Download Is structural flexibility of antigen-binding loops

International Immunology, Vol. 9, No. 5, pp. 771–777
© 1997 Oxford University Press
Is structural flexibility of antigen-binding
loops involved in the affinity maturation of
anti-DNA antibodies?
Satoru Miyazaki1, Junko Shimura1,2, Sachiko Hirose2, Reiko Sanokawa2,
Hiromichi Tsurui2, Midori Wakiya2, Hideaki Sugawara1 and Toshikazu Shirai2
1Life
Science Research Information Division, The Institute of Physical and Chemical Research (Riken), 2-1
Hirosawa, Wako, Saitama 351-01, Japan
2Department of Pathology, Juntendo University School of Medicine, 2-1-1 Bunkyo-ku, Hongo, Tokyo 113,
Japan
Keywords: anti-DNA antibodies, autoantibodies, autoimmunity, Ig V gene somatic mutation, Ig V region
gene, molecular dynamics simulation, systemic lupus erythematosus
Abstract
Effects of somatic mutations in Ig variable region genes on the affinity maturation of
autoantibodies were investigated using single precursor B cell-derived anti-double-stranded DNA
mAb generated from an autoimmune disease-prone (NZB H NZW)F1 mouse. Analyses of DNA
sequences, homology modeling on a graphic computer and molecular dynamics simulation of
antigen-binding sites showed that any single site of mutation and changes in the electrostatic or
hydrogen-bonding potential of the residues and in the three-dimensional structure could not solely
explain the difference in DNA-binding activities. However, a significant increase in the flexibility of
antigen-binding Fv loops, particularly VL CDR1 and VH CDR3, was associated with affinitymaturated anti-DNA antibodies. Such high flexibility of the Fv loops may provide the environment
where the antibodies could effectively interact with antigen DNA, a model consistent with the
‘induced-fit’ hypothesis of antigen–antibody interactions.
Introduction
The affinity maturation of B cells involves a high rate of
somatic point mutations in rearranged Ig variable (V) region
genes during the process of proliferation and differentiation
including class-switching (1–6). However, the correlation
between structural changes in the antigen-binding site due
to somatic mutations in V region genes and changes in
antigen-binding activities of antibodies remain unknown.
Based on sequence analyses of V genes in specific acquired
immune responses to foreign antigens, somatic hypermutations were found to occur mainly in complementarity-determining regions (CDR) of V genes during the process of affinity
maturation (1,3,4). In contrast, studies of our own as well as
those of other workers with naturally occurring anti-DNA
autoantibodies in systemic lupus erythematosus (SLE)-prone
mice showed that mutations in V region genes frequently
occur in both CDR and framework regions (FR) (5–11). In our
preceding studies, because replacement mutations occurred
at random, and differed in numbers and positions in CDR
and FR, we could not correlate changes between primary
nucleotide sequences and DNA-binding activities of the antibodies (5,6).
Studies using X-ray crystallography have demonstrated
that structures of variable and constant domains of Ig of
different sources are highly conserved and that FR of these
domains are essentially superimposable (12–14). Thus, the
effect of point mutations in V region genes on conformational
changes in Ig does not seem to be so large as previously
assumed. However, as the antigen–antibody interaction is
dynamic, further analyses on changes in molecular dynamics
of the antigen-binding site are necessary to better understand
the molecular basis of affinity maturation of antibodies. In this
regard, studies using molecular dynamics simulation on a
graphic computer through a conformational homology search
and refinement with energy minimization procedures provided
reliable models of Ig V domains (15,16). Based on this
approach, we analyzed the effects of somatic hypermutations
in V region genes on the affinity maturation of autoantibodies,
using single precursor B cell-derived monoclonal anti-DNA
Correspondence to: T. Shirai
Transmitting editor: J. Berzofsky
Received 26 August 1996, accepted 3 February 1997
772 Affinity maturation and Fv flexibility
antibody clones with a distinct double-stranded DNA
(dsDNA)-binding activity.
Methods
Hybridomas
NZB and NZW strains of mice, originally obtained from Japan
SLC (Shizuoka, Japan), were mated to produce
(NZB3NZW)F1 hybrid mice. Spleen cells from a 6-month-old
female mouse were fused to the non-secreting myeloma cell
line P3-X63-Ag8.653, as described (5). Culture supernatants
with binding activities to DNA were screened using ELISA
(5). Positive clones were subsequently subcloned at least
twice by limiting dilution.
DNA sequencing
Total cellular RNA was isolated from ~107 hybridoma cells as
described in our previous study (5), and 1 µg of RNA was
used for each VH and VL cDNA amplification. The first stranded
cDNA was synthesized using 200 U of reverse transcriptase
and 20 pmol of the C primer (for Cγ, 59-GGOCAGTGGATAGAC-39 and for Cκ, 59-GCTCACTGGATGGTGGGAAGATG39). Single-stranded cDNA was then amplified using 100 pmol
each of 59 primer (for VH, 59-AGGT(C/G)(A/C)A(A/G)CTGCAG(G/C)AGTC(A/T)GG-39 or for Vκ, 59-GAAATTGTGCT
(G/C)ACCCA(G/A)(T/A)CIC(C/A)A-39) and the corresponding
C primer. The primers were synthesized using a Gene
Assembler (Pharmacia, Uppsala, Sweden). The PCR mixtures
containing 30 mM Tris–HCl (pH 8.3), 5 mM MgCl2, 0.01%
gelatin, 0.2 mM of each primer, 2.5 U Tag polymerase (Takara
Shuzou, Kyoto, Japan) and 0.1– 1 µg of cDNA were prepared,
and amplification was carried out using a DNA thermal cycler
(Perkin-Elmer-Cetus, Norwalk, CT) for 30 cycles. Each cycle
consisted of denaturation at 94°C for 30 s, annealing at 55°C
for 30 s and polymerase reaction at 72°C for 1 min. The
length of the polymerase reaction was increased at each
cycle by 15 s. PCR products of ~400 bp were isolated, using
agarose gel electrophoresis, and then cloned in the pUC18
vector. Cloned and selected PCR products were sequenced
using the dideoxy chain termination method and Taq dye
primer cycle sequencing kits (Applied Biosystems, Foster
City, CA), according to the manufacturer’s instruction. At least
three independent clones of each hybridoma were sequenced
to preclude misincorporated nucleotides during PCR.
Measurement of DNA binding activities using DNA-coupled
silica gel particles and flow cytometry
To determine changes in DNA-binding activities of anti-DNA
mAb, a new method using oligonucleotide-coupled silica
gel particles and flow cytometry was used, as described
elsewhere (6). Briefly, diol-silica, 5 µm in diameter, was
activated with tresyl chloride and linked with Aminolink 2
(Applied Biosystems)-labeled oligonucleotide (59-GTGTTATAGAAATTTGATATGGAG-39). Double-stranded oligonucleotide
was produced by incubating the gel particles with the complementary synthetic strand. For the measurement of DNAbinding activity of anti-DNA mAb, 53105 dsDNA-coupled
silica gel particles were incubated with 30 µl of culture
supernatant, with a known Ig concentration at room temper-
ature for 1 h with mild shaking. After three washings with PBS
(pH 7.5) supplemented with 1% BSA and 0.05% Tween 20,
particles were incubated with 30 µl of appropriately diluted
FITC-labeled goat anti-mouse Ig for 30 min at room temperature, washed twice and resuspended with 400 µl of PBS
containing 0.2% BSA. The fluorescence intensity of DNAcoupled silica gel particles was then measured using a
FACStar (Becton Dickinson, Mountain View, CA) in logarithmic
amplification. The channel number showing the mean fluorescence intensity on the FACS profile was regarded as a
value of the potential affinity of anti-dsDNA mAb.
Model building and molecular dynamics simulation
The computer graphic models of VH and VL chains of antidsDNA mAb were built, using the graphics program Insight
II and Homology (Version 2.3.0, Biosym Technologies, San
Diego, CA), on a Silicon Graphics (Indigo2) computer system.
The general strategy for the model building was as follows:
(i) selecting a reference molecule, (ii) loop searching and
loop replacement, when insertion is required, (iii) checking
chirality, and (iv) energy minimization (15,16). As a reference
graphic model for the building of VH and VL, we used the
coordinates of the reported mAb BV04-01, which is reactive
with single-stranded DNA (14), because of high sequence
homology to V region genes of our clonally related antidsDNA mAb. After refinement by 1000 energy minimization
steps, the potential energy of all models was reduced to
less than –8000 kcal/mol. Molecular dynamics of the threedimensional Fv structure models of Ig was performed for 50
ps at the temperature of 298 K and at 1 atm. The coordinates
of atomic positions in the dynamics trajectory were obtained
using the program Discover (Version 2.3.0, Biosym Technologies), programmed based on the report of Dauber-Osguthorpe et al. (17). To make initial loop conformations natural,
the lowest energy conformations were taken. Cut-off distance
for non-bonded pair interactions was set 10 Å. The extent of
the protein motions seen in the simulations was assessed by
calculating the fluctuations of root mean square of atomic
positions in the dynamics trajectory.
Results
V gene mutations and DNA-binding activities in clonally
related anti-DNA hybridomas
We cloned 12 anti-dsDNA hybridomas (five IgM and seven
IgG2b) from a single, 6-month-old (NZB3NZW)F1 mouse and
sequenced V regions of heavy (VH) and light chain (VL) genes
of each mAb. Among these, five IgG2b clones, BW9-7,
BW9-8, BW9-9, BW9-11 and BW9-13, were thought to have
originated from a single precursor, for two reasons: (i) these
mAb commonly used VH J558 and JH3 as the VH region gene,
and Vκ21 and Jκ2 as the VL gene, and (ii) they had identical
or related junctional areas in CDR3, including potential N
sequences in VH genes.
Because of the lack of information on germ-line sequences
of these anti-DNA clones, the nucleotide sequences of both
VH and VL genes in each mAb were aligned with the consensus
sequence deduced from VH and VL sequences of individual
mAb to determine the site and number of somatic mutations
Affinity maturation and Fv flexibility 773
Fig. 1. Nucleotide and deduced amino acid sequences for heavy (a) and light chain (b) V regions of clonally related anti-DNA mAb BW9-7,
BW9-8, BW9-9, BW9-11 and BW9-13. These sequences are aligned with the consensus precursor sequence (for details, see text). Dashes
indicate identity with the consensus sequence deduced from VH and VL sequences of individual mAb. The sequence of 59 end (8 bases) of
the 59 primer for PCR is not shown. Amino acid sequences of CDR are boxed according to the definition of Kabat et al. (26).
in the V regions. The consensus sequence may not necessarily
represent the exact germline sequence. However, because
members of a clone must express the same genes, any
sequence differences between them are considered to be
caused by a somatic mutation. As shown in Fig. 1, these five
clones had different numbers of somatic mutations, with or
without replacement mutations throughout CDR and FR. Thus,
it was evident that somatic mutations in V region genes
accumulate even after the IgM to IgG class switching of antiDNA antibodies.
Comparisons of dsDNA-binding activities among mAb were
made using 24mer synthetic dsDNA-coupled silica gel par-
ticles as substrate and a fluorescence-activated cell sorter.
The difference of DNA-binding activity was based on the
difference in mean peak channels of fluorescence intensity.
As shown in Fig. 2, each mAb showed a considerable
difference in dsDNA-binding activity. It was of note that
compared to BW9-7, four other clones, BW9-8, BW9-9, BW911 and BW9-13, showed a substantially higher dsDNA-binding
activity. However, there were no common mutational positions
which could account for the increased dsDNA-binding activities in higher affinity clones, compared to the activity of BW97. Using the neighbor joining method with Kimura’s distance
(18), the clones, BW9-7 and BW9-9, were geneologically
774 Affinity maturation and Fv flexibility
Fig. 2. Comparisons of dsDNA-binding activities between clonally
related IgG2b anti-DNA mAb. dsDNA-binding activities were
examined using synthetic dsDNA-coupled silica gel particles and
flow cytometry, as described in Methods. Each dsDNA-coupled
particle was first stained with culture supernatant of mAb at given Ig
concentrations and then stained with FITC-labeled anti-mouse Ig
antibodies. dsDNA-binding activities were expressed by mean
channels of fluorescence intensity.
suggested to be the closest pair (6) (data not shown).
Nonetheless, BW9-9 showed a considerably higher dsDNAbinding activity than did BW9-7.
Three-dimensional structure of antigen-binding sites of
clonally related anti-DNA antibodies
To determine whether changes in the three-dimensional structure of antigen-binding sites contribute to the affinity maturation of anti-DNA clones, we prepared three-dimensional
structure models of the antigen-binding site consisting of Nterminal domains of VL and VH (Fv). Each Fv model was
constructed by coordinating the reported VH and VL template
structures of the reference anti-DNA mAb BV04-01 (14), which
has a highly homologous sequence to our anti-DNA mAb,
followed by refinement, using energy minimization (15,16).
Figure 3 illustrates the model structure of Fv fragments of
BW9-7 as a representative model structure of five clones.
The attached data shows the root mean square values of
deviations between the reference structure, BV04-01, and
each three-dimensional model of BW9-7, BW9-8, BW9-9,
BW9-11 and BW9-13, which were within the range of 2.15–
2.54 Å. In contrast, the root mean square of deviations of
backbone atoms between these models of single-cell-derived
anti-DNA mAb were within the range of 0.07–0.89 Å (Table 1).
Thus, three-dimensional structures of antigen-binding sites,
including the antigen-binding groove, were not so different
and may not explain the difference in DNA-binding activities
between these anti-DNA clones.
Changes in flexibility of CDR loops in anti-DNA antibodies
with different affinities
According to the ‘induced fit’ hypothesis of the antigen–
antibody interaction, it is possible that the difference in
antigen-binding activity of anti-DNA mAb clones in the present
Fig. 3. A representative three-dimensional structure model of anti-DNA
mAb (BW9-7) refined using energy minimization. FR are displayed in
trace lines and CDR in ribbons. The attached data indicates root
mean square values of deviations between the reference structure
(BV04-01) and each three-dimensional model (BW9-7, BW9-8, BW99, BW9-11 and BW9-13).
Table 1. Root mean square of deviations (Å) of backbone
atoms between models
Clone
BW9-7
BW9-8
BW9-9
BW9-11
BW9-8
BW9-9
BW9-11
BW9-13
0.32
0.31
0.31
0.49
0.80
0.80
0.87
0.07
0.89
0.41
study may be due to differences in their flexibility of antigenbinding loops. To investigate differences in flexibility of antigen-binding loops between low-affinity and high-affinity
clones, molecular dynamics simulations were performed. The
extent of protein motions seen in the simulations was assessed
by calculating root mean square fluctuations of atomic positions in the molecular dynamics trajectory. Coordinates of all
atoms of Fv structure models including each of three CDR
loops and four FR in VH and VL were obtained at the interval
of 500 fs during 50 ps molecular dynamics, and deviations
of atomic positions among 100 simulated structure (snapshots
of the simulation), expressed by root mean square, were
Affinity maturation and Fv flexibility 775
Fig. 4. Comparison of the flexibility of Fv loops during molecular
dynamics, as determined by cluster graphs. Coordinates of all atoms
were obtained at the time interval of 500 fs during the simulation
period of 50 ps, each set of coordinates was saved as a frame and
100 frames were stored in chronological order. The extent of protein
motions seen in the simulations was estimated by calculating the root
mean square fluctuations of all atomic positions including side chains
among 100 frames. Results are displayed in cluster graph, in which
ranges of root mean square values of 0.00–1.00 Å are shown in the
black area and those of .1.00 Å are in the white area.
plotted in the cluster graphs shown in Fig. 4. Deviations of
atomic positions among simulated structures of 1.0 Å of root
mean square value or more were regarded as a high extent
of vibration (white areas in Fig. 4) and those less than 1.0 Å
were regarded as a low extent of vibration (black areas). As
shown by narrower black areas in Fig. 4, Fv structures of the
higher affinity clones, BW9-8, BW9-9, BW9-11 and BW9-13,
were more dynamic, thus more flexible than that of the lowest
affinity, BW9-7.
Comparisons of animated frames of all six CDR loops in
VH and VL of BW9-7 showed that the VH CDR3 was the most
vibration-limited loop (Fig. 5a). We then compared the extent
of vibration of VH CDR3 among five clones and found that
such vibration of BW9-7 was most restrained (Fig. 5). Because
amino acid sequences of VH CDR3 of BW9-7 were identical
to those of BW9-9, the observed limited vibration of BW9-7
VH CDR3 may be due to interaction with other loops located
close to VH CDR3. Thus, we compared the distance from VH
CDR3 to each VL CDR loop between BW9-7 and BW9-9. As
shown in Fig. 6, while the distances between the α carbon of
Lys102 in VH CDR3 and either one of the α carbons of Asn57
in VL CDR2 or that of Arg96 in VL CDR3 did not differ between
the two clones (Fig. 6b and c), the distance between VH
CDR3 and VL CDR1 differed between them, in which while
the distance between the α carbons of Lys102 in VH CDR3
and of Asp33 in VL CDR1 of BW9-7 ranged from 10.7 to 16.1
Å, the distance between Lys102 and Gly33 of BW9-9 ranged
from 8.6 to 17.0 Å (Fig. 6a). Thus, while the maximum movable
distance in the low-affinity clone BW9-7 was 5.4 Å, the
distance in the high-affinity BW9-9 was 8.4 Å, suggesting that
VL CDR1 in the low-affinity clone BW9-7 restrains the flexibility
of VH CDR3 and that certain amino acid changes in VL CDR1
are responsible for the difference in such Fv loop flexibilities
between the two clones.
Fig. 5. Comparison of flexibility between Fv structures of five clones
was expressed by the difference in width of ribbon trace. Ribbon
width corresponded to the extent of vibration in the structure during
the molecular dynamics. Note that the ribbon width of VH CDR3 in
the low-affinity clone BW9-7 is narrower than those in high-affinity
clones. The width of this loop correlates well with the increase in
DNA-binding activity.
Discussion
In the present study, we investigated the effects of somatic
mutations in Ig V region genes on the affinity maturation of
anti-DNA antibodies, using single precursor B cell-derived
anti-DNA antibody clones with a distinct DNA-binding activity.
The results suggested that somatic mutations in Ig V region
genes affected the flexibility of antigen-binding Fv loops, and
that there was correlation between an increase in flexibility of
Fv loops and affinity maturation of anti-DNA antibodies.
It has been reported that a common and characteristic
feature of anti-DNA mAb from SLE-prone mice is the appearance of the cationic amino acid residue Arg in VH (7–11), and
the process of both selection and somatic mutation may be
involved in the enrichment of Arg in VH CDR3 (8). Arg is the
most versatile amino acid for DNA binding, in that it can form
ionic bonds with the negatively charged phosphodiester
backbone of DNA and also donate up to five hydrogen bonds.
Thus, it is suggested that the increase in the number of Arg
in the antigen-binding site contributes to the increase in
affinity of antibodies to DNA. This is indeed the case with the
highest affinity clone BW9-11, which had three repeated Arg
in the junction of VH FR3 and CDR3, in contrast to two Arg in
the same region in the lowest affinity clone BW9-7 (Fig. 1).
However, such was not necessarily the case for other higher
affinity clones BW9-8, BW9-9 and BW9-13, i.e. these three
clones had no additional substitutions to Arg in CDR, compared to the lowest affinity clone, BW9-7.
In addition to Arg, Asn and Lys also contribute to DNA
binding (10). Compared to the presence of two Asn in CDR
of BW9-7, higher affinity clones BW9-8, BW9-11 and BW9-13
had three, four and five Asn respectively. However, another
high-affinity clone BW9-9 had only one Asn. As for Lys, the
776 Affinity maturation and Fv flexibility
Fig. 6. Comparison of the transition of distance between VH CDR3
and VL CDR loops over the period of molecular dynamics. The
cationic amino acid residue located in each loop was selected as an
effective point for the antigen-binding site and the distance between
α carbons of the selected amino acids was measured within the
trajectory and plotted against the frame. The selected positions of
amino acids on each loop are as follows: VL CDR2, Lys57; VL CDR3,
Arg96; VH CDR3, Lys102. Since the VL CDR1 lacks cationic residues
at the top portion of the loop, Asp33 or Gly33 was taken to measure
the distances. (a) Comparison of distance between VL CDR1 and VH
CDR3. (b) Comparison of distance between VL CDR2 and VH CDR3.
(c) Comparison of distance between VL CDR3 and VH CDR3. Solid
line indicates the results of BW9-7 and dotted line BW9-9.
number in CDR of low-affinity clone BW9-7 was the largest of
all the five clones (Fig. 1). Substitutions of certain amino acids
to negatively charged Glu and/or Asp may be related to the
low DNA-binding activity. The lowest affinity clone BW9-7
contained Asp33 instead of Gly33 in VL CDR1. However, the
higher affinity clones BW9-8, BW9-9 and BW9-13 had an
additional Glu56 in VH CDR2, Glu27 in VL CDR1 and Asp56
in VH CDR2 respectively. Taken collectively, the amino acid
substitutions to Arg, Asn, Lys, Glu and Asp in CDR loops do
not simply explain the difference in DNA-binding activities of
anti-dsDNA antibodies in the present study.
Both the lock-and-key (19) and induced fit (20–25) type
models have been used to describe antibody–antigen recognition. Wilson et al. (21) analyzed the shape of the antigenbinding site, based on studies of a number of Fab complexes
with antigens ranging from small haptens to proteins. Although
total surface areas buried by various antigens were often
similar, the surface contours seen by the antigen differed and
the difference in the two conformational forms of the Fab
(liganded and unliganded) illustrated the induced fit of an
antibody to an antigen (20–25). Based on these hypotheses,
we examined the relationship between flexibilities of antigenbinding Fv loops, as determined using molecular dynamics
simulations and DNA-binding activities of clonally related anti-
DNA mAb. Our findings were: (i) Fv structures of higher
affinity clones were more dynamic, thus more flexible than
the structure of the lowest affinity clone BW9-7, and (ii) the
flexibility of VH CDR3 and VL CDR1 seems to correspond to
the affinity change. Molecular dynamics simulations in the
present studies were carried out in vacuo. Further studies
using simulations in the presence of a 5 Å layer of water are
ongoing in our laboratories and our preliminary results are in
principle consistent with the observations in vacuo.
As the low-affinity BW9-7 and the high-affinity BW9-9 originated from a single precursor B cell, and as amino acid
sequences of VH CDR3 of BW9-7 and BW9-9 were identical,
certain somatic replacement mutations which occurred in VL
CDR1 must be responsible for the observed difference in the
loop flexibility. There are three amino acid changes in VL
CDR1 between BW9-7 and BW9-9, one Lys27 versus Glu27,
one Asp33 versus Gly33 and one Asn35 versus Ser35 (Fig.
1). There are at least two plausible interpretations for the
observed difference in flexibilities of VL CDR1 and VH CDR3
between BW9-7 and BW9-9. First, changes in the loop flexibility may be caused by the occurrence of certain amino acids.
The amino acid substitution to Gly may increase the loop
flexibility, because Gly residues have more conformational
freedom than any other amino acids. The side chain of
Asn frequently hydrogen bonds to the main chain, thereby
stabilizing the local conformation. The substitutions from Gly33
and Ser35 in BW9-9 to Asp33 and Asn35 in BW9-7 may
decrease the flexibility of VL CDR1 in BW9-7. Such restricted
flexibility of VL CDR1 in BW9-7 may also be related to the
observed decrease in flexibility of VH CDR3.
Alternatively, the electrostatic interaction between the negatively charged Asp33 in VL CDR1 and the cationic residue
Lys102 in VH CDR3 in BW9-7, which is located close to Asp33
in VL CDR1, may form the basis of the restricted flexibility of
VL CDR1 and VH CDR3. In BW9-9, this restriction may be
released because of Asp33 to Gly33 substitution in VL CDR1.
Although there is additional substitution from Lys27 to negatively charged Glu27 in VL CDR1 of BW9-9, the site is not
close enough to interact with positively charged Lys102 in
VH CDR3.
Taken collectively, our results suggest that in addition to
the reported importance of cationic amino acid residues in
Fv loops, the loop flexibilities may also influence the DNAbinding reaction of antibodies. The high flexibility of VH CDR3
and VL CDR1 loops may provide the environment where
electrophilic amino acid residues along the antigen-binding
groove of antibody could effectively interact with antigen. The
Fv loop flexibility model we have described may be consistent
with the ‘induced fit’ hypothesis of antibody–antigen interactions (20–25).
Acknowledgements
We thank M. Ohara for the critical comments, and M. Tsukamoto and
T. Okumura for the repeated computations.
Abbreviations
CDR
FR
complementarity-determining regions
framework regions
Affinity maturation and Fv flexibility 777
SLE
VH
VL
systemic lupus erythematosus
heavy chain variable region
light chain variable region
14
References
1 Berek, C. and Milstein, C. 1987. Mutation drift and repertoire shift
in the maturation of the immune response. Immunol. Rev. 96:23.
2 Shlomchik, M. J., Marshak-Rothstein, A., Wolfowicz, C. B.,
Rothstein, T. L. and Weigert, M. G. 1987. The role of clonal
selection and somatic mutation in autoimmunity. Nature 328:805.
3 Allen, D., Cumano, A., Dildrop, R., Kocks, C., Rajewsky, K.,
Rajewsky, N., Roes, J., Sablitzky, F. and Siekevitz, M. 1987.
Timing, genetic requirements and functional consequences of
somatic hypermutation during B-cell development. Immunol.
Rev. 96:5.
4 Jacob, J., Kelsoe, K., Rajewsky, K. and Weiss, U. 1991. Intraclonal
generation of antibody mutants in germinal centers. Nature
354:389.
5 Taki, S., Hirose, S., Kinoshita, K., Nishimura, H., Shimamura, T.,
Hamuro, J. and Shirai, T. 1992. Somatically mutated IgG antiDNA antibody clonally related to germ-line encoded IgM antiDNA antibody. Eur. J. Immunol. 22:987.
6 Hirose, S., Wakiya, M., Kawano-Nishi, Y., Yi, J., Sanokawa, R.,
Taki, S., Shimamura, T., Kishimoto, T., Tsurui, H., Nishimura, H.
and Shirai, T. 1993. Somatic diversification and affinity maturation
of IgM and IgG anti-DNA antibodies in murine lupus. Eur. J.
Immunol. 23:2813.
7 Marion, T. N., Tillman, D. M., Jou, N.-T. and Hill, R. J. 1992.
Selection of immunoglobulin variable regions in autoimmunity to
DNA. Immunol. Rev. 128:123.
8 Shlomchick, M., Moscelli, M., Shan, H., Radic, M. Z., Pisetsky, D.,
Marshak-Rothstein, A. and Weigert, M. 1990. Anti-DNA antibodies
from autoimmune mice arise by clonal expansion and somatic
mutation. J. Exp. Med. 171: 265.
9 Eilat, D., Webster, D. M. and Rees, A. R. 1988. V region sequences
of anti-DNA and anti-RNA auto-antibodies from NZB/NZW F1
mice. J. Immunol. 141:1745.
10 Radic, M. Z. and Weigert, M. 1994. Genetic and structural
evidence for antigen selection of anti-DNA antibodies. Annu. Rev.
Immunol. 12:487.
11 O’Keefe, T. L., Bandyopadhyay, S., Datta, S. K. and ImanishiKari, T. 1990. V region sequences of an idiotypically connected
family of pathogenic anti-DNA autoantibodies. J. Immunol.
144:275.
12 Padlan, E. A., Silverton, E. W., Sheriff, S., Cohen, G. H., SmithGill, S. J. and Davies, D. R. 1989. Structure of an antibody–
antigen complex: crystal structure of the HyHEL-10 Fab–lysozyme
complex. Proc. Natl Acad. Sci. USA 86:5938.
13 Chothia C., Lesk, A. M., Tramontano, A., Levitt, M., Smith-Gill, S.
15
16
17
18
19
20
21
22
23
24
25
26
J., Air, G., Sheriff, S., Padlan, E. A., Davies, D., Tulip, W. R.,
Colman, P. M., Spinelli, S., Alzari, P. M. and Poljak, R. J. 1989.
Conformations of immunoglobulin hypervariable regions. Nature
342:877.
Herron, J. N., He, X. M., Ballard, D. W., Blier, P. R., Pace, P. E.,
Bothwell, A. L. and Voss, E. W., Jr. 1991. An autoantibody
to single-stranded DNA: comparison of the three-dimensional
structures of the unliganded Fab and a deoxynucleotide–Fab
complex. Proteins: Struct. Funct. Genet. 11:159.
Wang, D., Hubbard, J. M. and Kabat, E. A. 1993. Modeling study
of antibody combining site to (alpha1–6)dextrans. Predictions of
the conformational contribution of VL-CDR3 and Jκ segments to
groove-type combining sites. J. Biol. Chem. 268:20584.
Mas, M. T., Smith, K. C., Yarmush, D. L., Aisaka, K. and Fine, R.
M. 1992. Modeling the anti-CEA antibody combining site by
homology and conformational search. Proteins: Struct. Funct.
Genet. 14:483.
Dauber-Osguthorpe, P., et al. 1988. A. Structure and energetics
of ligand binding to proteins: E. coli dihydrofolate reductase–
trimethoprim, a drug–receptor system. Proteins: Struct. Funct.
Genet. 4: 31.
Saito, N. and Nei, M. 1987. The neighbor-joining method: a
new method for reconstructing phylogenetic trees. Mol. Biol.
Evol. 4:406.
Amit, A. G., Mariuzza, R. A., Phillips, S. E. and Poljak, R. J. 1986.
Three-dimensional structure of an antigen–antibody complex at
2.8 A resolution. Science 233:747.
Colman, P. M., Tulip, W. R., Varghese, J. N., Tulloch, P. A., Baker,
A. T., Laver, W. G., Air, G. M. and Webster, R. G. 1989. Threedimensional structures of influenza virus neuraminidase–antibody
complexes. Phil. Trans. R. Soc. London [Biol.] 323:511.
Wilson, I. A., Stanfield, R. L., Rini, J. M., Arevalo, J. H., SchulzeGahmen, U., Fremont, D. H. and Stura, E. A. 1991. Structural
aspects of antibodies and antibody–antigen complexes. Catalytic
antibodies. Ciba Foundation Symp. 159:13.
Colman, P. M., Laver, W. G., Varghese, J. N., Baker, A. T., Tulloch,
P. A., Air, G. M. and Webster, R. G. 1987. Three-dimensional
structure of a complex of antibody with influenza virus
neuraminidase. Nature 326:358.
Rini, J. M., Schulze-Gahmen, U. and Wilson, I. A. 1992. Structural
evidence for induced fit as a mechanism for antibody–antigen
recognition. Science 225:959.
Colman, P. M. 1988. Structures of antibody–antigen complexes:
implications for immune recognition. Adv. Immunol. 43:99.
Sheriff, S., Silverton, E. W., Padlan, E. A., Cohen, G. H., SmithGill, S. J., Finzel, B. C. and Davies, D. R. 1987. Three-dimensional
structure of an antibody–antigen complex. Proc. Natl Acad. Sci.
USA 84:8075.
Kabat, E. A., Tai, T. W., Perry, H. M., Gottesman, K. S. and Foller,
C. 1991. Sequences of Proteins of Immunological Interest, 5th
edn. US Government Printing Office, Bethesda, MD.