Full text of DOI:10.1078/0171-2985-00107

Immunobiol. (2002) 205, pp. 1–16
© 2002 Urban & Fischer Verlag
http://www.urbanfischer.de/journals/immunobiol
Clinical Research Unit for Rheumatology, Albert-Ludwigs-University Medical Center, Freiburg,
Germany
Monoclonal Autoantibodies from Patients with Autoimmune
Diseases: Specificity, Affinity and Crossreactivity of MAbs
Binding to Cytoskeletal and Nucleolar Epitopes, Cartilage
Antigens and Mycobacterial Heat-Shock Protein 60
TIL
MENGE*, RITA
RZEPKA, and INGA
MELCHERS
Received January 29, 2001 · Accepted in revised form November 5, 2001
Abstract
Serum autoantibodies produce typical immunofluorescence staining patterns on HEp-2 cells, which
are frequently used for diagnostic purposes. These include antibodies recognizing cytoskeletal and
nuclear epitopes. The detailed analysis of human monoclonal antibodies (MAbs) should help to
understand which antigens or autoantigens were involved in the generation of these immune respons-
es. Here, three MAbs are described staining HEp-2 cells in a characteristic pattern. They were derived
from peripheral blood B cells of two patients with rheumatic diseases (rheumatoid arthritis and relaps-
ing polychondritis). Their binding reactivities were characterized in detail in several assay systems and
their affinities measured. Although the antibodies differed in their fine specificity and crossreactivity,
all three MAbs (2 IgM, 1 IgA) bound to purified cytoskeletal antigens (desmin) and, in addition, to
cartilage antigens (human collagen type II, proteoglycans). The binding to HEp-2 cells could be in-
hibited specifically with soluble antigens as shown by intracellular flow cytometry. The affinities for
both groups of antigens were relatively high (examples: KD (desmin) = 0.1 × 10–7 M; KD (collagen) =
3.5 × 10–7 M). Two of the MAbs also bound to heat-shock protein 60 (HSP60) derived from
Mycobacterium tuberculosis. The results prove that antibodies and B cells with reactivity to both intra-
cellular cytoskeletal and nuclear antigens and exogenous antigens (e. g. HSP60) exist in patients with
rheumatic diseases. Similar to an animal model such human B cells may be induced by the exogenous
antigen (HSP60) and crossreact with local auto-antigens related to the disease (cartilage). In this way
they might contribute to pathogenic processes. Due to their additional crossreactivity with intracellu-
lar cytoskeletal and nuclear antigens, these antibodies simultaneously can be detected in the HEp-2
immunofluorescence assay.
Abbreviations: ANA = antinuclear antibodies; HSP60 = heat-shock protein 60; IMF = intermediate fil-
aments; MAb = monoclonal antibody; NAA = natural autoantibodies; MFI = mean fluorescence inten-
sity; RA = rheumatoid arthritis; rPC = relapsing polychondritis.
* Present address: Neuroimmunology Laboratories, Dept. of Neurology, University of California, San Fran-
cisco, USA.
0171-2985/02/205/01-001 $ 15.00/0

2 · T. MENGE et al.
Introduction
The screening for antinuclear antibodies (ANA) and antibodies binding to cytoplasmic
and cytoskeletal proteins is often performed on HEp-2 cells, a line originated from a
human laryngeal carcinoma (epithelioma type 2), grown as monolayers on microscope
slides (1). The over 30 different nuclear and cytoplasmic patterns known to occur on
HEp-2 cells are produced by many different autoantibodies all contributing to the spec-
ificity of the analyzed sera.
Autoantibodies to cytoskeleton proteins are present in a wide range of diseases, includ-
ing several connective tissue diseases such as rheumatoid arthritis (RA), SLE or systemic
sclerosis (2–8). However, they are also present in healthy blood donors, and are then
regarded as “natural autoantibodies” (NAA), crossreacting with many bacterial and viral
antigens (9–11). Recently, the development of autoantibodies to intermediate filaments
(IMF), both keratins and vimentin, was demonstrated in sera of rats, injected with heat
killed Mycobacterium butyricum in mineral oil, which developed adjuvants arthritis (12).
In many cases it is difficult to understand the relationship between typical staining pat-
terns of autoantibodies and the disease they are correlated with, especially if intracellular
antigens are involved (13, 14). It seems unlikely that these antigens themselves are the
antigens originally activating the antibody producing B cell. However, to detect the “true”
disease related activating antigen – and possible pathogenic mechanisms hidden behind
it – may be difficult employing polyclonal body fluids, such as sera. We therefore started
to produce and characterize in detail human monoclonal autoantibodies (MAbs) gener-
ated from patients with systemic autoimmune diseases (15, 16).
The cytoskeleton consists of three types of filaments – the microfilaments, microtu-
bules and intermediate filaments (IMF) – which differ in diameter and composition (17,
18). Microfilaments contain mainly actin, microtubules tubulin and IMF keratins (type
I and II IMF), vimentin, desmin (type III IMF) or other members of the IMF superfam-
ily (type IV, V). IMF are expressed differentially in different cell types and developmen-
tal stages, e.g. desmin is synthesized in muscle cells. Type V IMF proteins are the only
IMF proteins known to occur in the nucleus where they compose the nuclear lamina.
The aim of the present study was to characterize the binding specificities and affinities
of patient derived human MAbs specific for cytoskeleton proteins and for disease related
local autoantigens. Here we describe 3 MAbs binding to intracellular antigens as detect-
ed on HEp-2 cells. These hybridomas were derived from peripheral blood B cells of two
patients, one with rheumatoid arthritis (RA), the other with relapsing polychondritis
(rPC) – both diseases with involvement of cartilage.
Materials and Methods
Patients
Patient M (59 years old male) suffered since 17 years from rPC, diagnosed according to the McAdam
criteria (19) and by nasal septum biopsy. Additionally he had diabetes mellitus. Patient RA (70 years
old male) suffered since 5 years from RA, diagnosed according to the ACR criteria (20). Synovial tissue
was obtained at the time of clinically indicated surgery from another patient with RA (70 years old
female). Synovial adherent cells were prepared as described (21). Chondrocytes were prepared as
described (22) from non-rheumatic autopsy samples of knee cartilage and used before the first passage.

Human MAbs binding to intermediate filaments · 3
B cell hybridomas
Peripheral blood mononuclear cells (PB-MNC) were T cell depleted by rosetting with sheep erythro-
cytes and PB B cells transformed with Epstein Barr Virus (EBV) according to standard protocols. From
patient RA, two lines were generated from 1 × 106 cells. In the case of patient M, the cells were incu-
bated with the virus for 2 h and then distributed into 3 × 96 microtiter wells (NUNC, Denmark) con-
taining 10.000, 5.000 or 1.000 cells/well. All cultures were grown up into lines, but 25 were subse-
quently lost. One line per patient was chosen and used for cell fusion. Briefly, B cells and the hetero-
myeloma cells CB-F7 (23) (kindly provided by S. Jahn, Berlin, Germany) were washed twice in serum-
free culture medium (RPMI 1640, 25 mM HEPES, 2 mM L-glutamine, 0.5% penicillin/streptomy-
cin; GIBCO) and mixed in a 5:1 ratio. PEG1500 (Boehringer, Mannheim, Germany) was added
(2 ml/105 lymphocytes) and stepwise diluted with warm PBS. Cells were centrifuged, resuspended in
culture medium (including 10% FCS) and distributed into microtiter plates (4 × 104/culture in 150 ml
medium). The next day 50 ml of 4-fold concentrated HAT medium was added. Hybridoma lines were
subcloned several times until stable clones were established. The clones M-O and M-P were derived
from the same EBV-transformed B cell line (I-A6), but from different hybridoma lines.
Indirect immunofluorescence and flow cytometry analysis
HEp-2 cells and 3T3 cells were obtained from ATCC (Rockville, Maryland, USA). For microscopic
analysis cells were grown on glass slides for 12–24 h, washed with serum-free culture medium, air dried
and fixed in 100% methanol for 10 min at –20°C. They were then incubated with the first antibodies
for 30 min at 20°C in a moist chamber, washed twice in PBS, then again incubated with the second
antibodies (FITC-labelled), washed and examined with a fluorescence microscope (Axioscope, Zeiss,
Germany). Antibodies were diluted in PBS, 2% FCS, 0.1% NaN3. In addition, commercially available
HEp-2 cell preps were used (Hiss, Freiburg, Germany).
For intracellular flow cytometry, cells were washed in PBS, 3% FCS, 0.1% NaN3. They were then
placed in round bottom 96-well plates (5 × 105 cells/well; NUNC), pelleted and resuspended in ice
cold PBS containing 1% paraformaldehyde and 0.25% saponin (Sigma). After 15 min. on ice, cells
were washed twice in PBS containing 0.1% saponin and 3% FCS (saponin solution), at 4°C. The pel-
let was resuspended in antibody- containing saponin solution and incubated 30 min. on ice, followed
by washing and incubation with FITC-labelled second reagents. Cells were analyzed by FACScan
applying the Lysis II software program (BectonDickinson, Mountain View, CA, USA).
Antigens and antibodies
The following antigens were used: chicken desmin, bovine vimentin, bovine keratin K8, rat actin, hen
egg lysozyme (all Boehringer, Mannheim, Germany), keyhole limpet hemocyanin (KLH, Calbiochem,
USA), bovine gelatine (Merck, Darmstadt, Germany), human IgG-Fc-fragments (Dianova, Hamburg,
Germany). Human collagens Type I, II and V, a human cartilage extract (containing proteoglycans,
devoid of collagens) and Mycobacterium tuberculosis (M. tub.) HSP60 were prepared as described
(24–26) and kindly provided by K. VON DER
MARK, Erlangen, and S. KAUFMANN, Berlin, respectively.
Isotype control antibodies (myeloma IgM, pooled serum IgA) were bought from Dianova
(Germany). K8.13, V9 (Sigma, Deisenhofen, Germany) and DE-B-5 (Calbiochem) are murine IgG
MAbs specific for keratin K8, vimentin and desmin, respectively. Secondary reagents were species-spe-
cific goat-anti-Ig antibodies, labelled with FITC for indirect immunofluorescence or with alkaline phos-
phatase (AP) for ELISAs.
ELISA
ELISAs were performed as described (27). Plates (NUNC) were coated with antigens (usually 5 mg/ml),
and blocked with gelatine (0.01%), when appropriate. For quantitation of Ig contents, anti-Ig reagents
and isotype controls (Dianova) were used. As detection system we used AP, with para-nitrophenyl phos-
phate as substrate and read optical densities (OD) at l = 405 nm with an ELISA reader (EAR 400 AT,
SLT, Austria).

4 · T. MENGE et al.
Inhibition assays and affinity measurement
Experiments were performed according to the protocol of FRIGUET et al. (28). After determination of
useful antibody concentrations, the antibodies were incubated 15 h at room temperature with different
concentrations of soluble antigen (up to 6 × 10–7 M). Unbound antibody subsequently was detected
on ELISA plates coated with the same antigen. Dissociation constants KD were calculated from the OD
values obtained in the absence and presence of the inhibitor as described (28).
Preparation of protein extracts and immunoblots
Protein extracts were prepared from 1 × 107 HEp-2 cells, washed twice with ice cold PBS and incubat-
ed for 30 min on ice with 100 ml lysis buffer (150 mM NaCl, 50 mM Tris pH 8, 1% NP40, protease
inhibitors [Complete]; Boehringer, Mannheim, Germany). The supernatant after centrifugation (15
min. 20000 g) was used as extract 1. The pellet was resuspended in the same volume of lysis buffer plus
1.33 M KCl, incubated on ice for 45 min and sonified (10 cycles at 50% power; Sonoplus HD 70,
Bandeln, Germany). Then 0.1% SDS was added, the mixture was heated to 95°C for 1 min., and cen-
trifuged as before. The supernatant was used as extract 2. The pellet was resuspended in Laemmli buf-
fer and used as extract 3. Immunoblots were prepared as described (21), using 10% SDS-PAGE, nitro-
cellulose membranes (Hybond ECL nitrocellulose; Amersham-Buchler, Braunschweig, Germany),
blocking with PBS + 5% milk powder + 0.1% Tween 20. The antibodies were added at 5 mg/ml. As
secondary reagents goat-anti-human-Ig labelled with HRPO, and as substrate Super signal (Pierce,
Rockford, IL, USA) were used.
Results
Generation of human B cell hybridomas
PB B cells derived from a patient with RA (patient RA) were transformed with EBV, and
two B cell lines generated, one of which secreted mainly IgA, stained HEp-2 cells (dilu-
tion 1:5) and was used for cell fusion. PB B cells from a patient with rPC (patient M)
were purified and 263 B cell lines generated by EBV transformation. Supernatants of 5
lines each were mixed in equal volumes (dilution 1:5) and screened on HEp-2 cells with
anti-hIgMFITC, -hIgAFITC and -hIgGFITC. All pools contained IgM staining HEp-2 cells,
with a large variety of cytoplasmatic and nuclear patterns, not including homogeneous
staining of nuclei. IgG and IgA were only detected in 3 pools, staining nuclei (dots), or
cytoplasm. One line (M-I-A6) was chosen and used for cell fusion (from 1× 104 B cells
seeded). From both EBV transformed B cell lines several hybridomas were generated and
stable monoclonal sublines established. We here present data for 2 MAbs from patient M
(M-O and M-P, both IgMl) and 1 MAb from patient RA (RAB-A, IgA1Í).
Staining patterns of M-O, RAB-A and M-P on adherent cells detected by indirect
immunofluorescence
Both original B cell lines produced antibodies which stained the cytoskeleton of HEp-2
cells in a similar way, as shown for the line M-I-A6 in Figure 1a. These patterns resulted
from a mixture of antibodies, as both lines were derived from ≥10000 cells originally
seeded, and they were later shown to contain at least 5 different clones (by isoelectric
focussing and clonal analyses, data not shown). After hybridoma clones were established,
we observed that M-O as well as RAB-A bound to 2 different cellular structures, one of

Human MAbs binding to intermediate filaments · 5
Figure 1. Staining patterns of human monoclonal antibodies (indirect immunofluorescence). a:
Supernatant of the EBV transformed B cell line M-I-A6 on HEp-2 cells, anti-IgMFITC. b–d: M-O on
HEp-2 cells (b) and on synovial adherent cells from a RA patient (c, d), anti-IgMFITC, nucleolar (b),
cytoplasmatic (c) or nuclear (d) staining. e, f: RAB-A on HEp-2 cells, anti-IgAFITC, nucleolar (e) or
cytoskeletal (f) staining. g: M-P on HEp-2 cells, anti-IgMFITC, cytoskeletal staining. HEp-2 cells were
either prepared as described in Materials and Methods (a, b and e) or bought (f, g). Original magnifi-
cation × 400.
these apparently was a cytoskeletal molecule (Fig. 1c, 1f), whereas the other one was
located within the nucleus and/or nucleolus (Fig. 1b, 1d, 1e). The cytoskeletal staining
closely resembled the anti-vimentin pattern (1). The clone M-P only bound to cyto-
skeletal epitopes (Fig. 1g) but not to the nucleus. All three antibodies stained epithelial
cells (HEp-2 cells) and mesenchymal cells (murine 3T3 cells, human skin fibroblasts, syn-
ovial adherent cells and chondrocytes) with similar patterns (Fig. 1c, 1d). Both patterns

6 · T. MENGE et al.
were usually present in the same cell, however, they could also be seen separate from each
other, i.e. in different cells of the same preparation (Fig. 1c, 1d). In general, the staining
of human cells was stronger than that of rodent cells; the nucleolar pattern seen with
M-O was stronger than that seen with RAB-A. Commercially available HEp-2 cells pref-
erentially showed the cytoskeletal pattern, whereas cells prepared in our own laboratory
showed both, the cytoskeletal and nuclear pattern.
Quantitative analysis of intracellular epitope binding by cytometry
In addition to the microscopic analyses, the monoclonal antibodies were tested for bind-
ing by FACS analyses. Neither M-O nor M-P or RAB-A bound to the cell surface of
HEp-2 cells or of PB-MNC (not shown). In contrast, the binding to intracellular epi-
topes could be quantitatively analysed by flow cytometry. As an example, a histogram of
the fluorescence intensities of M-O binding to fixed and permeabilized HEp-2 cells is
depicted in Figure 2a, in comparison to an isotype control antibody. Figure 2b illustrates
the correlation between the mean fluorescence intensities (MFI) of binding and the con-
centration of the monoclonal antibodies.
Figure 2. Binding to intracellular epitopes analyzed by cytometry. HEp-2 cells were fixed, permeabi-
lized and then incubated with M-O (dark line) or an isotype control antibody (light line) at a concen-
tration of 28 mg/ml, followed by goat anti-human-IgMFITC (a). The mean fluorescence intensities
(MFI) were 375 310.0 (M-O) and 10 4.5 (control), respectively. In addition, titration curves are
shown (b).
Antigen binding specificity of M-O, RAB-A and M-P analyzed by ELISA
To characterize the specific binding profiles of the monoclonal antibodies, they were test-
ed at a fixed concentration for binding to a number of cytoskeletal and control antigens
in an ELISA. The most interesting results are depicted in Figure 3. M-O was rather spe-
cific, binding only to desmin, and not to other cytoskeletal antigens tested like vimentin,
keratin K8 or actin. Neither did it bind to a variety of other antigens tested, including

Human MAbs binding to intermediate filaments · 7
Figure 3. Antigen binding profiles of three MAbs. Antibody containing hybridoma supernatants were
standardized to a concentration of 6 mg/ml and incubated on ELISA plates coated with the antigens.
Binding was monitored with anti-IgMAP or anti-IgAAP, respectively.
insulin, collagens, albumins, transferrin, lactoferrin or pepsin. In contrast, RAB-A and
M-P were polyreactive, binding to all cytoskeletal antigens tested, and in addition to IgG-
Fc-fragments as well as most of the other antigens tested including hen egg lysozyme and
keyhole limpet hemocyanin. M-O never bound to M. tub. HSP60; RAB-A and M-P
sometimes did, indicating an instability of the epitope detected in this assay (not shown).
None of the three MAbs bound to gelatine or DNA (not shown).
Binding inhibition assays
In an ELISA also low affinity reactivities are detected. In contrast, in the immunofluores-
cence assay the antibody will bind preferentially to those epitopes which react with the

8 · T. MENGE et al.
Table 1. Affinities of MAbs
Antigen
KD (M)
RAB-A
M-O
M-P
Desmin
Actin
Keratin K8
Collagen Typ II
KLH
0.1 × 10–7
n.d.
1.9 × 10–7
0.1 × 10–7
5.2 × 10–7
4.1 × 10–7
2.4 × 10–7
52.0 × 10–7
15.0 × 10–7
12.5 × 10–7
3.5 × 10–7
1.4 × 10–7
n.d.
n.d.
n.d.
Dissociation constants (KD) were determined by inhibition with the same antigen in solution as used
as coat in the ELISA. n. d. = not determined, as M-O only bound to desmin. In the same way the KD
of the commercial MAb K8.13 binding to keratin K8 was determined (2.3 × 10–7 M). For each MAb
the lowest KD (highest affinity) is indicated by boldface type.
Table 2. Binding inhibition measured by cytometric analyses
Inhibitor
MFI (% Inhibition)
M-O
RAB-A
320
130 (84.4%)
230 (37.5%)
80 (100%)
non
desmin
400
70 (100%)
90 (98.4%)
n.d.
desmin (1:10)
actin
keratin K8
collagen Type II
cartilage extract
M. tub. HSP 60
150 (79.4%)
n.d.
180 (69.8%)
n.d.
280 (16.7%)
220 (41.7%)
210 (45.8%)
90 (96.9%)
Controls:
isotype control + anti-IgFITC
85
70
80
70
anti-IgFITC
The antibodies were used in a concentration of 2.5 mg/ml (M-O) and 7 mg/ml (RAB-A), respectively.
Inhibitors were used in the highest concentration used in the ELISA inhibition assays (3.7 – 6 × 10–7
M). Desmin in addition was used in a 10× lower concentration. Inhibition was calculated as 100 ×
[100 – (MFI+Inhibitor – MFIisotype control/(MFI-inhibitor – MFIisotype control)]. n. d. = not determined.
highest affinity. This might explain certain discrepancies between both assays, e. g. bind-
ing of RAB-A to actin in the ELISA but not producing a “typical” actin-binding staining
pattern on HEp-2 cells. We therefore measured the binding affinities for the most inter-
esting antigens according to a protocol by FRIGUET et al (28). Within the range of con-
centrations tested, inhibition could not be obtained with a number of antigens, includ-
ing IgG-Fc-fragments, albumins, lysozyme and transferrin (not shown). The binding of
the MAbs to these antigens therefore did only occur with low affinity. Homologous inhi-
bition could be obtained with desmin (all 3 MAbs) and with actin, keratin K8, collagen

Human MAbs binding to intermediate filaments · 9
Figure 4. Specific binding of M-P and RAB-A to M. tub. HSP60 in solution. The binding of RAB-A
and M-P to desmin was inhibited by soluble M. tub. HSP60. Bound MAbs were detected with goat-
anti-human-IgMAP or -IgAAP and PNPP as substrate.
type II and KLH (RAB-A and M-P). The affinities calculated from these data are shown
in Table 1. Vimentin was not available for inhibition assays, and therefore could not be
included here. To confirm a high affinity binding of RAB-A and M-P to M. tub. HSP60,
we performed a heterologous inhibition assay using desmin as coating antigen and
HSP60 as soluble inhibitor, which showed reproducibly a strong inhibition (Fig. 4).
In addition to ELISA inhibition assays, we studied the inhibition of the binding to
cytoskeletal antigens of HEp-2 cells by cytometry. MAbs were preincubated in appropri-
ate concentrations with the soluble antigens, and then tested for binding to permeabilized
cells (Table 2). Binding of M-O and RAB-A to HEp-2 cells was inhibited with desmin in
a concentration-dependent way. Interestingly, M-O was also inhibited by keratin K8 and
by a cartilage extract, containing only proteoglycans. As might be expected from the affin-
ity measurements, actin was the best inhibitor for RAB-A, followed by M. tub. HSP60 and
desmin, whereas keratin K8 was of lower efficiency. Interestingly, RAB-A binding to HEp-

10 · T. MENGE et al.
Figure 5. Binding of MAbs to desmin and to proteins extracted from HEp-2 cells detected by immu-
noblotting. Desmin (a) or proteins solubilized from HEp-2 cells (b) were applied to SDS-PAGE fol-
lowed by immunoblotting. M-P (lane a1), M-O (lane a2) and RAB-A (lane a3) bound to desmin
(detection with anti-IgM or anti-IgA, respectively). Binding to proteins extracted from HEp-2 cells by
the consecutive application of lysis buffers and sonication as described in Materials and Methods is
demonstrated for M-O (b). The lanes contained extract 1 (lane b3), extract 2 (lane b2), extract 3 (lane
b1) or a mixture of all 3 extracts (lane b4). Blots were first incubated with M-O (a) or the isotope con-
trol (c), then stripped and retested with antibodies specific for vimentin (b) or desmin (d). The bands
detected in (a) and (b) were identical.
2 cells was also inhibited by collagen type II and the collagen-free cartilage extract. The
binding of the commercial antibody K8.13 was completely inhibited by keratin K8 only.
Binding to proteins extracted from HEp-2 cells
All three MAbs were selected for analysis because they stained HEp-2 cells in a specific
pattern. The cytoskeleton of HEp-2 cells is supposed to contain type I and II intermedi-
ate filaments (IMFs), the type III IMF vimentin, but not desmin, and microfilaments
containing actin. The cytoskeletal staining patterns we observed showed similarities to
patterns obtained with antibodies exclusively binding to vimentin or keratins. The spe-
cificities of RAB-A and M-P detected in ELISA and in the inhibition assays suggest that
the binding to either vimentin or keratins or to both molecules produced the specific
cytoskeletal staining pattern of these two antibodies with different fine specificities. The
cytoskeletal binding of M-O was reflected in the ELISAs by binding to desmin, in the
binding inhibition assays in addition by binding to keratin K8. M-O did not bind to
vimentin in the ELISA. All three MAbs bound to desmin in immunoblots (Fig. 5a), but
not to keratin K8 or actin (not shown), indicating that they are able to recognize certain
but not all epitopes on denatured molecules. We therefore prepared protein extracts from
HEp-2 cells, using 3 different buffers with increasing capacity to solubilize cytoskeletal
and nuclear proteins, and analysed these in immunoblots. Here we show that M-O
bound to a major band with an apparent MW of 65 kD, present in 2 of the 3 protein
extracts, and in their mixture. However, only in 1 of the extracts the same band was also
detected by a MAb specific for vimentin (Fig. 5b). A MAb specific for desmin did not
bind to any protein in these HEp-2 extracts (Fig. 5b). Thus, most probably the cytoskel-

Human MAbs binding to intermediate filaments · 11
Figure 6. Binding of serum IgG (patient M) to HEp-2 cells
and vimentin. Serum from patient M (diluted 1:50) stained
HEp-2 cells in indirect immunofluorescence detected with
anti-IgGFITC (a). HEp-2 protein extract 3 was immunopre-
cipitated with serum and Staph. aureus protein A, coupled to
sepharose beads. The immunoprecipitate and controls were
tested by SDS-PAGE and immunoblots, using anti-vimentin
or an isotype control antibody. The extract contained vimen-
tin (lane 1, detection with anti-vimentin), as did the immu-
noprecipitate with IgGs from patient M (lane 3). The con-
trols were negative (lane 2, control precipitate, anti-vimentin;
lane 4, immunoprecipitate, isotype control antibody).
etal staining pattern observed with M-O was due to binding to vimentin. In addition, at
least one other protein – of the same size – might be recognized.
Serum antibodies of similar specificity detected in the patient
Serum IgG as well as IgM (not shown) from patient M stained HEp-2 cells in indirect
immunofluorescence with a similar pattern as M-O and RAB-A, especially regarding the
nucleolar staining (Fig 6a). In addition, proteins extracted from HEp-2 cells were immu-
noprecipitated with the serum and Staph. aureus protein A. The analysis in immunoblots
demonstrated the presence of vimentin in these immunoprecipitates (Fig. 6b).

12 · T. MENGE et al.
Discussion
Three MAbs derived from two patients with RA and rPC, respectively, were studied for
their specific binding capacities in several assay systems to further understand how the
appearance of such antibodies might be related to disease. These MAbs were chosen
because they showed similar staining patterns on HEp-2 cells, but differed in other
respects. One of them (RAB-A) is a human monoclonal IgA and therefore can be expect-
ed to be derived from a B cell which in vivo was selected and driven by antigen. One
of the IgMs (M-O) produced a very similar immunofluorescence staining pattern on
HEp-2 cells as did RAB-A. The other IgM (M-P) was derived from the same EBV-trans-
formed B cell line as M-O, and showed a related but different specificity. The results of
this study are summarized in Table 3.
The data illustrate that one antibody may show different crossreactivities dependent on
the assay system used. The differences concern mainly the availability of the epitopes rec-
ognized, i. e. most probably their stability under different conditions. As an example, bind-
ing of M-O to desmin could be detected in all assays; desmin thus contains an epitope not
sensitive to conformational changes (solid phase, soluble or denatured). The reactivity with
vimentin, however, was not detected in ELISA, although vimentin most probably was the
molecule detected by the staining of HEp-2 cells which do not contain desmin.
Crossreactions between desmin and vimentin have been described before and might be
expected as both molecules belong to the same protein family and are highly homologous
(more than 60% identity (18,29)). Also murine and human MAbs reacting with desmin
and nuclear antigens were described, and it was suggested that the nuclear antigens might
be DNA (30–33). We do not yet know the identity of the nucleolar epitope detected by
M-O and RAB-A. However, both MAbs did not bind to DNA, neither in ELISA, nor
in the Crithidia assay (not shown). The data clearly show that one antibody may stain
HEp-2 cells with two distinct patterns, i. e. react with cytoskeletal and nucleolar epitopes,
although the exact molecules detected in the nucleolus are still not identified.
In addition to recognizing desmin and vimentin, all three antibodies bound to keratin
K8; RAB-A and M-P also to actin. Nevertheless, the immunofluorescence pattern
observed on HEp-2 cells did not resemble a typical actin-specific staining, although espe-
cially RAB-A bound to actin with high affinity (KD = 0.1 × 10–7 M). None of the anti-
bodies reacted with actin or keratin K8 when these molecules were denatured as in SDS-
PAGE. In immunoblots with protein extracts M-O bound to a band also detected by an
antibody specific to vimentin. In addition – and to our surprise – M-O also bound to a
yet undefined protein of the same apparent molecular weight present in an extract not
expected to contain cytoskeletal components.
Most interesting are the crossreactions of RAB-A and M-P with M. tub. HSP60 and
all three MAbs with cartilage antigens, which were detected in ELISA and/or in inhibi-
tion assays. RAB-A and M-P bound to human collagen type II, with relative high affin-
ities (KD = 4.1 × 10–7 M and 3.5 × 10–7 M, respectively). M-O did not bind to collagen,
however inhibition assays were not performed. All three MAbs reacted with an antigen
present in the collagen-free cartilage extract. Some preliminary evidence suggests this to
be cartilage link protein (34). HSP60 molecules have been implied in arthritis models,
and the titers of antibodies binding to HSP60 molecules might be enhanced in patients
with RA (35–37). Furthermore, crossreactions between HSP60 and cartilage components
appear to play a role in the induction of experimental arthritis (35, 38).

Human MAbs binding to intermediate filaments · 13

14 · T. MENGE et al.
Affinities of human autoreactive antibodies were only rarely reported in the literature
(39–44). Especially, affinity measurements of autoantibodies reacting with cytoskeletal
autoantigens were still missing. The affinities we measured are relatively high and place the
MAbs RAB-A, M-O and M-P well within the group of antibodies induced by experimen-
tal immunizations. This is exemplified by the KD measured for the commercially available
murine antibody specific for keratin K8 (2.3 × 10–7 M). MAbs derived from newborn
mice, which were regarded as “natural autoantibodies” because they bound to a variety of
self antigens including actin and DNA, reacted with KDs in the range of 3 × 10–5 to 5 ×
10–6 M (11). These data are directly comparable as all reports used the same method for
affinity measurements (9, 11, 28, 39–44). However, as affinities of “natural autoantibod-
ies” fall into a broad range (KDs between 5 × 10–3 M and 5 × 10–11 M), affinity cannot
serve to define an antibody as “natural” – in addition an unmutated v-gene is required.
Mono- or polyreactivity in ELISA does not relate to affinity (9). RAB-A and M-P
showed a very similar binding profile in ELISA, whereas M-O specifically only bound to
desmin. The highest affinity for desmin was measured for M-O (KD = 0,1 × 10–7 M)
closely followed by RAB-A (KD = 1.9 × 10–7 M), whereas M-P bound with 500fold lower
affinity to desmin than M-O (KD = 52 × 10–7 M).
To our knowledge this is the first compilation of human MAbs describing their fine
specificities to different antigens in several assay systems as well as their binding affinities.
We provide evidence that the staining patterns detected in the HEp-2 immunofluores-
cence assay routinely employed in many rheumatology laboratories can be produced by
antibodies binding with high affinity to autoantigens relevant for the disease (cartilage) as
well as foreign antigens (mycobacterial HSP60). To the best of our knowledge human
MAbs showing these crossreactivities have not been described before.
The data do not answer the question whether these antibodies or the B cells synthesiz-
ing them are involved in the pathogenesis of the disease. As shown here, the blood of
patient M contained not only the B cells from which M-O and M-P derived, but also an
IgG and IgM population with similar binding characteristics. It is therefore well conceiv-
able, that these were induced in a similar way as the antibodies binding to cytoskeletal
antigens in rats developing arthritis after immunization with mycobacterial HSP60 (12).
Acknowledgements
We thank U. LANGENBERG, Freiburg, for helpful comments, C. BEREK, Berlin, and H. PETER
,
Freiburg, for critical discussion of the manuscript and R. HUMMES for excellent technical assistance.
This work was supported by a fellowship of the Hans-Hench-Stiftung to T. M.
References
1. BRADWELL, A. R., R. P. STOKES, and G. D. JOHNSON. 1995. Atlas of HEp-2 patterns, KNP
Group Ltd. Redditch, 118 pp.
2. BORG A. A., P. T. DAWES, and D. L. MATTEY. 1993. Increased levels of IgA antibodies to cytoker-
atin-18 and epidermal keratin in rheumatoid arthritis. Arthr. Rheum. 36: 229.
3. JACOBS R., M. G. BUTLER, and D. L. SCOTT. 1990. Antibodies to cardiolipin and intermediate fil-
aments: A study of autoimmunity in rheumatoid arthritis. Clin. Rheumatol. 9: 509.
4. KATAAHA P. K., S. M. MORTAZAVI-MILANI, G. RUSSELL, and E. J. HOLBOROW. 1985. Anti-inter-
mediate filament antibodies, antikeratin antibody, and antiperinuclear factor in rheumatoid
arthritis and infectious mononucleosis. Ann. Rheum. Dis. 44: 446.

Human MAbs binding to intermediate filaments · 15
5. KURKI O., T. HELVE, and A. I. VIRTANEN. 1983. Antibodies to cytoplasmic intermediate filaments
in rheumatic diseases. J. Rheumatol. 10: 558.
6. MAYET W., E. HERMANN, M. BACHMANN, M. MANNS, and K.-H. MEYER ZUM BUESCHENFELDE.
1990. Correlation of anti-cytoskeleton antibody activities in synovial fluid with interleukin-6 in
patients with osteoarthritis and inflammatory joint disease. Klin. Wochenschr. 68: 685.
7. OLIVER J. M., J. L. SENÉCAL, and N. L. ROTHFIELD. 1985. Autoantibodies to the cytoskeleton in
human sera, In: Shay JW. ed. Cell and Muscle Motility, New York: Plenum Press 55.
8. OSUNG O. A., M. CH. and R. A., and E. J. HOLBOROW. 1982. Antibody to intermediate fila-
ments of the cytoskeleton in rheumatoid arthritis. Ann. Rheum. Dis. 41: 69.
9. AVRAMEAS S. 1991. Natural autoantibodies – from ‘horror autotoxicus’ to ‘gnothi seauton’.
Immunol. Today 12: 154.
10. CASTELL, M. 1996. Cytoskeletal autoantibodies. In: Peter JB & Shoenfeld Y. ed. Autoantibodies.
Amsterdam:, Elsevier Science 217.
11. TERNYNCK T., and S. AVRAMEAS. 1986. Murine natural monoclonal autoantibodies: A study of
their polyspecificities and their affinities. Immunol. Rev. 94: 99.
12. FRANCH A., C. CASTELLOTE, J. L. VILA, S. VILARO, and M. CASTELL. 1994. Anticytoskeletal
autoantibody development in Adjuvant Arthritis. J. Rheumatol. 21: 489.
13. NAPARSTEK Y., and P. H. PLOTZ. 1993. The role of autoantibodies in autoimmune disease. Ann.
Rev. Immunol. 11: 79.
14. HASKARD D. O., V. GUL, A. MORGAN, P. KATAAHA, N. A. SZAINES, and J. R. ARCHER. 1985.
Human monoclonal antibodies from patients with rheumatoid arthritis:crossreactions against cel-
lular constituents. Clin. Exp. Immunol. 61: 496.
15. LANDENBERG P. V., R. RZEPKA, and I. MELCHERS. 1999. Monoclonal autoantibodies from patients
with autoimmune diseases: Synovial fluid B lymphocytes of a patient with rheumatoid arthritis pro-
duced an IgG antibody recognizing J-sequences of IgÍ chains in a conformation dependent way.
Immunobiol. 200: 205.
16. LANDENBERG P. V., C.V. LANDENBERG, M. GRUNDDL, G. SCHMITZ, J. SCHÖLMERICH, and
I. MELCHERS. 1999. A new antigenic epitope localized within human kappa light chains specific
for RA and SLE. J. Autoimmunity 13: 83.
17. STEINERT P. M., and D. R. ROOP. 1988. Molecular and cellular biology of intermediate filaments.
Annu. Rev. Biochem. 57: 593.
18. FUCHS E., and K. WEBER. 1994. Intermediate filaments: Structure, dynamics, function, and
disease. Ann. Rev. Biochem. 63: 345.
19. MCADAM L. P., M. A. O’HANLAN, R. BLUESTONE, and C. M. PEARSON. 1976. Relapsing poly-
chondritis. Prospective study of 23 patient and a review of the literature. Medicine 55: 193.
20. ARNETT F. C., S. M. EDWORTHY, and D. A. BLOCH, et al. 1988. The American Rheumatism
Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthr. Rheum. 31: 315.
21. KURZIK-DUMKE U., C. SCHICK, R. RZEPKA, and I. MELCHERS. 1999. Overexpression of human
homologues of the bacterial DnaJ chaperone in the synovial tissue of arthritis patients. Arthr.
Rheum. 42: 210.
22. KEMPIS J. V., and G. KIENZLE. 1998. VCAM-1 (CD106) on primary human articular chondroc-
ytes: functional regulation of expression by cytokines and comparison with ICAM-1 (CD54) and
VLA-2. Arthr. Rheum. 41: 1296.
23. GRUNOW R., S. JAHN, and T. PORSTMANN, et al. 1988. The high efficiency, human B cell immor-
talizing heteromyeloma CB-F7. J. Immunol. Meth. 106: 257.
24. YAN T., H. BURKHARDT, T. RITTER, and F. EMMRICH. 1992. Specificity and T cell receptor b
chain usage of a human collagen type II-reactive T cell clone derived from a healthy individual. Eur.
J. Immunol. 22: 51.
25. MELCHERS I., J. JOOSS-RÜDIGER, and H. H. PETER. 1997. Reactivity patterns of synovial T-cell
lines derived from a patient with Rheumatoid Arthritis. I. Reactions with defined antigens and
autoantigens suggest the existence of multireactive T cell clones. Scand. J. Immunol. 46: 187.
26. KOGA T., A. WAND-WÜRTTEMBERGER, J. DERBRUYN, M. E. MUNK, B. SCHOEL, and S. H. E.
KAUFMANN. 1989. T cells against a bacterial heat shock protein recognize stressed macrophages.
Science 245: 1112.

16 · T. MENGE et al.
27. RUDOLPHI U., A. HOHLBAUM, B. LANG, H. H. PETER, and I. MELCHERS. 1998. The B cell rep-
ertoire of patients with rheumatoid arthritis. Frequencies and specificities of peripheral blood B
cells reacting with human IgG, human collagens, a mycobacterial heat shock protein and other
antigens. Clin. Exp. Immunol. 92: 404.
28. FRIGUET B., A. F. CHAFFOTTE, L. DJAVANDI-OHANIANCE, and M. E. GOLDBERG. 1985.
Measurements of the true affinity constant in solution of antigen-antibody complexes by enzyme-
linked immunosorbent assay. J. Immunol. Meth. 305.
29. VARON D., S. LINDER, E. GEMBOM, L. GUEDG, H. LANGEHEIM, A. BERREBI, and Z. ESHHAR
.
1990. Human monoclonal antibody derived from an Autoimmune Thrombocytopenic Purpura
patient, recognizing an intermediate filament’s determinant common to vimentin and desmin.
Clin. Immunol. Immunopathol. 54: 454.
30. SENÉCAL J. L., and J. RAUCH. 1988. Hybridoma lupus autoantibodies can bind major cytoskele-
tal filaments in the absence of DNA-binding activity. Arthr. Rheum. 31: 864.
31. KAMEI H. 1990. Location of nuclear antigen(s) recognized by DSB389 MAb, a monoclonal anti-
body against desmin, observed by confocal laser scanning fluorescence microscopy. Cell Biol. Int.
Rep. 14: 727.
32. ICHIKI S., STEMARIE, and J. L. SENÉCAL. 1992. Cross-reactive and non cross-reactive subsets of
autoantibodies to intermediate filament proteins: Analysis of sera from 379 connective tissue
patients and controls. Arthr. Rheum. 35: R42 (Abstract).
33. KAMEI H. 1996. Nuclear Antigens, as DNA, of DSB389 MAB and some other Anti-Desmin
Monoclonal Antibodies. Cell Biol. Intern. 20: 769.
34. MENGE, T., HUMMES, R., SCHLEDZ, M., LANGENBERG, U., PETER, H. H., and I. MELCHERS
.
1995. Functional characterization of human monoclonal antibodies with specificities related to
rheumatoid arthritis. Immunobiol. 194: 145. (Abstract).
35. FEIGE U., and W. VAN
359.
36. HANDLEY H. H., M. D. NGUYEN, D. T. Y
E
DEN. 1996. Infection, autoimmunity and autoimmune disease. EXS 77:
U, R. S. GUPTA, and J. H. VAUGHAN. 1995.
Purification of recombinant human Hsp60: use of a GroEL-free preparation to assess autoimmu-
nity in rheumatoid arthritis. J. Autoimmunity 8: 659.
37. RUDOLPHI U., R. RZEPKA, S. BATSFORD, S. H. E. KUAFMANN, K. VON DER
MARK, H. H. PETER,
and I. MELCHERS. 1997. The B cell repertoire of patients with rheumatoid arthritis. II. Increased
frequencies of IgG+ and IgA+ B cells specific for mycobacterial heat-shock protein 60 or human
type II collagen in synovial fluid and tissue. Arthr. Rheum. 40: 1409.
38. VAN EDEN W., J. HOLOSHITZ, Z. NEVO, A. FRENKEL, A. KLAJMAN, and I. R. COHEN. 1985.
Arthritis induced by a T-lymphocyte clone that responds to mycobacterium tuberculosis and to car-
tilage proteoglycans. Proc. Natl. Acad. Sci. USA 82: 5117.
39. CASALI P. , and A. L. NOTKINS. 1989. CD5+ B lymphocytes, polyreactive antibodies and the
human B-cell repertoire. Immunol. Today 10: 364.
40. EHRENSTEIN M. R., C. LONGHURST, and D. A. ISENBERG. 1993. Production and analysis of IgG
monoclonal anti-DNA antibodies from systemic lupus erythematodes (SLE) patients. Clin. Exp.
Immunol. 92: 39.
41. EHRENSTEIN M. R., C. LONGHURST, and D. A. ISENBERG. 1992. Production and analysis of
human monoclonal antibodies to myeloperoxidase. Immunol. 76: 617.
42. LU E. W., M. DEFTOS, H. TIGHE, D. A. CARSON, and P. P. CHEN. 1992. Generation and charac-
terization of two monoclonal self-associating IgG rheumatoid factors from a rheumatoid synovium.
Arthr. Rheum. 35: 101.
43. LU E. W., and M. DEFTOS. 1993. Generation and molecular analysis of two rheumatoid arthritis
synovial fluid derived IgG rheumatoid factors. Arthr. Rheum. 36: 927.
44. DIAW L., C. MAGNAC, O. PRITSCH, M. BUCKLE, P. M. ALZARI, and G. DIGHIERO. 1997.
Structural and affinity studies of IgM polyreactive natural autoantibodies. J. Immunol. 158: 968.
Dr. INGA
MELCHERS, Klinische Forschergruppe für Rheumatologie, Klinikum der Albert-Ludwigs-
Universität, Breisacher Str. 64, D-79106 Freiburg, Germany. Phone: 0049 761 270 5295, Fax: 0049
761 270 5298, e-mail: Melchers@nz11.ukl.uni-freiburg.de.