Proc. Nati. Acad. Sci. USA
Vol. 84, pp. 9150-9154, December 1987
Immunology
Structure and function of anti-DNA autoantibodies derived from a
single autoimmune mouse
MARK J. SHLOMCHIK*, ANN H. AUCOINt, DAVID S. PISETSKYt,
AND
MARTIN G. WEIGERT*
*The Institute for Cancer Research, Fox Chase Cancer Center, Philadelphia, PA 19111; and tDepartment of Medicine, Durham Veterans Administration
Hospital, Division of Rheumatic and Genetic Diseases and Immunology, Duke University Medical Center, Durham, NC 27705
Communicated by D. Bernard Amos, August 19, 1987
Downloaded from https://www.pnas.org by 50.214.187.246 on August 3, 2023 from IP address 50.214.187.246.
ABSTRACT
Four monoclonal anti-DNA antibodies derived from a single autoimmune MRL/lpr mouse were studied.
Three of these antibodies showed similarities in DNA binding;
the fourth had a much higher specific activity for singlestranded DNA and, in addition, was unique in binding doublestranded DNA and cardiolipin. Complete nucleotide sequences
of heavy- and light-chain variable regions demonstrated that all
four antibodies are clonally related. The sequences also showed
numerous somatic mutations, the distribution of which suggests that positive selection by antigen operated on these
clonally related autoantibodies.
chain isotype of all four antibodies, y3 K, was determined
using subclass-specific reagents (Litton Bionetics, Charleston, SC).
Specificity Analyses. The specificity of antibodies for DNA
antigens was tested as described (22). Antibodies to cardiolipin were assayed by ELISA using bovine heart cardiolipin
(Sigma) adhered to plates by evaporation of a solution of 100
,ug/ml in 95% ethanol. Results are reported as optical density
(OD) units. To determine the specific activity of antibodies
for DNA antigens, monoclonal antibodies were purified from
tissue culture supernatants by affinity chromatography on
columns containing rabbit anti-mouse IgG antibodies and
protein concentrations were determined by OD280. Antibodies were assayed over a range of protein values; the concentration of antibody producing an OD280 of0.5 was determined
as specific activity. Antibody specificity for double-stranded
DNA (dsDNA) was evaluated by staining of Crithidia luciliae
(23) (Kallestad Labs, Austin, TX), using fluorescein-conjugated anti-mouse IgG antiserum as a developing reagent.
Nucleotide Sequencing and DNA Blot Hybridization. These
procedures were carried out as described (24, 25).
Analysis of Mutations. We used a binomial probability
model to calculate whether the number of replacement (R)
mutations in the complementarity-determining regions (26)
(RCDR) was significantly higher than the number expected
randomly. Specifically, the probability of k RCDR [P(k)], given
n total mutations, with an expected value of p [P(k) n, p] was
determined. We assumed that V-region mutations occur at
random (ref. 27; K. Huppi, R. Kleinfield, S. Litwin, and
M.G.W., unpublished observations). Then p is the product of
the relative size of the CDRs (0.25 of the total V-region
length) and the fraction of random mutations that are of the
R type (0.75), or p = 0.19. The value n is the total number of
V-region mutations that occurred, which is the number of
observed RCDR Plus the number of silent (S) mutations plus
2 times the number of R mutations in framework regions
(RFR) [n = RCDR + S + (2RFR)1. The reason for multiplying
the number of RFR by 2 is that some RFR events will result in
deleterious mutations that will be lost from a clone of
antibody-producing cells and thus will not be among the
observed RFR. We estimated that half of all RFR events will
fall into this category as follows. The fraction of RFR events
that would be lost was determined by analyzing the variability of FR sites in the collection of mouse VK sequences
compiled by Kabat et al. (26). These sites fall into three
categories: invariant, conservative, or nonconservative. Invariant positions showed no variation among the sequences;
Antibodies to DNA (anti-DNA) are prominent autoantibodies in the sera of patients with systemic lupus erythematosus
(SLE) and of MRL/Mp-lpr/lpr (MRL/lpr) mice, a murine
model of SLE. The levels of these autoantibodies have
diagnostic and prognostic significance. Moreover, a direct
role of anti-DNA in disease pathogenesis has been established by correlation of antibody levels with disease activity
and identification of anti-DNA at sites of tissue injury (1-5).
Two competing models to explain anti-DNA production have
emerged. The first suggests that anti-DNA result from polyclonal B-lymphocyte activation and are the consequence of
antigen-nonspecific disturbances of B and/or T cells (6-8).
The extensive diversity of anti-DNA has been interpreted as
evidence for the polyclonal activation model. In addition,
reports of idiotypes expressed on a large fraction of antiDNA suggest that anti-DNA are encoded by germ-line
variable (V)-region genes (9-13). The second model proposes
that antigen stimulates anti-DNA production. Although nonspecific immune disturbances may exist in SLE patients, in
this model their role is to allow or promote induction of
autoantibodies by antigen (14-17). It is reasonable to postulate DNA as the inciting antigen, although a non-DNA self or
foreign antigen could fulfill this role (8, 17-19). To provide
further insight into the etiology and structure of anti-DNA,
we investigated the specificity and primary structure of
monoclonal anti-DNA obtained from spleen cells of an
autoimmune MRL/lpr mouse.t The goal was to determine
the genetic basis of anti-DNA diversity and thereby to
distinguish these models of anti-DNA production.
MATERIALS AND METHODS
Antibodies. Antibodies used in this study were derived
from the fusion of spleen cells from a 6-month-old MRL/lpr
mouse (obtained from The Jackson Laboratory, Bar Harbor,
ME) with the myeloma cell line NS1 as described (20).
Antibodies initially were identified by ELISA using as an
antigen source rabbit thymus extract (Pel-Freeze, Rogers,
AR) as previously described (21) and subsequently were
shown to be anti-DNA (see below). The heavy- and light-
Abbreviations: SLE, systemic lupus erythematosus; dsDNA, double-stranded DNA; ssDNA, single-stranded DNA; anti-DNA, antibody(ies) to DNA; CDR, complementarity-determining region; FR,
framework region; R, replacement (mutation); S, silent (mutation);
V, variable; D, diversity; J, joining; subscript H, heavy chain.
tThe heavy- and light-chain sequences reported in this paper are
being deposited in the EMBL/GenBank data base (Bolt, Beranek,
and Newman Laboratories, Cambridge, MA, and Eur. Mol. Biol.
Lab., Heidelberg) (accession nos. J03595 and J03596, respectively).
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
9150
Proc. Natl. Acad. Sci. USA 84 (1987)
Immunology: Shlomchik et al.
we assume that R mutations at these positions would not be
allowed and would be lost (a frequency of 0 for allowed R
mutations). Conservative positions were defined as those
which have only conservative amino acid substitutions according to the criteria of Grantham (28). For each conservative site, we calculated the fraction of single-base R mutations of the codons of the consensus amino acid that would
yield any one of the other observed (conservative) amino acid
substitutions; this was taken to be the frequency of allowed
R mutations. Nonconservative sites had at least one substitution that was nonconservative relative to the consensus
residue. For these positions, we assumed that all possible R
mutations would be allowed (a frequency of 1 for allowed R
mutations). An average of these frequencies for all FR sites
gives 0.48 as the overall frequency of allowable R mutations.
This frequency predicts that the ratio RFR/SFR would be
about 1.5 rather than the random value of about 3. (A survey
of 97 somatic FR mutations taken from a collection of
sequences of IgG anti-IgG and anti-influenza hemagglutinin
antibodies shows a ratio of 1.6, providing further basis for the
idea that half of RFR mutations are eliminated.)
Downloaded from https://www.pnas.org by 50.214.187.246 on August 3, 2023 from IP address 50.214.187.246.
RESULTS
Specificity of Monoclonal Antibodies. The specific binding
activities of the antibodies for single-stranded DNA (ssDNA)
were compared by determining the concentration of each
antibody that produced an OD value of 0.5 in the ELISA (Fig.
1). Antibody 3H9 has a specific activity for DNA 50- to
100-fold greater than the other antibodies in this group, which
themselves are indistinguishable. The characteristics of DNA
binding were assessed by determining the ability of synthetic
DNA and RNA homopolymers to inhibit the interaction of
these antibodies with ssDNA. Qualitatively, the inhibition
patterns for each of the antibodies were similar (data not
shown). Differences were seen only with poly(dC), which
inhibited 3H9 significantly better than other antibodies and
with poly(dU), which inhibited 2F2 slightly better and lA11
slightly worse than 4H8 and 3H9. These latter differences
were not sufficient to allow a statistically significant distinction of the four antibodies by polynucleotide inhibition
assays.
3H9 was unique among this set in displaying kinetoplast
binding in the Crithidia assay, an indication of anti-dsDNA
reactivity (23). The other antibodies tested at the same
0
OD
rl(
0
0
10
Protein conc.
1.0
0.1
(jig/mi)
FIG. 1. Binding of monoclonal antibodies to DNA and cardiolipin. Affinity-purified antibodies at various protein concentrations
were tested by ELISA for binding to ssDNA (solid lines) and
cardiolipin (broken lines). Results of representative determinations
are reported in terms of OD380, with each point the mean of triplicate
determinations. Calculated specific activities (protein concentration
required for OD380 = 0.5) were as follows: 3H9, 0.3 ,ug/ml; 2F2, 24
jug/ml; lA11, 28 j.g/ml; 4H8, 40 Ag/ml.
9151
protein concentration failed to display this immunofluorescence pattern. The interaction of 3H9 with dsDNA was
confirmed by ELISAs using both dsDNA and poly(dGdC)poly(dG-dC) as antigens (data not shown). The antibodies were also tested by ELISA for binding to cardiolipin,
since this property has been observed for some other antiDNA (17). As shown in Fig. 1, only 3H9 displayed significant
binding to cardiolipin. Thus, on the basis of several features
of reactivity, 3H9 was distinguishable from other antibodies
in this set.
Molecular Analysis of V Genes. The complete V-region
sequences of heavy- and light-chain mRNAs expressed by
the hybridomas (Figs. 2 and 3) were determined. Several
features of these sequences indicate that these four antibodies are descendants of a single B-cell precursor. First, the VH
and VK sequences are 98-100% homologous and all four cell
lines use the same heavy-chain diversity (DH) and joining
(JH2) and light-chain joining (JK3) gene segments. Second, and
most important for establishing clonal relatedness, all four
hybridomas share nucleotide sequence in the heavy chain at
the DH/JH and VH/DH junctions (Fig. 2). Because of germline DH diversity along with the flexibility of joining and de
novo nucleotide (N) addition (29), even antibodies with
indistinguishable specificity or amino acid sequence but of
independent origin have different nucleotide sequences in
this region (25, 30). The sequence identity in this region of
3H9, lA11, 2F2, and 4H8 thereby provides a strong argument
for their being the progeny of a single cell. Further, all four
share the same VK4/JK3 junctional nucleotide sequence.
DNA blot hybridization analysis of heavy- and light-chain
rearrangements (data not shown) showed that the restriction
fragment lengths of both the productive alleles and the
nonproductive alleles of the heavy-chain (H) locus are the
same (only the identical productive K chain allele is retained
in these cells). Rearrangements to form nonproductive genes
(H-, K-) occur independently of those forming the productive genes (H', K+). Hence, the contexts of H- and K- alleles
of B-lineage cells of independent origin rarely overlap, as has
been shown in extensive surveys of hybridomas and plasmacytomas (31). The identity of the H- restriction fragment
lengths of these cells establishes their clonal relatedness.
These antibodies are similar to other clonally related
examples (25, 30) in that (i) because of somatic mutations
they are not all identical and (ii) among the mutations, there
is a hierarchy regarding the extent to which they are shared
by more than one member of the clone. Some mutations are
found only in a single member. At the next level are mutations
shared by 2F2 and 4H8 but not the others. Next are a series
of nucleotide substitutions shared by lA11, 2F2, and 4H8 but
not 3H9. Also, some differences from the original germ-line
genes are probably shared by all examples, but these can only
be identified through comparison to the germ-line donor
sequences.
These germ-line VH and VK sequences are not available,
nor are the JH and JK sequences of the MRL strain. In lieu of
this information we have compared these expressed V-region
sequences to "consensus" sequences derived from examples
in which the same V gene or gene segment is expressed in
lines of independent origin. Such consensus sequences correspond to the germ-line sequence except for sites at which
the same mutation has occurred independently. Since independent, parallel mutations are rare even in antibodies of
similar specificities, consensus sequences are usually a
reliable representation of a germ-line sequence. One such
consensus sequence can be derived for the VK region, since
the same VK gene is rearranged and expressed in a non-DNAbinding MRL/lpr hybridoma, 2H9, which is of independent
clonal origin. We believe this is the case because of sequence
similarity (99% homology, Fig. 3) and because the restriction
fragment lengths of these Ka alleles are consistent with the
9152
Proc. Natl. Acad. Sci. USA 84 (1987)
Immunology: Shlomchik et al.
FR I
10
20
20
40
70
60
50
Gin Val Gln Leu Gin Gin Ser Gly Pro Glu Leu Val Lys Pro Gly Ala Ser Val Lys Ile Ser Cys Lys Ala Ser
CAG GTT CAG CTG CAG CAG TCT GGA CCT GAG CTG GTG AAG CCT GGG GCC TCA GTG AAG ATT TCC TGC AAG GCT TCT
--A .-.__
lAl 1
2F2
3H9
4H8
---
---
--A
---
___--
---
---
-
-
-
-
-
---
---
---
---
---
---
-
-
-
f R2
CDR
1 00
-
---
---
---
---
---
---
-
---
-__
-__T_ -__
-
-
-
-
CDR2
1 10
1 20
1 30
1 40
1 SO
220
230
240
I1 O
I
Gly Tyr Ala Phe Ser Ser Ser Trp Met Asn Trp Val Lys Gln Arg Pro Gly Lys Gly Leu Clu Trp Ile Gly Arg lie Tyr Pro Ciy Asp
TGG ATG AAC TGG GTG AAG CAG AGG CCT GGA AAG GGT CTT GAG TGG ATT GGA CGG ATT TAT CCT GGA GAT
GGC
___ TAT GCA TTC AGT AGC
--_T TCC
--A-- ___
___
-_T --90
80
---
---
___
---
---
_
---
_
---
_
_
-
_
FR3
1 70
1
2 11 0
90
22 S 0
Cly Asp Thr Thr Asn Asn Gly Lys Phe Lys Asp Lys Ala Thr LLeu Thr Ala Asp Lys Ser Ser Ser Thr Ala Tyr Met Gln Leeu Ser Ser
GGA GAT ACT ACT AAC AAT GGG AAG TTC AAG GAC AAG GCC ACA CCTG ACT GCA GAC AAA TCC TCC AGC ACA GCC TAC ATG CAA CTC AGC AGC
-AC T-- --.
___ --- - T- -AT-- --- --____-_
---
-AC T--
---
-_
---
CDR3
290
280
270
260
310
300
f R4
330
320
3 41 0
Leu Thr Ser Glt Asp Ser Ala Val Tyr Phe Cys Ala Arg Ala AArg Ser Lys Tyr Ser Tyr Val Met Asp Tyr Trp Gly Gln Gly Thr
jAGG AGT AAA TAT TCC TAT GTT ATG GAC TAC TGG GGT CAA GGG ACC
CTG ACA TCT GAG GAC TCT GCG GTC TAC TTC TGT GCA AGA GCG
---_
---_ --C --- --- --- T__ ___ __T .-___
---
---
---_
___
---
_
_
-_
_
---
__-
-_--C
--
---
---
---
---
T-- --- --T
.--
Downloaded from https://www.pnas.org by 50.214.187.246 on August 3, 2023 from IP address 50.214.187.246.
FIG. 2. Nucleotide sequences of heavy-chain V (VH) regions of anti-DNA hybridomas. All sequences are compared to lAll; identities are
indicated by dashes. As lAll is not the germ-line sequence, see text and Fig. 4 for interpretation of germ-line and mutant nucleotides. The
translation of lAll is given above the nucleotide sequences. The start of each of the domains defined by Kabat et al. (26) is demarcated by a
vertical line and designated either FR (for framework region) or CDR (for complementarity-determining region) followed by the appropriate
number. Sequence from 342 to 353 is not shown. Al sequences in this region are identical to the BALB/c JH2 germ-line sequence (26).
cell having rearranged the same VK gene (data not shown). A
second consensus sequence can be derived for JH, a JH we
presume based on homology considerations to be the MRL
allele of BALB/c JH2, since a closely related JH sequence is
also expressed in 2H9 (data not shown) and other MRL
heavy-chain sequences (16). From comparisons to these
2F2/4H8. All of the anti-DNA differ from 2H9 by a Ser--Cys
substitution in CDR3, but here we cannot be certain whether
this mutation occurred in the anti-DNA or 2H9. Since
cysteine is extremely rare at this site but serine is found in two
independent genes from this VK homology group, it is likely
that serine is the germ-line-encoded residue (ref. 27; M.J.S.
conclude that the amino acid
substitutions threonine, asparagine, and isoleucine in the VK
CDR1 are the result of somatic mutations shared by lA11 and
and M.G.W., unpublished data).
A genealogy (Fig. 4) shows the nature and reconstructs an
order of mutations that occurred during the growth of this
consensus sequences, we
FR I
CDRI
20
10
30
40
60
so
Gltu Asn Val Leu Thr Cin Ser Pro Ala lie Met Ala Ala Ser Pr Gly Gtu Lys Val Thr Met Thr Cys Ser Thr
GAA AAT GTG CTG ACC CAG TCT CCA GCA ATC ATG GCT GCA TCT CCA GGG GAG AAG GTC ACC ATG ACC TGC AGT ACC
A-___
--- G-o
lAl
1
2F2
3H9
4H8
2H9
---__
__-
- --
--
-- -
__--_
---
---__
__-
- --
-
-
---
A-
-
___
---
___
---
---
---
---
---
---
---
---
---
---
---
G--
FR2
J0
60
1 00
CDR2
11
0
1 20
1 40
13
s0
1
1
60
Asn Ser Ser Ile Ser Ser Gly Asn Phe His Trp Tyr Gin Gin Lys Pro Gil Thr Ser Pro Lys Leu Trp lie Tyr Arg Thr Ser Asn Leu
AAC TCA AGT ATA AGT TCT GGT AAC TTT CAC TGG TAC CAG CAG AAG CCA GGC ACT TCT CCC AAA CTC TGG ATT TAT AGG ACA TCC AAC CTG
-_
-__
-G - --- --- G-- --- --- --- --- --- --- -__
___
___
___
-G -
---
---
--- ---
---
G--
---
---
---
---
--- --- --- ---
---
---
--- ---
-__
_
_
_
-----
-
-
---
---
---
---
---
---
-
-
-
-
-
-
-
-
-
-
_
___
---
__
___
---
___
_
-
-
-
-
_
_
_
-
-
-
-
-
-
-
-
-
-
-
-__
___
---
---
---
---
---
---
___
---
---
---
---
---
---
___
---
---
---
---
---
---
_
_
_
-
-
-
-
-
-
-
-
-
-
FR3
1 70
1 0
1 90
200
210
230
220
240
250
Ala Ser Gly Val Pro Ala Arg Phe Ser Gly Ser Cly Ser Cly Thr Ser Tyr Ser Leu Thr lie Ser Ser Met Glu Ala Glu Asp Ala Ala
GCT TCT GGA GTC CCC GCT CGC TTC AGT GGC AGT GGG TCT GGG ACC TCT TAC TCT CTT ACA ATC AGC AGC ATG GAG GCC GAA GAT GCT GCC
---
-
-
---
--_
CDR3
270
260
-
-
FR4
280
290
300
310
320
Thr Tyr Tyr Cys Gin Gin Trp Cys Gly Tyr Pro Phe Thr Phe Gly Thr Gly Thr Lys Leu Ctsu le Lys
ACT TAT TAC TGC CAG CAG TGG TGT GGT TAC CCA TTC ACG TTC GGC ACG GGG ACA AAA TTG GAA ATA AAA
.A_
---------__
--G
___
---
___
---
___
---
---
---
_--
---
---
---
---
---
---
---
-__
___
---
---
---
---
---
---
---
---
---
-__
---
---
---
---
---
---
---
A--
-AT
---
--- --- -- G
---
--- -- A
---
---
---
---
T--
---
--C --G G--
FIG. 3. Nucleotide sequences of VK regions of anti-DNA hybridomas. Sequences are presented as in Fig. 2. Below the group offour anti-DNA
sequences is the sequence of 2H9, a non-DNA-binding hybridoma of separate clonal origin. This sequence is 99% identical to the anti-DNA
sequences in the V region. The differences at the 3' end of CDR3 and in FR4 can be accounted for if 2H9 uses JK2 and the anti-DNA hybridomas
use JK3 (numbering does not count the pseudo-JK gene). The 2H9 sequence was used to derive a consensus sequence to represent a germ-line
VK gene expressed by all of these hybridomas (see text).
Immunology: Shlornchik et al.
Proc. Natl. Acad. Sci. USA 84 (1987)
T-C(H-C3)
C-T(H-C3)
C-T (H-C2)
T-C(H-Cl)
Alo-Thr(L-Fl)
<-Met (H-0)1
i
3H)
\ ys-Thr
(H-C2)
Tyr-Asn(H-C2)
Val-Ile(L-CI)
Ser-Asn(L-CI)
Ala-Thr(L-CI),
Gly-Arg(H-C2)
Thr-Ile (H-C2)
Ala-0Vl(H-fI)
-I
-A
TSer-Cys(L-C3)
Downloaded from https://www.pnas.org by 50.214.187.246 on August 3, 2023 from IP address 50.214.187.246.
Vh J558/ VK 4
FIG. 4. Genealogy relating the anti-DNA hybridomas. This
genealogy was constructed from the pattern of shared and unique
mutations, assuming that shared mutations represent single events
and not independent parallel mutations. The lengths of branches are
roughly proportional to the number of mutations. Mutations along
each branch are given in the direction of germ-line to mutant;
replacement mutations are shown as amino acid changes and silent
mutations as nucleotide changes. The chain (H, heavy; L, light) and
domain [either F (framework) or C (complementarity-determining)
along with the appropriate number] are indicated in parentheses. The
consensus sequence derived from 2H9 was used to assign the
direction of mutations that occurred in the light chain prior to the
branching of lA11 from 2F2/4H8 (see text for details). The gap in the
first stem indicates that mutations that occurred in the heavy chain
before the first branch would not be recognized; however, the
consensus sequence allows identification of such mutations in the
light chain (e.g., the Ser-*Cys change). "VhJ558/VK4" indicates that
the precursor of this clone had productively rearranged members of
the VhJ558 family and VK4 group.
clone. The direction of mutations beyond points of divergence is inferred from the hierarchy of shared mutations; the
opposite direction of mutation would require postulating
multiple independent occurrences of the same mutation in all
of the proximal branches. This reasoning does not apply to
mutations prior to branch points; assigning the direction of
these mutations would require knowing the (undetermined)
germ-line sequences, but the direction of mutations could
nonetheless-be reasonably interpreted for the VK mutations
by using the consensus sequence. The placement of the four
VH mutations that occurred prior to the branch point is less
certain in the absence of a consensus sequence. Our placing
them in the 3H9 branch is based only on the fact that the
putative germ-line residue is in each case the predominant
residue at that site among collections of sequences from other
homologous genes in this VH family, whereas the putative
mutant residue is either never or rarely found (32). It is
important to note that, regardless of the direction of mutations, their type (R or S) and their location relative to branch
points are unaffected.
DISCUSSION
The derivation of the anti-DNA from this mouse from a single
precursor provides a unique opportunity to interpret the
genetics and structure of these antibodies. In particular, from
the analysis of the pattern of mutations, we can distinguish
whether antigen or polyclonal activation was the stimulus for
division of the B cells in this clone. Further, by correlating
somatically acquired differences in amino acid sequences
with binding properties, we can infer the basis of functional
differences between 3H9 and the other members of the clone.
Finally, as these differences-increased binding capacity and
9153
specificity for dsDNA-are potentially relevant to pathogenesis by anti-DNA, they provide insight into the role of
somatic mutation and selection in the process of autoimmunity.
Somatic Mutation and Clonal Evolution. The types of
mutations found in a clone of B cells reflect the nature of
selective forces operating on that clone during its growth.
Somatic mutations are thought to occur randomly within V
regions, as evidenced by the random distribution of S and
noncoding mutations found in a large number of V region
sequences. Although the result of a random event, the
expression of R-type mutations could result in a selective
advantage or disadvantage to cells that harbor them, depending on how the R mutation affects the immunoglobulin
receptor interaction with its ligand (antigen). Selection would
be mediated by preferential expansion (positive selection) or
failure to divide (negative selection) of cells containing
certain types of mutations, resulting after time in a population
of antigen-specific cells with highly nonrandom distributions
of R mutations. For example, R mutations in the antigen
combining site (CDR) that improve affinity for antigen would
be positively selected, but mutations that destroy immunoglobulin protein folding (usually in FR) would be selected
against. Positive selection by antigen would be detected by a
significantly higher frequency of RCDR events than expected
randomly. On the other hand, polyclonal activation-a receptor-independent event-would not predict positive selection (i.e., it would predict a random pattern of RCDR).
That all four of these anti-DNA antibodies have a high
percentage of RCDR (lAll, 100%; 3H9, 40o; and 2F2/4H8,
50%) suggests that positive selection operated on this clone.
Using a binomial probability model to assess the significance
of this trend (see Materials and Methods), we obtained the
following p values for finding the number of observed RECDR
at random: lA11, 0.000006; 3H9, 0.076; and 2F2/4H8, 0.034.
These small p values for all four lines are indicative of
positive selection operating on the clone. This conclusion
was confirmed by performing the same analysis on the
mutations in toto (counting each mutation once only), which
gave a p value of 0.007.
These anti-DNA antibodies are similar to a variety of
foreign antigen-stimulated B-cell clones that also have patterns of mutation suggesting positive selection (25, 30). They
are also similar in this regard to a number of IgG anti-IgG
autoantibodies from lpr mice (16). Evidence for positive
selection in two different kinds of autoantibody specificities
may mean that antigen-driven clonal expansion is a common
feature of autoantibodies from autoimmune mice.
Effect of Mutation on Specificity. The enhanced binding of
3H9 must be due to the CDR replacements that distinguish
this antibody from lA11 and 2F2/4H8. (The possibility that
enhanced binding is the result of constant-region deletion or
insertion mutations has been ruled out by NaDodSO4/PAGE
analysis of purified proteins; data not shown). One of these,
the replacement of glycine with arginine, is an excellent
candidate, since arginine can have strong interactions with
DNA. As described by Seeman et al. (33), this interaction
involves the formation, through the guanidinium group of
arginine, of two hydrogen bonds with either phosphate,
cytosine (in ssDNA only), or guanine (in both ssDNA or
dsDNA). The ability to form two additional hydrogen bonds
with one of these moieties should lead to a significant
increase in affinity for DNA and could in part account for the
100-fold difference in specific activity of 3H9 seen in the
direct binding assay (Fig. 1).
The unique features of the 3H9 fine-specificity profile are
in agreement with the interactions that are predicted between
arginine and DNA. These include binding to cardiolipin,
which contains phosphodiester bonds, to dsDNA, and probably enhanced binding to poly(dC). Direct binding of 3H9 to
Downloaded from https://www.pnas.org by 50.214.187.246 on August 3, 2023 from IP address 50.214.187.246.
9154
Immunology: Shlomchik et al.
cardiolipin suggests that the extra arginine may contribute to
3H9 DNA binding through interaction with the DNA backbone. This acquisition of specificity for both the base
cytosine and the phosphodiester linkage is a paradox, since
in a combining site the extra arginine could contact only one
or the other moiety. However, an interesting explanation for
this paradox is that both interactions occur and the bivalency
of IgG allows the two combining sites to simultaneously
participate in different kinds of interactions with DNA.
Conclusion. This study reveals a number of important
features of anti-DNA and potential mechanisms for the
generation of autoantibody specificities characteristic of
SLE. First, it establishes that anti-DNA can contain mutations and strongly suggests that these mutations are selected
for by antigen. The nonrandom pattern of mutations is
incompatible with a model by which anti-DNA arise as a
result of immunoglobulin receptor-independent ("polyclonal") activation. Second, it shows that antibody avidity and
specificity for DNA can be altered by somatic mutation.
Third, 3H9, in which enhanced ssDNA and dsDNA binding
are the result of one or a few amino acid differences vs. the
other members of the clone, provides a paradigm for explaining the origin of dsDNA antibodies. That is, these could arise
by somatic mutation (34) [with the dsDNA binding phenotype
resulting from either enhanced capacity for ssDNA binding
and/or the acquisition of a unique second specificity for
dsDNA (35)]. In principle, this mechanism for creating
anti-dsDNA could operate in normal as well as autoimmune
situations, since clonal expansion and mutation occur in
normals. However, anti-dsDNA are not generally observed
in sera of normal individuals (35, 36). The propensity of SLE
patients to express anti-dsDNA may be the result of a greater
extent of clonal expansion found in autoimmune individuals.
Indeed, we have observed extensive clonal expansion in
studying rheumatoid factor (RE) autoantibodies from autoimmune mice (16). Most of the RFs in a single mouse derive
from a single precursor. Since these RFs comprised 10-20%
of all hybridomas obtained from individual mice, they must
have resulted from an unusual amount of clonal expansion.
Antibodies to DNA and other autoantigefts may derive from
a similar amount of clonal expansion; in the mouse studied
here at least some of the.anti-DNA were also derived from a
clonally expanded set. As clonal expansion, mutation, and
selection are connected (25), the linkage of anti-dsDNA and
high-avidity anti-ssDNA antibodies to autoimmune situations might relate to greater opportunities for generating a
mutant with these characteristics.
We thank Anita Cywinski for outstanding technical work and
Donna Hnosko, Annmarie Shepherd, and Gloria Szymanski for
expert preparation of the manuscript. This work was supported by
grants from the National Institutes of Health to M.G.W. and D.S.P.
and an appropriation from the Commonwealth of Pennsylvania.
M.J.S. is a trainee of the Medical Scientist Training Program at the
University of Pennsylvania.
1. Tan, E. M., Schur, P. H., Carr, R. I. & Kunkel, H. G. (1966)
J. Clin. Invest. 45, 1732-1740.
2. Tan, E. M. (1982) Adv. Immunol. 33, 167-240.
3. Dixon, F. J., Oldstone, M. A. & Tonietti, G. (1971) J. Exp.
Med. 134, 65s.
4. Yoshida, H., Kohno, A., Ohta, K., Hirose, S., Maruyama, N.
Proc. Natl. Acad. Sci. USA 84
(1987)
& Shirai, T. (1981) J. Immunol. 127, 433-437.
5. Morimoto, C., Sano, H., Abe, T., Homma, M. & Steinberg, A.
(1982) J. Immunol. 129, 1960-1965.
6. Hang, L., Slack, J. H., Amundson, C., Izui, S., Theofilopoulos,
A. N. & Dixon, F. J. (1983) J. Exp. Med. 157, 874-883.
7. Theofilopoulos, A. N., Shawler, D. L., Eisenberg, R. A. &
Dixon, F. J. (1980) J. Exp. Med. 151, 446-466.
8. Schwartz, R. S. & Stollar, B. D. (1985) J. Clin. Invest. 75,
321-327.
9. Marion, T. N., Lawton, A. R., Kearney, J. F. & Briles, D. E.
(1982) J. Immunol. 128, 668-674.
10. Rauch, J., Murphy, E., Roths, J. B., Stollar, B. D. & Schwartz,
R. S. (1982) J. Immunol. 129, 236-241.
11. Solomon, G., Schiffenbauer, J., Keiser, H. D. & Diamond, B.
(1982) Proc. Nati. Acad. Sci. USA 80, 850-854.
12. Shoenfeld, Y., Isenberg, D. A., Rauch, J., Madaio, M.,
Stollar, B. D. & Schwartz, R. S. (1983) J. Exp. Med. 158,
718-730.
13. Andrzejewski, C., Rauch, J., Lafer, E., Stdllar, B. D. &
Schwartz, R. S. (1981)1J. Immunol. 126, 226-231.
14. Shores, E. A., Eisenberg, R. A. & Cohen, P. L. (1986) J.
Immunol. 136, 3662-3667.
15. Rappard-Van Der Veen, F. M. (1984) J. Immunol. 132, 18141819.
16. Shlomchik, M. J., Marshak-Rothstein, A., Wolfowicz, C. B.,
Rothstein, T. & Weigert, M. G. (1987) Nature (London) 328,
805-811.
17. Lafer, E. M., Rauch, J., Andrzejewski, C., Mudd, D., Furie,
B., Schwartz, R. S. & Stollar, B. D. (1981) J. Exp. Med. 153,
897-909.
18. Carrol, P., Stafford, D., Schwartz, R. S. & Stollar, B. D.
(1985) J. Immunol. 135, 1086-1090.
19. Emlen, W., Pisetsky, D. S. & Taylor, R. P. (1986) Arthritis
Rheum. 29, 1417-1426.
20. Warren, R. W., Sailstad, D. M., Caster, S. A. & Pisetsky,
D. S. (1984) Arthritis Rheum. 27, 545-551.
21. Darwin, B. S., Grudier, J. P., Klatt, C. L. & Pisetsky, D. S.
(1986) J. Immunol. 137, 3796-3801.
22. Pisetsky, D. S. & Peters, D. V. (1982) J. Immunol. Methods
41, 187-200.
23. Swaak, A. J. G., Groenwold, J., Aarden, L. A. & Feltkamp,
T. E. W. (1981) Ann. Rheum. Dis. 40, 45-49.
24. Shlomchik, M. J., Nemazee, D. A., Sato, V., Van Snick, J.,
Carson, D. & Weigert, M. G. (1986) J. Exp. Med. 164, 407427.
25. Clarke, S., Huppi, K., Ruezinsky, D., Staudt, L., Gerhard, W.
& Weigert, M. G. (1985) J. Exp. Med. 161, 687-704.
26. Kabat, E., Wu, T., Bilofsky, H., Reid-Miller, M. & Perry, H.
(1983) Sequences of Proteins of Immunological Interest (U.S.
Government Printing Office, Bethesda, MD).
27. Gearhart, P. J. & Bogenhagen, D. (1983) Proc. Nati. Acad.
Sci. USA 80, 3439-3443.
28. Grantham, R. (1974) Science 185, 862-864.
29. Alt, F. W. & Baltimore, D. (1982) Proc. Nati. Acad. Sci. USA
79, 4118-4122.
30. Caton, A. J., Brownlee, G. G., Staudt, L. M. & Gerhard, W.
(1986) EMBO J. 5, 1577-1587.
31. Coleclough, C., Perry, R. P., Karialainen, K. & Weigert,
M. G. (1981) Nature (London) 290, 372-378.
32. Dildrop, R. (1984) Immunology Today 5, 85-86.
33. Seeman, N. C., Rosenberg, J. M. & Rich, A. (1976) Proc.
Natl. Acad. Sci. USA 73, 804-808.
34. Diamond, B. & Scharff, M. D. (1984) Proc. Nati. Acad. Sci.
USA 81, 5841-5844.
35. Arana, R. & Seligmann, M. (1967) J. Clin. Invest. 46, 18671882.
36. Notman, D. D., Kurata, N. & Tan, E. M. (1975) Ann. Intern.
Med. 83, 464-469.