding treatments with
piperidine that previously has been reported to epimerize CR
of other ketomethylene isosteres.67
The glycopeptides were cleaved from the solid support
followed by deacetylation of the galactose moiety by treatment
with methanolic sodium methoxide; a reaction that was carefully
monitored by analytical reversed-phase HPLC. Unfortunately,
the deprotection conditions led to epimerization of the base-
sensitive stereocenter in the ketomethylene isostere in glyco-
peptides 28 and 31. However, it was found that the rate of
deacetylation was much faster than the epimerization.68 Hence,
by performing the deacetylation in several small portions under
a Reagents and conditions: (a) TFAA, TEA, CH2Cl2, 0 °C, 78%; (b) I2,
PPh3, imidazole, toluene, rt, 79%; (c) PPh3, toluene, reflux, quantitative
yield; (d) (i) 2 equiv n-BuLi, THF, -78 °C f rt f -78 °C; (ii)
TBDPSO(CH2)2CHO,62-64 THF, -78 °C f 0 °C, 80-89% as isomeric
mixtures of E/Z 3:2-2:3; (e) TBAF, THF, rt, (E)-21 (29%) and (Z)-21
(30%); (f) CrO3, 4.4 M H2SO4 (aq), acetone, rt, 92% for 22, 87% for 24;
(g) K2CO3, MeOH, H2O, rt; (h) FmocOSu, MeCN/Na2CO3 (10% aq) 1:1,
rt, 89% for 3 from 22, 86% for 26 from 24.
alcohol (E)-21 was oxidized to carboxylic acid 22 in 92% yield
by treatment with Jones’ reagent. This reaction was carefully
monitored to minimize the formation of unidentified byproducts
that could not be removed in the following purification step.
Deprotection of the amino group using aqueous potassium
carbonate and methanol, followed by immediate Fmoc-protec-
tion afforded the desired building block 3 (89% from 22). Due
to lack of E-selectivity in the Wittig reaction, this reaction
sequence could also provide the Alaψ[(Z)-CHdCH]Gly building
block 26, which was subsequently incorporated in the glyco-
peptide backbone. Thus, building block 26 was synthesized from
Scheme 5. Structures of the Backbone-Modified Glycopeptides Derived from Type II Collagen

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5633
peptide at the 100 µM concentration, just as native 30. However,
at 4 µM concentration only ketomethylene 31 bound to Aq as
efficiently as the nonmodified glycopeptide 30. At this con-
centration, a reduction in the binding efficiency was observed
for (E)-alkene 33, whereas methyleneamine 32 displayed almost
no Aq binding. As expected, the (Z)-alkene modification in 34
resulted in a complete loss of binding to Aq even at the highest
tested concentrations.
In the Aq/CII260-267 model, two hydrogen bonds were
formed between the carbonyl group in the Ala261-Gly262 amide
bond and the â82Asn and R24His residues in Aq (Figure 3b).
The hydrogen bond involving the â82Asn is conserved in other
class II MHC A/peptide complexes.47-49 The ketomethylene
isostere in 31 is able to retain these hydrogen-bonding interac-
tions, and its increased flexibility compared to the amide bond
did not seem to affect the Aq binding ability. The hydrophobic
and rigid (E)-alkene in 33, on the other hand, is successful in
mimicking the geometry of the amide bond but is not able to
participate in any hydrogen-bonding interactions and is not
favored by the polar character of the binding site at this position.
(E)-Alkene 33 still displayed good binding to Aq, despite the
loss of two hydrogen bonds, suggesting that this glycopeptide
is bound in a conformation that successfully retains other
important interactions. However, the loss of binding that was
observed at low concentrations of 33 indicates that the carbonyl
group in the Ala261-Gly262 amide bond is engaged in interac-
tions that, although not critical for Aq binding, do contribute to
the affinity of the glycopeptide/Aq complex. The importance
of hydrogen bonding to the peptide backbone has previously
been studied for a diverse set of peptides that bind to the Ad
molecule.69 It was found that selective mutations of residues in
Ad that hydrogen bonded to the peptide ligand increased the
dissociation rates of the complexes by 4- to 1000-fold. The
hydrogen-bonding interactions at the peptide N-terminal were
particularly important for the binding stability.
Figure 5. Structure of the major product formed in the initial attempt
to synthesize glycopeptide 28, which contains the ketomethylene
isostere.
Figure 6. Binding of a fixed concentration of biotinylated CLIP peptide
to recombinant Aq proteins in a competitive binding assay with
increasing concentrations of the backbone-modified glycopeptides 28,
29, and 31-34, or the native glycopeptides 27 and 30. Aq-bound
biotinylated CLIP was detected in a time-resolved fluoroimmunoassay
using europium-labeled streptavidin. The values represent the average
from duplicates.
a limited reaction time, epimerization was minimized and the
minor diastereomer could successfully be removed during the
subsequent purification by reversed-phase HPLC. Analytical
reversed-phase HPLC also revealed that crude glycopeptides
29 and 32 each consisted of two main products in a ratio of
4:1, and after separation, both compounds were established to
have the correct mass. This isomeric mixture was most likely
due to racemization of the sensitive stereocenter in aldehyde
13 during synthesis of the methyleneamine building block 2.
Importantly, all traces of the minor diastereomers were suc-
cessfully removed in the purification to give both 29 and 32 as
pure isomers.
The significant reduction in binding observed for methyl-
eneamine 32 at 20 µM and lower concentrations is most likely
caused to a large extent by localization of a charged group in
an unfavorable position in the peptide-binding groove of Aq.
In addition, 32 is more flexible than the amide bond and is
unable to form the hydrogen bonds that the carbonyl group of
the Ala261-Gly262 amide bond is involved in. As expected, (Z)-
alkene 34 displayed only a very weak ability to bind to Aq at
high concentrations, which is most likely explained by the
stereochemistry of the alkene that distorts the extended backbone
conformation in between the two anchoring residues. The
simultaneous positioning of the Ile260 and Phe263 side chains in
the P1 and P4 pockets of Aq is thereby prevented.
After purification by reversed-phase HPLC, the backbone-
modified glycopeptides 28, 29, and 31-34 were obtained in
7-26% overall yields based on the resin capacities and in 98-
100% purity. All glycopeptides were homogeneous according
to analytical reversed-phase HPLC using two eluent systems,
The truncated CII260-267 backbone-modified glycopeptides
28 and 29 displayed the same trends in binding to Aq as their
extended counterparts. The effects were, however, more pro-
nounced in these shorter sequences, probably because fewer
interaction points in total are present that may compensate for
any loss of interactions, as a consequence of the introduced
isosteres. Compared to the corresponding reference 27, a
reduction in binding was observed for ketomethylene 28 at all
tested concentrations. The methyleneamine isostere in 29 was
found to only have a very weak ability to bind to Aq.
1
and their structures were confirmed by H NMR spectroscopy
and MALDI-TOF mass spectrometry.
Binding to the Class II MHC Aq Molecule. The ability of
the backbone-modified (28, 29, and 31-34) and native (27 and
30) glycopeptides to bind to the class II MHC Aq molecule was
studied in a competitive binding assay. Increasing concentrations
of each glycopeptide were incubated with a fixed concentration
of biotinylated CLIP peptide and recombinant Aq protein, and
the amount of Aq-bound biotinylated CLIP was then quantified
in a time-resolved fluoroimmunoassay (Figure 6). It was found
that the CII259-273 glycopeptides incorporating the ketom-
ethylene (31), methyleneamine (32), and (E)-alkene (33) isos-
teres completely inhibited binding of the biotinylated CLIP
Recognition by T-Cell Hybridomas. The ability of the
isosteric glycopeptides 28, 29, and 31-34, as well as the native
27 and 30, to induce an immune response was studied using a
panel of carbohydrate-dependent and Aq-restricted T-cell hy-
bridomas obtained from mice with CIA. The hybridomas were
selected from five groups previously established to have different

5634 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
Table 1. Immune Response of T-Cell Hybridomas When Incubated with Antigen-Presenting Spleen Cells Expressing Aq and Increasing Concentrations
of Glycopeptides 27-34a,b
sequence
CII259-273
CII260-267
30
(CONH)
31
(COCH2)
32
(CH2NH2
33
34
27d
(CONH)
28
(COCH2)
29
+
+
hybridoma
groupc
)
((E)-CHdCH))
((Z)-CHdCH))
(CH2NH2 )
HD13.9
HD13.10
HNC.1
HCQ.10
HCQ.2
HCQ.3
HCQ.6
HM1R.2
22a1-7E
1
1
1
2
3
3
3
4
5
+++
+++
+++
++
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
-
-
-
-
-
-
-
-
-
-
-
++
++
+++
+
-
-
-
-
-
-
-
-
-
++++
++++
+++
+++++
+++
-
++++
+++++
++++++
++++++
+++
-
++++
+++++
+++++
+++
+++
+++
+++
++
++
++
++
+
+++(+)
++
+++
+
a The T-cell response data and the corresponding dose-response curves are provided in the Supporting Information. b The magnitude of the responses
were determined from the concentration of antigen required to induce secretion of IL-2 corresponding to 6% of the max response for the native CII259-273
glycopeptide 30: - ) no response; + ) 20 µM; ++ ) 4 µM; +++ ) 0.8 µM; ++++ ) 0.16 µM; +++++ ) 0.032 µM; ++++++ ) 0.0064 µM;
(+) ) just below the threshold. c Groups of hybridomas with different specificity for the hydroxyl groups on the GalHyl264 moiety.15 d The native CII260-
267 glycopeptide 27 was tested on a separate plate and the magnitude of its response was determined compared to the response induced by reference 30 on
the same plate according to the criteria stated above.
specificity for the hydroxyl groups on the GalHyl264 moiety.15
The glycopeptides were incubated with antigen-presenting
spleen cells expressing Aq and each of the selected hybridomas.
Recognition of the Aq/glycopeptide complex on the cell surface
by a T-cell hybridoma results in secretion of interleukin-2 (IL-
2) into the medium, which was quantified in an ELISA.
As expected, all T-cell hybridomas elicited strong responses
when stimulated with the native glycopeptide 30 in complex
with the Aq molecule (Table 1). Presentation of the keto-
methylene-modified glycopeptide 31 by Aq molecules led to
IL-2 secretion by eight out of nine T-cell hybridomas, represent-
ing four out of five groups but, compared to the native Aq/30
complex, the responses were weaker. Interestingly, the Aq/31
complex failed to stimulate HCQ.10, indicating that the interac-
tions between its TCR and Aq/31 have been altered as compared
to Aq/30. The inability of 31 to activate HCQ.10 could indicate
that its TCR depends on interactions with the proton in the
Ala261-Gly262 amide bond of 30. Although this proton is
pointing up toward the TCR in the modeled Aq/CII260-267
complex, it is still positioned deep down in the peptide-binding
groove, making it somewhat uncertain if direct contact is
possible. The reduced stimulation may also be explained by a
subtle change of the conformation of the Aq-bound glycopeptide
induced by the isostere that, as a consequence, results in
presentation of a slightly different epitope, affecting other
interaction points with the TCR of HCQ.10. The T-cell
responses of the different hybridomas displayed the same trends
for the truncated ketomethylene-modified glycopeptide 28 when
bound to Aq as for 31, but the responses were even weaker,
most likely due to the lower affinity of 28 for the Aq molecule.
Although the (E)-alkene in 33 was well-tolerated in binding
to Aq, it was found to have a drastic effect on the interactions
with the different T-cell hybridomas. Only the three hybridomas
belonging to group 3 secreted low amounts of IL-2 upon
stimulation at the highest concentration of 33. The remaining
five hybridomas, which recognized both 31 and 28 when bound
to Aq, displayed a complete loss of T-cell response when instead
presented to the Aq/33 complex. Isostere 33 is unable to retain
the two hydrogen bonds that, according to the Aq/CII260-267
model, were formed between residues in Aq and the carbonyl
group in the Ala261-Gly262 amide bond. As a consequence, we
propose that the conformation of 33 is adjusted in the Aq
peptide-binding groove to compensate for positioning of the
alkene in the polar environment that is normally occupied by
the Ala261-Gly262 amide bond, thus resulting in the presentation
of an altered T-cell epitope. A contributing factor to the lack of
signal transduction could also be the inability of the TCRs to
directly interact with an amide proton at the modified position
in 33.
A previous study has shown that Ek binding glycopeptides
with amino acid substitution and small differences in the TCR
exposed carbohydrate moiety induced drastically altered TCR
signaling.70 It was found that the levels of secreted IL-2 that
was triggered by the analogues could be linked to different
phosphorylation patterns in the signal transduction machinery
of the TCR. Other studies have correlated the ability of a class
II MHC/peptide complex to activate a TCR with the dissociation
rate of the MCH/peptide/TCR complex.71-73 A somewhat slower
dissociation rate was required for complete phosphorylation and
thereby efficient T-cell stimulation, while dissociation rates at
the faster end of the scale resulted in partial activation or even
deactivation. Studies to elucidate the phosphorylation patterns
and the rates of dissociation have not been performed for the
backbone-modified glycopeptides described in this article.
However, based on the findings from such studies, it is possible
that an adjustment of the Aq-bound conformation of (E)-alkene
33 could confer presentation of a slightly altered T-cell epitope
that in turn affect the interactions with the TCRs leading to
faster dissociation rates for the MHC/glycopeptide/TCR com-
plexes. Thus, incomplete or lack of phosphorylation then results
in partial or no activation of the TCRs, respectively, as was
seen for Aq/33 and the T-cell hybridomas included in the present
study. Methyleneamine 32, which displayed a significant
reduction in Aq binding, failed to stimulate any of the T-cell
hybridomas. As expected, the T-cell hybridomas did not show
any stimulation either on incubation with the corresponding
truncated methyleneamine analogue 29 or the (Z)-alkene
containing 34, which displayed very weak and almost no binding
to Aq, respectively.
In summary, our data show that very small structural changes
in the glycopeptide backbone can have striking effects on the
interactions with different TCRs and thereby on T-cell activa-
tion. Such sensitivity of TCRs when interacting with various
backbone-modified and amino acid-substituted peptides bound
to other murine class II MHC proteins have been encountered
in several recent studies.35-38,70,73 The different responses
displayed by the T-cell hybridomas investigated in this study
have not yet been correlated to their effect in vivo. Thus, it
would be highly interesting to evaluate the ability of ketom-
ethylene 31 and (E)-alkene 34, that both bind well to Aq, to
induce tolerance in mice with CIA in vaccination studies. These

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5635
and by calculation of RMSD values before selecting a template.
The coordinates for seven amino acids (residues from â105Arg to
â111His) in the â-chain of the template crystal structure (PDB
code: 1MUJ47) were missing. These amino acids were located far
away from the binding site and were therefore considered to be of
no relevance for the modeling. The sequences of the R- and â-chain
of Aq were separately aligned with their corresponding template
sequence using the blosum62 substitution matrix.74 Ten independent
models were constructed for each chain based on a Boltzmann-
weighted randomized modeling procedure as implemented in MOE
with no energy minimization. The coordinates for the backbone
and the side chains of conserved amino acids were directly copied
from the template crystal structure. For the amino acids in Aq that
differed from the template, 10 side chain conformations were
suggested using a rotamer library in MOE. All models of the R-
and â-chains were superposed and side chain conformations were
selected for those amino acids that differed from the template in
the binding site. This selection was guided by the indicated rotamer
preference found in all crystal structures with more than 90%
sequence identity. In those cases where the side chain conformations
could not be unambiguously determined, several rotamers were
chosen to represent the suggested conformations. The R- and
â-chains were combined to give all possible combinations of the
side chain conformations and subjected to further analysis. Reduce77
was used to add hydrogens to the molecules followed by energy
minimization using a combination of three methods; steepest
descent, conjugate gradient, and truncated Newton, as implemented
in MOE. The minimizations were terminated when the root-mean-
square gradient (RMSG) was <0.1, current geometry was used as
chiral constraint, and a quadratic force was applied on the backbone
atoms with a force constant of 100. The side chain of â86Val was
kept fixed during the energy minimization to prevent it from
readopting its initial conformation (see Results and Discussion).
After visual inspection of the minimized models, a final comparative
model of the Aq molecule was chosen based on the ability of the
amino acid side chains to participate in hydrogen bonds, hydro-
phobic and π-interactions, as well as steric factors. The numbering
of amino acids in the Aq molecule is consistent with those used for
Ab, Ak, HLA-DR, and I-E. Hydrogen bonds were defined to have
a distance of e3.0 Å between a hydrogen of a donor atom D and
an acceptor atom A, and to have a D-H‚‚‚A angle of 120-180°.
Modeling of the CII260-267 Glycopeptide. Docking: The
BABEL78 file format converter was used to represent the CII260-
267 glycopeptide as chiral SMILES,79 from which the 3D atomic
coordinates were generated using the CORINA80 software. This
conformation was used as input in a conformational search using
OMEGA81 with the parameters ewindow, rms, and maxconfs set to
14, 1.0, and 5000, respectively. The generated ensemble of ligand
conformations was docked into the binding site of the comparative
Aq model using the FRED82 program. The parameter settings tstep
1.25, rstep 1.0, clash_checking 0.5, neg_img_size large, and
chemgauss as scoring function were selected based on previous
optimization studies.83 The 20 highest-ranked conformations by the
scoring function were subjected to visual inspection and manual
ranking based on the following criteria: the side chains of Ile260
and Phe267 positioned deeply in the P1 and P4 pockets, respectively,
an extended backbone conformation, and the galactose moiety
pointing out of the binding site as supported by experimentally
determined data.15-17 The highest manually ranked conformation
was energy-minimized with the Aq molecule using MOE until the
RMSG was <0.01 and applying current geometry as chiral
constraint. Comparative Modeling: The sequence of CII260-267
(with Lys264 instead of GalHyl264) was manually aligned with the
template CLIP47 sequence so that the anchoring residues in the
respective P1 and P4 pocket matched. A total of 100 independent
comparative models were then constructed for CII260-267 using
the same parameter settings as previously described for the modeling
of the Aq molecule. Selection of side chain conformations in
CII260-267 was guided by the indicated rotamer preference of
the ligand found in the template and two other crystal structures,48,49
as well as in the conformation of CII260-267 obtained when
could be performed either with the glycopeptide alone or in
complex with soluble Aq proteins.
Conclusions
In this study, a new comparative model of the murine class
II MHC Aq molecule in complex with the CII260-267
glycopeptide has provided a detailed structural basis for
understanding antigen presentation by Aq. The CII260-267
glycopeptide, positioned deep in the binding groove of Aq, was
firmly anchored by the Ile260 and Phe263 residues in the P1 and
P4 pockets, respectively. This left the GalHyl264 side chain,
which is a critical TCR contact point, partially extending out
of the groove. On the basis of the Aq/CII260-267 model, the
Ala261-Gly262 amide bond in the glycopeptide backbone was
selected for replacement by ψ[COCH2], ψ[CH2NH2+], and ψ-
[(E)-CHdCH] isosteres. Fmoc-protected Ala-Gly isostere build-
ing blocks were successfully synthesized and introduced in two
sets of glycopeptides from rat CII based on the CII259-273
and the CII260-267 sequences.
The ability of the modified glycopeptides to bind to Aq and
to stimulate T-cell hybridomas was evaluated and then inter-
preted in view of the structural features of the Aq/glycopeptide
model. It was found that the ketomethylene isostere, being the
only isostere able to retain the two hydrogen bonds formed to
the Ala261-Gly262 amide bond, bound to Aq as efficiently as
the native glycopeptide. The (E)-alkene isostere also bound well,
while the other modifications had various degrees of detrimental
effects on binding. Evaluation of the ability of the modified
glycopeptides to stimulate a panel of T-cell hybridomas revealed
that the TCRs were very sensitive for the isosteric modifications.
While the ketomethylene isostere was able to stimulate eight
out of nine T-cell hybridomas (four out of five groups) to various
extents, the (E)-alkene gave only weak responses for three
hybridomas belonging to the same group. It was reasoned that
the introduction of an (E)-alkene isostere probably conferred
changes in the conformation of the glycopeptide when bound
by Aq, thereby causing a different epitope to be presented to
the TCRs. It is also possible, but less likely, that a direct loss
of hydrogen-bonding interactions with the TCRs at the position
of the original Ala261-Gly262 amide bond contributes to the
altered T-cell response.
In conclusion, novel backbone-modified glycopeptides have
been used to probe the interactions in ternary Aq/glycopeptide/
TCR complexes to investigate the molecular mechanisms
governing activation of autoreactive T cells in CIA. The results
obtained in this study can be further used in the design of new
glycopeptide mimetics with increased stability toward proteolytic
degradation in vivo, potentially rendering structures that can
give protective effects when used in vaccination studies in CIA.
Experimental Section
Comparative Modeling. The modeling was performed using
the Molecular Operating Environment74 (MOE) software and the
AMBER94 force field with an implicit distance dependent dielectric
solvation model. Default parameter settings were used in the
computational operations, unless otherwise stated. The amino acid
sequence of the Aq protein was retrieved from the Swiss-Prot/
TrEMBL75 database (accession numbers, R-chain: P04227,
â-chain: P06342). Amino acids DHVGFY were added at the
N-terminal of the R-chain. The BLAST76 program was used to
search the PDB46 database for crystal structures with more than
90% sequence-identity with both the R- and â-chain of the Aq
protein. The resulting complexes (PDB codes: 1MUJ,47 2IAD,48
and 1LNU)49 were superposed by backbone alignment using the
MOE74 software and their similarity was compared both visually

5636 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
docking with FRED. The ability to participate in hydrogen bonds,
hydrophobic and π-interactions, as well as steric factors with the
Aq molecule were also considered. Manual hydroxylation and
subsequent glycosylation of the Lys264 side chain, acetylation of
the N-terminal and amidation of the C-terminal, were followed by
energy minimization with the Aq molecule until the RMSG was
<0.01.
4.52 (m, 1H, NCH), 1.49 (s, 9H, C(CH3)3), 1.43 (s, 9H, C(CH3)3),
1.34 (d, J ) 7.1 Hz, 3H, CH3); 13C NMR (CDCl3) δ 198.4 (CO),
164.3 (COO), 155.1 (NCO), 135.0 (COCH), 134.5 (CHCOO), 82.1
(C(CH3)3), 80.0 (C(CH3)3), 54.3 (NCH), 28.3 (C(CH3)3), 27.9
(C(CH3)3), 17.6 (CH3).
1
(Z)-6: H NMR (CDCl3) δ 6.46 (d, J ) 12.1 Hz, 1H, COCH),
6.04 (d, J ) 12.1 Hz, 1H, CHCOO), 5.34-5.21 (m, 1H, NH), 4.50-
4.41 (m, 1H, NCH), 1.46 (s, 9H, C(CH3)3), 1.43 (s, 9H, C(CH3)3),
1.38 (d, J ) 7.2 Hz, 3H, CH3); 13C NMR (CDCl3) δ 202.1 (CO),
164.6 (COO), 155.1 (NCO), 137.4 (COCH), 128.9 (CHCOO), 82.4
(C(CH3)3), 79.7 (C(CH3)3), 55.2 (NCH), 28.3 (C(CH3)3), 27.9
(C(CH3)3), 17.3 (CH3).
General Synthetic Procedures. All reactions were carried out
under an inert atmosphere with dry solvents under anhydrous
conditions, unless otherwise stated. CH2Cl2 and MeCN were
distilled from calcium hydride, whereas THF was distilled from
potassium. DMF was distilled under reduced pressure and then dried
over 3 Å molecular sieves. MeOH was dried over 3 Å molecular
sieves. TLC analysis was performed on silica gel 60 F254 (Merck)
with detection by UV light and staining with phosphomolybdic acid
in ethanol, phosphomolybdic acid and Ce(SO4)2 in aqueous
H2SO4 or alkaline aqueous KMnO4, followed by heating. After
workup, organic phases were dried over Na2SO4 before being
concentrated under reduced pressure. Flash column chromatography
was performed on silica gel (Matrex, 60 Å, 35-70 µm, Grace
Amicon). Optical rotations were measured with a Perkin Elmer
(5S)-5-tert-Butoxycarbonylamino-4-oxo-hexanoic Acid tert-
Butyl Ester (7). The isomeric mixture of alkenes 6 (133 mg, 0.447
mmol) and Pd/C (15 mg) in EtOAc (15 mL) were stirred under an
atmospheric pressure of H2 for 16 h at rt. The catalyst was removed
by filtration through a pad of Celite followed by concentration under
reduced pressure and purification by flash chromatography (n-
heptane/EtOAc 5:1 f 4:1) to afford the saturated analogue 7 (109
mg, 81%) as a colorless oil. [R]20D +13.3 (c 2.0, CHCl3); 1H NMR
(CDCl3) δ 5.26 (d, J ) 5.8 Hz, 1H, NH), 4.33-4.23 (m, 1H, CH),
2.83-2.72 (m, 1H, COCH2), 2.70-2.62 (m, 1H, COCH2), 2.56-
2.41 (m, 2H, CH2COO), 1.39 (s, 9H, C(CH3)3), 1.38 (s, 9H,
C(CH3)3), 1.31 (d, J ) 7.1 Hz, 3H, CH3); 13C NMR (CDCl3) δ
208.1 (CO), 171.6 (COO), 155.1 (NCO), 80.6 (C(CH3)3), 79.6
(C(CH3)3), 55.0 (CH), 33.8 (COCH2), 28.9 (CH2COO), 28.2
(C(CH3)3), 27.9 (C(CH3)3), 17.7 (CH3); HRMS (FAB) calcd for
C15H28NO5, 302.1967 [M + H]+; found, 302.1980.
1
model 343 polarimeter at 20 °C. H and 13C NMR spectra for the
Ala-Gly isostere building blocks and their intermediates were
recorded at 298 K on a Bruker DRX-400 spectrometer at 400 and
100 MHz, respectively, and calibrated using the residual peak of
solvent as internal standard [CDCl3 (CHCl3 δH 7.26 ppm, CDCl3
δC 77.0 ppm) or CD3OD (CD2HOD δH 3.31 ppm, CD3OD δC 49.0
ppm)]. First-order chemical shifts and coupling constants were
obtained from one-dimensional spectra; carbon and proton reso-
nances were assigned from COSY and HETCOR experiments.
Signals that could not be assigned are not reported. Spectra for the
glycopeptides were recorded at 298 K on a Bruker Avance
spectrometer at 500 MHz in H2O/D2O (9:1) with H2O (δH 4.76) as
internal standard. COSY, TOCSY, and ROESY experiments were
used for assignment of signals and determination of chemical shifts.
Analytical reversed-phase HPLC was performed on a Beckman
System Gold HPLC equipped with either a Kromasil C8 column
(250 × 4.6 mm, 5 µm) or a Supelco Discovery Bio Wide Pore
C18 column (250 × 4.6 mm, 5 µm) using a flow-rate of 1.5 mL/
min and detection at 214 nm. Preparative reversed-phase HPLC
was performed on either a Kromasil C8 column (250 × 21.2 mm,
5 µm) or a Supelco Discovery Bio Wide Pore C18 column (250 ×
21.2 mm, 5 µm) using the same eluent as for the analytical HPLC,
a flow-rate of 11 mL/min, and detection at 214 nm. HRMS data
were recorded with fast atom bombardment (FAB+) ionization or
electro spray (ES) ionization. n-Butyllithium was titrated84 with
diphenyl acetic acid as indicator. Compound 5,50 tert-butyl gly-
oxylate,51 N-Fmoc alaninol,55,56 and TBDPSO(CH2)2CHO,62-64 were
synthesized as described in the cited literature references. (5R)-
NR-(Fluoren-9-ylmethoxycarbonyl)-Nꢀ-benzyloxycarbonyl-5-O-
(2,3,4,6-tetra-O-acetyl-â-D-galactopyranosyl)-5-hydroxy-L-lysine was
prepared according to a published procedure66 but with some
modifications (see Supporting Information). Boc-Gly-Ile-OH was
synthesized from commercially available H-Gly-Ile-OH using
standard conditions85 for Boc-protection.
(5S)-5-(9H-Fluoren-9-ylmethoxycarbonylamino)-4-oxo-hex-
anoic Acid (1). Trifluoroacetic acid (0.7 mL) was added to a stirred
solution of 7 (103 mg, 0.342 mmol) in CH2Cl2 (3.3 mL) at rt. After
90 min, the solution was concentrated under reduced pressure and
the residue was then concentrated from CHCl3 three times. The
crude trifluoroacetate salt of 8 was dissolved in H2O/acetone (1:1,
4 mL), and NaHCO3 (64 mg, 0.762 mmol) was added followed by
addition of N-(9-fluorenylmethoxycarbonyloxy)succinimide (122
mg, 0.362 mmol). After stirring for 90 min, a second portion of
NaHCO3 (32 mg, 0.381 mmol) was added followed by stirring for
another 3 h. The acetone was removed under reduced pressure,
CHCl3 was added to the aqueous phase and the solution was
acidified to pH 2 by addition of HCl (10%, aq) at 0 °C. The organic
layer was separated, and the aqueous phase was extracted with
CHCl3. The combined organic layers were washed with brine, dried,
and concentrated under reduced pressure. Purification by flash
chromatography (n-heptane/EtOAc/AcOH 3:1:1% f 1:1:1%) af-
forded 1 (110 mg, 88%) as a white amorphous solid. [R]20D -15.2
(c 1.0, MeOH); 1H NMR (CD3OD; broad, minor peaks due to the
existence of rotamers are not reported) δ 7.79 (d, J ) 7.4 Hz, 2H,
Fmoc-arom), 7.69-7.63 (m, 2H, Fmoc-arom), 7.41-7.35 (m, 2H,
Fmoc-arom), 7.34-7.28 (m, 2H, Fmoc-arom), 4.46 (dd, J ) 10.5
and 6.8 Hz, 1H, Fmoc-CH2), 4.35 (dd, J ) 10.5 and 6.8 Hz, 1H,
Fmoc-CH2), 4.24-4.14 (m, 2H, Fmoc-CH, NCH), 2.73 (t, J ) 6.1
Hz, 2H, COCH2), 2.56-2.50 (m, 2H, CH2CO2H), 1.29 (d, J ) 7.2
Hz, 3H, CH3); 13C NMR (CD3OD, rotamers) δ 210.7 (CO), 176.2
(CO2H), 158.4 (NCO), 145.3, 145.2, 142.7, 128.8, 128.2, 128.1,
126.3, 126.1, 120.9, 67.7 (Fmoc-CH2), 57.0 (NCH), 48.5 (Fmoc-
CH), 34.2 (COCH2), 28.5 (CH2CO2H), 16.6 (CH3); HRMS (FAB)
calcd for C21H22NO5, 368.1498 [M + H]+; found, 368.1492.
(5S)-4-Oxo-5-((2S)-3,3,3-trifluoro-2-methoxy-2-phenylpropio-
nylamino)-hexanoic Acid Methyl Ester (9). Compound 7 (41 mg,
0.133 mmol) in CH2Cl2 (8.5 mL) was treated with TFA (1.5 mL)
for 2 h at rt. The solution was concentrated under reduced pressure,
and the residue was then concentrated from CHCl3 four times. The
crude product was dissolved in MeOH (2.0 mL), trimethylsilyl
chloride (59 µL, 0.465 mmol) was added, and the reaction was
stirred for 3 h.86 The solution was concentrated under reduced
pressure and the crude methyl ester was dissolved in CH2Cl2 (2.0
mL). (S)-(+)-R-Methoxy-R-trifluoromethylphenylacetyl chloride
((S)-MTPA-Cl, 30 µL, 0.160 mmol) was added dropwise followed
by addition of DIPEA (35 µL, 0.199 mmol), and the reaction was
stirred for a further 80 min. The reaction was quenched by addition
(5S)-5-tert-Butoxycarbonylamino-4-oxo-hex-2-enoic Acid tert-
Butyl Ester (6). Lithium chloride (162 mg, 3.82 mmol) and
triethylamine (530 µL, 3.80 mmol) were added to a solution of
â-ketophosphonate 550 (1.12 g, 3.80 mmol) in MeCN (30 mL) at
0 °C. tert-Butyl glyoxylate51 (705 mg, 5.41 mmol) dissolved in
MeCN (34 mL) was added dropwise, and the reaction was stirred
for a further 90 min at 0 °C after completed addition. The reaction
was quenched by addition of citric acid (10%, aq), Et2O, and brine,
and the organic layer was separated. The aqueous phase was
extracted with Et2O and the combined organic layers were washed
with brine, dried, and concentrated under reduced pressure. The
residue was purified by flash chromatography (n-heptane/EtOAc
8:1 f 2:1) to afford 6 (1.11 g, 97% as an isomeric mixture of E/Z
3:1) as a slightly yellow oil.
1
(E)-6: H NMR (CDCl3) δ 7.08 (d, J ) 15.8 Hz, 1H, COCH),
6.74 (d, J ) 15.8 Hz, 1H, CHCOO), 5.34-5.21 (m, 1H, NH), 4.61-

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5637
of NH4Cl (1 M, aq), and the phases were separated. The aqueous
phase was extracted with CH2Cl2, and the combined organic layers
were dried and concentrated under reduced pressure to give Mosher
amide 9. The diastereomeric ratio could not be determined from
the crude product due to overlapping signals. The crude product
was therefore filtered through a silica column with care to minimize
any changes in the isomeric ratio. Integration of the 1H NMR spectra
displayed a diastereomeric ratio of >100:1.
matography (n-heptane/EtOAc/AcOH 3:1:2% f 2:1:2%) to give
carboxylic acid 2 (252 mg, 89%) as a white foam. [R]20 -5.1 (c
D
1
1.0, CHCl3); H NMR (CDCl3, rotamers) δ 7.75 (d, J ) 7.5 Hz,
2H, Fmoc-arom), 7.61-7.54 (m, 2H, Fmoc-arom), 7.41-7.36 (m,
2H, Fmoc-arom), 7.33-7.26 (m, 2H, Fmoc-arom), 5.55 (d, J )
6.6 Hz, 0.6H, NH), 5.11-5.02 (m, 0.4H, NH), 4.41 (d, J ) 5.1
Hz, 0.8H, Fmoc-CH2), 4.32 (d, J ) 6.8 Hz, 1.2H, Fmoc-CH2),
4.22-4.14 (m, 1H, Fmoc-CH), 4.01-3.91 (m, 2H, CH2CO2H),
3.91-3.80 (m, 0.8H, NCH), 3.70-3.58 (m, 0.7H, CHCH2), 3.58-
3.46 (m, 0.2H, NCH), 3.44-3.33 (m, 0.3H, CHCH2), 3.26-3.15
(m, 0.4H, CHCH2), 3.11-2.98 (m, 0.6H, CHCH2), 1.44 (s, 3.2H,
C(CH3)3), 1.41 (s, 5.8H, C(CH3)3), 1.21-1.14 (m, 2.5H, CH3),
1.08-0.93 (m, 0.5H, CH3); 13C NMR (CDCl3, rotamers [δ for minor
rotamer when assigned]) δ 174.2 [173.9] (CO2H), 156.5 (NCOO),
156.3 (NCOO), 155.9 (NCOO), 143.9, 143.8, 141.3, 141.2, 127.6,
127.0, 125.2, 124.9, 119.9, 119.9, 81.2 [81.5] (C(CH3)3), 66.8 [66.5]
(Fmoc-CH2), 52.8 [53.3] (CHCH2), 49.8 (CH2CO2H), 47.2
(Fmoc-CH), 46.9 [46.7] (NCH), 28.1 [28.2] (C(CH3)3), 18.6 (CH3);
HRMS (FAB) calcd for C25H31N2O6, 455.2182 [M + H]+; found,
455.2189.
(5S/R)-4-Oxo-5-((2S)-3,3,3-trifluoro-2-methoxy-2-phenylpro-
pionylamino)-hexanoic Acid Methyl Ester (10). DBU (98 µL,
0.650 mmol) was added to a solution of 7 (15 mg, 0.049 mmol) in
THF (1.0 mL) followed by heating at 80 °C for 1800 s by
microwave irradiation in a closed vessel. The solution was
concentrated under reduced pressure, and the residue was taken up
in CH2Cl2 and washed with citric acid (10% aq). The combined
aqueous phases were extracted with CH2Cl2, and the combined
organic layers were then washed with brine, dried, and concentrated
under reduced pressure to give the crude racemate rac-7, [R]20
D
0.0 (c 1.0, CHCl3). The corresponding diastereomeric mixture of
Mosher amides 10 was synthesized from rac-7 using the same
procedure as described for 9, and integration of the 1H NMR spectra
confirmed a 1:1 diastereomeric mixture.
N-Trifluoroacetyl-(2S)-alaninol (17). L-Alaninol (2.50 g, 33.3
mmol) dissolved in CH2Cl2 (200 mL) was treated with triethylamine
(16.3 mL, 117 mmol) at rt. The solution was cooled to 0 °C,
trifluoroacetic anhydride (6.5 mL, 46.6 mmol) was slowly added
over 40 min, and the reaction was stirred for 6 h after complete
addition. The solution was concentrated under reduced pressure at
low temperature and the crude product was dissolved in EtOAc
and washed with NaHCO3 (aq, satd). The aqueous phase was
extracted with EtOAc, and the combined organic layers were
washed with HCl (0.1 M, aq) and brine. The organic layer was
dried and concentrated under reduced pressure to afford alcohol
17 (4.43 g, 78%) as a white amorphous solid, which was used
without further purification. [R]20D -16.3 (c 1.0, CHCl3); 1H NMR
(CDCl3) δ 6.77 (br s, 1H, NH), 4.17-4.06 (m, 1H, CH), 3.74 (dd,
J ) 11.2 and 3.8 Hz, 1H, CH2), 3.61 (dd, J ) 11.2 and 4.9 Hz,
[(2S)-2-(9H-Fluoren-9-ylmethoxycarbonylamino)-propylamino]-
acetic Acid tert-Butyl Ester (14). Dess-Martin periodinane (9.50
mL, 4.57 mmol, 15 wt % solution in CH2Cl2) was added to a
solution of alcohol 1255 (800 mg, 2.69 mmol) in CH2Cl2/DMSO
(1:1, 24 mL). After stirring for 40 min at rt, the reaction was
quenched by addition of Na2S2O5 (5.37 g, 28.2 mmol) dissolved
in NaHCO3 (aq, satd). The organic layer was separated and washed
with NaHCO3 (aq, satd) and H2O. The combined aqueous phases
were extracted with CH2Cl2, and the combined organic layers were
washed with brine and dried. The solution was concentrated under
reduced pressure at low temperature to provide crude aldehyde 13
(643 mg) as a white amorphous solid, which was directly dissolved
in CH2Cl2 (15 mL) and treated with glycine tert-butyl ester
hydrochloride (438 mg, 2.61 mmol) at rt for 40 min. NaBH(OAc)3
(738 mg, 3.48 mmol) and MeOH (3.6 mL) were added to the
reaction. After 90 min, NaHCO3 (aq, satd) and brine were added,
the organic layer was separated, and the aqueous phase was
extracted with CH2Cl2. The combined organic layers were dried
and concentrated under reduced pressure. Purification by flash
chromatography (silica gel pretreated with AcOH, n-heptane/EtOAc/
triethylamine 2:3:0% f 1:4:2%) afforded secondary amine 14 (468
1H, CH2), 2.38 (br s, 1H, OH), 1.26 (d, J ) 6.9 Hz, 3H, CH3); 13
NMR (CDCl3) δ 157.1 (q, JC-F ) 37 Hz, NCO), 115.8 (q, JC-F
C
)
287 Hz, CF3), 65.0 (splitted, CH2), 47.8 (splitted, CH), 16.5 (splitted,
CH3); HRMS (ES) calcd for C5H9F3NO2, 172.0585 [M + H]+;
found, 172.0570.
N-((1S)-2-Iodo-1-methyl-ethyl)-trifluoroacetamide (18). Triph-
enylphosphine (7.51 g, 28.6 mmol) and imidazole (2.20 g, 32.3
mmol) were added to alcohol 17 (2.04 g, 11.9 mmol) dissolved in
toluene (140 mL). The solution was treated with I2 (5.80 g, 22.9
mmol) and was slightly heated during addition. After 75 min, Et2O
was added, and the organic phase was washed with Na2S2O3 (5%,
aq). The aqueous phase was extracted with Et2O, and the combined
organic layers were dried and concentrated under reduced pressure.
The residue was purified by flash chromatography (n-heptane/
mg, 43% from 12) as a slightly yellow oil. [R]20 +2.1 (c 1.0,
D
CHCl3); 1H NMR (CDCl3) δ 7.76 (d, J ) 7.5 Hz, 2H, Fmoc-arom),
7.64-7.59 (m, 2H, Fmoc-arom), 7.42-7.37 (m, 2H, Fmoc-arom),
7.34-7.29 (m, 2H, Fmoc-arom), 5.25 (br s, 1H, CONH), 4.39 (d,
J ) 6.9 Hz, 2H, Fmoc-CH2), 4.23 (t, J ) 6.9 Hz, 1H, Fmoc-CH),
3.86-3.74 (m, 1H, NCH), 3.38-3.22 (m, 2H, COCH2), 2.70-2.61
(m, 2H, NCH2CH), 1.96 (br s, 1H, NH), 1.48 (s, 9H, C(CH3)3),
1.22-1.14 (m, 3H, CH3); 13C NMR (CDCl3, rotamers) δ 171.7
(COO), 156.1 (NCO), 144.0, 144.0, 141.3, 127.6, 127.0, 125.1,
119.9, 81.3 (C(CH3)3), 66.5 (Fmoc-CH2), 54.2 (NCH2CH), 51.6
(COCH2), 47.3 (NCH), 46.8 (Fmoc-CH), 28.1 (C(CH3)3), 18.9
(CH3); HRMS (FAB) calcd for C24H31N2O4, 411.2284 [M + H]+;
found, 411.2275.
EtOAc 40:1 f 2:1) to afford iodide 18 (2.64 g, 79%) as a white
1
amorphous solid. [R]20 -47.2 (c 1.0, CHCl3); H NMR (CDCl3)
D
δ 6.56 (br s, 1H, NH), 3.96-3.85 (m, 1H, CH), 3.43 (dd, J ) 10.5
and 4.9 Hz, 1H, CH2), 3.29 (dd, J ) 10.5 and 4.5 Hz, 1H, CH2),
1.32 (d, J ) 6.7 Hz, 3H, CH3); 13C NMR (CDCl3) δ 156.5 (q, JC-F
) 37 Hz, NCO), 115.6 (q, JC-F ) 288 Hz, CF3), 45.8 (CH), 20.5
(CH3), 12.0 (CH2); HRMS (ES) calcd for C5H8F3INO, 281.9603
[M + H]+; found, 281.9621.
{tert-Butoxycarbonyl-[(2S)-2-(9H-fluoren-9-ylmethoxycarbo-
nylamino)-propyl]-amino}-acetic Acid (2). Secondary amine 14
(255 mg, 0.621 mmol) was dissolved in CH2Cl2 (20 mL) and treated
with trifluoroacetic acid (8.6 mL). The reaction was stirred for 14
h at rt followed by concentration under reduced pressure. The
residue was concentrated from CHCl3 three times and was then
dissolved in CH2Cl2 (12 mL). Diisopropylethylamine (238 µL, 1.37
mmol) was added followed by addition of Boc2O (186 mg, 0.853
mmol) dissolved in CH2Cl2 (2.0 mL). After 40 min, a second portion
of DIPEA (66 µL, 0.379 mmol) and Boc2O (177 mg, 0.811 mmol)
were added. After an additional 15 min, the reaction was quenched
by addition of citric acid (10%, aq), the phases were separated,
and the aqueous phase was extracted with CH2Cl2. The combined
organic layers were washed with brine, dried, and concentrated
under reduced pressure. The residue was purified by flash chro-
((2S)-2-Trifluoroacetamido-propyl)-triphenyl Phosphonium
Iodide (19). Iodide 18 (1.76 g, 6.26 mmol) and PPh3 (6.63 g, 25.3
mmol) were refluxed in toluene (22 mL) for 18 h. The solvent was
evaporated under reduced pressure and the residue was purified
by flash chromatography (CH2Cl2/MeOH 1:0 f 30:1) to give
phosphonium salt 19 (3.40 g, quantitative yield) as a white foam.
[R]20D -17.2 (c 1.0, CHCl3); 1H NMR (CD3OD) δ 7.93-7.85 (m,
8H, Ph), 7.80-7.73 (m, 6H, Ph), 4.65-4.53 (m, 1H, CH), 4.03-
3.82 (m, 2H, CH2), 1.47 (dd, J ) 6.6 and 1.9 Hz, 3H, CH3); 13C
NMR (CD3OD) δ 157.5 (q, JC-F ) 37 Hz, NCO), 136.3 (d, JC-P
) 2.9 Hz, Ph), 134.9 (d, JC-P ) 10 Hz, Ph), 131.5 (d, JC-P ) 13
Hz, Ph), 119.1 (d, JC-P ) 108 Hz, Ph), 116.6 (q, JC-F ) 288 Hz,
CF3), 42.6 (d, JC-P ) 4.4 Hz, CH), 29.2 (d, JC-P ) 51 Hz, CH2),

5638 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
23.5 (d, JC-P ) 14 Hz, CH3); HRMS (FAB) calcd for C23H22F3-
NOP, 416.1386 [M]+; found, 416.1400.
(E)-(5S)-5-(Trifluoroacetamido)-hex-3-enoic Acid (22). The
general procedure described above applied on (E)-21 (134 mg,
0.635 mmol) in acetone (13 mL) followed by flash chromatography
(n-heptane/EtOAc/AcOH 4:1:1% f 3:1:1%) afforded carboxylic
N-[(1S)-5-(tert-Butyl-diphenyl-silyloxy)-1-methyl-pent-2-enyl]-
trifluoroacetamide (20). Phosphonium salt 19 (1.46 g, 2.94 mmol)
was dried under vacuum overnight before being dissolved in THF
(25 mL) and cooled to -78 °C. Titrated84 n-BuLi (4.20 mL, 1.4 M
in hexanes, 5.88 mmol) was slowly added whereby the solution
turned bright yellow. The reaction was allowed to reach rt during
20 min and was then cooled to -78 °C again followed by addition
of3-(tert-butyl-diphenyl-silyloxy)-propionaldehyde62-64(TBDPSO(CH2)2-
CHO, 919 mg, 2.94 mmol) dissolved in THF (8 mL). The reaction
was allowed to reach 0 °C over a period of 110 min, followed by
addition of Et2O and 0.1 M HCl saturated with NH4Cl. The organic
layer was separated and the aqueous phase was extracted with Et2O.
The combined organic layers were washed with brine, dried, and
concentrated under reduced pressure. The residue was purified by
flash chromatography (n-heptane/EtOAc 10:1 f 8:1) to afford 20
(1.17 g, 89%) as an isomeric mixture of E/Z 1:1 (this reaction
typically gave mixtures of alkenes of E/Z 3:2-2:3). 1H NMR
(CDCl3, E/Z isomers) δ 7.69-7.63 (m, 8H, Ph), 7.45-7.35 (m,
12H, Ph), 6.18 (br s, 1H, NHZ), 6.12 (br s, 1H, NHE), 5.74-5.66
(m, 1H, NCHCHCHE), 5.66-5.58 (m, 1H, NCHCHCHZ), 5.48 (dd,
J ) 16.4 and 5.9 Hz, 1H, NCHCHE), 5.40-5.31 (m, 1H, NCHCHZ),
4.82-4.71 (m, 1H, NCHZ), 4.60-4.50 (m, 1H, NCHE), 3.77-3.63
(m, 4H, CH2OSi), 2.49-2.34 (m, 2H, CH2CH2OSiZ), 2.33-2.25
(m, 2H, CH2CH2OSiE), 1.30 (d, J ) 6.8 Hz, 3H, CH3 E), 1.26 (d,
J ) 6.7 Hz, 3H, CH3 Z), 1.06 (s, 18H, C(CH3)3); 13C NMR (CDCl3,
acid 22 (132 mg, 92%) as a white amorphous solid. [R]20 -72.0
D
(c 1.0, CHCl3); 1H NMR (CDCl3) δ 6.35-6.23 (m, 1H, NH), 5.82-
5.73 (m, 1H, CHCH2CO2H), 5.62 (dd, J ) 15.6 and 5.7 Hz, 1H,
NCHCH), 4.65-4.55 (m, 1H, NCH), 3.14 (d, J ) 6.8 Hz, 2H, CH2-
CO2H), 1.35 (d, J ) 6.9 Hz, 3H, CH3); 13C NMR (CDCl3) δ 176.6
(CO2H), 156.3 (JC-F ) 37 Hz, NCO), 133.6 (NCHCH), 123.4
(CHCH2CO2H), 115.8 (JC-F ) 288 Hz, CF3), 47.1 (NCH), 37.0
(CH2CO2H), 19.9 (CH3); HRMS (FAB) calcd for C8H11F3NO3,
226.0691 [M + H]+; found, 226.0689.
(Z)-(5S)-5-(Trifluoroacetamido)-hex-3-enoic Acid (24). The
general procedure described above applied on (Z)-21 (144 mg, 0.682
mmol) in acetone (14 mL) followed by flash chromatography (n-
heptane/EtOAc/AcOH 4:1:1%) afforded carboxylic acid 24 (133
mg, 87%) as a colorless oil. [R]20D +13.3 (c 1.0, CHCl3); 1H NMR
(CDCl3) δ 10.8 (br s, 1H, CO2H), 6.44 (br s, 1H, NH), 5.77-5.69
(m, 1H, CHCH2CO2H), 5.54-5.46 (m, 1H, NCHCH), 4.80-4.69
(m, 1H, NCH), 3.29 (dd, J ) 7.4 and 1.6 Hz, 2H, CH2CO2H), 1.33
(d, J ) 6.7 Hz, 3H, CH3); 13C NMR (CDCl3) δ 176.7 (CO2H),
156.4 (JC-F ) 37 Hz, NCO), 132.7 (NCHCH), 124.2 (CHCH2-
CO2H), 115.7 (JC-F ) 288 Hz, CF3), 43.7 (NCH), 32.9 (CH2CO2H),
20.6 (CH3); HRMS (FAB) calcd for C8H11F3NO3, 226.0691 [M +
H]+; found, 226.0701.
General Procedure for Hydrolysis of 22 and 24 and Subse-
quent Fmoc-Protection to Afford 3 and 26. Trifluoroacetamide
22 or 24 (1.0 equiv) was dissolved in MeOH (19 mL/mmol
trifluoroacetamide), and K2CO3 (10% aq, 7.5 mL/mmol trifluoro-
acetamide) was added followed by stirring for 19-36 h at rt. The
solvent was removed under reduced pressure, and the residue was
dissolved in Na2CO3 (10%, aq)/MeCN (1:1, 29 mL/mmol), followed
by addition of N-(9-fluorenylmethoxycarbonyloxy)succinimide
(1.05 equiv). After stirring for 8-16 h, the MeCN was evaporated
under reduced pressure, CHCl3 was added, and the aqueous phase
was acidified to pH 2 by addition of HCl (10% aq) at 0 °C. The
organic layer was separated, and the aqueous phase was extracted
with CHCl3. The combined organic layers were washed with brine,
dried, and concentrated under reduced pressure, followed by
purification by flash chromatography (n-heptane/EtOAc/AcOH 3:1:
1%).
E/Z isomers) δ 156.1 (q, JC-F ) 37 Hz, NCO), 156.0 (q, JC-F
)
37 Hz, NCO), 135.5 (Ph), 134.8 (Ph), 133.8 (Ph), 133.8 (Ph), 133.7
(Ph), 130.9 (NCHCHE), 130.5 (NCHCHCHZ), 130.3 (NCHCHZ),
129.6 (Ph), 129.6 (Ph), 129.5 (NCHCHCHE), 127.6 (Ph), 115.8
(q, JC-F ) 288 Hz, CF3), 115.8 (q, JC-F ) 288 Hz, CF3), 63.1
(CH2OSi), 63.1 (CH2OSi), 47.4 (NCHE), 43.9 (NCHZ), 35.5 (CH2-
CH2OSiE), 31.1 (CH2CH2OSiZ), 26.8 (C(CH3)3), 26.8 (C(CH3)3),
21.1 (CH3 Z), 20.1 (CH3 E), 19.2 (C(CH3)3), 19.2 (C(CH3)3).
(5S)-5-(Trifluoroacetamido)-hex-3-en-1-ol (21). The isomeric
mixture of 20 (1.10 g, 2.46 mmol, E/Z 1:1) in THF (50 mL) was
treated with tetrabutylammonium fluoride trihydrate (2.33 g, 7.37
mmol) for 6 h at rt. The solvent was evaporated under reduced
pressure and the residue was purified by flash chromatography (n-
heptane/EtOAc 4:1 f 2:1) twice to completely separate (E)-21 (148
mg, 29%) and (Z)-21 (156 mg, 30%).
1
(E)-21: A colorless oil; [R]20 -75.6 (c 1.0, CHCl3); H NMR
D
(E)-(5S)-5-(9H-Fluoren-9-ylmethoxycarbonylamino)-hex-3-
enoic Acid (3). The general procedure described above applied on
22 (122 mg, 0.543 mmol) afforded the Fmoc-protected building
(CDCl3) δ 6.90-6.80 (m, 1H, NH), 5.68-5.59 (m, 1H, CHCH2),
5.51 (dd, J ) 15.6 and 5.9 Hz, 1H, NCHCH), 4.56-4.46 (m, 1H,
NCH), 3.64 (t, J ) 6.2 Hz, 2H, CH2OH), 2.37 (br s, 1H, OH),
2.30-2.24 (m, 2H, CH2CH2OH), 1.30 (d, J ) 6.9 Hz, 3H, CH3);
13C NMR (CDCl3) δ 156.5 (q, JC-F ) 37 Hz, NCO), 131.9
(NCHCH), 128.8 (CHCH2), 115.8 (q, JC-F ) 288 Hz, CF3), 61.4
(CH2OH), 47.6 (NCH), 35.4 (CH2CH2OH), 20.0 (CH3); HRMS
(FAB) calcd for C8H13F3NO2, 212.0898 [M + H]+; found,
212.0887.
block 3 (169 mg, 89%) as a white amorphous solid. [R]20 -18.8
D
(c 1.0, MeOH); 1H NMR (CDCl3) δ 7.76 (d, J ) 7.6 Hz, 2H, Fmoc-
arom), 7.59 (d, J ) 7.4 Hz, 2H, Fmoc-arom), 7.42-7.37 (m, 2H,
Fmoc-arom), 7.34-7.29 (m, 2H, Fmoc-arom), 5.75-5.49 (m, 2H,
NCHCH, CHCH2CO2H), 4.92 (br s, 1H, NH), 4.50-4.38 (m, 2H,
Fmoc-CH2), 4.35-4.25 (m, 1H, NCH), 4.22 (t, J ) 6.5 Hz, 1H,
Fmoc-CH), 3.15-3.04 (m, 2H, CH2CO2H), 1.34-1.12 (m, 3H,
CH3); 13C NMR (CDCl3) δ 176.4 (CO2H), 155.6 (NCO), 143.9,
141.4, 135.9 (NCHCH), 127.7, 127.0, 125.0, 121.5 (CHCH2CO2H),
120.0, 66.6 (Fmoc-CH2), 48.0 (NCH), 47.3 (Fmoc-CH), 37.1 (CH2-
CO2H), 20.8 (CH3); HRMS (FAB) calcd for C21H22NO4, 352.1549
[M + H]+; found, 352.1552.
1
(Z)-21: A colorless oil; [R]20 -12.4 (c 1.0, CHCl3) H NMR
D
(CDCl3) δ 6.73 (br s, 1H, NH), 5.61-5.53 (m, 1H, CHCH2CH2-
OH), 5.45-5.39 (m, 1H, NCHCH), 4.82-4.71 (m, 1H, NCH),
3.75-3.61 (m, 2H, CH2OH), 2.55-2.45 (m, 1H, CH2CH2OH),
2.36-2.27 (m, 1H, CH2CH2OH), 2.23 (br s, 1H, OH), 1.31 (d, J
) 6.7 Hz, 3H, CH3); 13C NMR (CDCl3) δ 156.5 (q, JC-F ) 37 Hz,
NCO), 131.6 (NCHCH), 129.8 (CHCH2), 115.7 (q, JC-F ) 288
Hz, CF3), 61.5 (CH2OH), 44.0 (NCH), 31.0 (CH2CH2OH), 20.8
(CH3); HRMS (FAB) calcd for C8H13F3NO2, 212.0898 [M + H]+;
found, 212.0888.
(Z)-(5S)-5-(9H-Fluoren-9-ylmethoxycarbonylamino)-hex-3-
enoic Acid (26). The general procedure described above applied
on 24 (100 mg, 0.445 mmol) afforded the Fmoc-protected building
block 26 (134 mg, 86%) as a white amorphous solid. [R]20D +71.2
(c 1.0, MeOH); 1H NMR (CDCl3) δ 7.76 (d, J ) 7.6 Hz, 2H, Fmoc-
arom), 7.57 (d, J ) 7.6 Hz, 2H, Fmoc-arom), 7.42-7.37 (m, 2H,
Fmoc-arom), 7.31 (m, 2H, Fmoc-arom), 5.67-5.56 (m, 1H, CHCH2-
CO2H), 5.46-5.38 (m, 1H, NCHCH), 4.79 (br s, 1H, NH), 4.52-
4.37 (m, 3H, Fmoc-CH2, NCH), 4.20 (t, J ) 6.3 Hz, 1H, Fmoc-
CH), 3.45-3.31 (m, 1H, CH2CO2H), 3.27-3.14 (m, 1H, CH2CO2H),
1.30-1.16 (m, 3H, CH3); 13C NMR (CDCl3, rotamers) δ 174.6
(CO2H), 155.9 (NCO), 143.8, 143.8, 141.3, 134.9 (NCHCH), 127.7,
127.1, 124.9, 124.9, 122.4 (CHCH2CO2H), 120.0, 66.8 (Fmoc-CH2),
General Procedure for Oxidation of (E)-21 and (Z)-21 to
Afford 22 and 24. Jones’ reagent (1 M CrO3 (3.0 equiv) in 4.4 M
H2SO4) was added to alcohol (E)-21 or (Z)-21 (1.0 equiv) dissolved
in acetone (20.5 mL/mmol alcohol) at 0 °C, and the solution was
stirred for 45 min, while allowed to attain rt. i-PrOH (1 mL) was
added, and the pH was adjusted to 3 by addition of NaHCO3 (aq,
satd). The aqueous phase was extracted with Et2O, and the
combined organic layers were dried and concentrated under reduced
pressure.

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5639
Table 2. 1H NMR Chemical Shifts for CII260-267 with GalHyl264 and Ala261ψ[COCH2]Gly262 (28)a
residue
NH
HR
Hâ
1.81
Hγ
others
Ile260
8.08
8.42
4.11
4.40
2.46c
4.57
4.26
3.88c
4.31
4.26
1.45, 1.18, 0.91 (CH3)
0.86 (Hδ)b
Ala261ψ[COCH2]
Gly262
1.29
2.83 and 2.75 (COCH2)
Phe263
8.21
8.35
7.91
8.26
8.44
3.07, 3.01
2.00, 1.73
7.25 (Hδ), 7.34 (Hꢀ), 7.31 (Hú)
4.01 (Hδ), 3.17 and 2.97 (Hꢀ)d
Hyl264
1.59c
Gly265
Glu266
2.08, 1.93
2.08, 1.96
2.41c
2.34c
Gln267
7.47 and 6.82 (δNH2)e
a Measured at 500 MHz and 298 K in water containing 10% D2O with H2O (δH 4.76 ppm) as internal standard. b Chemical shift for the N-terminal acetate
group: δ 2.01. c Degeneracy has been assumed. d Chemical shifts for the galactose moiety: δ 4.42 (H1), 3.91 (H4), 3.75 (H6), 3.68 (H5), 3.63 (H3), and
3.52 (H2). e Chemical shifts for the C-terminal amide: δ 7.55, 7.08.
Table 3. 1H NMR Chemical Shifts for CII260-267 with GalHyl264 and Ala261ψ[CH2NH2+]Gly262 (29)a
residue
NH
HR
Hâ
1.87
Hγ
others
Ile260
8.13
8.12
4.07
4.20
3.86c
4.72
4.26
3.90c
4.31
4.26
1.18, 1.42, 0.90 (CH3)
0.86 (Hδ)b
3.08 and 3.00 (CH2NH2 )
+
+
Ala261ψ[CH2NH2
Gly262
]
1.22
Phe263
8.56
8.52
8.13
8.21
8.47
3.11, 2.99
1.99, 1.76
7.26 (Hδ), 7.34 (Hꢀ), 7.31 (Hú)
4.01 (Hδ), 3.16 and 2.97 (Hꢀ)d
Hyl264
1.60c
Gly265
Glu266
2.07, 1.88
2.08, 1.97
2.43c
2.33c
Gln267
7.48 and 6.82 (δNH2)e
a Measured at 500 MHz and 298 K in water containing 10% D2O with H2O (δH 4.76 ppm) as internal standard. b Chemical shift for the N-terminal acetate
group: δ 2.03. c Degeneracy has been assumed. d Chemical shifts for the galactose moiety: δ 4.43 (H1), 3.91 (H4), 3.76 (H6), 3.69 (H5), 3.64 (H3), and
3.53 (H2). e Chemical shifts for the C-terminal amide: δ 7.57, 7.09.
Table 4. 1H NMR Chemical Shifts for CII259-273 with GalHyl264 and Ala261ψ[COCH2]Gly262 (31)a
residue
NH
HR
3.84b
Hâ
Hγ
others
Gly259
Ile260
8.44
8.53
4.21
1.82
1.30
1.44, 1.17, 0.92 (CH3)
0.85 (Hδ)
Ala261ψ[COCH2]
Gly262
4.42
2.84 and 2.74 (COCH2)
2.47b
4.59
Phe263
8.18
8.35
7.97
8.19
8.48
8.26
3.06, 3.00
1.99, 1.73
7.24 (Hδ), 7.33 (Hꢀ), 7.31 (Hú)
4.00 (Hδ), 3.16 and 2.98 (Hꢀ)c
Hyl264
4.26
1.59b
Gly265
3.88b
4.35
Glu266
2.11, 1.94
2.10, 1.96
2.44b
2.35b
Gln267
4.36
4.11, 3.98
4.40
7.48 and 6.81 (δNH2)
Gly268
Pro269
2.26, 1.90
1.82, 1.74
1.98b
1.44b
3.58b (Hδ)
Lys270
8.44
8.34
8.20
8.10
4.28
1.66b (Hδ), 2.97b (Hꢀ)
Gly271
3.93b
4.46
Glu272
2.14, 1.96
4.33
2.45b
1.16
Thr273
4.33
a Measured at 500 MHz and 298 K in water containing 10% D2O with H2O (δH 4.76 ppm) as internal standard. The glycopeptide existed in two forms
in solution due to cis/trans isomerization of the Gly268-Pro269 amide bond. Shifts are reported for the major trans-isomer (displayed NOEs between Gly268
HR and Pro269 Hδ,δ′), while the minor cis-isomer (<15%) could not be completely assigned due to spectral overlap. b Degeneracy has been assumed. c Chemical
shifts for the galactose moiety: δ 4.42 (H1), 3.90 (H4), 3.75 (H6), 3.67 (H5), 3.62 (H3), and 3.51 (H2).
47.3 (Fmoc-CH), 44.3 (NCH), 33.6 (CH2CO2H), 21.0 (CH3);
HRMS (FAB) calcd for C21H22NO4, 352.1549 [M + H]+; found,
352.1545.
were coupled for 26-46 h using O-(7-azabenzotriazol-1-yl)-1,1,3,3-
tetramethyluronium hexafluorophosphate (HATU, 1.6-2.2 equiv)
and 2,4,6-collidine (3.2-3.5 equiv), except in the synthesis of 28,
where building block 1 (2.0 equiv) was instead activated with HOAt
(4.7 equiv) and DIC (2.1 equiv). Fmoc-deprotection after each
coupling cycle was accomplished by treatment with 20% piperidine
in DMF for 10 min. Cleavage of the glycopeptides from the solid
phase by treatment with trifluoroacetic acid/H2O/thioanisole/
ethanedithiol (35:2:2:1) at 40 °C for 3 h and subsequent workup
was performed essentially as described elsewhere.19 This was
followed by purification by reversed-phase HPLC using a 0-100%
linear gradient of MeCN (0.1% TFA) in H2O (0.1% TFA) over 60
min, except in the synthesis of 29, where the crude acetylated
glycopeptide was directly deacetylated. Deacetylation was ac-
complished by treatment with NaOMe in MeOH (20 mM, 1 mL/
mg peptide) for 2-3 h. Neutralization by addition of AcOH and
concentration under reduced pressure was followed by purification
using reversed-phase HPLC and the same eluent as stated previ-
ously.
General Procedure for Solid-Phase Glycopeptide Synthesis.
All glycopeptides were synthesized in mechanically agitated reactors
using standard solid-phase methodology according to the Fmoc-
protocol essentially as described elsewhere.19 Glycopeptides 31, 32,
33, and 34 were synthesized on Tentagel-S-PHB-Thr-Fmoc resins,
whereas 28 and 29 were synthesized as C-terminal amides on
Tentagel-S-NH2 resins by first coupling the linker Fmoc-2,4-
dimethoxy-4′-(carboxymethyloxy)-benzhydrylamine (Rink). All
couplings were performed in DMF, unless otherwise stated, and
monitored either by using bromophenol blue as indicator87 or by
performing a Kaiser test.88 NR-Fmoc amino acids (4 equiv) with
standard side chain protecting groups and the Rink linker (4 equiv)
were activated with 1-hydroxy-benzotriazole (HOBt, 6 equiv) and
diisopropyl carbodiimide (DIC, 3.9 equiv). (5R)-NR-(Fluoren-9-
ylmethoxycarbonyl)-Nꢀ-benzyloxycarbonyl-5-O-(2,3,4,6-tetra-O-
acetyl-â-D-galactopyranosyl)-5-hydroxy-L-lysine66 (1.5-1.6 equiv)
was coupled for 24-37 h using 1-hydroxy-7-azabenzotriazole
(HOAt, 3.0-3.4 equiv) and DIC (1.5-1.6 equiv) as coupling
reagents. Isostere building blocks 1, 2, 3, and 26 (1.5-1.8 equiv)
Ac-L-isoleucyl-L-alanylψ[COCH2]glycyl-L-phenylalanyl-5-O-
(â-D-galactopyranosyl)-5-hydroxy-L-lysylglycyl-L-glutam-1-yl-L-
glutamine (28). Synthesis was performed with building block 1

5640 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
Table 5. 1H NMR Chemical Shifts for CII259-273 with GalHyl264 and Ala261ψ[CH2NH2+]Gly262 (32)a
residue
NH
HR
3.87b
Hâ
Hγ
others
Gly259
Ile260
8.50
8.27
4.16
4.18
3.91, 3.84
4.72
4.28
3.96, 3.87
4.33
4.37
4.11, 3.98
4.39
1.87
1.23
1.41, 1.16, 0.90 (CH3)
0.85 (Hδ)
+
+
Ala261ψ[CH2NH2
Gly262
]
3.10 and 2.98 (CH2NH2 )
Phe263
8.56
8.52
8.17
8.18
8.50
8.27
3.13, 3.00
2.00, 1.80
7.24 (Hδ), 7.33 (ꢀH), 7.31 (Hú)
4.03 (Hδ), 3.12 and 2.99 (Hꢀ)c
Hyl264
1.61b
Gly265
Glu266
2.06, 1.87
2.12, 1.88
2.40b
2.36b
Gln267
7.49 and 6.82 (δNH2)
Gly268
Pro269
2.26, 1.90
1.84, 1.76
1.98b
1.45b
3.58b
Lys270
8.44
8.34
8.19
8.04
4.29
1.67b (Hδ), 2.99b (Hꢀ)
Gly271
3.94b
4.46
Glu272
2.16, 1.97
4.30
2.45b
1.16
Thr273
4.30
a Measured at 500 MHz and 298 K in water containing 10% D2O with H2O (δH 4.76 ppm) as internal standard. The glycopeptide existed in two forms
in solution due to cis/trans isomerization of the Gly268-Pro269 amide bond. Shifts are reported for the major trans-isomer (displayed NOEs between Gly268
HR and Pro269 Hδ,δ′), while the minor cis-isomer (<15%) could not be completely assigned due to spectral overlap. b Degeneracy has been assumed. c Chemical
shifts for the galactose moiety: δ 4.42 (H1), 3.91 (H4), 3.76 (H6), 3.68 (H5), 3.63 (H3), and 3.52 (H2).
Table 6. 1H NMR Chemical Shifts for CII259-273 with GalHyl264 and Ala261ψ[(E)-CHdCH]Gly262 (33)a
residue
NH
HR
3.84b
Hâ
Hγ
others
Gly259
Ile260
8.39
8.24
4.11
1.78
1.19
1.43, 1.14, 0.87 (CH3)
0.82 (Hδ)
Ala261ψ[(E)-CHdCH]
Gly262
4.36
5.51b (CHdCH)
2.95b
4.59
Phe263
8.07
8.42
7.99
8.21
8.48
8.26
3.08, 2.98
2.01, 1.75
7.22 (Hδ), 7.33 (Hꢀ), 7.29 (Hú)
4.02 (Hδ), 3.07 and 2.97 (Hꢀ)c
Hyl264
4.27
3.92, 3.86
4.36
4.37
4.12, 3.98
4.39
1.61b
Gly265
Glu266
2.09, 1.93
2.11, 1.97
2.44b
2.35b
Gln267
7.48 and 6.81 (δNH2)
Gly268
Pro269
2.26, 1.90
1.84, 1.76
1.98b
1.45b
3.57b (Hδ)
Lys270
8.44
8.34
8.19
8.08
4.37
1.66b (Hδ), 2.99b (Hꢀ)
Gly271
3.93b
4.46
Glu272
2.15, 1.97
4.32
2.46b
1.17
Thr273
4.32
a Measured at 500 MHz and 298 K in water containing 10% D2O with H2O (δH 4.76 ppm) as internal standard. The glycopeptide existed in two forms
in solution due to cis/trans isomerization of the Gly268-Pro269 amide bond. Shifts are reported for the major trans-isomer (displayed NOEs between Gly268
HR and Pro269 Hδ,δ′), while the minor cis-isomer (<7%) could not be completely assigned due to spectral overlap. b Degeneracy has been assumed. c Chemical
shifts for the galactose moiety: δ 4.42 (H1), 3.90 (H4), 3.75 (H6), 3.68 (H5), 3.63 (H3), and 3.51 (H2).
Table 7. 1H NMR Chemical Shifts for CII259-273 with GalHyl264 and Ala261ψ[(Z)-CHdCH]Gly262 (34)a
residue
NH
HR
3.83b
Hâ
Hγ
others
Gly259
Ile260
8.41
8.30
4.07
1.76
1.16
1.43, 1.13, 0.83 (CH3)
0.82 (Hδ)
Ala261ψ[(Z)-CHdCH]
Gly262
4.58
5.47 and 5.44 (CHdCH)
3.11b
4.58
Phe263
8.11
8.41
7.91
8.21
8.47
8.26
3.08, 3.01
2.00, 1.74
7.23 (Hδ), 7.32 (Hꢀ), 7.30 (Hú)
4.04 (Hδ), 3.06 and 2.98 (Hꢀ)c
Hyl264
4.28
3.90, 3.86
4.37
4.37
4.13, 3.98
4.39
1.60b
Gly265
Glu266
2.10, 1.94
2.11, 1.97
2.45b
2.36b
Gln267
7.48 and 6.81 (δNH2)
Gly268
Pro269
2.26, 1.90
1.83, 1.76
1.98b
1.44b
3.58b (Hδ)
Lys270
8.44
8.34
8.18
8.16
4.29
1.67b (Hδ), 2.98b (Hꢀ)
Gly271
3.93b
4.47
Glu272
2.15, 1.98
4.34
2.48b
1.17
Thr273
4.38
a Measured at 500 MHz and 298 K in water containing 10% D2O with H2O (δH 4.76 ppm) as internal standard. The glycopeptide existed in two forms
in solution due to cis/trans isomerization of the Gly268-Pro269 amide bond. Shifts are reported for the major trans-isomer (displayed NOEs between Gly268
HR and Pro269 Hδ,δ′), while the minor cis-isomer (<11%) could not be completely assigned due to spectral overlap. b Degeneracy has been assumed. c Chemical
shifts for the galactose moiety: δ 4.42 (H1), 3.91 (H4), 3.75 (H6), 3.68 (H5), 3.63 (H3), and 3.51 (H2).
attached to the solid phase (48 µmol) according to the general
procedure described above, with the modification of coupling Ac-
Ile-OH (4.0 equiv) at the N-terminal using HOBt (6 equiv) and
DIC (3.9 equiv) as coupling reagents. Deacetylation was performed
in several small scales (1.5-7 mg) by treatment with NaOMe in
MeOH (5 mM, 1 mL/mg peptide) and the reaction time was limited
to 10 min. The final purification using reversed-phase HPLC was
performed twice, the second time using a 0-100% linear gradient
of MeOH (0.1% TFA) in H2O (0.1% TFA) over 40 min to afford
28 (3.8 mg, 7% yield based on the amount of resin used) as a white
amorphous solid after lyophilization. MS (MALDI-TOF) calcd for
1
[M + Na]+, 1089.5; found, 1089.4. H NMR data are given in
Table 2.
Ac-L-isoleucyl-L-alanylψ[CH2NH2+]glycyl-L-phenylalanyl-5-
O-(â-D-galactopyranosyl)-5-hydroxy-L-lysylglycyl-L-glutam-1-yl-
L-glutamine (29). Synthesis was performed with building block 2

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5641
attached to the solid phase (48 µmol) according to the general
procedure described above, with the modification of acetylating
the N-terminal by treatment with Ac2O/DMF (2:1) for 3 h. This
afforded 29 (16 mg, 26% yield based on the amount of resin used)
as a white amorphous solid after lyophilization. MS (MALDI-TOF)
calcd 1054.5 [M + H]+, found 1054.7. 1H NMR data are given in
Table 3.
Glycyl-L-isoleucyl-L-alanylψ[COCH2]glycyl-L-phenylalanyl-
5-O-(â-D-galactopyranosyl)-5-hydroxy-L-lysylglycyl-L-glutam-1-
yl-L-glutaminylglycyl-L-prolyl-L-lysylglycyl-L-glutam-1-yl-L-thre-
onine (31). Synthesis was performed with building block 1 attached
to the solid phase (50 µmol) according to the general procedure
described above, with the modification of coupling Boc-Gly-Ile-
OH85 (2.0 equiv) at the N-terminal using HATU (2.1 equiv) and
2,4,6-collidine (4.2 equiv) in CH2Cl2. This coupling was repeated
using 1.0 equiv of Boc-Gly-Ile-OH. Deacetylation was
performed in several small scales (0.5-8 mg) and the reaction time
was limited to 10 min. The final purification using reversed-phase
HPLC was performed twice, the second time using a 0-100%
linear gradient of MeOH (0.1% TFA) in H2O (0.1% TFA) over
40 min to afford 31 (7 mg, 7% yield based on the amount of
resin used) as a white amorphous solid. MS (MALDI-TOF) calcd
1652.8 [M + H]+, found 1652.8. 1H NMR data are given in
Table 4.
instructions. All glycopeptides were evaluated on the same plate
using biological material from the same batch. The experiments
were performed in duplicates.
Determination of T-Cell Hybridoma Responses. The response
of each glycopeptide-specific and Aq-restricted T-cell hybridoma
line, that is, the amount of IL-2 secreted following incubation of
the hybridoma with antigen-presenting spleen cells expressing Aq
and increasing concentrations of the glycopeptides, was determined
in a standard assay using ELISA. In brief, 5 × 104 T-cell
hybridomas were co-cultured with 5 × 105 syngeneic spleen cells
and increasing concentrations of the glycopeptides 27-34 (0 µM,
0.00128 µM, 0.0064 µM, 0.032 µM, 0.16 µM, 0.8 µM, 4 µM, and
20 µM) in a volume of 200 µL in 96-well flat-bottom microtiter
plates. After 24 h, 100 µL aliquots of the supernatants were removed
and frozen to kill any transferred T cells hybridoma. The contents
of IL-2 in the culture supernatant were measured by sandwich
ELISA (purified rat anti-mouse IL-2 as capturing mAb and the
biotin rat anti-mouse IL-2 as detecting mAb, both from PharMingen,
Los Angeles, CA) using the DELFIA system (Wallac, Turku,
Finland). Recombinant mouse IL-2 served as a positive control and
standard.
Acknowledgment. This work was funded by grants from
the Swedish Research Council and the Biotechnology Fund at
Umeå University. We are grateful to Drs. David Blomberg,
Tomas Gustafsson, and Lotta Holm for helpful discussions, Dr.
Mattias Hedenstro¨m for assistance with NMR spectroscopy, and
David Andersson for providing docking scripts.
Glycyl-L-isoleucyl-L-alanylψ[CH2NH2+]glycyl-L-phenylalanyl-
5-O-(â-D-galactopyranosyl)-5-hydroxy-L-lysylglycyl-L-glutam-1-
yl-L-glutaminylglycyl-L-prolyl-L-lysylglycyl-L-glutam-1-yl-L-thre-
onine (32). Synthesis was performed with building block 2 attached
to the solid phase (50 µmol) according to the general procedure
described above. This afforded 32 (19.5 mg, 19% yield based on
the amount of resin used) as a white amorphous solid after
lyophilization. MS (MALDI-TOF) calcd 1639.8 [M + H]+, found
Note Added after ASAP Publication. This manuscript was
released ASAP on October 18, 2007 with a typographical error
in the Results and Discussion. The correct version posted on
October 23, 2007.
1
1639.8. H NMR data are given in Table 5.
Glycyl-L-isoleucyl-L-alanylψ[(E)-CHdCH]glycyl-L-phenyla-
lanyl-5-O-(â-D-galactopyranosyl)-5-hydroxy-L-lysylglycyl-L-glutam-
1-yl-L-glutaminylglycyl-L-prolyl-L-lysylglycyl-L-glutam-1-yl-L-
threonine (33). Synthesis was performed with building block 3
attached to the solid phase (50 µmol) according to the general
procedure described above. This afforded 33 (19 mg, 19% yield
based on the amount of resin used) as a white amorphous solid
after lyophilization. MS (MALDI-TOF) calcd 1636.8 [M + H]+,
Supporting Information Available: Coordinates of the com-
parative model of Aq, details of the molecular modeling, the
modified experimental procedures for the galactosylated hydroxy-
lysine derivative, purity of the glycopeptides, HPLC chromatograms
for the backbone-modified glycopeptides, inhibition data for
glycopeptide binding to Aq, T-cell response data with dose-
1
response curves, and H NMR and 13C NMR spectra for all new
1
found 1636.9. H NMR data are given in Table 6.
isolated compounds. This material is available free of charge via
the Internet at http://pubs.acs.org.
Glycyl-L-isoleucyl-L-alanylψ[(Z)-CHdCH]glycyl-L-phenyla-
lanyl-5-O-(â-D-galactopyranosyl)-5-hydroxy-L-lysylglycyl-L-glutam-
1-yl-L-glutaminylglycyl-L-prolyl-L-lysylglycyl-L-glutam-1-yl-L-
threonine (34). Synthesis was performed with building block 26
attached to the solid phase (50 µmol) according to the general
procedure described above. This afforded 34 (22.6 mg, 23% yield
based on the amount of resin used) as a white amorphous solid
after lyophilization. MS (MALDI-TOF) calcd 1636.8 [M + H]+,
References
(1) Arnett, F. C.; Edworthy, S. M.; Bloch, D. A.; McShane, D. J.; Fries,
J. F.; Cooper, N. S.; Healey, L. A.; Kaplan, S. R.; Liang, M. H.;
Luthra, H. S.; Medsger, T. A. Jr.; Mitchell, D. M.; Neustadt, D. H.;
Pinals, R. S.; Schaller, J. G.; Sharp, J. T.; Wilder, R. L.; Hunder, G.
G. The American Rheumatism Association 1987 Revised Criteria
for the Classification of Rheumatoid Arthritis. Arthritis Rheum. 1988,
31, 315-324.
(2) Astorga, G. P.; Williams, Jr, R. C. Altered Reactivity in Mixed
Lymphocyte Culture of Lymphocytes from Patients with Rheumatoid
Arthritis. Arthritis Rheum. 1969, 12, 547-554.
(3) Stastny, P. Mixed Lymphocyte Cultures in Rheumatoid Arthritis. J.
Clin. InVest. 1976, 57, 1148-1157.
(4) Gregersen, P.; Silver, J.; Winchester, R. The Shared Epitope
Hypothesis. An Approach to Understanding the Molecular Genetics
of Susceptibility to Rheumatoid Arthritis. Arthritis Rheum. 1987, 30,
1205-1213.
(5) McFarland, B. J.; Beeson, C. Binding Interactions Between Peptides
and Proteins of the Class II Major Histocompatibility Complex. Med.
Res. ReV. 2002, 22, 168-203.
(6) Londei, M.; Savill, C. M.; Verhoef, A.; Brennan, F.; Leech, Z. A.;
Duance, V.; Maini, R. N.; Feldmann, M. Persistence of Collagen
Type II-Specific T-Cell Clones in the Synovial Membrane of a Patient
with Rheumatoid Arthritis. Proc. Natl. Acad. Sci. U.S.A. 1989, 86,
636-640.
(7) Kim, H. Y.; Kim, W. U.; Cho, M. L.; Lee, S. K.; Youn, J.; Kim, S.
I.; Yoo, W. H.; Park, J. H.; Min, J. K.; Lee, S. H.; Park, S. H.; Cho,
C. S. Enhanced T Cell Proliferative Response to Type II Collagen
and Synthetic Peptide CII (255-274) in Patients with Rheumatoid
Arthritis. Arthritis Rheum. 1999, 42, 2085-2093.
1
found 1636.8. H NMR data are given in Table 7.
Inhibition Assay. The ability of the backbone-modified glyco-
peptides to bind to the Aq molecule was studied in an inhibition
assay, performed essentially as described elsewhere.17,42 Briefly, a
96-well microtiter plate was first coated with the mAb Y3-P (10
µg/mL) by incubating for 2 h at room temperature. Blocking was
achieved by treatment with PBS containing 2% low fat milk for 1
h and this was followed by washing. Recombinant Aq proteins (0.1
µM) were then captured by incubating overnight at 4 °C. After
washing, 100 µL of a mixture containing a fixed concentration of
biotinylated CLIP peptide (2.5 µM) and increasing concentrations
of the (modified) glycopeptides 27-34 (0 µM, 0.8 µM, 4 µM, 20
µM, 100 µM, and 500 µM) in 100 mM phosphate buffer (pH 7.0)
was then incubated for 48 h at room temperature. This mixture
also contained a cocktail of protease inhibitors (Complete, Boe-
hringer, Mannheim). Excess of peptide was removed by washing
with PBS, and the amount of biotinylated CLIP peptide bound to
the recombinant Aq proteins was quantified using the dissociation-
enhanced lanthanide fluoroimmunoassay (DELFIA) system based
on the time-resolved fluoroimmunoassay technique with europium-
labeled streptavidin (Wallac, Turku), according to the manufacturers

5642 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23
Andersson et al.
(8) Ba¨cklund, J.; Carlsen, S.; Ho¨ger, T.; Holm, B.; Fugger, L.; Kihlberg,
J.; Burkhardt, H.; Holmdahl, R. Predominant Selection of T Cells
Specific for the Glycosylated Collagen Type II Epitope (263-270)
in Humanized Transgenic Mice and in Rheumatoid Arthritis. Proc.
Natl. Acad. Sci. U.S.A. 2002, 99, 9960-9965.
(9) Menzel, J.; Steffen, C.; Kolarz, G.; Eberl, R.; Frank, O.; Thumb, N.
Demonstration of Antibodies to Collagen and of Collagen-Anticol-
lagen Immune Complexes in Rheumatoid Arthritis Synovial Fluids.
Ann. Rheum. Dis. 1976, 35, 446-450.
(27) Dzhambazov, B.; Nandakumar, K. S.; Kihlberg, J.; Fugger, L.;
Holmdahl, R.; Vestberg, M. Therapeutic Vaccination of Active
Arthritis with a Glycosylated Collagen Type II Peptide in Complex
with MHC Class II Molecules. J. Immunol. 2006, 176, 1525-1533.
(28) Ba¨cklund, J.; Treschow, A.; Bockermann, R.; Holm, B.; Holm, L.;
Issazadeh-Navikas, S.; Kihlberg, J.; Holmdahl, R. Glycosylation of
Type II Collagen is of Major Importance for T Cell Tolerance and
Pathology in Collagen-Induced Arthritis. Eur. J. Immunol. 2002, 32,
3776-3784.
(10) Cook, A. D.; Stockman, A.; Brand, C. A.; Tait, B. D.; Mackay, I.
R.; Muirden, K. D.; Bernard, C. C. A.; Rowley, M. J. Antibodies to
Type II Collagen and HLA Disease Susceptibility Markers in
Rheumatoid Arthritis. Arthritis Rheum. 1999, 42, 2569-2576.
(11) Burkhardt, H.; Koller, T.; Engstro¨m, Å.; Nandakumar, K. S.; Turnay,
J.; Kraetsch, H. G.; Kalden, J. R.; Holmdahl, R. Epitope-Specific
Recognition of Type II Collagen by Rheumatoid Arthritis Antibodies
Is Shared With Recognition by Antibodies that are Arthritogenic in
Collagen-Induced Arthritis in the Mouse. Arthritis Rheum. 2002, 46,
2339-2348.
(29) Rosloniec, E. F.; Brandstetter, T.; Leyer, S.; Schwaiger, F. W.; Nagy,
Z. A. Second-Generation Peptidomimetic Inhibitors of Antigen
Presentation Effectively Treat Autoimmune Diseases in HLA-DR-
Transgenic Mouse Models. J. Autoimmunity 2006, 27, 182-195.
(30) Sloan-Lancaster, J.; Allen, P. M. Altered Peptide Ligand-Induced
Partial T Cell Activation. Annu. ReV. Immunol. 1996, 14, 1-27.
(31) Smilek, D. E.; Wraith, D. C.; Hodgkinson, S.; Dwivedy, S.; Steinman,
L.; McDevitt, H. O. A Single Amino Acid Change in a Myelin Basic
Protein Peptide Confers the Capacity to Prevent Rather Than Induce
Experimental Autoimmune Encephalomyelitis. Proc. Natl. Acad. Sci.
U.S.A. 1991, 88, 9633-9637.
(32) Gautam, A. M.; Lock, C. B.; Smilek, D. E.; Pearson, C. I.; Steinman,
L.; McDevitt, H. O. Minimum Structural Requirements for Peptide
Presentation by Major Histocompatibility Complex Class II Mol-
ecules: Implications in Induction of Autoimmunity. Proc. Natl. Acad.
Sci. U.S.A. 1994, 91, 767-771.
(12) Courtenay, J. S.; Dallman, M. J.; Dayan, A. D.; Martin, A.; Mosedale,
B. Immunisation Against Heterologous Type II Collagen Induces
Arthritis in Mice. Nature 1980, 283, 666-668.
(13) Wooley, P. H.; Luthra, H. S.; Stuart, J. M.; David, C. S. Type II
Collagen-Induced Arthritis in Mice. I. Major Histocompatibility
Complex (I Region) Linkage and Antibody Correlates. J. Exp. Med.
1981, 154, 688-700.
(14) Brunsberg, U.; Gustafsson, K.; Jansson, L.; Michae¨lsson, E.; A¨ hrlund-
Richter, L.; Pettersson, S.; Mattsson, R.; Holmdahl, R. Expression
of a Transgenic Class II Ab Gene Confers Susceptibility to Collagen-
Induced Arthritis. Eur. J. Immunol. 1994, 24, 1698-1702.
(15) Holm, B.; Ba¨cklund, J.; Recio, M. A. F.; Holmdahl, R.; Kihlberg, J.
Glycopeptide Specificity of Helper T Cells Obtained in Mouse
Models for Rheumatoid Arthritis. ChemBioChem 2002, 3, 1209-
1222.
(16) Rosloniec, E. F.; Whittington, K. B.; Brand, D. D.; Myers, L. K.;
Stuart, J. M. Identification of MHC Class II and TCR Binding
Residues in the Type II Collagen Immunodominant Determinant
Mediating Collagen-Induced Arthritis. Cell. Immunol. 1996, 172, 21-
28.
(17) Kjelle´n, P.; Brunsberg, U.; Broddefalk, J.; Hansen, B.; Vestberg, M.;
Ivarsson, I.; Engstro¨m, Å.; Svejgaard, A.; Kihlberg, J.; Fugger, L.;
Holmdahl, R. The Structural Basis of MHC Control of Collagen-
Induced Arthritis; Binding of the Immunodominant Type II Collagen
256-270 Glycopeptide to H-2Aq and H-2Ap Molecules. Eur. J.
Immunol. 1998, 28, 755-767.
(18) Michae¨lsson, E.; Broddefalk, J.; Engstro¨m, Å.; Kihlberg, J.; Holm-
dahl, R. Antigen Processing and Presentation of a Naturally Glyco-
sylated Protein Elicits Major Histocompatibility Complex Class II-
Restricted, Carbohydrate-Specific T Cells. Eur. J. Immunol. 1996,
26, 1906-1910.
(19) Broddefalk, J.; Ba¨cklund, J.; Almqvist, F.; Johansson, M.; Holmdahl,
R.; Kihlberg, J. T Cells Recognize a Glycopeptide Derived from Type
II Collagen in a Model for Rheumatoid Arthritis. J. Am. Chem. Soc.
1998, 120, 7676-7683.
(20) Corthay, A.; Ba¨cklund, J.; Broddefalk, J.; Michae¨lsson, E.; Gold-
schmidt, T. J.; Kihlberg, J.; Holmdahl, R. Epitope Glycosylation Plays
a Critical Role for T Cell Recognition of Type II Collagen in
Collagen-Induced Arthritis. Eur. J. Immunol. 1998, 28, 2580-2590.
(21) Holm, B.; Baquer, S. M.; Holm, L.; Holmdahl, R.; Kihlberg, J. Role
of the Galactosyl Moiety of Collagen Glycopeptides for T-Cell
Stimulation in a Model for Rheumatoid Arthritis. Bioorg. Med. Chem.
2003, 11, 3981-3987.
(22) Wellner, E.; Gustafsson, T.; Ba¨cklund, J.; Holmdahl, R.; Kihlberg,
J. Synthesis of a C-Glycoside Analogue of â-D-Galactosyl Hydrox-
ynorvaline and its Use in Immunological Studies. ChemBioChem
2000, 1, 272-280.
(33) Hart, M.; Beeson, C. Utility of Azapeptides as Major Histocompat-
ibility Complex Class II Protein Ligands for T-Cell Activation. J.
Med. Chem. 2001, 44, 3700-3709.
(34) Bolin, D. R.; Swain, A. L.; Sarabu, R.; Berthel, S. J.; Gillespie, P.;
Huby, N. J. S.; Makofske, R.; Orzechowski, L.; Perrotta, A.; Toth,
K.; Cooper, J. P.; Jiang, N.; Falcioni, F.; Campbell, R.; Cox, D.;
Gaizband, D.; Belunis, C. J.; Vidovic, D.; Ito, K.; Crowther, R.;
Kammlott, U.; Zhang, X.; Palermo, R.; Weber, D.; Guenot, J.; Nagy,
Z.; Olson, G. L. Peptide and Peptide Mimetic Inhibitors of Antigen
Presentation by HLA-DR Class II MHC Molecules. Design, Structure-
Activity Relationships, and X-ray Crystal Structures. J. Med. Chem.
2000, 43, 2135-2148.
(35) Me´zie`re, C.; Viguier, M.; Dumortier, H.; Lo-Man, R.; Leclerc, C.;
Guillet, J. G.; Briand, J. P.; Muller, S. In Vivo T Helper Cell Response
to Retro-Inverso Peptidomimetics. J. Immunol. 1997, 159, 3230-
3237.
(36) Cotton, J.; Herve´, M.; Pouvelle, S.; Maille`re, B.; Me´nez, A.
Pseudopeptide Ligands for MHC II-Restricted T Cells. Int. Immunol.
1998, 10, 159-166.
(37) Ettouati, L.; Salvi, J. P.; Trescol-Bie´mont, M. C.; Walchshofer, N.;
Gerlier, D.; Rabourdin-Combe, C.; Paris, J. Substitution of Peptide
Bond 53-54 of HEL(52-61) with an Ethylene Bond Rather Than
Reduced Peptide Bond is Tolerated by an MHC-II Restricted T Cell.
Pept. Res. 1996, 9, 248-253.
(38) Ettouati, L.; Casimir, J. R.; Raynaud, I.; Trescol-Bie´mont, M. C.;
Carrupt, P. A.; Gerlier, D.; Rabourdin-Combe, C.; Testa, B.; Paris,
J. Synthesis, Biological Activity and Molecular Modelling of the 53-
54 Ketomethylene Analogue of HEL(52-61). Protein Pept. Lett.
1998, 5, 221-229.
(39) Smith, A. B., III.; Benowitz, A. B.; Sprengeler, P. A.; Barbosa, J.;
Guzman, M. C.; Hirschmann, R.; Schweiger, E. J.; Bolin, D. R.;
Nagy, Z.; Campbell, R. M.; Cox, D. C.; Olson, G. L. Design and
Synthesis of a Competent Pyrrolinone-Peptide Hybrid Ligand for the
Class II Major Histocompatibility Complex Protein HLA-DR1. J.
Am. Chem. Soc. 1999, 121, 9286-9298.
(40) Koehler, N. K. U.; Yang, C. Y.; Varady, J.; Lu, Y.; Wu, X.; Liu,
M.; Yin, D.; Bartels, M.; Xu, B.; Roller, P. P.; Long, Y.; Li, P.;
Kattah, M.; Cohn, M. L.; Moran, K.; Tilley, E.; Richert, J. R.; Wang,
S. Structure-Based Discovery of Nonpeptidic Small Organic Com-
pounds To Block the T Cell Response to Myelin Basic Protein. J.
Med. Chem. 2004, 47, 4989-4997.
(41) Hedenstro¨m, M.; Holm, L.; Yuan, Z.; Emtena¨s, H.; Kihlberg, J.
Stereoselective Synthesis of Ψ[CH2O] Pseudodipeptides and Con-
formational Analysis of a PheΨ[CH2O]Ala Containing Analogue of
the Drug Desmopressin. Bioorg. Med. Chem. Lett. 2002, 12, 841-
844.
(23) Marin, J.; Blaton, M. A.; Briand, J. P.; Chiocchia, G.; Fournier, C.;
Guichard, G. Synthesis of Glycopeptides from Type II Collagen-
Incorporating Galactosylated Hydroxylysine Mimetics and Their Use
in Studying the Fine Specificity of Arthritogenic T Cells. ChemBio-
Chem 2005, 6, 1796-1804.
(24) Gustafsson, T.; Hedenstro¨m, M.; Kihlberg, J. Synthesis of a C-
Glycoside Analogue of â-D-Galactosyl Hydroxylysine and Incorpora-
tion in a Glycopeptide from Type II Collagen. J. Org. Chem. 2006,
71, 1911-1919.
(25) Holm, L.; Kjelle´n, P.; Holmdahl, R.; Kihlberg, J. Identification of
the Minimal Glycopeptide Core Recognized by T Cells in a Model
for Rheumatoid Arthritis. Bioorg. Med. Chem. 2005, 13, 473-482.
(26) Holm, L.; Frech, K.; Dzhambazov, B.; Holmdahl, R.; Kihlberg, J.;
Linusson, A. Quantitative Structure-Activity Relationship of Peptides
Binding to the Class II Major Histocompatibility Complex Molecule
Aq Associated with Autoimmune Arthritis. J. Med. Chem. 2007, 50,
2049-2059.
(42) Holm, L.; Bockermann, R.; Wellner, E.; Ba¨cklund, J.; Holmdahl,
R.; Kihlberg, J. Side-Chain and Backbone Amide Bond Requirements
for Glycopeptide Stimulation of T-Cells Obtained in a Mouse Model
for Rheumatoid Arthritis. Bioorg. Med. Chem. 2006, 14, 5921-5932.
(43) Pietersz, G. A.; Pouniotis, D. S.; Apostolopoulos, V. Design of
Peptide-Based Vaccines for Cancer. Curr. Med. Chem. 2006, 13,
1591-1607.
(44) Korber, B.; LaBute, M.; Yusim, K. Immunoinformatics Comes of
Age. PLoS Comput. Biol. 2006, 2, e71.
(45) Davies, M. N.; Flower, D. R. Harnessing Bioinformatics to Discover
New Vaccines. Drug DiscoVery Today 2007, 12, 389-395.

Molecular Interactions Associated with Arthritis
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 23 5643
(46) RCSB Protein Data Bank; http://www.rcsb.org/pdb/home/home.do
(accessed September 18, 2006).
(47) Zhu, Y.; Rudensky, A. Y.; Corper, A. L.; Teyton, L.; Wilson, I. A.
Crystal Structure of MHC Class II I-Ab in Complex with a Human
Clip Peptide: Prediction of an I-Ab Peptide-Binding Motif. J. Mol.
Biol. 2003, 326, 1157-1174.
(48) Scott, C. A.; Peterson, P. A.; Teyton, L.; Wilson, I. A. Crystal
Structures of Two I-Ad-Peptide Complexes Reveal That High Affinity
Can Be Achieved without Large Anchor Residues. Immunity 1998,
8, 319-329.
(49) Liu, X.; Dai, S.; Crawford, F.; Fruge´, R.; Marrack, P.; Kappler, J.
Alternate Interactions Define the Binding of Peptides to the MHC
Molecule IAb. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 8820-8825.
(50) De´ziel, R.; Plante, R.; Caron, V.; Grenier, L.; Llinas-Brunet, M.;
Duceppe, J. S.; Malenfant, E.; Moss, N. A Practical and Diastereo-
selective Synthesis of Ketomethylene Dipeptide Isosteres of the Type
AAΨ[COCH2]Asp. J. Org. Chem. 1996, 61, 2901-2903.
(51) Våbenø, J.; Brisander, M.; Lejon, T.; Luthman, K. Diastereoselective
Reduction of a Chiral N-Boc-Protected δ-Amino-R,â-unsaturated
γ-Keto Ester Phe-Gly Dipeptidomimetic. J. Org. Chem. 2002, 67,
9186-9191.
(52) Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. P.;
Masamune, S.; Roush, W. R.; Sakai, T. Horner-Wadsworth-Emmons
Reaction: Use of Lithium Chloride and an Amine for Base-Sensitive
Compounds. Tetrahedron Lett. 1984, 25, 2183-2186.
(53) Våbenø, J.; Lejon, T.; Uhd Nielsen, C.; Steffansen, B.; Chen, W.;
Ouyang, H.; Borchardt, R. T.; Luthman, K. Phe-Gly Dipeptidomi-
metics Designed for the Di-/Tripeptide Transporters PEPT1 and
PEPT2: Synthesis and Biological Investigations. J. Med. Chem. 2004,
47, 1060-1069.
(54) Dale, J. A.; Dull, D. L.; Mosher, H. S. R-Methoxy-R-trifluorometh-
ylphenylacetic Acid, a Versatile Reagent for the Determination of
Enantiomeric Composition of Alcohols and Amines. J. Org. Chem.
1969, 34, 2543-2549.
(55) Boeijen, A.; van Ameijde, J.; Liskamp, R. M. J. Solid-Phase Synthesis
of Oligourea Peptidomimetics Employing the Fmoc Protection
Strategy. J. Org. Chem. 2001, 66, 8454-8462.
(56) Kokotos, G. A Convenient One-Pot Conversion of N-Protected
Amino-Acids and Peptides into Alcohols. Synthesis 1990, 4, 299-
301.
(57) Dess, D. B.; Martin, J. C. Readily Accessible 12-I-5 Oxidant for the
Conversion of Primary and Secondary Alcohols to Aldehydes and
Ketones. J. Org. Chem. 1983, 48, 4155-4156.
(58) Lubell, W. D.; Rapoport, H. Configurational Stability of N-Protected
R-Amino Aldehydes. J. Am. Chem. Soc. 1987, 109, 236-239.
(59) Myers, A. G.; Zhong, B.; Movassaghi, M.; Kung, D. W.; Lanman,
B. A.; Kwon, S. Synthesis of Highly Epimerizable N-protected
R-Amino Aldehydes of High Enantiomeric Excess. Tetrahedron Lett.
2000, 41, 1359-1362.
(60) Abdel-Magid, A. F.; Carson, K. G.; Harris, B. D.; Maryanoff, C.
A.; Shah, R. D. Reductive Amination of Aldehydes and Ketones with
Sodium Triacetoxyborohydride. Studies on Direct and Indirect
Reductive Amination Procedures. J. Org. Chem. 1996, 61, 3849-
3862.
(61) Wiktelius, D.; Luthman, K. Taking Control of P1, P1′ and Double
Bond Stereochemistry in the Synthesis of Phe-Phe (E)-Alkene Amide
Isostere Dipeptidomimetics. Org. Biomol. Chem. 2007, 5, 603-605.
(62) McDougal, P. G.; Rico, J. G.; Oh, Y. I.; Condon, B. D. A Convenient
Procedure for the Monosilylation of Symmetric 1,n-Diols. J. Org.
Chem. 1986, 51, 3388-3390.
(68) The deacetylation of 28 and 31 was initially performed in a small
test scale and was allowed to proceed for a longer period of time (2
h) with careful monitoring by analytical reversed-phase HPLC. It
was found that the deacetylation was complete after approximately
10 min, displaying one major product peak corresponding to the
desired glycopeptide and one smaller peak with a slightly different
retention time. With increasing reaction time, the ratio between the
two peaks approached, and finally reached, 1:1. After separation,
both peaks were found to have the correct mass when analyzed with
LCMS, indicating that a mixture of diastereomers resulting from
racemization of the stereocenter in the isostere had been obtained.
(69) McFarland, B. J.; Katz, J. F.; Sant, A. J.; Beeson, C. Energetics and
Cooperativity of the Hydrogen Bonding and Anchor Interactions that
Bind Peptides to MHC Class II Protein. J. Mol. Biol. 2005, 350,
170-183.
(70) Jensen, T.; Nielsen, M.; Gad, M.; Hansen, P.; Komba, S.; Meldal,
M.; Ødum, N.; Werdelin, O. Radically Altered T Cell Receptor
Signaling in Glycopeptide-Specific T Cell Hybridoma Induced by
Antigen with Minimal Differences in the Glycan Group. Eur. J.
Immunol. 2001, 31, 3197-3206.
(71) Matsui, K.; Boniface, J. J.; Steffner, P.; Reay, P. A.; Davis, M. M.
Kinetics of T-Cell Receptor Binding to Peptide/I-Ek Complexes:
Correlation of the Dissociation Rate with T-Cell Reponsiveness. Proc.
Natl. Acad. Sci. U.S.A. 1994, 91, 12862-12866.
(72) Lyons, D. S.; Lieberman, S. A.; Hampl, J.; Boniface, J. J.; Chien,
Y.; Berg, L. J.; Davis, M. M. A TCR Binds to Antagonist Ligands
with Lower Affinities and Faster Dissociation Rates Than to Agonists.
Immunity 1996, 5, 53-61.
(73) Kersh, G. J.; Kersh, E. N.; Fremont, D. H.; Allen, P. M. High- and
Low-Potency Ligands with Similar Affinities for the TCR: The
Importance of Kinetics in TCR Signaling. Immunity 1998, 9, 817-
826.
(74) Molecular Operating EnVironment (MOE), version 2005.06; Chemi-
cal Computing Group Inc.: 1010 Sherbrooke St. West Suite, 910
Montreal, Canada H3A 2R7.
(75) Swiss-Prot/TREMBL; http://www.expasy.org/sprot/ (accessed Sep-
tember 18, 2006).
(76) Protein-protein BLAST; http://www.ncbi.nlm.nih.gov/BLAST/ (ac-
cessed September 18, 2006).
(77) Word, J. M.; Lovell, S. C.; Richardson, J. S.; Richardson, D. C.
Asparagine and Glutamine: Using Hydrogen Atom Contacts in the
Choice of Side-Chain Amide Orientation. J. Mol. Biol. 1999, 285,
1735-1747.
(78) BABEL, version 1.6; OpenEye Scientific Software Inc.: 3600
Cerrillos Road, Suite 1107, Santa Fe, NM 87507.
(79) Weininger, D. SMILES, a Chemical Language and Information-
System. 1. Introduction to Methodology and Encoding Rules. J.
Chem. Inf. Comput. Sci. 1988, 28, 31-36.
(80) CORINA; Molecular Networks GmbH, Na¨gelsbachstraâe 25, 91052
Erlangen, Germany, http://www2.chemie.uni-erlangen.de/software/
corina/free_struct.html (accessed November 1, 2006).
(81) OMEGA, version 2.0; OpenEye Scientific Software Inc.: 3600
Cerrillos Road, Suite 1107, Santa Fe, NM 87507.
(82) FRED, version 2.0.1; OpenEye Scientific Software Inc.: 3600
Cerrillos Road, Suite 1107, Santa Fe, NM 87507.
(83) Andersson, C. D.; Thysell, E.; Lindstro¨m, A.; Bylesjo¨, M.; Raubacher,
F.; Linusson, A. A Multivariate Approach to Investigate Docking
Parameters’ Effects on Docking Performance. J. Chem. Inf. Model.
2007, 47, 1673-1687.
(63) Harnden, M. R.; Jarvest, R. L. Analogues of the Antiviral Acyclo-
nucleoside 9-(4-Hydroxy-3-hydroxymethylbutyl)guanine. Part 4.
Substitution on the 2-Amino Group. J. Chem. Soc., Perkin Trans. I
1989, 12, 2207-2213.
(64) Barry, C. S.; Bushby, N.; Harding, J. R.; Willis, C. L. Stereoselective
Synthesis of the Tetrahydropyran Core of Polycarvernoside A. Org.
Lett. 2005, 7, 2683-2686.
(65) Andersson, E. C.; Hansen, B. E.; Jacobsen, H.; Madsen, L. S.;
Andersen, C. B.; Engberg, J.; Rothbard, J. B.; Sønderstrup McDevitt,
G.; Malmstro¨m, V.; Holmdahl, R.; Svejgaard, A.; Fugger, L.
Definition of MHC and T Cell Receptor Contacts in the HLA-DR4-
Restricted Immunodominant Epitope in Type II Collagen and
Characterization of Collagen-Induced Arthritis in HLA-DR4 and
Human CD4 Transgenic Mice. Proc. Natl. Acad. Sci. U.S.A. 1998,
95, 7574-7579.
(66) Syed, B. M.; Gustafsson, T.; Kihlberg, J. 9-BBN as a Convenient
Protecting Group in Functionalisation of Hydroxylysine. Tetrahedron
2004, 60, 5571-5575.
(67) Gonza´lez-Mun˜iz, R.; Garc´ıa-Lo´pez, M. T.; Go´mez-Monterrey, I.;
Herranz, R.; Jimeno, M. L.; Sua´rez-Gea, M. L.; Johansen, N. L.;
Madsen, K.; Thøgersen, H.; Suzdak, P. Ketomethylene and (Cya-
nomethylene)amino Pseudopeptide Analogues of the C-Terminal
Hexapeptide of Neurotensin. J. Med. Chem. 1995, 38, 1015-1021.
(84) Kofron, W. G.; Baclawski, L. M. A Convenient Method for
Estimation of Alkyllithium Concentrations. J. Org. Chem. 1976, 41,
1879-1880.
(85) Bodanszky, M.; Bodanszky, A. Protecting Groups. In The Practice
of Peptide Synthesis, 1st ed.; Hafner, K., Rees, C. W., Trost, B. M.,
Lehn, J. M., von Rague´ Schleyer, P., Zahradn´ık, R., Eds.; Springer-
Verlag: New York, 1984; Vol. 21, pp 20.
(86) Brook, M. A.; Chan, T. H. A Simple Procedure for the Esterification
of Carboxylic Acids. Synthesis 1983, 3, 201-203.
(87) Krchnak, V.; Vagner, J.; Lebl, M. Noninvasive Continous Monitoring
of Solid-Phase Peptide Synthesis by Acid-Base Indicator. Int. J. Pept.
Prot. Res. 1988, 32, 415-416.
(88) Kaiser, E.; Colescott, R. L.; Bossinger, C. D.; Cook, P. I. Color Test
for Detection of Free Terminal Amino Groups in the Solid-Phase
Synthesis of Peptides. Anal. Biochem. 1970, 34, 595-598.
JM0705410