Analysis of anticentromere autoantibodies using cloned autoantigen

Proc. Natl. Acad. Sci. USA
Vol. 84, pp. 4979-4983, July 1987
Immunology
Analysis of anticentromere autoantibodies using cloned
autoantigen CENP-B
WILLIAM C. EARNSHAW*, PAULA S. MACHLINt, BONNIE J. BORDWELLt, NAOMI F. ROTHFIELD*,
AND DON W. CLEVELANDt
Departments of *Cell Biology and Anatomy, and tBiological Chemistry, The Johns Hopkins University School of Medicine, 725 N. Wolfe Street, Baltimore,
MD 21205; and tDivision of Rhetzmatic Diseases, Department of Medicine, University of Connecticut Health Center, Farmington, CT 60305
Communicated by John W. Littlefield, March 16, 1987 (received for review January 16, 1987)
A cDNA clone encoding CENP-B, the 80-kDa
ABSTRACT
human centromere autoantigen, was used to construct a panel
of hybrid proteins containing four different regions of CENPB. These have allowed us to identify three independent epitopes
on CENP-B that are targets of autoantibodies. Two of these are
recognized concurrently in -90% of patient sera containing
anticentromere autoantibodies (ACA), conclusively demonstrating that this autoimmune response is polyclonal. When
present and previous data are combined, ACA are shown to
recognize at least five independent epitopes on CENP-B. A
radioimmunoassay based on cloned CENP-B has demonstrated
that sera from 296% of patients with ACA recognize the
cloned antigen, thus defining a region of the protein that is
recognized by virtually all patients with ACA. These rmdings
have significant implications for models that seek to explain the
origin of ACA and for the future detection of this group of
autoantibodies in the clinical setting.
The rheumatic diseases are characterized by the production
of autoantibodies directed against nuclear and cytoplasmic
autoantigens (reviewed in refs. 1-8). The reasons for
autoantibody expression are generally unknown, and many
theories seeking to explain the phenomenon are currently
under consideration. In particular, it is not known if the
autoimmune response results from a classical antigen-driven
immunization or is a result of aberrations of the mechanisms
that normally control the immune system.
We have chosen the anticentromere autoantibody (ACA)
response for study since this involves the production of
high-titer, high-affinity autoantibodies that recognize protein
antigens. ACA were discovered in 1980, when it was found
that certain patients with the calcinosis/Raynauds phenomenon/esophogeal dysmotility/sclerodactyly/telangiectasiae
(CREST) variant of scleroderma produce autoantibodies that
recognize the centromere region of chromosomes (9-11).
Though ACA are closely associated with the CREST syndrome, the only clinical finding common to all ACA+ individuals is Raynauds phenomenon (12).
Our prior immunoblotting analysis revealed that >96% of
a test group of 39 ACA+ sera recognized three chromosomal
polypeptides of 17 kDa (CENP-A), 80 kDa (CENP-B), and
140 kDa (CENP-C) (refs. 12 and 13; see also refs. 14-19).
Antibodies affinity purified from CENP-B cross-reacted with
CENPs A and C, indicating that these antigens are structur
ally related (13). Antibodies to CENP-B are present at high
titer in all ACA+ sera examined, whereas the titer of
antibodies to CENPs A and C is occasionally lower (12).
We describe below a detailed examination of the binding of
ACA to subdomains of CENP-B that have been cloned and
expressed in bacteria. Our experiments reveal that the
autoimmune response against this protein is multifocal: as
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.
many as five distinct determinants are recognized. Contrary
to our prior expectations, the data suggest that ACA arise
from a specific polyclonal immune response directed against
centromeres.
MATERIALS AND METHODS
General Methods. All cloning methods and procedures
have been described in detail elsewhere (20). NaDodSO4/
PAGE was performed using the method of Lewis and
Laemmli (21). Electrophoretic transfer of proteins to nitrocellulose (22) was performed at 340 mA for 6 hr at 4°C. The
immunoblotting protocol has been described (12), as has the
method for affinity purification of antibodies from nitrocellulose strips (13). The antibody-blocking experiments were
described in ref. 20.
RIA. Fusion protein granules were isolated by differential
centrifugation from induced (23, 24) cultures of lysogen
X-CENP-B1 (20, 36). Granules were then solubilized in hot
urea and dialyzed into 10 mM Tris-HCl, pH 7.7/50 mM
NaCl/2 mM EDTA. The yield was 1 mg of Cte,,nCENP-B[,8gal] per liter of bacterial culture. CtlmCENP-B[p-gal] in 10
mM imidazole buffer was adsorbed to microtiter plates (0.1
,ug per well; Removawell strips, Immulon) that were then
probed with a 1:500 dilution of patient serum followed by
1251I-labeled protein A; radioactivity was determined in a y
counter as described (25). All assays were performed in
triplicate and sera were used in random order.
RESULTS
Molecular Cloning of CENP-B. We previously obtained a
series of overlapping cDNA clones corresponding to =95% of
the mRNA encoding CENP-B (20). The availability of the
clones permitted us to produce the following series of
chimeric proteins as fusions with the bacterial TrpE protein
(using the pATH plasmid series of expression vectors, gift of
T. J. Koerner, Duke University). With the exception of
CtermCENP-B[,8-ga1] (described below), all chimeric proteins
used in the studies presented here were TrpE fusions. (The
relative sizes and locations of these proteins are presented in
Fig. 5.)
CtermCENP-B[13-gal] consists of the amino-terminal 113kDa portion of P-galactosidase linked to the carboxylterminal 147 amino acids of CENP-B. This hybrid gene was
carried by bacteria lysogenic for the recombinant X bacteriophage originally detected by autoantibody (20). CtemCENPB[,B-gal] was used to elicit production of two monoclonal
ACA, m-ACA1 and m-ACA2, which recognize two nonoverlapping determinants on CENP-B (20).
Abbreviations: ACA, anticentromere autoantibody(ies); CREST,
calcinosis/Raynauds phenomenon/esophogeal dysmotility/sclerodactyly/telangiectasiae.
4979
4980
Immunology: Earnshaw et al.
Proc. Natl. Acad. Sci. USA 84
CtermCENP-B consists of the 147 carboxyl-terminal amino
acid residues of cloned CENP-B fused to the bacterial TrpE
protein. CtermCENP-BL and CternCENP-BR were produced
by further subdividing the human portion of CtermCENP-B
into segments of 104 amino-terminal and 43 carboxyl-terminal amino acid residues, respectively, and expressing these as
TrpE fusion proteins.
NproxCENP-B is comprised of TrpE linked to 347 amino
acid residues from the amino-terminal region of CENP-B
(20). (The cDNA encoding the -50 amino-terminal residues
has not yet been isolated.) The human portion of the protein
encoded by this clone is separated in the CENP-B sequence
by 98 amino acid residues from that encoded by CtermCENPB (Fig. 5).
Localization of Centromere Epitope 1 (CE1), a Major
Autoepitope in the Carboxyl-Terminal Region of CENP-B. To
identify the epitope(s) in CENP-B that are targets for autoantibodies, we have examined the interactions of a panel of
37 ACA+ patient sera and 3 ACA- control sera with the
cloned CENP-B polypeptide. [All patient sera were previously characterized by indirect immunofluorescence and
immunoblotting against the proteins of isolated mitotic chromosomes (12).] Remarkably, all 37 ACA' sera bound significantly to CtermCENP-B (Fig. 1A). Strong binding was evident for 86% (32 sera) and weaker binding characterized the
remaining 14% (5 sera). Control sera (lanes 12, 18, and 40) did
not bind the cloned antigen.
In our previous studies of cloned CENP-B, we isolated a
murine monoclonal antibody, m-ACA1, that recognizes a site
(CE1) present in CtermCENP-B (20). We now show that the
site recognized by m-ACA1 overlaps a major autoepitope on
CENP-B. Nitrocellulose blots containing CtermCENP-B were
cut into 40 strips, each ofwhich was incubated overnight with
a 1:50 dilution of human serum. Each strip was then probed
with m-ACA1, the binding of which was detected with
125I-labeled goat anti-mouse IgG (Fig. 1D). Pretreatment of
the blot strips with any of the 37 ACA+ sera blocked the
binding of m-ACA1 to some extent, with 19 sera showing
.80% blocking (Fig. 1D). The ACA+ sera least effective at
blocking the binding of m-ACA1 (sera 8, 20, and 33) were also
3 of the poorest at binding to CtermCENP-B (Fig. LA),
suggesting that the variability of blocking of m-ACA1 may be
largely due to variations in the titer of the anti-CtermCENP-B
autoantibodies.
CE2, a Second Autoepitope in the Carboxyl-Terminal Region
of CENP-B. A minority of the panel of patient sera recognizes
(1987)
at least one other epitope (CE2) in CterrCENP-B (identified
by immunoblotting of bacterial lysates expressing Cterm
CENP-BL, Fig. 1C). Sera 6 and 9 recognized this hybrid
protein (and a proteolytic fragment) strongly, whereas weaker binding was exhibited by sera 3, 13, 19, 24, 32, and 38
(making 22% in all).
The location of CE2 was defined by a second murine
monoclonal antibody, m-ACA2, previously shown to bind to
CtermCENP-BL, (20). Three patient sera showed significant
blocking of the binding of m-ACA2 (Fig. 1E, lanes 6, 9, and
19). The other five sera that bound to CtermCENP-BL in
immunoblots failed to block the binding of m-ACA2 to a
significant degree, perhaps due to low titers of anti-CE2
antibodies.
Even though all ACA' sera bound to CtermCENP-B, only
a handful (22%) recognized CtemCENP-BL and none recognized CterCENP-BR (not shown). Thus, most ACA' sera
apparently recognize only one epitope in CtermCENP-B.
CE3, an Epitope Present on N-Proximal CENP-B. To determine whether additional autoantibody binding sites occur
outside the carboxyl-terminal region of CENP-B, we examined the binding of the panel of 40 sera to NprOXCENP-B in
immunoblots (Fig. 1B). Ninety percent (33) of the ACA' sera
showed significant binding, although the intensity was much
more variable than that observed with Cte~rmCENP-B (compare Fig. 1 A and B). Thus, a third autoepitope(s), CE3, is
localized within the amino-terminal 60% of CENP-B. The
precise location and number of epitopes recognized in this
region are not known.
Titer of ACA. We have measured the titers of antibodies
against chromosomal CENPs A, B, and C and against two
regions of cloned CENP-B for one serum (KG). The titer of
antibodies specific for each antigen was determined by
probing parallel blot strips of chromosomal proteins with
serial dilutions of the patient serum (Fig. 2). Positive signals
were obtained in overnight exposures for the following serum
dilutions: CENP-A, 1:1,638,400; CENP-B, >1:3,276,800;
and CENP-C, 1:12,800 (lanes 1-9). Thus, though the titers of
antibodies against CENPs A and B are comparable, antibodies to CENP-C are less abundant (roughly by a factor of 200).
(Note, however, that the three chromosomal antigens may be
present in differing amounts.) When the serum was titered
against cloned CtrmCENP-B, a positive signal was obtained
at an antibody dilution of 1:4,096,000 (Fig. 2, lane 14). The
titer of antibodies recognizing NproxCENP-B was also high,
with a positive signal being observed at an antibody dilution
A
5
10
0.
15
20
see as
25
30
35
easeo so *a a0a es
x
a
40
a
FIG. 1. Binding of
B
e
0
-*,
*
18, and 40, indicated by a *) to various
of cloned CENP-B. (A-C) Binding of the sera to blots containing
CtermCENP-B (A), NPrOXCENP-B (B),
and Ctr,,CENP-BL (C) (see Fig. 5 for a
^
^
@
| a *@ *" "
*2*
C
s
A|
X
S||
Xll
§
"
"
C emCENP-BL
D
5
10
20
25
30
15
35
40
>
*
*
_ *
_
_*
m-ACA1
E
X*
..5
panel of 37
regions
NproxCENP-B
i
a
ACA' sera and 3 control sera (lanes 12,
CermCENP-B
me.
***||*
eee
._
00._
polypeptides).
diagram
case a nitrocellulose
strip cut from a blot
of the appropriate bacterial lysate was
incubated with the indicated patient serum. Blots were processed for antibody
detection
as described
(12).
Only the
is shown.
region of antibody
binding
(D
and E) Binding of the autoantibodies to
sites on CtermCENP-B defined by monoclonal antibodies m-ACA1 (D) and mACA2 (E). In these blocking experiments,
a
strong signal
serum
m-ACA2
is
seen
when
the
* does
*human
** not recognize the site
(i.e., does not block the binding of the
monoclonal antibody).
Immunology:
CENP-C
Earnshaw et al.
Proc. Natl. Acad. Sci. USA 84 (1987)
.
_
CENP-B'.
CENP-B * 0e
CENP-B.
-
-
I
10 11121314
W
CENP-A@
e e
1 2 34 5678
9 15 161718
FIG. 2. Titer of autoantiserum KG against chromosomal antigens
and cloned CENP-B. Lanes 1-8, mitotic chromosomes isolated from
HeLa cells probed with the following dilutions of antibody: 1:800,
1:3200, 1:6400, 1:12,800, 1:102,400, 1:204,800, 1:409,600, and
1:819,200. Lane 9, Coomassie blue-stained gel. Lane 10, Coomassie
blue-stained gel of Ct,,mCENP-B. Lanes 11-14, parallel nitrocellulose strips probed with antibody dilutions 1:64,000, 1:512,000,
1:2,048,000, and 1:4,096,000. Lane 15, Coomassie blue-stained gel of
NproxCENP-B. Lanes 16-18, parallel nitrocellulose strips probed
with the following antibody dilutions: 1:128,000, 1:256,000, and
1:512,000.
of 1:512,000 (Fig. 2, lane 18). The extraordinary titers
measured in these experiments are substantially greater than
the 10- to 30-fold increase observed subsequent to a nonspecific polyclonal lymphocyte activation (26, 27), suggesting
that ACA do not arise as a result of generalized derepression
of the immune system.
Development of a Solid-Phase Binding Assay to Detect ACA.
We wished to determine whether a solid-phase binding assay
using cloned CENP-B might be suitable for detection of ACA
in the clinic. We therefore purified CternCENP-B[f-gal] to
>40% homogeneity (Fig. 3 Inset). [Details of the procedure
are described elsewhere (36).] Immunoblotting experiments
confirmed that the soluble CtmCENP-B[P-gal] retains reactivity with the autoantibodies (Fig. 3 Inset, lane 7).
We then used a solid-phase RIA to screen 48 ACA+ patient
10
A
B13- GALACTOSIDASE
EDC termCENP-BI
S
_8
x 6
C~.2
i1LLfI~ ~lri
4
2
-~~~~~~~0
6
5
29*
4
X
3
1 234
567
:L
()
sera for binding to Cte,,CENP-B[,8-gal]. We also analyzed 20
sera from normal individuals and 92 ACA- patient sera. The
latter included all those available to us from patients with
Raynauds phenomenon and scleroderma (with or without full
or partial CREST) as well as randomly chosen sera from
individuals with systemic lupus erythematosus and Sj0gren
syndrome. Virtually every ACA' patient serum binds to this
portion of CENP-B [Fig. 3, except for two sera, both
characterized by the presence of substantial levels of antiCtermCENP-B of the IgG3 subtype, which is not recognized
by protein A (R. A. Eisenberg, B.J.B., W.C.E., and N.F.R.,
unpublished data)]. The average values for the three serum
classes shown were normal control sera, 195 ± 100 cpm;
ACA- patient sera, 284 ± 187 cpm; and ACA' patient sera,
6013 ± 4809 cpm (a 22-fold stimulation for ACA' sera over
ACA- control sera).
In a control experiment, the binding of a number of
randomly chosen ACA' and ACA- sera to CtemCENP-B[,Pgal] and to f-galactosidase was examined (Fig. 4A). (The
latter comprises 87% of the mass of CtermCENP-B[,8-gal]). A
rabbit antiserum (elicited by injection with CtemCENP-B[PBgal], ref. 20) was included as a positive control. This serum
bound to both substrates in this assay (Fig. 4A, lane 4).
However, none of the patient or control sera exhibited
significant binding to 8-galactosidase.
A quantitative immunoblotting assay also confirmed that
the binding observed in the RIA was specific for the human
portion of Cte,,CENP-B[,3-gal]. Bacterial lysates containing
Cte1,CENP-B (fused to TrpE and therefore containing no
bacterial sequences in common with CtemCENP-B[,8-gal])
were subjected to NaDodSO4/PAGE and immunoblotting
(Fig. 1A). The region of the nitrocellulose containing the
fusion protein with its bound autoantibody and 1251-labeled
protein A was excised and counted in a y counter. The results
for the panel of 40 sera are shown in Fig. 4B along with the
values obtained for these same sera by RIA. The striking
200-
>6
0
4981
2
_*..
:
ACA
:@**.** A~b@@S~~:
NORMAL ACA-NEGATIVE
PATIENTS
FIG. 3. Use of C,,,.,,CENP-B[3-gal] in a RIA to detect ACA. The
RIA shows binding of various patient and control sera to CtermCENPB[3-gal]. Each dot is the average of three measurements for the
serum from a different individual. (Inset) Isolation of partly purified
Ctrn,CENP-B[P-gal] from induced lysogens. Lane 1, marker proteins
(molecular masses indicated in kDa to the left of the gel). Lanes 2-4,
Coomassie blue staining of proteins of whole cell lysate (lane 2),
soluble protein fraction (lane 3), and final fraction (lane 4). Lanes
5-7, immunoblotting analysis of the samples of lanes 2-4 using
patient serum GS (1:1000).
140
120
100
80
60
40
20
B
4
6
8
10 12 14 16 18 20
US
j:]
5
10
RIA
IMMUNOBLOT
25
20
15
SERUM NUMBER
30
35
40
FIG. 4. Demonstration that binding in the RIA is specific for
cloned human CENP-B. (A) Test for the binding of a random panel
of control and patient sera to 3-galactosidase and CtemnCENP-B[fgal] by RIA. The sera used were normal controls (1-3), serum from
a rabbit immunized with CtermCENP-B[P-gal] (contains both antiCENP-B and anti-/3-galactosidase) (4), ACA- patient control sera
(5-9), and ACA+ patient sera (10-21). (B) Comparison of the RIA of
Fig. 3 and the immunoblotting results of Fig. 1A. The relevant
portions of the immunoblots shown in Fig. 1A were excised and
counted in a y counter. The two sets of results were normalized to
give an equivalent average value.
4982
Proc. Natl. Acad. Sci. USA 84
Immunology: Earnshaw et al.
correlation exhibited by the two data sets confirms that the
RIA detects the binding of autoimmune sera to the human
portion of CENP-B.
DISCUSSION
Classes of ACA. Our analysis of the binding of ACA to
cloned portions of CENP-B expressed in bacteria identifies
three epitopes, CE1, CE2, and CE3, that are targets for the
ACA autoimmune response. These results, in conjunction
with our previous work, define at least six independent
classes of ACA (summarized in Fig. 5).
CE1 is located in the carboxyl-terminal region of CENP-B
and overlaps the site of binding of a murine monoclonal
antibody, m-ACA1. Because disruption of CtermCENP-B at a
site 43 amino acids from the carboxyl terminus destroys the
epitope recognized by m-ACA1 (20), we postulate (Fig. 5)
that CE1 may span this region of the protein (although loss of
antigenicity could also result from a protein folding defect).
Because of the close correspondence between the levels of
binding observed with CtermCENP-B[,8-gal] by RIA and with
CtermCENP-B[TrpE] by immunoblotting (Fig. 4B), it is likely
that CE1 is the principal determinant being recognized in the
RIA. CE1 is recognized by virtually all ACA' patient sera
previously identified by indirect immunofluorescence.
CE2 is located within a 104 amino acid stretch starting 43
amino acids upstream from the carboxyl terminus of CENPB. This epitope is recognized by only =20% of the patient
sera tested, and, even in the sera that interact most significantly with it (Fig. 1C, lanes 6 and 9), anti-CE2 comprises
only a small fraction of the antibodies that recognize
CtermCENP-B.
CE3 is located somewhere within the amino-terminal 60%
of CENP-B. It is strongly recognized by a minority of ACA+
sera, but lower titers of anti-CE3 ACA are present in most (or
all) of the remaining ACA+ sera. Overall, -90% of the ACA+
sera exhibit detectable binding to CE3. CE1 and CE3 are
distinct structural determinants, since affinity-purified antiCE1 and anti-CE3 do not cross-react (20).
CLONED ANTIGENS: |
CENP-aB
";`3
N
CR
it
CENP-B
prox
(347
(547 aa)
CENP-B
term
(147
aa)
aa)
term
(104
CENP-B
c
CENP-B
C
L
term
aa)
(43
R
aa)
ICHROMOSOMAL ANTIGENS:I
6
5
4
CENP-C
(140 kDa)
CENP-B
(80 kDa)
CENP-A (17 kDa)
FIG. 5. Distribution of epitopes recognized by ACA. Cloned
antigens: All are present only on CENP-B and are localized to the
hybrid proteins as shown. The precise location of the epitope within
a given shaded segment is not known. aa, Amino acids. Chromosomal antigens: These determinants are defined by our earlier studies
(13). None of them is present on cloned CENP-B, although epitopes
4 and 5 are present on chromosomal CENP-B (13). None of these
determinants has been precisely mapped.
(1987)
Three additional autoepitopes recognized by ACA were
previously identified by analysis of the binding of various
affinity-purified patient sera to chromosomal antigens in
immunoblots (13). The first of these, CE4, is present on
CENPs A and B and is absent from CENP-C (defined by sera
GS and SN, ref. 13). We demonstrated that -97% (38/39) of
a panel of 39 ACA' patient sera had anti-CENP-A detectable
by immunoblotting (12). We assume that all sera binding to
CENP-A bind to CE4, although the structure of this antigen
may be more complex. Epitope CE5, shared by CENPs B and
C, was defined by antibodies from serum JR, affinity-purified
from CENP-C, that were subsequently found to cross-react
strongly with CENP-B (13). Finally, CE6, found solely on
CENP-C, was defined when antibodies were affinity purified
from CENP-C (using serum GS) that showed no rebinding to
CENP-B (13). The pattern of binding observed using antibodies affinity purified from CE1, CE3, CE4, CE5, and CE6
indicates that all of these determinants are structurally
independent (13, 20).
Solid-Phase Binding Assay for ACA. Antibodies to CENP-B
appear to be diagnostic for ACA, since all ACA' patient sera
we have tested (105 sera, from Farmington, Baltimore,
Montreal, Houston, La Jolla, Nijmegen, and Heidelberg),
recognize chromosomal CENP-B in immunoblots. We have
yet to observe anti-CENP-B in any ACA- patient (365 tested)
or normal control (32 tested) serum (W.C.E., B.J.B., and
N.F.R., unpublished). Moreover, a RIA based on cloned
CtermCENP-B[,8-gal] provides a sensitive, reliable method for
the detection of ACA. This assay may eventually provide an
alternative method for screening large numbers of patient
sera for ACA in the clinic.
Origin of ACA. Although much progress is being made in
identifying and characterizing autoantigens, the origin of
antinuclear autoantibodies remains obscure. It has been
suggested that autoantibodies might arise as a result of
fortuitous cross-reactions exhibited by normal antibodies
(28, 29), as a result of chance mutations causing normal B-cell
clones elicited by foreign antigen to change specificity and
recognize self components (30, 31), or as antiidiotypes
elicited during an immune response against a viral protein (7).
Our data are inconsistent with all such models. These models
could explain an autoimmune response against any single
epitope, but they cannot account for the ACA response, since
multiple structurally independent epitopes are targeted in
virtually every affected individual (requiring multiple chance
events).
Two of our observations presented above suggest that
ACA might arise from an antigen-driven response. (i) Cloned
CENP-B [unlike double-stranded DNA (32)] is immunogenic
in rabbits and mice (20). [The Sm antigen (a marker antigen
for systemic lupus erythematosus) is also immunogenic in
rabbits and mice (33).] (ii) ACA are polyclonal. Serum from
single patients recognizes at least four independent epitopes
on CENP-B. We have shown here that >90% of ACA+ sera
recognize CE1 and CE3 and have shown elsewhere that .95%
of the same sera also recognize CE4 and either CE5 or CE6
(12). Thus, CENP-B appears to be the target of a diversified
polyclonal response. This was postulated earlier to be true for
anti-Sm (33, 34) and antiribonucleoprotein (35) autoantibodies, but in neither case was the conclusion confirmed by
isolating multiple noncross-reactive specificities from patient
serum.
In a recent review, Hardin (6) noted that most major
autoantibodies found in systemic lupus erythematosus recognize nucleoprotein structures (small nuclear ribonucleoproteins, cytoplasmic ribonucleoproteins, and nucleosomes).
Our data suggest that ACA fit this pattern and that the ACA
response may result from self immunization with centromeres. This is surprising, since centromeres are extremely
minor cellular components and as such seem unlikely can-
Immunology: Earnshaw et al.
didates for the progenitors of a high-titer immune response
(particularly since ACA occur in only a small fraction of the
autoimmune patients who make antinuclear antibodies).
Monoclonal antibody m-ACA1, which we have shown here to
bind to a major autoepitope in CENP-B, should provide an
important tool for future studies that attempt to locate the
source of centromere antigen in affected individuals.
We acknowledge the collaboration of K. Sullivan in isolation of the
original cDNA clone; Carol Cooke and D. Kaiser for isolation of
monoclonal antibodies; H. Langevin (Baltimore), J.-L. Senecal
(Montreal), E. Tan (La Jolla), W. J. van Venrooij (Nijmegan), and H.
Guldner (Heidelberg) for the gift of sera; and J. Stobo for helpful
discussions. D.W.C. is the recipient of a National Institutes of Health
Career Development Award. This work has been supported by
National Institutes of Health Grant GM 35212 to W.C.E. and N.F.R.,
the Arthritis Foundation Devil's Bag Award to W.C.E. and D.W.C.,
and National Institutes of Health Grant GM 29513 to D.W.C.
1. Reichlin, M. & Mattioli, M. (1974) Bull. Rheum. Dis. 24,
756-760.
2. McDuffie, F. C. & Bunch, T. W. (1977) Bull. Rheum. Dis. 27,
900-905.
3. Tan, E. M. (1982) Adv. Immunol. 33, 167-240.
4. Shoenfeld, Y. & Schwartz, R. S. (1984) N. Engl. J. Med. 311,
1019-1029.
5. Schwartz, R. S. & Stollar, B. D. (1985) J. Clin. Invest. 75,
321-327.
6. Hardin, J. A. (1986) Arthritis Rheum. 29, 457-460.
7. Plotz, P. H. (1983) Lancet ii, 824-826.
8. Grabar, P. (1983) Immunol. Today 4, 337-340.
9. Moroi, Y., Peebles, C., Fritzler, M. J., Steigerwald, J. & Tan,
E. M. (1980) Proc. Nati. Acad. Sci. USA 77, 1627-1631.
10. Fritzler, M. J., Kinsella, T. D. & Garbutt, E. (1980) Am. J.
Med. 69, 520-526.
11. Tan, E. M., Rodnan, G. P., Garcia, I., Moroi, Y., Fritzler,
M. J. & Peebles, C. (1980) Arthritis Rheum. 23, 617-625.
12. Earnshaw, W. C., Bordwell, B. J., Marino, C. & Rothfield,
N. F. (1986) J. Clin. Invest. 77, 426-430.
13. Earnshaw, W. C. & Rothfield, N. F. (1985) Chromosoma
(Berlin) 91, 313-321.
14. Cox, J. V., Schenk, E. A. & Olmsted, J. B. (1983) Cell 35,
331-339.
Proc. Natl. Acad. Sci. USA 84 (1987)
4983
15. Guldner, H. H., Lakomek, H.-J. & Bautz, F. A. (1985) Clin.
Exp. Immunol. 58, 13-20.
16. van Venrooij, W. J., Stapel, S. O., Houben, H., Habets,
W. J., Kallenberg, C. G. M., Penner, E. & van de Putte, L. B.
(1985) J. Clin. Invest. 75, 1053-1060.
17. Valdivia, M. M. & Brinkley, B. R. (1985) J. Cell Biol. 101,
1124-1134.
18. Spowart, G., Forster, P., Dunn, N. & Cohen, B. B. (1985) Dis.
Markers 3, 103-112.
19. Nyman, Y., Hallman, H., Hadlaczky, G., Pettersson, I.,
Sharp, G. & Ringertz, N. R. (1986) J. Cell Biol. 102, 137-141.
20. Earnshaw, W. C., Sullivan, K. F., Machlin, P. S., Cooke,
C. A., Kaiser, D. A., Pollard, T. D., Rothfield, N. F. &
Cleveland, D. W. (1987) J. Cell Biol. 104, 817-829.
21. Lewis, C. D. & Laemmli, U. K. (1982) Cell 17, 849-858.
22. Towbin, H., Staehelin, T. & Gordon, J. (1979) Proc. Natl.
Acad. Sci. USA 76, 4350-4354.
23. Young, R. A. & Davis, R. B. (1983) Proc. Natl. Acad. Sci.
USA 80, 1194-1198.
24. Young, R. A. & Davis, R. B. (1983) Science 222, 778-782.
25. Shero, J. H., Bordwell, B., Rothfield, N. F. & Earnshaw,
W. C. (1986) Science 231, 737-740.
26. Budman, D. R., Merchant, E. B., Steinberg, A. D., Doft, B.,
Gershwin, M. E., lizzio, E. & Reeves, J. P. (1977) Arthritis
Rheum. 20, 829-833.
27. Izui, S., McConahey, P. J. & Dixon, F. J. (1978) J. Immunol.
121, 2213-2219.
28. Cooke, A., Lydyard, P. M. & Roitt, I. M. (1983) Immunol.
Today 4, 170-175.
29. Klinman, D. M. & Steinberg, A. D. (1986) Arthritis Rheum.
29, 697-705.
30. Eilat, D., Hochberg, M., Pomphrey, J. & Rudikoff, S. (1984) J.
Immunol. 133, 489-494.
31. Diamond, B. & Scharff, M. D. (1984) Proc. Natl. Acad. Sci.
USA 81, 5841-5844.
32. Madaio, M. P., Hodder, S., Schwartz, R. S. & Stollar, B. D.
(1984) J. Immunol. 132, 872-876.
33. Shores, E. W., Eisenberg, R. A. & Cohen, P. L. (1986) J.
Immunol. 136, 3662-3667.
34. Eisenberg, R. A., Dyer, K., Craven, S. Y., Fuller, C. R. &
Yount, W. J. (1985) J. Clin. Invest. 75, 1270-1277.
35. Maddison, P. J. & Reichlin, M. (1977) Arthritis Rheum. 20,
819-824.
36. Rothfield, N. F., Whitaker, D., Bordwell, B., Weiner, E., Senecal, J.-L. & Earnshaw, W. (1987) Arthritis Rheum., in press.