Dynamics Determine Signaling in a Multicomponent System Associated with Rheumatoid Arthritis

Article abstract of DOI:10.1021/acs.jmedchem.7b01880

Strategies that target multiple components are usually required for treatment of diseases originating from complex biological systems. The multicomponent system consisting of the DR4 major histocompatibility complex type II molecule, the glycopeptide CII259-273 from type II collagen, and a T-cell receptor is associated with development of rheumatoid arthritis (RA). We introduced non-native amino acids and amide bond isosteres into CII259-273 and investigated the effect on binding to DR4 and the subsequent T-cell response. Molecular dynamics simulations revealed that complexes between DR4 and derivatives of CII259-273 were highly dynamic. Signaling in the overall multicomponent system was found to depend on formation of an appropriate number of dynamic intramolecular hydrogen bonds between DR4 and CII259-273, together with the positioning of the galactose moiety of CII259-273 in the DR4 binding groove. Interestingly, the system tolerated modifications at several positions in CII259-273, indicating opportunities to use analogues to increase our understanding of how rheumatoid arthritis develops and for evaluation as vaccines to treat RA.

Full text of DOI:10.1021/acs.jmedchem.7b01880

Subscriber access provided by Kaohsiung Medical University
Article
Dynamics determine signaling in a multicomponent
system associated with rheumatoid arthritis
Cecilia Lindgren, Mohit Tyagi, Johan Viljanen, Johannes Toms, Changrong
Ge, Naru Zhang, Rikard Holmdahl, Jan Kihlberg, and Anna Linusson
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01880 • Publication Date (Web): 04 May 2018
Downloaded from http://pubs.acs.org on May 5, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Dynamics determine signaling in a multicomponent
system associated with rheumatoid arthritis
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
‡
,†
‡,§
§
§
∥
Cecilia Lindgren, Mohit Tyagi, Johan Viljanen, Johannes Toms, Changrong Ge, Naru
≌
≌
§,*
†,*
Zhang,∥, Rikard Holmdahl,∥, Jan Kihlberg, Anna Linusson.
1
§
Department of Chemistry, Umeå University, SEꢀ901 87 Umeå, Sweden. Department of
∥
Chemistry – BMC, Uppsala University, Box 576, SEꢀ751 23 Uppsala, Sweden. Medical
Inflammation Research, Department of Medical Biochemistry and Biophysics, Karolinska
≌
Institute, SEꢀ171 77 Stockholm, Sweden. School of Pharmaceutical Science, Southern Medical
‡
University, Guangzhou, China. These authors contributed equally.
KEYWORDS. Rheumatoid arthritis, glycopeptide, DR4, Tꢀcell, multicomponent system,
MHCII, molecular dynamics, hydrogen bonding
ABSTRACT: Strategies that target multiple components are usually required for treatment of
diseases originating from complex biological systems. The multicomponent system consisting of
the DR4 major histocompatibility complex type II molecule, the glycopeptide CII259–273 from
type II collagen, and a Tꢀcell receptor is associated with development of rheumatoid arthritis
(RA). We introduced nonꢀnative amino acids and amide bond isosteres into CII259–273 and
investigated the effect on binding to DR4 and the subsequent Tꢀcell response. Molecular
dynamics simulations revealed that complexes between DR4 and derivatives of CII259–273
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
were highly dynamic. Signaling in the overall multicomponent system was found to depend on
formation of an appropriate number of dynamic intramolecular hydrogen bonds between DR4
and CII259–273, together with the positioning of the galactose moiety of CII259–273 in the DR4
binding groove. Interestingly, the system tolerated modifications at several positions in CII259–
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
73, indicating opportunities to use analogues to increase our understanding of how rheumatoid
arthritis develops and for evaluation as vaccines to treat RA.
INTRODUCTION
Modern medicinal chemistry has mainly focused on the discovery of drugs that act at one
specific target with high potency and specificity. However, this can result in a minimal impact on
the biological system and failure of the drug candidate when the system is regulated through
multiple components and pathways. Such, increasingly complex biological systems are now
often being investigated in drug discovery and pose significant challenges for medicinal
chemists. For instance, there is an increasing interest in the design of drugs that will modulate
1
,2
multiple targets of a biological system, i.e. drugs that display polypharmacology. Alternatively,
drugs targeting multicomponent systems are sought for, where the successful drug must be able
to interact with, or influence, two or more macromolecular species simultaneously to obtain the
desired effect..
One example of a multicomponent system is the complex formed between a Tꢀcell receptor
(TCR), a peptide and a major histocompatibility complex class II (MHCII) protein present on an
antigenꢀpresenting cell. This multicomponent system is essential for the function of the immune
system, and enables it to distinguish between the body’s own constituents and foreign antigens
so that an immune response is initiated towards the latter. The detailed mechanisms behind the
2
ACS Paragon Plus Environment

Page 3 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
initiation of an immune response are still not fully understood, but are known to require
additional interactions between CD3ꢀ and CD4 coꢀreceptors on the two interacting immune
3
4
5
cells. However, as for many other biological systems, dynamics have been suggested to play
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
an important role for signaling in the TCR/peptide/MHCII complex and have been studied by
6,7
molecular dynamics (MD) simulations in a few cases.
Development of autoimmune diseases are assumed to originate from a failure of the immune
system to differentiate between self and nonꢀself, resulting in that an immune response is
mounted towards the body’s own constituents. Rheumatoid arthritis (RA) is such an autoimmune
disorder that mainly affects the peripheral joints of hands and feet, leading to a chronic
inflammation that destroys bone and cartilage, involving the degradation of type II collagen
8
(
CII). The cause of the disease is not yet known, but it has been associated to the expression of
9
the MHCII allelic DRA/DRB1*0101 (DR1) and DRA/DRB1*0401 (DR4) proteins.
Interestingly, the glycopeptide CII259–273 (1, Figure 1A), which is derived from CII, is an
10
MHCIIꢀrestricted selfꢀantigen that is targeted in the immune response associated with RA.
When presented by MHCII DR1 or DR4 proteins on antigen presenting cells, glycopeptide 1 is
recognized by Tꢀcells isolated from patients that suffer from RA, and by Tꢀcell hybridomas from
10
transgenic mice that express human DR4. Moreover, studies have shown that vaccination with
glycopeptide 1 alone, or with MHCII complexes of 1 can reduce the progression of arthritis in
1
1
collagen induced arthritis (CIA), an established mouse model for RA. Vaccination with MHCII
complexes of 1 was particularly effective and reduced the incidence of arthritis in CIA from
12
>
90% to <20%. Most recently it was found that T cells specific for 1 escape tolerance
1
3
induction in the thymus, which explains why 1 is a target for the autoimmune response in CIA
1
0
and also in patients with RA.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Structural models suggest that glycopeptide 1 binds to the human MHCII proteins between the
7
c, 14
q 15
α ꢀ and β ꢀhelices,
consistent with how 1 is bound by the murine MHCII allelic protein A .
1
1
It is kept in the binding groove by a hydrogen bond network that is formed to the backbone of
2
63
266
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
the peptide, but also through anchoring of Phe in the deep hydrophobic P1 pocket and Glu
16
264
in the shallow P4 pocket (Figure 1B). The galactosylated hydroxylysine (GalHyl ) then
1
0, 17
protrudes from the MHC binding groove, forming a crucial contact point for the TCR,
which
will approach from above (Figure 1C). Thus, the MHCII/peptideꢀcomplex will interact with the
variable domain of the TCR, forming a multicomponent system that is important for
development of RA.
Figure 1. Structure of glycopeptide 1, the DR4/1 complex, and the multicomponent system
consisting of the DR4/peptide/TCR complex. A) Chemical structure of glycopeptide CII259–
4
ACS Paragon Plus Environment

Page 5 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
73, 1. B) A three dimensional structure of the multicomponent system consisting of the
1
4a
DR4/peptide/TCR complex (based on PDB:1J8H , where an influenza virus peptide is
positioned in the binding groove). The peptide is located between the α ꢀ and β ꢀhelices of DR4,
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
being presented to the TCR that is approaching from above. DR4 is displayed in grey, the
peptide in cyan, and the TCR in pink. C) A structural model of the glycopeptide 1 bound to DR4.
2
63
268
271
Phe is positioned in the P1 pocket, Gly and Gly are positioned next to the to P6 and P9
2
64
pockets, respectively. The carbohydrate moiety of GalHyl is protruding out from the binding
pocket (displayed in ball and stick). A surface has been added to the DR4 protein to display the
binding pockets. Carbon, oxygen, and nitrogen atoms in 1 are displayed in cyan, red, and blue,
respectively.
In previous studies, we have focused on the multicomponent system consisting of the murine
q
MHCII protein A , glycopeptide 1 and TCRs as this system regulates development of disease in
7
, 11a, 17
collagen induced arthritis (CIA), a validated model for RA.
Glycopeptide 1 was modified
through introduction of nonꢀnative amino acids or amide bond isosteres in different positions in
order to probe molecular interactions and to increase resistance to degradation. We found that
q
such modifications could affect the binding to A as well as recognition by TCRs, i.e. signaling
in the multicomponent system. Here, we explore how chemical variations of 1 influence the
binding to DR4 and the subsequent Tꢀcell response of the multicomponent system linked to
development of RA in humans. This information is important in future investigations of how
tolerance may be induced. Amino acids in 1 important for binding to DR4 were varied, and
amide bond isosteres of the backbone were introduced. The dynamics of the resulting complexes
were studied by MD simulations, and correlated to experimental data for binding of the different
glycopeptide variants of 1 to DR4 and to subsequent Tꢀcell responses.
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
RESULTS AND DISCUSSION
Design of Modified Glycopeptides and Peptide Isosteres
2
63
268
271
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Three amino acids in glycopeptide 1 (Phe , Gly , and Gly ) were selected for substitution
with nonꢀnative amino acids. The purpose was to investigate how the variations influence the
binding affinity of glycopeptides to DR4, and investigate how that would affect the subsequent
2
63
7c
Tꢀcell response. Phe is an important anchor residue positioned in the P1 pocket of DR4,
268
271
while Gly and Gly are positioned next to the P6 and P9 pockets, respectively (Figure 1B).
1
4
Visual inspection of structures of DR4 in complex with other peptides showed that the latter
two pockets could accommodate larger lipophilic or polar side chains, which we hypothesized
would provide stronger binding to DR4.
7
c
In a previous study, it has been shown that Phe at position 263 can be substituted for nonꢀ
native amino acids with maintained DR4 binding and subsequent Tꢀcell response. Here, we
decided to explore mꢀfluoroꢀphenylalanine (mFPhe), cyclohexylalanine (Cha), and Trp together
with the native Phe in this position. Positions 268 and 271 were investigated by incorporation of
Val or Thr and Leu or Gln, respectively, in addition to the native Gly. A Dꢀoptimal design was
applied to select a smaller subset of glycopeptides for synthesis (Gꢀefficiency of 64.6, condition
number of 2.4), resulting in a final selection of 12 modified glycopeptides (2–13, Figure 2).
267
268
270
271
Positions Gln –Gly and Lys –Gly were further investigated through the introduction of
alkane and alkene amide bond isosteres (14–16, Figure 2). The alkane isostere was chosen as it
cannot form backbone hydrogen bonds and has an increased flexibility as compared to an amide
bond. Because of these differences it is suitable to probe molecular interactions in the
DR4/glycopeptide/TCR system. A Zꢀalkene isostere became available for incorporation at
6
ACS Paragon Plus Environment

Page 7 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
267
268
position Gln –Gly as a result of the selected synthetic route, and glycopeptide 15 was
investigated as a “negative control”.
HO OH
O
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
HO
NH2
OH
Gly259-Ile-Ala-Gly-Aa1
264
Gly-Glu-Gln-Aa2-Pro-Lys-Aa3-Glu-Thr273
N
H
O
Modified Glycopeptides
Substitutions
ID
Aa1 (p263)
Aa2 (p268)
Aa3 (p271)
Gly
1
2
3
4
5
6
7
8
9
Phe
mFPhea
Chab
Trp
Gly
Val
Leu
Gln
Gly
Gly
Gly
Phe
Thr
Gly
Phe
Val
Gly
Phe
Gly
Leu
Gly
mFPhe
Gly
Cha
Val
Gly
1
0
1
2
3
mFPhea
Cha
Thr
Gln
1
1
1
Thr
Leu
Gln
Trp
Val
Trp
Thr
Leu
Amide bond isosteres
Gln267[CH CH ]Gly
268
1
4
5
2
2
268
1
Gln267[(Z)-CH =CH ]Gly
2
2
271
16
Lys270[CH CH ]Gly
2 2
Figure 2. Structure of native CII259–273 (1), modified glycopeptides 2–13 with substituted
amino acids at three positions, and isosteric glycopeptides 14–16. Apart from the amide bond
a
b
isostere, 14–16 had identical sequences as native 1. mFPhe = m-fluoroꢀphenylalanine. Cha =
cyclohexylalanine.
Synthesis
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The designed glycopeptides 2–13 (Figure 2) were synthesized by Fmocꢀbased solidꢀphase
1
8
peptide synthesis using commercially available amino acids and protected βꢀDꢀgalactosylated
1
9
hydroxylysine, prepared as reported previously. After cleavage from the solid support and
purification by reversedꢀphase HPLC the galactose moiety was deacetylated by treatment with
methanolic sodium methoxide. Final purification by reversedꢀphase HPLC then afforded 2–13 in
6–52% overall yield based on the capacity of the resin. All glycopeptides had a purity of >95%
according to analytical reversedꢀphase HPLC, and their structures were confirmed by MALDIꢀ
TOF mass spectrometry (Table 3).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Synthesis of glycopeptides 14 and 16 required access to alkane isosteres 17 as Fmoc protected
2
67
268
270
271
dipeptide building blocks, ready to be incorporated at the Gln –Gly
and Lys –Gly
positions of 1. A Wittig reaction between Gln and Lys derived aldehydes 19 and 1ꢀ(2ꢀ
2
0
triphenylphosphoniummethyl)ꢀ4ꢀmethylꢀ2,6,7ꢀtrioxaꢀbicycloꢀ[2,2,2]ꢀoxetane bromide (20)
constitutes the key step in synthesis of these isosteres (Figure 3).
2
67
268
270
271
Figure 3. Retrosynthetic analysis of Gln –Gly and Lys –Gly dipeptide isosteres 17.
8
ACS Paragon Plus Environment

Page 9 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
267
268
Preparation of the Gln –Gly isostere began by reaction of protected glutamine 21 with
N,Oꢀdimethylhydroxylamine, using HATU as promoter and TEA as a base, to give Weinreb
2
1
amide 22 in 92% yield (Scheme 1). Compound 22 was then reduced to aldehyde 23 using
2
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
DIBALꢀH at ꢀ78 °C in dry THF (90% yield). Triphenylphosphonium bromide (20) was reacted
with KHMDS in dry THF to generate the corresponding ylide, which was used in a Wittig
2
3
reaction with aldehyde 23. As aldehyde 23 and the alkene obtained in the Wittig reaction could
not be separated by flash column chromatography on silica gel, reversedꢀphase HPLC was used
for purification. Interestingly, the orthoester functionality underwent hydrolysis under the acidic
conditions of the purification resulting in isolation of alkene 24 as an ester in 62% yield and with
1
00% Zꢀalkene selectivity. Changing the base to NaHMDS, or LiHMDS did not change the
selectivity of the Wittig reaction and no traces of Eꢀisomer could be detected in the crude
products by NMR spectroscopy. As (Z)ꢀisomer 24 can be used for synthesis of the desired alkane
isostere it was subjected to hydrogenation over Pd/C in order to obtain simultaneous reduction of
the alkene and deprotection of the Cbz group. Subsequent Fmoc protection using Nꢀ(9ꢀ
fluorenylmethoxycarbonyl)oxysuccinimide (FmocꢀOSu) then gave 25 in 68% yield over two
7
a
steps. Finally, the trityl and orthoester moieties were removed by reacting 25 with acetyl
2
67
268
chloride in a mixture of tꢀBuOH and water to give the desired Gln –Gly alkane isostere 26 in
8
3% yield. Since Fmoc protected (Z)ꢀisostere 27 can be made in two steps by acid catalyzed
hydrolysis of 24, followed by Fmoc protection, it was also prepared for incorporation into
glycopeptide 1.
2
67
268
a
Scheme 1. Synthesis of Gln –Gly dipeptide isosteres 26 and 27
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
TrHN
O
TrHN
O
TrHN
O
R
c
CbzHN
O
d
FmocHN
CbzHN
O
+ 20
a
b
21 R = OH
O
O
O
2
2 R = N(OMe)Me
23 R = H
OH
OH
OH
2
5
2
4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
OH
H2N
O
e
H2N
O
f
FmocHN
FmocHN
2
7
CO2H
26
CO2H
a
ꢀ
+
Reagents and conditions: (a) TEA, HATU, Cl NH OMe, 3 h, 92%; (b) DIBALꢀH, THF, ꢀ78
C, 4 h, 90%; (c) 20, KHMDS, THF ꢀ30 °C to rt to ꢀ30 °C again (1.5 h), then 23 was added (ꢀ30
C for 1 h, then 1 h at rt.), 62%; (d) (i) Pd/C, H , MeOH, (ii) FmocꢀOSu, 10% Na CO ,
H O:acetonitrile (1:1), 68% over two steps; (e) tꢀBuOH, H O, acetyl chloride, 0 °C to rt. 24 h,
3%; (f) (i) tꢀBuOH, H O, acetyl chloride, 0 °C to rt. 24 h, (ii) FmocꢀOSu, 10% Na CO
(H O:acetonitrile 1:1), 82% over two steps.
3
°
°
2
2
3
2
2
8
2
2
3
2
2
70
271
Lys –Gly alkane isostere 34 was synthesized using a similar sequence as used for the two
2
67
268
Gln –Gly isosteres (Scheme 2). Protected lysine 28 was first converted into aldehyde 30, via
the corresponding Weinreb amide in 59% yield over two steps. A Wittig reaction of 30 with the
ylid obtained from 20, followed by purification by reversedꢀphase HPLC and concomitant acidꢀ
catalyzed rearrangement of the orthoester, gave alkene 31 in 84% yield and 100% (Z)ꢀselectivity.
Reduction of the double bond and cleavage of the Cbz group of 31 by hydrogenation over Pd/C
and subsequent Fmoc protection of the αꢀamino group gave 32 in 98% yield. Acid catalyzed
hydrolysis of the ester of 32 also resulted in cleavage of the Boc group at the αꢀamino group (33,
2
70
271
24
7
9%), which was therefore reprotected to give Lys –Gly alkane isostere 34 in 55% yield.
2
70
271
a
Scheme 2. Synthesis of Lys –Gly dipeptide isostere 34
10
ACS Paragon Plus Environment

Page 11 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
ꢀ
+
Reagents and conditions: (a) TEA, HATU, Cl NH OMe, 3 h, 81%; (b) DIBALꢀH, THF, ꢀ78 °C,
3
4
h, 73%; (c) 20, KHMDS, THF ꢀ30 °C to rt (1.5 h), then 30 was added (ꢀ30 °C for 1 h then 2 h
at rt.), 84%; (d) (i) Pd/C, H , MeOH, 76% (ii) FmocꢀOSu, 10% Na CO , H O:acetonitrile (1:1),
2
2
3
2
98%; (e) tꢀBuOH, H O, acetyl chloride, 0 °C to rt, 24 h, 79%; (f) (Boc) O, NaHCO ,
2
2
3
H O:dioxane (3:5), 55%.
2
The isosteric amino acids 26, 27 and 34 were utilized in Fmocꢀbased solidꢀphase glycopeptide
synthesis using HATU as coupling agent and DIPEA as base. After completion of the solidꢀ
phase synthesis, cleavage from the resin, purification by reversedꢀphase HPLC, deacetylation
and purification again by HPLC, glycopeptides 14, 15 and 16 were obtained in 11–23% yield
based on the capacity of the resin (Table 3). The three glycopeptides had a purity of >95%
according to analytical reversedꢀphase HPLC and their structures were confirmed by MALDIꢀ
TOF mass spectrometry (Table 3).
Binding to the DR4 Protein and Tꢀcell Receptor Recognition
Glycopeptides 1–13 and glycopeptide isosteres 14–16 were evaluated for their binding to DR4
and subsequent Tꢀcell response using a competitive ELISA and a cellꢀbased assay, respectively.
All of the sideꢀchain modified glycopeptides 2–13 bound to DR4, while none of the glycopeptide
isosteres 14–16 displayed any binding (Figure 4). Among 2–13, six glycopeptides (5, 6 and 10–
1
3) bound somewhat better to DR4, while 4 and 7 bound somewhat weaker.
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Binding of glycopeptide 1, modified glycopeptides 2–13, and glycopeptide isosteres
1
4–16, to DR4. Increasing concentrations of 1–16 and a fixed concentration of biotinylated CLIP
tracer were incubated with DR4. The amount of CLIP tracer bound to DR4 was thereafter
12
ACS Paragon Plus Environment

Page 13 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
measured in a timeꢀresolved fluoroimmunoassay. Each point represents the average of duplicates
and error bars show ± one standard deviation. A) Glycopeptide 1 compared to 2–3 and 6–9,
which are the glycopeptides that elicit Tꢀcell responses when bound to DR4. B) Glycopeptide 1
compared to 4–5 and 10–13, which did not elicit Tꢀcell responses in complexes with DR4. C)
Glycopeptide 1 compared to isosteres 14–16.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Stimulation of DR4 restricted Tꢀcell hybridoma mDR1.1 was determined through
measurements of ILꢀ2 secretion. Half of the modified glycopeptides, i.e. 2–3 and 6–9, elicited a
Tꢀcell response (Figure 4A and Figure 5A). For glycopeptides 3 and 8, the response was similar
to the native glycopeptide (1), while 6, 7, and 9 elicited a slightly lower Tꢀcell response, and
glycopeptide 2 had a significantly lower response than 1. The other glycopeptides (4–5, 10–13,
Figure 4B) did not stimulate any ILꢀ2 secretion at the concentrations investigated (Figure 5A).
Interestingly, the binding of the glycopeptides to DR4 did not correlate well with the Tꢀcell
response (cf. Figure 4A for glycopeptides that did elicit Tꢀcell responses, and Figure 4B for those
that did not). For instance, glycopeptides that bound weaker (7), somewhat better (6) or equally
well (3, 8 and 9, Figure 4A) to DR4 as 1 all elicited a similar or slightly lower Tꢀcell response as
1
(Figure 5A). In contrast, several glycopeptides that bound somewhat better than 1 to DR4, i.e.
5 and 10–13 (Figure 4B), where among those that did not elicit any Tꢀcell response at all (Figure
A). As expected, no ILꢀ2 secretion was detected at the tested concentrations for glycopeptide
5
isostere 14–16 that showed no binding to DR4 (Figure 5B).
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Recognition of glycopeptide 1, modified glycopeptides 2–13 (A), and glycopeptide
isosteres 14–16 (B), by the DR4 restricted Tꢀcell hybridoma mDR1.1, after incubation with
antigen presenting spleenocytes and increasing concentration of the glycopeptides. If the DR4ꢀ
glycopeptide complex is recognized by the Tꢀcell hybridoma, secretion of ILꢀ2 into the
supernatant will occur, which is quantified by an ELISA assay. Each point represents the average
of duplicates and error bars show ± one standard deviation.
StructureꢀActivity Relationships for DR4 Binding and TCR Recognition
In general, the P1 pocket of DR4 was able to accommodate a variation of amino acids at position
2
63 of the glycopeptides. Both the somewhat larger mFPhe and Trp, and the nonꢀaromatic Cha
7c
were tolerated by DR4. In a previous study, we found that mꢀmethylphenylalanine, pꢀ
14
ACS Paragon Plus Environment

Page 15 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
fluorophenylalanine, and tyrosine were tolerated in this position as well. The exchange of Phe for
mFPhe led to similar binding to DR4 compared to the native 1 (cf. 8 and 1), while replacement of
Phe for Trp (cf. 4 and 1) or Cha (cf. 9 and 6) resulted in a somewhat weaker binding. The Gly at
positions 268 and 271 positioned next to the pockets P6 and P9 of DR4, respectively, were
successfully exchanged for amino acids with larger sideꢀchains with similar or increased binding
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
68
268
268
to DR4. The change of Gly to Thr or Val somewhat increased the binding to DR4 (5 and
6
), while introducing Gln at position 271 appeared to result in similar binding affinity. The
271
exchange of Gly for Leu resulted in a decreased binding to DR4 (7).
The structureꢀactivity relationship (SAR) for the Tꢀcell response differed dramatically from the
SAR of DR4 binding, as mentioned above. Two of the modifications abolished the Tꢀcell
263
263
268
268
response, the exchange of Phe for Trp and Gly for Thr ; glycopeptides with these
modifications (4, 5 and 10–13) did not result in any ILꢀ2 secretion despite that their binding to
DR4 were similar to other glycopeptides that were recognized by the TCR (Figures 4B and 5A).
The other seven glycopeptides with nonꢀnative amino acids in positions 263, 268 and/or 271 did
result in Tꢀcell responses (2, 3 and 6–9, Figure 5A), although no clear SAR could be
distinguished among these.
In comparison, the glycopeptide isosteres 14–16 displayed neither binding to DR4, nor a Tꢀcell
response. The fact that they did not bind to DR4 probably originates from the increased
flexibility of 14 and 16, and/or loss of hydrogen bonding abilities. Glycopeptide 15 has a
substantially different conformational landscape compared to 1, which explains its lack of
binding to DR4.
In summary, glycopeptides 2–13 that had one, two or three nonꢀnative amino acids displayed
only minor variations in their binding to DR4 and there was no clear correlation between DR4
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
binding and their Tꢀcell responses. This enigma was further illustrated by contrasting the
glycopeptides that had Trp at position 263 to those that had Thr at position 268. The former all
bound weaker to DR4 than 1, while the latter bound somewhat stronger. However, both
categories of glycopeptides failed to elicit a Tꢀcell response.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Dynamics of Glycopeptides in Complex with DR4
As the SAR analysis failed to explain what governs the response of the multicomponent system
consisting of DR4ꢀexpressing antigenꢀpresenting cells, glycopeptide 1 and Tꢀcell hybridoma
mDR1.1, we decided to investigate the dynamics of the complex between DR4 and some of the
glycopeptides. Based on their Tꢀcell responses, four of the modified glycopeptides were selected
for MD simulations, just as the native 1. Glycopeptides 2 and 3 were included as they both
bound to DR4 and elicited a Tꢀcell response. Glycopeptide 2 displayed the lowest Tꢀcell
response of the peptides that were recognized, i.e. a response approximately two orders of
magnitude weaker than native 1, while 3 displayed a response comparable to 1. Neither of the
263
268
modified glycopeptides with the nonꢀnative amino acids Trp or Thr displayed any Tꢀcell
response, thus 4 and 5 were included as examples of glycopeptides that bound to DR4 but did not
give rise to ILꢀ2 secretion. It should be noted that 2 and 3 have nonꢀnative amino acids at all
three or two of the modified positions, respectively, while 4 and 5 only had single replacements
at positions 263 and 268.
Modified glycopeptides 2–5 were modeled in silico into DR4, based on four previously
published Xꢀray crystal structures of DR4 in complex with other peptides (see experimental
1
4a, 14c, 14d, 25
section for further details).
The resulting static models displayed very similar binding
modes as for the native 1 and could therefore not explain the differences observed in the
16
ACS Paragon Plus Environment

Page 17 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
biological assays. The dynamics of the complexes were instead studied over time using MD
simulations where the conformational space was sampled by six parallel 60 ns MD simulations
for each complex. In general, according to the rootꢀmeanꢀsquare deviation (RMSD, Figure S62),
all MD simulations were converged after 30 ns. The fast stabilizations of the complexes are
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
26
consistent with other MD studies of MHCII/peptide systems. We focused our analyses of the
MD simulations on the binding mode and dynamics of the glycopeptide in the binding groove,
and on the hydrogen bond network formed between the peptide backbone and DR4.
For the six MD simulations of the glycopeptide 1 in complex with DR4, the DR4 protein was
2
63
stable throughout the simulations (Figures S62–S66). Phe in the glycopeptide is positioned in
the P1 pocket (Figure S70), firmly anchoring the residues of the glycopeptide adjacent to
264
galactosylated hydroxylysine (positions 262–265) between the α ꢀ and β ꢀhelices, while the Cꢀ
1
1
terminus (positions 267–273) was more mobile (Figure 6A and S67–69).
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 6. Snapshots from the MD simulations displaying the dynamics of the DR4 systems. A)
Glycopeptide 1, B) glycopeptide 2, C) glycopeptide 3, D) glycopeptide 4, E) glycopeptide 5, all
in complex with DR4. For each simulation trajectory a cluster analysis of the glycopeptide was
performed based on snapshots over the last 30 ns. One representative glycopeptide conformation
for each cluster with at least 40 members is displayed. The complexes are displayed with the
proteins in gray and the glycopeptides in green. The conformation of the glycopeptide 1 from the
homology model is displayed in magenta, as a reference.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The MD simulations of DR4 in complex with modified glycopeptides 2 and 3 displayed that 2
was more firmly anchored in the binding groove than native 1, while the behavior of 3 was very
similar to that of 1 (Figure 6A–C and S67–69). For 2, this is most likely due to the larger side
chains of the nonꢀnative amino acids, positioned in the P1, P6 and P9 pockets of DR4, which led
to decreased mobility of the peptide in accordance with the design. Interestingly, the reduced
mobility of glycopeptide 2 in the DR4 binding groove also resulted in the lowest Tꢀcell response
of all glycopeptides that were recognized, while 3 that displayed a similar mobility as 1 gave an
almost identical Tꢀcell response. This suggests that rigidification of the peptide epitope prevents
it from adapting to the TCR and eliciting a response upon formation of the multicomponent
system, while modified peptide epitopes having similar flexibility as 1 elicit identical Tꢀcell
responses.
The complexes of DR4 and glycopeptides 4 and 5, respectively, had a different dynamic
2
63
263
pattern than 1–3 (Figure 6D–E). Glycopeptide 4, with Trp instead of Phe , displayed a much
higher degree of mobility in the binding groove than 1, and large variations were observed for
the multiple simulation trajectories (Figure S67–69). The high mobility did not only include the
Cꢀterminus of the peptide, but now also the Nꢀterminus, which may explain the somewhat lower
18
ACS Paragon Plus Environment

Page 19 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
binding affinity of 4. In addition, the DR4 protein also displayed an increased mobility, apparent
when analyzing the α ꢀ and β helices (Figures S64 and S65). Glycopeptide 5 showed a mobility
1
1
that was higher than 1 (Figures S67–S69) primarily in the central and Cꢀterminal part of the
2
68
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
epitope (cf. Figure 6A and Figure 6E). Interestingly, the introduction of Thr
affected the
dynamics of the α β βꢀsheet forming the bottom of the DR4 binding groove (Figure S66).
1
1
Overall, the binding strength of 5 to DR4 was not affected by the changed dynamic pattern, but it
appears to have led to a more flexible epitope being presented to the TCR in the center of the
glycopeptide, which may explain why 5 did not give a Tꢀcell response.
Hydrogen Bond Network between the Glycopeptides and DR4
Throughout the MD simulations, an extended hydrogen bond network was formed between the
residues of DR4 and the backbone of the native glycopeptide 1, keeping it anchored in the
binding groove. During the last 30 ns of the MD simulations there were on average 13.5
hydrogen bonds formed between the backbone of 1 and the residues of DR4. Out of these
hydrogen bonds, on average 7 had occupancy of at least 40% (Table 1). Notably, four of the
highest occupancy hydrogen bonds (>60%) were formed between DR4 and the backbone of
2
63
residues 262–264 of 1, which includes Phe that anchors the glycopeptide in the P1 pocket of
2
64
10, 17
DR4 and Gal–Hyl known to be essential for contacting the TCR (Table 2).
This finding
suggests that the exact positioning of these two residues is of high importance for DR4 binding
and Tꢀcell response, i.e. for signaling in the multicomponent system.
Table 1. The number of hydrogen bonds formed between residues of DR4 and the backbone of
glycopeptides 1–5, during the last 30 ns of the MD simulations
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
a
Occupancy
above 40%
Occupancy
Glycoꢀ
peptide
above 10%
Mean SD
Mean
7.0
SD
0.8
1.8
1.9
3.5
2.9
1
2
3
4
13.5
13.5
13.3
9.3
2.2
3.1
3.0
3.9
3.2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
9.3
7.7
4.8
5
6.0
11.2
a
The hydrogen bond occupancy has been calculated using the hydrogen bond extension tool in
2
7
VMD , with a donorꢀHꢀacceptor angle of minimum 140° and a donorꢀacceptor length of
maximum 3.4 Å.
Table 2. Occupancy of the four most populated hydrogen bonds formed between the backbone
of glycopeptides 1–5 and residues of DR4 over the last 30 ns of the MD simulations
262
a
263
c
264
264
Gly (a) –
Phe (d) –
αSer (m)
GalHyl (a)–
GalHyl (d)–
81
b
53
d
82
82
βHis (s)
Mean
63.0
βAsn (s)
βAsn (s)
Glycopeptide
SD
Mean
79.5
70.0
77.8
50.8
40.8
SD
Mean
88.7
95.3
64.8
61.8
58.2
SD
Mean
89.5
96.3
67.7
61.2
62.5
SD
1
2
3
4
5
31.6
7.1
18.6
32.1
20.6
41.2
35.1
7.1
10.4
2.4
74.3
3.7
65.7
24.2
36.0
37.5
35.8
44.2
41.9
35.2
43.4
44.3
37.2
34.7
The hydrogen bond occupancy has been calculated using the hydrogen bond extension tool in
VMD, with a donorꢀHꢀacceptor angle of minimum 140° and a donorꢀacceptor length of
maximum 3.4 Å. a = hydrogen bond acceptor. s = sidechain in DR4. d = hydrogen bond donor.
a
b
c
d
m = main chain in DR4.
2
0
ACS Paragon Plus Environment

Page 21 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The glycopeptides that were recognized by the Tꢀcell receptor (1–3) had a larger number of
average hydrogen bonds present over time compared to 4 and 5 that did not elicit ILꢀ2 secretion
(Tables 1 and 2). The least mobile glycopeptide in the DR4 binding groove (2) had the largest
average number of hydrogen bonds with an occupancy >40%, while 3 was more similar to 1. In
contrast, the glycopeptides that were not recognized by the TCR (4 and 5) had lower hydrogen
bond occupancies, in particular of the four important hydrogen bonds (Table 2). However, 5 had
higher hydrogen bond occupancies than 4 (Table 1), in line with the lower flexibility displayed
by 5 in the MD simulations.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
63
263
For glycopeptide 4, the exchange of Phe to Trp in the P1 pocket of DR4 led to a slight
repositioning of the backbone in the binding groove. This weakened the four important hydrogen
263
bonds (Table 2), which increased the dynamics of the complex and allowed Trp to leave the
P1 pocket occasionally (Figure S64 and S68). The modified glycopeptide 5 also had lower
2
63
occupancy of the four key hydrogen bonds to DR4, despite having the native Phe in the P1
pocket. Consequently, the reduced occupation of the key hydrogen bonds at positions 262–264
268
268
could be linked to the exchange of Gly for Thr , positioned in the P6 pocket of DR4. New
2
68
hydrogen bonds were formed between the hydroxyl group of Thr of 5 and residues in P6
6
2
66
pocket of DR4 (αAsn and/or αAsp ; Table S3), which altered the conformational landscape of
the complex and led to a changed hydrogen bonding network. The resulting altered dynamic
pattern of the complex of 5 with DR4 as compared to 1, apparently had a tremendous effect on
2
68
the Tꢀcell response. It is also possible that steric hindrance between Thr in 5 and the TCR
contribute to the lack of response to 5.
Challenging the Multicomponent System with Additional Glycopeptides
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Based on the MD simulations, we conclude that a retained biological response to 1 is highly
dependent on the dynamic pattern of the DR4/glycopeptide complexes. Appropriate dynamics
could be traced to a large number of hydrogen bonds of varying occupancy between the
glycopeptide and the binding groove of DR4, including four of high occupancy that involved
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
263
residue Phe and adjacent residues of the glycopeptides. The results indicated that a reduced
mobility of the glycopeptides, originating from an increased number of moderate (>40%)
occupancy hydrogen bonds, might lead to reduced Tꢀcell responses, just as an increased
mobility. In order to develop our model, and to further investigate the system, we designed three
additional glycopeptides (35, 36, and 37) that were modified at the same positions as 2–13 but
had no overlap with the previously selected nonꢀnative amino acids. All three positions were
modified in all three glycopeptides and the selection of amino acids was on purpose made to
challenge the limits of our understanding of this multicomponent system.
2
63
From the results obtained with 1–13, it was clear that Phe in the glycopeptides could be
2
63
replaced by larger amino acids with maintained binding to DR4. However, the indole of Trp
disrupted the dynamic pattern of the complex, which led to loss of the Tꢀcell response. Phe
263
was now replaced by mꢀbromoꢀphenylalanine or benzothiophenealanine (Figure 7) to explore
larger lipophilic amino acids, with the latter being equal in size to Trp. At position 268, a
replacement of Gly for Val was tolerated with maintained DR4 binding and Tꢀcell response; in
2
68
the new set Gly
was replaced by cyclopropylalanine in order to increase both size and
2
71
271
lipophilicity. The third position, Gly , tolerated both larger lipophilic (Leu ) and larger polar
271
(
Gln ) amino acids. To investigate the limits of this position further we incorporated pyridine
or His, both of which provide increased nonꢀpolar and polar surfaces for interactions with DR4
22
ACS Paragon Plus Environment

Page 23 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and the TCR. This resulted in glycopeptides 35–37 that contain significantly different side chains
at all three positions compared to native 1 and the modified glycopeptides 2–13.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Structure of native CII259–273 (1) and glycopeptides 35–37 substituted with nonꢀ
native amino acids at three positions.
The dynamics of the three complexes of DR4 and 35–37, respectively, were investigated using
MD simulations in an attempt to predict their DR4 binding and subsequent Tꢀcell response
before synthesis and biological evaluation. During the MD simulations of the DR4/35 complex,
263
268
both benzothiophenealanine
and cyklopropylalanine
remained in their respective DR4
occasionally dissociated from P9.
2
71
pockets throughout the MD simulations, while pyridine
Thus, the glycopeptide displayed a low mobility in the binding groove, similar to 2 in the DR4/2
2
63
complex (Figures S67–S69). The replacement of benzothiophenealanine
with mꢀbromoꢀ
2
63
phenylalanine to give glycopeptide 36 increased the dynamics of the glycopeptide in the
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
263
complex with DR4. The positioning of mꢀbromoꢀphenylalanine in P1 resulted in a decrease in
2
64
82
the hydrogen bond occupancy between GalHyl (a) of the glycopeptide and βAsn (s) of DR4,
2
68
and that the cyclopropylalanine of 36 dissociated occasionally from the P6 pocket (Figure 8
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
and Table S2).
Finally, during the simulations of the DR4/37 complex, the
263
benzothiophenealanine
was firmly anchored in the P1 pocket, while both
2
68
271
cyklopropylalanine and His left their respective pockets during the simulations, resulting in
a highly mobile Cꢀterminus and a somewhat larger overall mobility than displayed by 1 (Figure
8
, S67 and S68). In addition, the β ꢀhelix of DR4 occasionally unfolded around the Cꢀterminus
1
of 37 during the simulations (Figure 8 and S65).
The hydrogen bonding networks formed between DR4 and the backbone of glycopeptides 35–
3
7, respectively, differed somewhat from the DR4/1 complex. Although the average number of
hydrogen bonds between DR4 and 35–37, respectively, having occupancy higher than 40%, were
similar to DR4/1, a higher variability was observed between the different simulated trajectories
compared to native system. Furthermore, the total number of hydrogen bonds above 10%
occupancy was somewhat lower compared to the DR4/1 complex (Table S1). This reflects a
slight overall alteration in the dynamic patterns of the complexes formed by DR4 and 35–37,
respectively, compared to the native system. Still, the four essential hydrogen bonds were formed
in high occupancy (>60%) for all new complexes, except for 36 as mentioned above (Table S2).
Based on the MD simulations, we concluded that glycopeptide 35–37 should all bind to DR4,
and that 36 and 37 are likely to elicit a Tꢀcell response. The greater rigidity of the DR4/35
complex, which resembles that of the DR4/2 complex, should result in a low Tꢀcell response.
24
ACS Paragon Plus Environment

Page 25 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Snapshots from the MD simulations displaying the dynamics of the DR4 systems. A)
Glycopeptide 1, B) glycopeptide 35, C) glycopeptide 36 and D) glycopeptide 37. For each MD
simulation a cluster analysis of the glycopeptides has been performed based on snapshots over
the last 30 ns. One representative glycopeptide conformation for each cluster with at least 40
members is displayed. The tertiary structure is displayed with the DR4 protein in gray and the
glycopeptide in yellow. The conformation of the glycopeptide 1 from the prepared homology
model is displayed in magenta as a reference.
Determination of DR4 binding and TCR recognition revealed that 35–37 all bound to DR4
with similar strength as 1 (Figure 9A), and that both 36 and 37 gave a Tꢀcell response, although
decreased compared to 1, while the rigid DR4/35 complex did not elicit a significant response at
the tested concentrations (Figure 9). These data agree well with our predictions based on the MD
simulations of complexes between DR4 and 35–37. The elicited Tꢀcell response suggests that
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
substantial chemical modifications can be made to the glycopeptides, while still being
recognized by the TCRs. However, the lower Tꢀcell response towards DR4 in complex with 36
and 37, respectively, compared to DR4/1, indicates that the outer boarder of chemical alterations
has been reached.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 9. Binding to DR4 and TCR recognition for glycopeptide 1 compared to 35–37. A)
Binding of glycopeptide 1 to DR4 compared to modified glycopeptides 35–37. Increasing
concentrations of 1 and 35–37 and a fixed concentration of biotinylated CLIP tracer were
incubated with DR4. The amount of CLIP tracer bound to DR4 was thereafter measured in a
timeꢀresolved fluoroimmunoassay. Each point represents the average of duplicates and error bars
26
ACS Paragon Plus Environment

Page 27 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
show ± one standard deviation. B) TCR recognition of glycopeptide 1 and modified
glycopeptides 35–37 by the DR4 restricted Tꢀcell hybridoma mDR1.1, after incubation with
antigen presenting spleenocytes and increasing concentration of the glycopeptides. If the DR4ꢀ
glycopeptide complex is recognized by the Tꢀcell hybridoma, secretion of ILꢀ2 into the
supernatant will occur, which is quantified by a sandwich ELISA. Each point represents the
average of duplicates and error bars show ± one standard deviation.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Similarities and Differences between the Human and Murine the MHC/glycopeptide/TCR
Multicomponent Systems
In CIA, the glycopeptide CII259–273 (1) is presented to TCRs by the murine MHCII molecule
q 24
q
A . Just as for presentation of 1 to Tꢀcells by DR4, signaling in the A /glycopeptide/TCR
multicomponent system has been linked to development of disease. Although both MHCII
2
60
molecules bind the same glycopeptide, the epitope is shifted with Ile binding in the P1 pocket
q 15
263
28
264
of A , while Phe fills this pocket in DR4. As a consequence GalHyl , which has been
10, 17
found to be a crucial residue for contacting the TCR,
is shifted from the center of the binding
q
grove in A towards one of the ends of the DR4 groove. The results presented in this manuscript
now, for the first time allow us to compare signaling in the two related multicomponent systems
to each other.
The two analogous multicomponent systems displayed both similarities and important
differences. Just as for the complexes of glycopeptides with DR4, reported herein, static models
q
were unable to explain experimentally observed differences in glycopeptide binding to A and
subsequent Tꢀcell responses. Instead, insight into the dynamics of the systems was required to
7
a, 7b
understand their signaling.
Binding of 1 to both MHCII molecules relied on a network of
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 68
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
hydrogen bonds, that were in dynamic equilibrium so that just under 50% of the total number
was present at any single point in time. In addition, hydrogen bond networks in both systems
appear to be crucial also for eliciting a Tꢀcell response; i.e. the removal or weakening of even
just one important hydrogen bond can alter the response from the entire multicomponent
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
7a, 7b
system.
Interestingly, the most populated hydrogen bonds in both MHCII/glycopeptide complexes are
formed between the Nꢀterminal residues of 1 when bound in the MCHII groove. However, the
two complexes differ regarding the number and position of residues involved in formation of this
core of strong hydrogen bonds. For DR4 only three residues (262–264) in 1 participate in the
q
7a
core, as compared to the A complex that involves an additional three residues (259–264) of 1.
2
64
Both cores contain GalHyl , the residue crucial for contacting the TCR. This strengthens our
hypothesis that exact positioning of the galactose moiety is crucial for eliciting a Tꢀcell response.
Another difference is that the opportunities for modification of 1 with maintained MHCII
q
binding affinity is larger for the DR4 complexes than for A . Thus, out of 20 glycopeptides with
260
263
q
modifications of Ile and Phe in 1, which bind in the P1 and P4 pockets of A , only two had a
7
c
similar affinity as native 1. When varying the amino acids of 1 that bind in the P1, P6 and P9
pockets of DR4 all of the synthesized glycopeptides bound to DR4, and most bound with similar
affinity as 1. In addition, several of the analogues of 1 were also able to elicit a significant Tꢀcell
response that ranged from weak to equally strong as that generated by 1. This can be expected to
provide larger opportunities to make novel glycopeptides with maintained and modified Tꢀcell
response for further investigations of signaling in the DR4 multicomponent system as compared
q
to the A system.
28
ACS Paragon Plus Environment

Page 29 of 68
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CONCLUSIONS
We have investigated the effect on the MHCII/glycopeptide/TCR multicomponent system upon
modification of the glycopeptide CII259–273 (1). Signaling in the multicomponent system relies
on that glycopeptide 1, or analogues, binds to DR4 and that the resulting complex is recognized
by the TCR. Unexpectedly, the functional multicomponent system was compatible with
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
63
significant structural variation at Phe
of 1, the key anchor residue in DR4, including
replacement by a benzothiophenealanine moiety. It is equally surprising that major variations at
268
the P6 and P9 positions of 1 were also allowed, e.g. substitutions of Gly
with Val and
2
71
cyclopropylalanine, and of Gly with Leu or Gln. It is possible that the rather large variation
allowed at these positions depends on that the Tꢀcell hybridoma investigated herein is highly
264
specific for the GalHyl moiety in the P2 position of 1.
Static structural models of the multicomponent system were unable to explain why some
analogues of 1 elicited a Tꢀcell response and others failed to do so, in spite of displaying almost
identical affinities for DR4. Instead, MD simulations revealed that analogues of 1, such as 3, that
displayed similar dynamics of their complexes with DR4 as 1 were able to elicit an equal Tꢀcell
response. DR4/glycopeptide complexes that were more rigid or substantially more dynamic than
the DR4/1 complex, gave either a weak or no Tꢀcell response. Importantly, complexes with
similar dynamics as the DR4/1 complex, such as the DR4/3 complex, were characterized by
having a similar number of hydrogen bonds of high, medium and low occupancy between the
bound glycopeptide and DR4 as the DR4/1 complex.
Comparison of the DR4/glycopeptide/TCR multicomponent system with the
q
corresponding murine A ꢀbased system revealed both similarities and differences. Both MHCII
molecules bound 1 by a dynamic network of cooperative hydrogen bonds, in which loss of a
29
ACS Paragon Plus Environment