вход по аккаунту


Use of T cell receptorHLA-DRB1.17804004 molecular modeling to predict site-specific interactions for the DR shared epitope associated with rheumatoid arthritis

код для вставкиСкачать
Vol. 40, No. 7 , July 1997, pp 1316-1326
0 1997, American College of Rheumatology
Objective. To use molecular modeling tools to
analyze the potential structural basis for the genetic
association of rheumatoid arthritis (RA) with the major
histocompatibility complex (MHC) “shared epitope,” a
set of conserved amino acid residues in the third
hypervariable region of the DRP chain.
Methods. Homology model building techniques
were used to construct molecular models of the
arthritis-associated DRB1*0404 molecule and a T cell
receptor (TCR) from T cell clone EM025, which is
specific for DR4 molecules containing the shared
epitope sequence. Interactive graphics techniques were
used to orient the TCR on the DR molecule, guided by
surface complementarity analysis.
Results. The predicted TCR-MHC-peptide complex involved multiple interactions and specificity for
the shared epitope. TCR residues CDRlP D30, CDR2P
N51, and CDR3P Q97 were positioned to potentially
participate in hydrogen bond interactions with the
shared epitope DRP residues Q70 and R71.
Conclusion. These results suggest a structural
mechanism in which specific TCR recognition and possibly V, selection are directly influenced by the diseaseassociated MHC polymorphisms.
The interaction of specific T cell receptors
(TCR), major histocompatibility complex (MHC) molecules, and peptides forms the structural basis for the
principal antigen-specific activation event involving T
Supported by NIH grant AR-37296 and a grant from the
Whitaker Foundation. Ms Penzotti’s work was supported in part by
NIH training grant T32-GM08268.
Julie E. Penzotti, Terry P. Lybrand, PhD: University of
Washington, Seattle; Gerald T. Nepom, MD, PhD: Virginia Mason
Research Center, Seattle, Washington.
Address reprint requests to Terry P. Lybrand, PhD, Center
for Bioengineering, University of Washington, Box 351750, Seattle,
WA 98195-1750.
Submitted for publication September 27, 1996; accepted in
revised form January 22, 1997.
cell recognition. Although the details of this interaction
are not yet known for any immune-mediated or autoimmune disease, strong associations of some diseases
with specific HLA genes provide a basis for examining
the details of the disease-associated trimolecular TCRMHC-peptide complex. In rheumatoid arthritis (RA),
this approach is facilitated by a strong genetic association with specific alleles of the DRBl locus containing
the “shared epitope,” a conserved amino acid sequence
(LLEQRRAA or LLEQKRAA) that is located at positions 67-74 on the a-helical portion of the MHC class I1
chain (1,2).
Although neither the disease-associated TCR nor
the arthritogenic peptide(s) is yet known, mutagenesis
studies and structural analyses of the RA-associated
DRBl*O4 molecules have been performed in several
laboratories, and there has been close agreement on the
structural features of these molecules (3-6). Herein we
describe the analysis of a DRBl*0404-specific TCR, and
its use in developing a model for the trimolecular
TCR-MHC-peptide complex. Based on shape and
chemical surface complementarity analysis, we derived
an orientation for this complex, in which TCR V,
contacts DRP and TCR V, contacts DRa. More importantly, site-specific interactions between individual TCR
V, residues and disease-associated polymorphic MHC
residues were also predicted.
We have previously described a preliminary
model of this complex (7). In the present study, we
developed a refined version of this model and performed
a detailed analysis of the potential roles of each
complementarity-determining region (CDR) loop in
recognition of the MHC-peptide complex. Predictions
based on our modeling suggest a structural basis for the
genetic association of R A with the MHC shared epitope,
and indicate the potential contribution of this region to
the oligoclonal T cell selection and expansion process
observed in the disease. Our TCR-MHC-peptide model
also provides a framework for the structural interpretation of results from previous mutagenesis studies of this
TCR-MHC system, and for designing experiments t o
further investigate t h e role of the shared epitope in RA.
Modeling techniques and criteria. All minimization
and molecular dynamics analyses were performed with the
AMBER programs (8), using a standard all-atom potential
function (9) in vacuo. Structural models were constructed and
evaluated based on the following physical and stereochemical
criteria: 1) reasonable peptide backbone conformations; 2)
reasonable side-chain conformers (lo); 3) few or no cis peptide
bonds; 4) side-chain packing density similar to that observed in
known protein structures (11); and 5) logical partners for
hydrogen bonding, and charged functional groups that are not
solvent exposed.
Molecular modeling of EM025 TCR. T cell clone
EM025 is an alloreactive clone derived from a normal DR4/
Dw4 (*0401) donor by priming against the DR4/Dw14 (*0404)
alloantigen in a mixed lymphocyte culture, as described previously (12). We constructed 2 different models of the variable
domain of the EM025 a / p TCR. An initial EM025 model was
built based on homology to immunoglobulins of known structure. A second homology model for the EM025 TCR was
constructed using murine TCR a (13) and /3 (14) chain crystal
structures as templates.
For the TCR model structure based on homologous
immunoglobulin structures, we modeled the TCR /3 chain as
an immunoglobulin light chain and the a chain as a heavy
chain. The TCR framework structure and initial loop conformations were built using AbM, an antibody modeling software
package (Oxford Molecular, Campbell, CA). The template
used for the p-chain framework structure was the lBBD (15)
light chain (36% identity in the framework regions); for the a
chain, the 2FB4 (16) heavy chain was used (36% identity in the
framework regions). None of the 6 TCR CDR loops fit into the
canonical classes defined for antibody CDR loops, and they
were thus built using a conformational searching method in
AbM. The conformations of the /3 CDR2 and CDR3 loops
were modified considerably from their AbM assignment due to
the need to insert several residues in areas where antibody
CDRs were shorter in length. Single amino acid insertions
were also required in the third framework region of the /3 chain
and the first framework region of the a chain.
All insertions were performed using the molecular
graphics modeling package WHATIF (17). WHATIF inserts
backbone residues (polyglycine) by searching a fragment database and allowing the user to view the “hits” and select a
fragment. The inserted glycine residues were then replaced by
the corresponding EM025 amino acid residues using the
molecular graphics package MidasPlus (18). The resulting
structure was relaxed with limited energy minimization and low
temperature molecular dynamics.
A second EM025 TCR model structure was constructed using recently described crystal structures for murine
TCR a and /3 chains (13,14). Model construction and initial
refinement processes were done independently for each chain
38.4% identity (74.1% conservative replacements)
. ... . . . . . . . . . . . . . . . ... . . .. .. . . . .. .. . . ... . . .. .. . . . .. .. . . .
.. . ... . .. . . .. . .. . . . . .. .. ... .. .. . .. .. . .. .. .. . .. .. . .. .. ... .. .. .
EM025 VB / MURINE V 6 :
37.6% identity ( 8 0 . 3 % conservative replacements)
. .. ... .. ... .. . .. .. . .. ... .. ... . . . . .. . .. . .. . .. . .. . .. . .. . .. . .. . . . . . . . . . .
. . . . .. . ... . . . . . . . . . . . ... .. ... .. ... .. ..
. .. .. .. .. ..... .. .. .. .. .. .. ...
Figure 1. Sequence alignments for EM025 T cell receptor (TCR)
model construction, generated using the align program and the
PAM250 scoring matrix. Shown at the top is a sequence alignment of
the EM025 a chain, composed of the human V,,21 and a unique J,
gene segment (57), and the sequence of the murine a-chain crystal
structure, containing the murine V,4 and J,2B4 gene segments (13).
Shown at the bottom is a sequence alignment of the EM025 TCR p
chain with the sequence of the murine TCR p chain for which the
structure has been determined (14). The EM025 TCR p chain contains
the human V,8.2 and J,2.2 gene segments (57). The murine TCR p
chain contains the murine V,8.2 and J,2.1 gene segments (14). The
complementarity-determining regions are located as described by
Chothia et a1 (20). V A = V,; VB = V,.
prior to heterodimer construction. Using Midasplus, each
chain was built by replacing the amino acid residues in each
crystal structure with the corresponding amino acid from
EM025 according to the alignments shown in Figure 1. Sidechain conformations were assigned to the preferred conformer
(10) that was most similar to the conformation of the side chain
in the crystal structure templates. Construction of the EM025
/3 chain required a single amino acid insertion at position 64 in
the third framework region. A deletion of 1 amino acid from
the murine a chain was required in order to construct the
CDR3 of the EM025 a chain. Both the insertion and the
deletion were made using the molecular graphics modeling
program PSSHOW (19). For each chain, the resulting segments were reattached using a short, low temperature (1020K) molecular dynamics run with weak harmonic constraints
(15-30 kJ/mole/A), to form trans peptide linkages. Interactive
graphics techniques were used to manually adjust side-chain
torsion angles and backbone conformations to allowed values.
These adjustments were performed iteratively with short minimization (-200 steps) and molecular dynamics (typically
<1-2 ps) runs to improve side-chain packing and relieve steric
problems introduced by the model building process.
The a/P heterodimer was constructed based on homology to immunoglobulin VJV, dimers and the murine TCR
a-chain homodimer structure (13). Guided by conserved residues at the interface (20), we used interactive graphics techniques to superimpose the a and P chains on an array of
immunoglobulin heavy- and light-chain dimer structures and
the TCR a-chain homodimer structure. From this exercise we
obtained an initial starting structure for our TCR alp
heterodimer model.
In constructing the heterodimer, modifications were
made to the crystal structure-based backbone conformation of
the P-chain CDR3 in order to bury the Trp at position 100 in
the hydrophobic and largely aromatic interface region. This
was accomplished by manually adjusting backbone and sidechain dihedral angles using interactive molecular graphics
tools. The model of the alp heterodimer was refined with
limited minimization and molecular dynamics. This murine
TCR-based EM025 model was used in all subsequent modeling
of the ternary TCR-MHC-peptide complex.
HLA-DR4 model. A structural model for the DR*0404
(Dw14) molecule in complex with the influenza hemagglutinin
(HA) peptide 306-318 (PKYVKQNTLKLAT) was built based
on the crystal structure of the HLA-DR1-HA complex (21).
Dw14 is composed of the monomorphic D R a chain and a DRP
chain with 92% sequence homology to the D R l P chain. Only
16 amino acid replacements were needed to generate the
Dw14 P chain from the D R l crystal structure. Using MidasPlus (18), amino acid side chains were replaced and assigned to
the preferred conformer from the Ponder and Richards rotamer database (10) that was closest to the side-chain conformer
in the crystal structure template. The conformation and position of the HA peptide in the groove were not altered in
constructing the DR*0404-HA complex.
TCR-MHC-peptide complex. The model of the ternary complex was constructed using interactive graphics techniques to orient the EM025 TCR on the Dw14-HA-peptide
complex, taking into consideration the structural and chemical
surface complementarity. Visual inspection of the surface
residue types, generation of solvent-accessible surfaces using
MS (22), and performance of simple Coulomb electrostatics
calculations using PSSHOW (19) guided our visual analyses of
surface complementarity. The resulting trimolecular complex
was relaxed with limited minimization and molecular dynamics
refinement, keeping the constant domain of the D R a and
DRP chains fixed. Hydrogen bonding analysis of the complex
was performed using the carnal module of AMBER (8).
The general structures of our antibody-based and
murine TCR-based EM025 models were very similar,
with the V,/V, dimer assuming a framework structure
very similar to that of the Fv fragments of immunoglobulins. Due to this similarity,previous TCR models derived
from immunoglobulin structures have been useful in
describing general features of TCR recognition of the
MHC-peptide complex (20,23-25). However, none of
the 6 CDR loops of the EM025 TCR fit into the
canonical classes defined for antibody loops. Sites in the
TCR CDRl and CDR2 regions, which are highly con-
served particularly among subgroups of TCRs, are likely
to play a role in determining loop conformations (20).
The sequence similarity between the CDRl and CDR2
regions of EM025 and the murine TCR chains is shown
in Figure 1. There was also limited sequence similarity
between the highly variable CDR3 regions. In addition,
5 of the 6 EM025 CDRs were equal in length to the
murine TCR CDRs; the sixth CDR, EM025 CDR2,
differed in length by only 1residue. Thus, considering all
regions of the sequences, the homology between the
EM025 TCR and the murine TCR chains was greater
than the homology between EM025 and the immunoglobulin sequences. The murine TCR chains therefore
provided a better overall template for our model construction. Thus, we used only our murine TCR-based
model in subsequent modeling studies of the TCRMHC-peptide complex.
Each chain of the EM025 TCR was independently built and relaxed with minimization and molecular dynamics prior to construction of the a/p heterodimer.
In the course of refining the EM025 a-chain model, a cis
peptide linkage developed between S7 and P8 in the first
framework region of the EM025 a chain. Though no cis
peptide bond is present in TCR a-chain crystal structures (13,26) at this position, there is a cis bond at this
position in the p-chain crystal structures (14,26) and in
many immunoglobulin chains (e.g., see refs. 27-30). The
sequence of our a chain in this region showed greater
similarity to that of the murine p chain and immunoglobulins that have cis bonds at this position. Therefore,
we did not eliminate this cis bond in our model.
The EM025 heterodimer was constructed based
on homology to immunoglobulin V J V , dimers and the
murine TCR a-chain homodimer structure (13). This
was possible because the majority of critical residues for
immunoglobulin lightheavy chain association are conserved in TCR a and p chains (20). Recent reports of 2
crystal structures of different TCR alp heterodimers
have confirmed their similarity to antibody variable
domains (26,31). As in antibody V J V , dimers, the
resulting a/p dimer interface is dominated by hydrophobic interactions between highly conserved aromatic residues from each chain.
In constructing the heterodimer, we modified the
conformation of CDR3p from its initial crystal structurebased assignment to bury the Trp at position 100 in this
highly aromatic interface region. CDR3P residue WlOO
was oriented such that it packed with the aromatic
residues of the a-chain F33, F45, and Y100. We also
performed a homology search to identify immunoglobulins with CDR3 sequences similar to EM025 p CDR3. A
Figure 2. Upper, Solvent-accessible surfaces of the recognition regions of the EM025 T cell receptor (TCR) and major histocompatibility (MHC)
DR*0404-hemagglutinin (HA) peptide, indicating surface complementarity. Surfaces were generated using the Connolly algorithm (22) with a 1.4A
sphere. Color code for amino acid residues: red = acidic (D, E); light blue = basic (H, K, R); green = uncharged polar (C, N, Q, S, T); yellow =
hydrophobic (A, F, G, I, L, M, P, V, Y, W). Lower, A schematic of the combining site, viewed through the TCR directly onto the D R ~ 0 4 0 4 - m
surface. The outlines of the DR*0404-HA complex and EM025 TCR are in red and light blue, respectively. Complementarity-determining region
loops and hypervariable region 4 are labeled. DRA = DRa; DRB = DRP; VA = V,; VB = V,; N = N-terminus.
survey of crystal structures for immunoglobulins, with a
Pro followed by a Trp, or another aromatic residue, in
CDR3 or only a Trp in the mid-CDR3 region, revealed
loop conformations such that the aromatic residue could
interact with other aromatic residues composing the
dimer interface (e.g., see refs. 27, 32, and 33).
CDR3P W1OO also participated in several interchain hydrogen bonds: /3 WlOO N to a YlOO OH and S92
Figure 3. Side view of the EM025-hemagglutinin (HA) peptide-DR *0404 tricomplex model. Stick representations of
candidate contact residues between the T cell receptor (TCR) p chain and the DR-peptide complex are shown. The backbone
of the HA peptide is shown in purple. Color codes for the amino acid side chains are the same as in Figure 2. Color codes for
EM025 P amino acid side chains: B1 D30 in red, B2 N51 in green and B3 Q97 in green. and El03 in red. Color codes for DRP
shared epitope residues: Q70 in green and R71 in blue. Color codes for peptide side chains: K310 in blue and N312 in green.
OH, and p WlOO NE1 to a Y35.* In addition, there were
several other interchain hydrogen bonds in our model of
the EM025 TCR heterodimer. One was between the
highly conserved positions Q37 of the P chain and K37
of the a chain (Q is most common). A hydrogen bond
between these positions is conserved in many immunoglobulin structures (20) and is present in the 2 recently
described TCR alp heterodimer structures (26,31). The
NE and NH2 of P-chain R111 hydrogen bonded to the
carbonyl oxygen of A40 of the a chain. A fourth
hydrogen bond, likely specific to our alp dimer, was
found between the a and P CDR3 regions, P S95 to a Q98.
Overall, our TCR model corresponded well with
the recently described TCR alp heterodimer crystal
structures (26,31), with the most significant deviations
noted in the highly variable CDR3 a and P regions. The
root mean square deviation for the C, positions in our
TCR model compared with the human TCR crystal
structure (26) was 2.3A. This reduces to 1.86A if only the
framework regions are considered.
A model of the DRB1*0404 (Dw14) molecule
in complex with the H A peptide 306-318
(PKYVKQNTLKLAT) was constructed from the crystal
structure of the HLA-DR1-HA complex (21). Beginning with this structure and our murine TCR-based
EM025 TCR model, we derived a model for the EM025DR*0404-HA complex by manually docking the 2 structures, guided by surface complementarity. Shape and
chemical surface complementarity analyses suggested an
orientation in which the EM025 P chain interacts with
the DRP chain and the EM025 a chain with the D R a
chain. Surfaces for both the EM025 TCR and
DR *0404-HA models are shown in Figure 2.
The exposed residues along the a-helix of the
D R a chain consisted of mostly hydrophobic residues
(G58, L60, A61, 163, A64, V65), except near the
N-terminus of the bound peptide (DRa E55, Q57) (note
that in Figure 2, viewing of several of the exposed D R a
residues is partially blocked by exposed peptide residues
K307 near the N-terminal and K315 and T318 near the
C-terminal). The surface of the EM025 a-chain CDR
loops was similarly composed of a large hydrophobic
region and a polar region. M28, F29, and Y31 of CDR1,
I52 of CDR2, and CDR3 residues V94, G95, and G96
*International Union of Pure and Applied Chemistry
(IUPAC) standard-atom type abbreviations are defined as follows: N
backbone nitrogen atoms; NE, NE1, and NE2 are nitrogen atoms at
the E side-chain position of Arg, Trp, and Gln, respectively; NH1 and
NH2 are terminal nitrogen atoms of the Arg guanido group; OD1 and
OD2 are oxygen atoms at the 6 side-chain position of Asn or Asp; OEl
and OE2 are oxygen atoms at the E side-chain position of Gln or Glu;
and OH is an oxygen atom of alcohols and Tyr.
contributed to the hydrophobic patch. CDR2 residues
S50, S51, K53, and D54 as well as several charged
residues from the framework and hypervariable regions
formed the polar/charged patch of the a-chain surface.
In contrast, the D R P chain presented a largely
polar surface to the TCR due to exposed residues D66,
Q70, and R71 and other polar residues exposed along
the a-helix. The antigen recognition surface of our
EM025 TCR P chain was largely hydrophilic due to the
acidic, basic, and uncharged polar residues protruding
from the CDR regions. In V, CDR1, H29 and D30 were
exposed, V, CDR2 contained 3 exposed asparagines,
N50, N51, and N52, and CDR3 residues S96, Q97, and
El01 pointed up from the putative antigen recognition
surface of the EM025 P chain. In the TCR heterodimer,
the resulting surface along the interface of the EM025 a
and P chains alternated between polar and nonpolar
regions, which complemented the character of candidate
TCR contacts along the H A peptide. HA peptide residues K307, V309, K310, N312, and K315 pointed away
from the DR binding groove and were candidates for
TCR interaction (21).
The manually generated TCR-MHC-peptide
complex was relaxed with limited minimization and low
temperature molecular dynamics to improve the packing
interactions and relieve steric conflicts. The resulting
ternary complex is shown in Figure 3. The solventaccessible surface area buried by the interaction was
-792A’ for the EM025 TCR and -820A2 for the
DR*0404-HA complex, similar in size to the interaction
surface predicted by mutagenesis studies (34), other
TCR-MHC-peptide models (25), and a TCR-class I
MHC-peptide complex structure (26). The interaction
involved direct contact with 16 residues in the TCR,
consistent with theoretical predictions of the optimal
receptor-combining regions by Percus et a1 (35). All 6
CDR loops of the TCR interacted with the DR*0404HA complex (Table 1), although C D R l of the a chain
was minimally involved in the interaction, with only 1
residue, Y31, contacting the MHC D R a chain.
Residues for which the solvent-accessible area
was altered by complex formation included amino acid
positions in the DR molecule that affect T cell stimulation when mutated (DRa E55, Q57, G58, L60, A61,
A64, V65, K67, and A68 [36-381; D R P L67, Q70, and
R71 [3]) as well as residues from all 6 CDR regions of
the TCR (CDRIP S27, G28, H29, D30, and Y31;
CDR2P N50, N51, N52, V53, and P54; CDR3P S96,
Q97, G98, P99, W100, A101, G102, E103, and F105;
C D R l a Y31; CDR2a 147, S48, S50, and D54; and
CDR3a V94, G95, and G96) and peptide candidate
Table 1. Van der Waals and hydrogen bond interactions between the
DR*0404-HA complex and the EM025 TCR*
MHCipeptide residue
TCR residue
Shared epitope
HA K310
HA N312
HA K315
DRor K39
DRa Q57
DRor A61
DRa A64
DRa V65
Pl D30, P2 N51, P3 Q97t
P3 Q97t
P3 S96,t P3 E103$
Pl D30,t 33 S96,t P3 Q97
a 3 v94t
P N50, P2 N51,t P2 P54
a 2 S48,t or2 149, or2 DS4,$ a2 K5St
or2 S48,t P3 G102
P3 A101, P3 El03
a1 Y31t
p3 P99
* Hydrogen bonds were identified using the carnal module of the
AMBER programs, with a distance limit between heavy atoms of 4.0A
and an angle limit of 60". HA = hemagglutinin; TCR = T cell receptor;
MHC = major histocompatibility complex.
t Indicates a hydrogen bond.
$ Indicates a salt bridge.
TCR contact residues V309, K310, N312, and K315.
DRa E55 was well positioned to interact with the
N-terminus of the TCR P chain if we included the first 2
residues of the sequence in our model (the first residue
of the murine TCR P-chain crystal structure [14] corresponded with position 3 in our TCR). Ten hydrogen
bonds were identified between the TCR and MHC
molecules, including 2 that contributed to a salt bridge.
Recognition of the shared epitope region of the
DRP chain potentially involves all 3 CDR loops of the
TCR P chain. D30 of CDRlP and N51 of CDR2P were
van der Waals contacts to shared epitope residue Q70.
CDR3-specific interactions with the shared epitope consisted of a hydrogen bond from TCR residue P Q97 NE2
to Q70 O E l and 2 hydrogen bonds from TCR P Q97
OEl to R71 NH2 of the shared epitope region. This
involvement of 3 specific sites (D30, N51, and Q97) from
3 different CDR elements on the TCR P chain, as
suggested by the model, may have implications for the
selection and activation of T cells involved in recognition
of the DRB1*0404 molecule. A D or E residue at
position 30 of the P-chain CDRl was present in V,3.1,
9.1, 17.1, and 14.1, as well as in V,8.2. In the TCR V,
alignments of Arden et a1 (39), N was present in V,9.1
and V,14.1, at a position that corresponds to CDR2 N51
in our EM025 V,8.2 chain; a homologous Q was present
in V,17.1 at position 50. Thus, many of the specific V,
sequences frequently associated with RA (for review,
see ref. 40) or with allorecognition of the DRB1*0404
gene carry this "signature motif' of DE30 and/or
N5 1/Q50.
To test the likelihood that these residues occur
nonrandomly or are selected in association with RA, we
evaluated 45 V, sequences derived from CD4-t T cells
that have undergone clonal expansion in patients with
RA (41) (we did not include the 5 V,8 sequences
because the identification of the specific V,8 alleles was
required for our analysis). Among the TCRs from these
in vivo clonally expanded T cells, 39 of 45 (87%) carried
DE30 and 29 of 45 (64%) carried N5liQ.50. In contrast,
analysis of a published set of V, sequences present in
myelin basic protein-specific T cell clones from patients
with multiple sclerosis (42) indicated 33% (4 of 12)
positive for DE30 and 50% (6 of 12) for N51/Q50 (P <
0.001 and P not significant, respectively).
The possible association between CDR3P Q97
and the HLA shared epitope region is harder to evaluate
due to substantial variability in length and amino acid
composition of CDR3 regions among different T cells.
These factors and the nature of the residues composing
the associated CDR3a are likely to influence the conformation of CDR3P and thereby alter the residue
positions that can interact with the MHC molecule or
peptide. Interestingly, however, a similar CDR3 motif
containing a Q at position 97 has been previously
reported in the context of V,17 CDR3 sequences associated with RA (43). In our panel of 45 CD4-t clonally
expanded T cells from RA patients, however, only 2
contained Q97 (41).
Consistent with models proposed for the general
TCR-MHC-peptide orientation (20,23,24), our a- and
/3-chain CDR3 regions contacted the peptide. In our
model, CDR3P was positioned over the N-terminal half
of the peptide, and the TCR a-chain CDR3 region was
positioned to interact with residues toward the
C-terminal end of the peptide. There were 6 hydrogen
bonds between the TCR and HA peptide (see Table 1),
including 2 that were part of a salt bridge between
CDR3P El03 and K310 of the HA peptide. A similar
rotational orientation has been proposed for models of
different TCR-class I1 MHC-peptide systems (44,45).
However, the alignment of our TCR and MHC molecules in the trimolecular complex was oriented -180"
from that reported by Jorgensen et a1 (46) and from that
in another model of this TCR-class I1 MHC-peptide
complex (5C.C7/1-Ek/moth cytochrome c) (25).
The nature of our predicted contacts for the
shared epitope region was improved by a small rotation
of our TCR model on the DR*0404-HA complex to
align our V, chain with the crystal structure V, chains
and our V, chain with the crystal structure V, chains
(26,31) (such that our V, was over the N-terminal half of
the HA peptide and our V, was over the C-terminal
half). In this orientation, CDR2P residues N50, N51,
and/or N52 were better positioned to hydrogen bond to
shared epitope residue Q70, and specific contacts between CDRlP D30 and CDR3P Q97 were maintained.
Potential contacts to the HA peptide in this orientation
existed between HA K310 and CDR3P E103, between
N312 and CDR3P Q97, and between K315 and CDRla
D30. This adjustment in the rotational orientation also
enlarged the surface area buried in the interaction to
-9OOA’ for both the TCR and the DR*0404-HA complex, and thus increased the total number of TCR
residues contacting the MHC-peptide complex. If we
adjusted our TCR-MHC-peptide model to the same
rotational orientation as the TCR-class I MHC-peptide
structures (such that our V, was over the N-terminal
half of the HA peptide and our V, was over the
C-terminal half) (26,31), there was no apparent specificity for the shared epitope region in the predicted
In the trimolecular complex between TCR, MHC
class 11, and peptide, a limited number of intermolecular
interactions account for the specificity of binding, which,
in turn, controls subsequent immune activation. Structural information derived from crystallographic analyses
of MHC class 11-peptide complexes (21,47,48) and TCR
a and P chains (13,14) and the recently described
TCR-MHC class I-peptide complexes (26,31) now
makes it possible to envision the molecular details of the
trimolecular recognition event involving class I1 MHC.
In vivo and in vitro mutagenesis studies (49-53) and
structural analyses (13,14,20,26,31) indicate that the
principal recognition surface of the TCR aIP
heterodimer is composed of the CDR1, CDR2, and
CDR3 loops from both the a and the p chains. While all
6 of these CDR elements can potentially be involved in
the specificity of interaction, in our model of the EM025
TCR-MHC-peptide complex, the specificity in recognition is conferred predominantly by the P chain via direct
interactions of CDRlP D30, CDR2P N51, and CDR3P
Q97 with the shared epitope residues 70 and 71 of the
class I1 P chain.
These specific interactions are likely to be the
dominant elements involved in TCR recognition. They
are accompanied by a large number of nonspecific
interactions, which may function to further stabilize the
trimolecular complex. Many relatively nonspecific contacts, including van der Waals interactions and hydrogen
bonds to backbone nitrogen and carbonyl oxygens, are
predicted in our model. Indeed, with the exception of a
salt bridge between DRa K39 and TCR CDR2a D54,
and a hydrogen bond between the side chains DRa Q57
and TCR CDR2a S48, the contacts that the TCR a
chain makes with the DRa chain and peptide are all
nonspecific in nature. CDRl of the a chain is minimally
involved in the interaction, forming only 1 hydrogen
bond to a backbone carbonyl oxygen of the DRa chain.
CDR2 of the a chain contacts several residues along the
a-helix of the monomorphic DRa chain, including the
salt bridge and hydrogen bond mentioned above. The
carbonyl oxygen of CDR3a residue V94 hydrogen bonds
to K315 of the HA peptide, but CDR3 is otherwise not
intimately involved in recognition of the MHC-peptide
complex. The other surface-exposed residues of CDR3a
include 2 Gly residues, G95 and G96, which would not
be expected to participate in highly specific interactions,
but are likely to be more important in determining the
loop conformation.
Among the multiple contacts between the MHCpeptide complex and TCR V, in the trimolecular model
(see Table 1), the most notable are the specific interactions with the MHC class I1 P residues 67-74 comprising
the “shared epitope” region associated with RA. The
arginine at position 71 of the DR*0404 /3 chain is part of
an extended hydrogen bonding network in the shared
epitope region, which may create a specific conformational unit recognized by the TCR. The R71 side chain
forms 4 good hydrogen bonds (<3.0A) as follows: NE to
OD1 of DRP D28, NH1 to OE1 of shared epitope
residue Q70, NH2 to OD2 of DRP D28, and NH2 to the
backbone carbonyl oxygen of the HA peptide residue
N312 (Figure 4). A fifth good hydrogen bond is formed
between R71 NH1 and TCRP Q97 OEl in the trimolecular complex. There are several other hydrogen bonding partners in close proximity (<4A) to the 3 R71
nitrogens, as well as a number of hydrogen bonds
involving the other shared epitope residues (Figure 4).
This key role for R71 in both TCR contact and
peptide interactions is consistent with the findings of
mutagenesis studies in which substitutions at this residue
position abrogate T cell recognition (3,434) and influence the peptide binding specificity by altering the
nature of pocket 4 (5,6,55,56). Even the conservative
replacement R71K ablates EM025 recognition (57).
However, in the context of the shared epitope, this may
not be a conservative replacement because a K side
chain, which has only 1 terminal amino group, would not
OD 1
Figure 4. Schematic of the hydrogen bonding network surrounding
shared epitope residues DRP Q70 and R71. Hydrogen bonds, represented by the dotted lines, are less than a distance of 3.OA between the
heavy atoms, and deviate (15” from linearity for the H-donoracceptor angle. For simplicity, not all of the surrounding potentit1
hydrogen bonding partners (identified using a cutoff distance of 4.0A
and an angle cutoff of 1 radian) are shown.
preserve the extensive hydrogen bonding network described above for R71. In addition, as part of this
hydrogen bonding network in our model, Q70 of the
shared epitope establishes direct interactions with
TCRP CDR1, CDR2, and CDR3 residues D30, N51,
and Q97, respectively.
The functional implication of this model is that,
in addition to influencing the MHC-peptide interaction,
the shared epitope sequence, especially at residues 70
and 71, is intimately involved in direct TCR contact.
Evidence for shared epitope interaction with CDRl and
CDR2 residues in vivo would likely be an influence on T
cell selection and/or amplification. Indeed, this is precisely what is observed. Several different V, subsets are
known to be expanded in oligoclonally amplified T cells
in patients with RA (for review, see ref. 40). Interestingly, the predominant V, expression in such expanded
T cells corresponds to the “D30 and/or N51/Q50” motif
predicted in our model. Though the specific hydrogen
bonds predicted by our model are between CDR3P Q97
and DRP Q70 and R71 with only slight adjustments to
side-chain conformations, CDRlP D30 and CDR2P
N51 could play the principal role in specific recognition
of the shared epitope via a scenario in which N51
hydrogen bonds to DRP Q70 and D30 forms a salt
bridge to DRP R71. Whereas others have suggested that
the oligoclonal V, expansion in RA may be the consequence of a superantigen stimulus (58,59), we propose
that an exogenous superantigen is not involved. The
DRP shared epitope may be acting as a surrogate
endogenous superantigen, eliciting the corresponding
TCR V, response.
The interaction between TCR CDR3P residue
Q97 and the DRP shared epitope in our model raises
another interesting point. TCR cw and /3 CDR3 regions
are generally thought to principally interact with the
bound peptide in the MHC-peptide complex (20,23,24).
However, the direct interactions between TCR CDR3P
Q97 and class I1 p-chain residues Q70 and R71 suggest
that in some cases shared epitope-positive class I1
molecules may be capable of triggering T cell recognition in much the same way as an antigenic peptide.
Indeed, both CDR3a and CDR3P contact polymorphic
and nonpolymorphic MHC class I heavy chain residues
as well as peptide residues in the TCR-HLA-A2-Tax
peptide structure (26). Our T cell clone, EM025, recognizes DRB1*0404 even when expressed on bare lymphocyte syndrome cell lines, which are defective in peptide
loading and presentation (7,60). Thus, this recognition is
relatively peptide-independent, and one explanation
may simply be that the direct contacts of TCR to the
shared epitope residues form interactions with CDR3
sufficient for TCR activation.
In conclusion, the principal MHC-encoded genetic determinant of susceptibility to RA is the shared
epitope, a conserved sequence in the amino acid positions 67-74 located on the a-helical portion of a class I1
/3 chain (1,2). Since the principal known function of
HLA class I1 molecules is to bind and present antigenic
peptides, it is often assumed that the role of the shared
epitope is to control this peptide binding specificity. In
the model described herein, however, a different role for
the shared epitope is envisioned. The shared epitope
itself is likely to be a discrete unit of immunologic
recognition, functioning as a dominant selection element
in direct TCR recognition for activation and amplification. Such direct recognition may be the basis by which
the autoreactive T cell response in RA is biased toward
oligoclonality based on direct selection by MHC contact
sites in the shared epitope region.
Specific recognition of the shared epitope is best
satisfied by a rotational orientation of our model such
that V, is positioned over the N-terminal region of the
peptide and V, is positioned over the C-terminal region.
Due to the structural symmetry of TCR alp
heterodimers, a 180" variation of the diagonal binding
orientation observed in both TCR-class I MHC-peptide
structures cannot be ruled out. Similar detailed structural data for TCR-class I1 MHC-peptide complexes
and additional TCR-class I MHC-peptide complexes
are needed to help resolve the issue of rotational
flexibility in the binding mode. It is intriguing to note
that if a single orientation emerges from subsequent
structural studies consistent with that observed in the
recently described TCR-class I MHC-peptide structures (26,31), our CDRlP D30 and/or CDR2P N51/Q50
motif residues may interact with a superantigen, since
these residue positions are contacts to the superantigen
in the recent TCR p chain-superantigen crystal structure (61). The involvement of a superantigen is an
alternate mechanism consistent with the oligoclonal V,
expansion observed in RA (57,58).
We thank Dr. R. A. Mariuzza (Center for Advanced
Research in Biotechnology, University of Maryland Biotechnology Institute, Rockville, M D ) for kindly providing a prerelease of the coordinates for the murine a- and P-chain
structures and Dr. C. M. Weyand (Mayo Clinic and Foundation, Rochester, MN) for the CD4+ TCRP amino acid sequences from RA patients. We also thank Drs. P. Ghosh and
D. C. Wiley for kindly providing the coordinates of the
TCR/HLA-A2/Tax complex.
1. Gregersen PK, Silver J, Winchester RJ: The shared epitope
hypothesis: an approach to understanding the molecular genetics
of susceptibility to rheumatoid arthritis. Arthritis Rheum 30:12051213, 1987
Nepom BS, Nepom GT: Immunogenetics and the rheumatic
diseases. In, Textbook of Rheumatology. Edited by WN Kelley,
ED Harris Jr, S Ruddy, CB Sledge. Philadelphia, WB Saunders,
Hiraiwa A, Yamanaka K, Kwok WW, Mickelson EM, Masewicz S,
Hansen JA, Radka SF, Nepom GT: Structural requirements for
recognition of the HLA-Dw14 class I1 epitope: a key HLA
determinant associated with rheumatoid arthritis. Proc Natl Acad
Sci U S A 87:8051-8055, 1990
Signorelli KL, Watts LM, Lambert LE: The importance of
DR4Dw4 beta chain residues 70,71, and 86 in peptide binding and
T cell recognition. Cell Immunol 162:217-224, 1995
Fu XT, Bono CP, Woulfe SL, Swearingen C, Summers NL,
Sinigaglia F, Sette A, Schwartz BD, Karr RW: Pocket 4 of the
HLA-DR (alpha, beta 1*0401) molecule is a major determinant of
T cell recognition of peptide. J Exp Med 181:915-926, 1995
Hill CM, Liu A, Marshall KW, Mayer J, Jorgensen B, Yuan B,
Cubbon RM, Nichols EA, Wicker LS, Rothbard JB: Exploration
of requirements for peptide binding to HLA DRB1*0101 and
DRB1*0401. J Immunol 152:2890-2898, 1994
7. Penzotti JE, Doherty D, Lybrand TP, Nepom G T A structural
model for TCR recognition of the HLA class I1 shared epitope
sequence implicated in susceptibility to rheumatoid arthritis. J
Autoimmun 9:287-293, 1996
8. Pearlman DA, Case DA, Caldwell JW, Ross WS, Cheatham TE
111, DeBolt S, Ferguson D, Seibel G, Kollman P: AMBER, a
package of computer programs for applying molecular mechanics,
normal mode analysis, molecular dynamics and free energy calculations to simulate the structural and energetic properties of
molecules. Comput Phys Commun 91:1-41, 1995
9. Weiner SJ, Kollman PA, Nguyen DT, Case DA: An all atom force
field for simulations of proteins and nucleic acids. J Comp Chem
7:230-252, 1986
10. Ponder JW, Richards FM: Tertiary templates for proteins: use of
packing in the enumeration of allowed sequences for different
structural classes. J Mol Biol 193:775-791, 1987
11. Gregoret LM, Cohen FE: Novel method for the rapid evaluation
of packing in protein structures. J Mol Biol 211:959-974, 1990
12. Mickelson EM, Masewicz S, Smith A, Petersdorf E, Nepom GT,
Martin PJ, Hansen JA: T-cell clones identify three distinct
epitopes associated with HLA-Dw14. Hum Immunol 32:229-233,
13. Fields BA, Ober B, Malchiodi EL, Lebedeva MI, Braden BC,
Ysern X, Kim JK, Shao X, Ward ES, Mariuzza RA: Crystal
structure of the Va domain of a T cell antigen receptor. Science
270:1821-1824, 1995
14. Bentley GA, Boulot G, Karjalainen K, Mariuzza RA: Crystal
structure of the j3 chain of a T cell antigen receptor. Science
267:1984-1987, 1995
15. Tormo J, Stadler E, Skern T, Auer H, Kanzler 0, Betzel C, Blaas
D, Fita I: Three-dimensional structure of the Fab fragment of a
neutralizing antibody to human rhinovirus serotype 2. Protein Sci
1:1154-1161, 1992
16. Marquart M, Deisenhofer J, Huber R, Palm W: Crystallographic
refinement and atomic models of the intact immunoglobulin
molecule Kol and its antigen-binding fragment at 3.0 A and 1.0 A
resolution. J Mol Biol 141:369-391, 1980
17. Vriend G: WHATIF: a molecular modeling and drug design
program. J Mol Graph 852-56, 1990
18. Ferrin TE. Huane CC. Jarvis LE. Laneridee R: The MIDAS
display system. J Go1 Graph 6:13-27, 19f8 "
19. Swanson E: PSSHOW: Silicon Graphics 4D Version. Seattle, 1990
20. Chothia C, Boswell DR, Lesk AM: The outline structure of the T
cell aj3 receptor. EMBO J 7:3745-3755, 1988
21. Stern LJ, Brown JH, Jardetzky TS, Gorga JC, Urban RG,
Strominger JL, Wiley DC: Crystal structure of the human class I1
MHC protein HLA-DR1 complexed with an influenza virus
peptide. Nature 368:215-221, 1994
22. Connolly ML: Solvent-accessible surfaces of proteins and nucleic
acids. Science 221:709-713, 1983
23. Davis MM, Bjorkman PJ: T-cell antigen receptor genes and T-cell
recognition. Nature 334:395-402, 1988
24. Claverie JM, Prochnicka-Chalufour A, Bougueleret L: Implications of a Fab-like structure for the T cell receptor. Immunol
Today 10:10-14, 1989
25. Almagro JC, Vargas-Madrazo E, Lara-Ochoa F, Horjales E:
Molecular modeling of a T-cell receptor bound to a major
histocompatibility complex molecule: implications for T-cell recognition. Protein Sci 4:1708-1717, 1995
26. Garboczi DN, Ghosh P, Utz U, Fan QR, Biddison WE, Wiley DC:
Structure of the complex between human T-cell receptor, viral
peptide and HLA-A2. Nature 384:134-141, 1996
27. Epp 0, Lattman EE, Schiffer M, Huber R, Palm W: The molecular structure of a dimer composed of the variable portions of the
Bence-Jones protein RE1 refined at 2.0-A resolution. Biochemistry 14:4943-4952, 1975
Rini JM, Stanfield RL, Stura EA, Salinas PA, Profy AT, Wilson
I A Crystal structure of a human immunodeficiency virus type 1
neutralizing antibody, 50.1, in complex with its V3 loop peptide
antigen. Proc Natl Acad Sci U S A 90:6325-6329, 1993
Stanfield RL, Takimoto-Kamimura M, Rini JM, Profy AT, Wilson
I A Major antigen-induced domain rearrangements in an antibody. Structure 1:83-93, 1993
Herron JN, He XM, Mason ML, Voss EW Jr, Edmundson AB:
Three-dimensional structure of a fluorescein-Fab complex crystallized in 2-methyl-2,4-pentanedioL Proteins 5:271-280, 1989
Garcia KC, Degano M, Stanfield RL, Brunmark A, Jackson MR,
Peterson PA, Teyton L, Wilson I A An alpha beta T cell receptor
structure at 2.5 angstrom and its orientation in the TCR-MHC
complex. Science 274:209-219, 1996
Chitarra V, Alzari PM, Bentley GA, Bhat TN, Eisele JL,
Houdusse A, Lescar J, Souchon H, Poljak RJ: Three-dimensional
structure of a heteroclitic antigen-antibody cross-reaction complex. Proc Natl Acad Sci U S A 90:7711-7715, 1993
Jeffrey PD, Schildbach JF, Chang CY, Kussie PH, Margolies MN,
Sheriff S: Structure and specificity of the anti-digoxin antibody
40-50. J Mol Biol 248:344-360, 1995
Ajitkumar P, Geier SS, Kesari KV, Borriello F, Nakagawa M,
Bluestone JA, Saper MA, Wiley DC, Nathenson SG: Evidence
that multiple residues on both the alpha-helices of the class I MHC
molecule are simultaneously recognized by the T cell receptor.
Cell 54:47-56, 1988
Percus JK, Percus OE, Perelson AS: Predicting the size of the
T-cell receptor and antibody combining region from consideration
of efficient self-nonself discrimination. Proc Natl Acad Sci U S A
9011691-1695, 1993
Peccoud J, Dellabona P, Allen P, Benoist C, Mathis D: Delineation of antigen contact residues on an MHC class I1 molecule.
EMBO J 9:4215-4223, 1990
Jorgensen JL, Reay PA, Ehrich EW, Davis MM: Molecular
components of T-cell recognition. Annu Rev Immunol 10:835873, 1992
Dellabona P, Peccoud J, Kappler J, Marrack P, Benoist C, Mathis
D: Superantigens interact with MHC class I1 molecules outside of
the antigen groove. Cell 62:1115-1121, 1990
Arden B, Clark SP, Kabelitz D, Mak TW: Human T-cell receptor
variable gene segment families. Immunogenetics 42:455-500, 1995
Struyk L, Hawes GE, Chatila MK, Breedveld FC, Kurnick JT, van
den Elsen PJ: T cell receptors in rheumatoid arthritis. Arthritis
Rheum 38:577-589, 1995
Waase I, Kayser C, Carlson PJ, Goronzy JJ, Weyand CM: Oligoclonal T cell proliferation in patients with rheumatoid arthritis and
their unaffected siblings. Arthritis Rheum 39:904-913, 1996
Wucherpfennig KW, Hafler D A A review of T-cell receptors in
multiple sclerosis: clonal expansion and persistence of human
T-cells specific for an immunodominant myelin basic protein
peptide. Ann N Y Acad Sci 756:241-258, 1995
Li Y, Sun GR, Tumang JR, Crow MK, Friedman SM: CDR3
sequence motifs shared by oligoclonal rheumatoid arthritis synovial T cells: evidence for an antigen-driven response. J Clin Invest
9412525-2531, 1994
Wucherpfennig KW, Hafler DA, Strominger JL: Structure of
human T-cell receptors specific for an immunodominant myelin
basic protein peptide: positioning of T-cell receptors on HLADRZipeptide complexes. Proc Natl Acad Sci U S A 92:8896-8900,
Hong SC, Chelouche A, Lin RH, Shaywitz D, Braunstein NS,
Glimcher L, Janeway CA Jr: An MHC interaction site maps to the
amino-terminal half of the T cell receptor alpha chain variable
domain. Cell 69:999-1009, 1992
46. Jorgensen JL, Esser U, Fazekas de St. Groth B, Reay PA, Davis
MM: Mapping T-cell receptor-peptide contacts by variant peptide
immunization of single-chain transgenics. Nature 355:224-230,
47. Ghosh P, Amaya M, Mellins E, Wiley DC: The structure of an
intermediate in class I1 MHC maturation: CLIP bound to HLADR3. Nature 378:457-462, 1995
48. Jardetzky TS, Brown JH, Gorga JC, Stern LJ, Urban RG,
Strominger JL, Wiley DC: Crystallographic analysis of endogenous peptides associated with HLA-DR1 suggests a common,
polyproline 11-like conformation for bound peptides. Proc Natl
Acad Sci U S A 93:734-738, 1996
49. Nalefski EA, Kasibhatla S, Rao A Functional analysis of the
antigen binding site on the T cell receptor alpha chain. J Exp Med
175:1553-1563, 1992
50. Lone YC, Bellio M, Prochnicka-Chalufour A, Ojcius DM, Boissel
N, Ottenhoff TH, Klausner RD, Abastado JP, Kourilsky P: Role of
the CDRl region of the TCR beta chain in the binding to purified
MHC-peptide complex. Int Immunol 6:1561-1565, 1994
51. Brawley JV, Concannon P: Modulation of promiscuous T cell
receptor recognition by mutagenesis of CDR2 residues. J Exp Med
183:2043-205 1, 1996
52. Reay PA, Kantor RM, Davis MM: Use of global amino acid
replacements to define the requirements for MHC binding and T
cell recognition of moth cytochrome c (93-103). J Immunol
152:3946-3957, 1994
53. Danska JS, Livingstone AM, Paragas V, Ishihara T, Fathman CG:
The presumptive CDR3 regions of both T cell receptor a and p
chains determine T cell specificity for myoglobin peptides. J Exp
Med 172:27-33, 1990
54. Geluk A, Fu XT, van Meijgaarden KE, Jansen YY, de Vries RR,
Karr RW, Ottenhoff TH: T cell receptor and peptide-contacting
residues in the HLA-DR17(3) beta 1 chain. Eur J Immunol
24:3241-3244, 1994
55. Hammer J, Gallazzi F, Bono E, Karr RW, Guenot J, Valsasnini P,
Nagy ZA, Sinigaglia F: Peptide binding specificity of HLA-DR4
molecules: correlation with rheumatoid arthritis association. J Exp
Med 181:1847-1855, 1995
56. Krieger JI, Karr RW, Grey HM, Yu W-Y, O’Sullivan D, Batovsky
L, Zheng Z-L, Coldn SM, Gaeta FCA, Sidney J, Albertson M, del
Guercio M-F, Chesnut RW, Sette A Single amino acid changes in
DR and antigen define residues crit cal for peptide-MHC binding
and T cell recognition. J Immunol 146:2331-2340, 1991
57. Yamanaka K, Kwok WW, Mickelson EM, Masewicz S, Nepom
GT: T-cell receptor V beta selectivity in T-cell clones alloreactive
to HLA-Dw14. Hum Immunol33:57-64, 1992
58. Howell MD, Diveley JP, Lundeen KA, Esty A, Winters ST, Carlo
DJ, Brostoff SW: Limited T-cell receptor beta-chain heterogeneity
among interleukin 2 receptor-positive synovial T cells suggests a
role for superantigen in rheumatoid arthritis. Proc Natl Acad Sci
U S A 88~10921-10925,1991
59. Paliard X, West SG, Lafferty JA, Clements JR, Kappler JW,
Marrack P, Kotzin B L Evidence for the effects of a superantigen
in rheumatoid arthritis. Science 253:325-329, 1991
60. Kovats S, Nepom GT, Coleman M, Nepom B, Kwok WW, Blum
JS: Deficient antigen-presenting cell function in multiple genetic
complementation groups of type I1 bare lymphocyte syndrome.
J Clin Invest 96:217-223, 1995
61. Fields BA, Malchiodi EL, Li HM, Ysern X, Stauffacher CV,
Schlievert PM, Karjalainen K, Mariuzza RA: Crystal structure of
a T-cell receptor beta-chain complexed with a superantigen.
Nature 384:188-192, 1996
Без категории
Размер файла
1 261 Кб
site, drb1, interactions, associates, molecular, epitopes, modeling, predict, cells, shared, specific, arthritis, receptorhla, 17804004, use, rheumatoid
Пожаловаться на содержимое документа