JOllRHAL OF

IMM~GICAL

ELSEVIER

Journal of Immunological Methods 186 (1995) 89-99

Radioimmunoassays for glutamic acid decarboxylase (GAD65)

and GAD65 autoantibodies using 35Sor 3H recombinant human
ligands *
Albert0 Falorni a~b**,Eva 6rtqvist

, Bengt Persson , he

Lernmark

a Department of Molecular Medicine, Karolinska Institute, Stockholm, Sweden

b Department of Internal Medicine and Endocrinological and Metabolic Sciences, UniLlersityof Perugia, Perugia. Italy
Department of Woman and Child Health, Karolinska Institute, Stockholm, Sweden
Received 14 December 1994; revised 20 April 1995; accepted 1 June 1995

Abstract

Autoantibodies are an important marker of human autoimmune diseases and the development of simple, precise
and reproducible immunoassays to detect autoantibodies is important to our understanding of human autoimmunity.
GAD65 autoantibodies occur frequently in insulin-dependent
diabetic patients and is a useful marker for IDDM. A
RIA to detect immunoreactive GAD65 has not been described. In the present study we describe a semi-automated
fluid-phase immunoassay for the rapid detection of GAD65 autoantibodies in human serum. We also developed a
sensitive RIA to determine immunoreactive human GAD65 in biological fluids and in vitro cell systems. Using in
vitro translated recombinant human GAD65 in a multiwell-adapted procedure, our GAD65Ab RIA combines high
specificity and sensitivity with a high capacity to analyze a large number of samples. In this report the three critical
steps in the GAD65Ab RIA, DNA preparation, in vitro translation and immunoprecipitation,
have been optimized.
In our RIA, GAD65Ab were detected in 116/155 (75%) new onset Swedish IDDM children and in l/85 (1.2%)
healthy controls. In an immunoassay to detect autoantibodies against the proinsulin converting enzyme 2 (PC-21 no
such antibodies were detected in IDDM patients. In the GAD65 RIA the lower detection limit was 2 ng/ml (31
fmol/ml). Our data demonstrate that autoantigen radioligands produced by in vitro translation are useful in RIA for
autoantibodies and autoantigens in studies of human autoimmunity.

Abbreviations: Ab, antibodies; BSA, bovine serum albumine; EDTA, ethylenediaminetetraacetic

acid; GAD, glutamic acid
decarboxylase; IAA, insulin autoantibodies; ICA, islet-cell antibodies; IDDM, insulin-dependent diabetes mellitus; IMP, immunoprecipitation; JDF, Juvenile Diabetes Foundation; LB, Luria-Bertani broth; PAGE, polyacrylamide gel electrophoresis; PC-2.
proinsulin converting enzyme-2; PCIAA, phenol-chloroform-isoamyl alcohol; RIA, radioimmunoassay; SDS, lauryl sulfate sodium
salt; TCA, trichloroacetic acid; Tris, tris(hydroxymethyl)aminomethane.
* Part of this paper was presented in abstract form at the First International Congress of the British Society of Immunology,
Brighton, UK December 1993.
* Corresponding author. At: Laboratory for Molecular Immunology, Department of Molecular Medicine, Karolinska Hospital,
M3:00, S-17176 Stockholm, Sweden. Tel.: (46)8-7296179; Fax: (46)-8-7296180.
0022-1759/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved
SSDI 0022-1759(95)00139-5

90

A. Falorni et al. /Journal

of Immunological Methods 186 (1995) 89-99

Keywords: Autoantibody; Radioimmunoassay; Insulin-dependent diabetes mellitus; Glutamic acid decarboxylase; In

vitro translation; Proinsulin converting enzyme-2

1. Introduction
Several autoantigens detected by autoantibodies in human autoimmune disorders have been
cloned and sequenced. Numerous immunoassays
with recombinant autoantigen have been developed to detect circulating autoantibodies as well
as the autoantigen itself. Solid phase assays such
as immunoblotting or enzyme-linked immunoassays are often used. This is possible when the
autoantibodies are high affinity and high titer and
not sensitive to conformational epitopes. In some
autoimmune disorders such as insulin-dependent
diabetes mellitus (IDDM) the loss of pancreatic
beta cells is thought to be a chronic autoimmune
process (Palmer and Lernmark, 1990) associated
with autoantibodies against insulin (Palmer et al.,
1983), glutamic acid decarboxylase (GAD) (Baekkeskov et al., 1982,199O; Karlsen et al., 1992b) or
islet cells (ICAXBottazzo et al., 1974; Bosi et al.,
1994). Serum exchange analyses have demonstrated that solid phase assays are unreliable in
the detection of insulin specific autoantibodies
(IAN (Greenbaum et al., 1992) or GAD autoantibodies (GAD-Ab) (Schmidli et al., 1994) in
IDDM.
The observation that antibodies against the M,
65000 isoform of glutamic acid decarboxylase
(GAD651 (Baekkeskov et al., 1982; Karlsen et al.,
1992a; Hagopian et al., 1993b) appear before
ICA or IAA (Atkinson et al., 1990) and that, in
first-degree
relatives
of diabetic
patients,
GAD65Ab correlate with progressive beta-cell
dysfunction better than ICA or IAA (Biirmeier et
al., 1991) suggests that GAD65Ab may have sufficient diagnostic specificity and sensitivity to predict IDDM. However, most IDDM patients occur
in the general population where the predictive
values of current immunological tests are low
(Bingley et al., 1993).
The cloning of the M, 65000 (GAD651 (Karlsen et al., 1991) and M, 67000 (GAD671 (Michelsen et al., 1991) isoforms made it possible to
develop recombinant antigen immunoassays to
detect GAD autoantibodies
(Hagopian et al.,

1993a; Velloso et al., 1993, Seissler et al., 1993;

Grubin et al., 1994; Petersen et al., 1994; Falorni
et al., 1994). These novel assays showed that
GAD65Ab, but not GAD67Ab, had a high diagnostic sensitivity for IDDM to support the hypothesis that GAD65 is a main autoantigen in
IDDM.
GAD65Ab screening of the general population
requires an assay able to analyse several hundred
samples in few days. Based on our previously
described radioligand binding assay to detect
GAD65Ab in human serum (Grubin et al.,
1992,1994), the aim of the present study was: (1)
to optimize in vitro transcription and translation
in order to radiolabel the recombinant human
autoantigens GAD65 and proinsulin converting
enzyme 2 (PC-2) (Smeekens and Steiner, 1990);
(2) to develop a high capacity, semi-automated
assay for autoantibodies; and (3) to develop a
sensitive radioimmunoassay to detect immunoreactive GAD65.

2. Materials and methods

2.1. Human sera and rabbit antisera
Sera from 155 newly diagnosed Swedish insulin-dependent
diabetic patients (male/female
ratio: 84/71; age 1-18 years) and from 85 healthy
Swedish individuals (male/female
ratio: 37/48;
age 8-25 years) were used in the study. The
diabetic patients were diagnosed at the St. GGrans
Hospital, Stockholm, Sweden, between 1986 and
1992. The serum samples were kept frozen at
- 20C until the analysis in 1993-1994. The study
was approved by the ethics committee at Karolinska Institute, Stockholm, Sweden.
Sera from eight GAD65Ab positive IDDM
patients (Grubin et al., 1994) already used extensively in the Immunology of Diabetes Workshop
to standardize ICA (Bonifacio et al., 1988) were
used, as in our first assay (Grubin et al., 1994), to
validate the semi-automated assay.

A. Falomi et al./Joumal

of Immunological Methods 186 (1995) 89-99

Rabbits were immunized

with synthetic
GAD65 peptides as described (Li et al., 1995).
Antiserum R7309 was raised against a synthetic
peptide corresponding to amino acid 4-22 of
human GAD65, antiserum R8100 was against
amino acid 250-269 of human GAD65 and
R10266 was against amino acid 2-19 of human
GAD67.
A rabbit antiserum generated against human
PCZpeptide
4 (a kind gift of Dr. Donald F.
Steiner, Chicago, IL, USA) was used to test the
immunoreactivity of the radiolabelled recombinant human PC2.
2.2. Plasmid DNA preparation
A full-length human GAD-2 cDNA clone
(pEx9), derived from two overlapping cDNA
clones (Grubin et al., 19941, inserted in the pcDNAII (Invitrogen, San Diego, CA, USA) plasmid
was used to transform competent DHSaF E. coli
cells.
The pEx9 plasmid DNA was prepared from an
overnight DHSaF bacterial suspension using two
different procedures. In the first procedure a 250
ml bacterial suspension in Luria-Bertani (LB)
broth was used and the plasmid DNA extracted
and purified using a commercially available kit
(Maxi-kit, Qiagen, Chatsworth, CA, USA) following the manufacturers instructions. The DNA
was resuspended in 200 ~1 Tris 10 mM, EDTA 1
mM (TE) buffer, and the DNA concentration
evaluated spectrofluorometrically
(Labarca and
Paigen, 1980) (typically between 1.5 and 3
mg/ml).
In three separate experiments,
small-scale
preparations of plasmid DNA were obtained by
lysing the cells from a 20 ml LB bacterial suspension in alkali and after overnight incubation at
37C using previously described standardised procedures (Sambrook et al., 1990). DNA was purified by phenol:chloroform:isoamil
alcohol
(25:24:1), precipitated in absolute ethanol (2 ~01s.)
and 0.1 ~01s. 3 M sodium acetate pH 5.2, washed
in 70% ethanol and resuspended in 50 ~1 of
Tris-EDTA (TE) buffer. The DNA concentration
was 320-500 *g/ml.
In both of the procedures the quality of the
pEx9 plasmid preparation was evaluated by lin-

91

earisation of the DNA from a 2 ~1 aliquot with

Uppsala, Sweden) and analysis
by standard agarose gel electrophoresis.
Full-length human PC2 cDNA clone (pPC21
inserted into a pBluescript vector (Stratagene
Cloning Systems, La Jolla, CA, USA) was kindly
donated by Donald F. Steiner, Chicago, IL, USA.
The pPC2 plasmid DNA was prepared as described above using 250 ml of overnight bacterial
suspension and the commercially available kit
from Qiagen.

XbaI (Pharmacia,

2.3. In vitro coupled transcription /translation

human radiolabelled GAD65 and PC2

of

Recombinant human 35S-GAD65 or 35S-PC2

was produced in an in vitro coupled transcription/translation
system with SP6 (35S-GAD65) or
T3 (3sS-PC2) RNA polymerase and nuclease
treated rabbit reticulocyte lysate (Promega, Madison, WI, USA). Plasmid DNA (2 pg) was incubated for 90 min at 30C in a 50 ~1 reaction
mixture containing 25 ~1 rabbit reticulocyte lysate
(Promega, Madison, WI, USA), 1 ~1 RNA polymerase (40 U/pi) (Promega), 1 ~1 1 mM amino
acid mixture (minus methionine) (Promega), 1 ~1
RNasin (20 U/p11 (Promega) and 4 ~1 translation grade [35S]methionine ( > 1000 Ci/mmol) at
10 mCi/ml (NEN Research Products, Boston,
MA, USA). After completion of the reaction, 2
~1 of the translation product were incubated for
10 min at 37C with 1 M NaOH/2% H,O, and
precipitated with 25% trichloroacetic acid (TCA)
for 30 min on ice. TCA precipitated proteins
were collected on a Whatman GF/A glass fiber
filter (Whatman, Maidstone, UK) and washed
five times with 5% TCA. Total precipitable radioactivity was evaluated in a liquid scintillation
analyzer (Tri-Carb 1550, Packard Instrument
Company, Meriden, CT, USA).
The in vitro translated 3sS-GAD65 and 3sS-PC2
were kept at -70C until used in the radioimmunoassays, within 5 weeks.
In three separate experiments, the in vitro
transcription and translation of 35S-GAD65 were
performed following a two step procedure as
previously described (Grubin et al., 1994). In another three experiments, 3H-GAD65 was pro-

92

A. Falorni et al. /Journal

of Immunological Methods 186 (1995) 89-99

duced by coupled in vitro transcription/translation using 1 ~1 1 mM amino acid mix without

leucine and 5 ~1 [3H]leucine (150 Ci/mmol) at 5
mCi/ml LAmersham) and following the procedure described above.
2.4. Radioimmunoassays for GAD65 and PC2 antibodies

Multiscreen-DV 96-well filtration plates, 0.65

pm hydrophilic PVDF (Millipore AB, Vastra
Frolunda, Sweden) were pretreated
with 200
pi/well of Tris/NaCl buffer (150 mM NaCl, 20
mM Tris, pH 7.4) for 1 h at 4C and then
incubated for 2 h at room temperature with 1%
bovine serum albumin (BSA) (Sigma Chemical
Co., St. Louis, MO, USA) in Tris/NaCl buffer.
After the BSA coating, the buffer was removed
and the filters washed twice with 200 ~1 0.05%
Tween 20 (Sigma) in Tris/NaCl buffer, and stored
for 2-3 days at 4C in the same buffer. The use of
Triton X-100 (Sigma) instead of Tween 20 was
associated with a higher non specific binding of
labelled GAD65.
In both the GAD65 antibody (GAD65Abl and
the PC2 antibody (PC2Ab) specific RIAs, the in
35S-radiolabelled recombinant
vitro translated
human antigen was directly immunoprecipitated
with human serum. In previous studies (Grubin et
al., 1994; Falorni et al., 1994) we showed that
using our cDNA clone (pEx9), more than 95% of
the translated GAD65 was full-length. Preliminary results (data not shown) indicated that the
use of more than 20 000 cpm of TCA precipitable
35S-GAD65 was associated with a higher non
specific precipitation of the radioligand and a
lower ratio between the immunoprecipitation
of
the positive and the negative standard sera.
Therefore, in each analysis, aliquots containing
20000 cpm of TCA precipitable 35S-GAD65 or
35S-PC2, in triplicate, were diluted in 60 ~1 immunoprecipitation
buffer (20 mM Tris, 150 mM
NaCl, pH 7.4, 0.15% Tween 20, 0.1% aprotinin,
10 mM benzamidine, 0.1% BSA) before addition
of 2.4 ~1 of human serum (final serum dilution
l/25). After an overnight incubation at 4C on a
rotating platform, 50 ~1 of the immunoprecipi-

tated samples were transferred, in triplicate, into

a treated multiwell plate. Immediately before addition of the samples, the plates were washed
twice with washing buffer (20 mM Tris, 150 mM
NaCl, pH 7.4, 0.1% Tween 20, 0.1% BSA), and
the buffer discarded. Overnight incubation in the
multiwell plate was evaluated in four separate
experiments and found to be associated with a
2-3-fold higher background, due to non-specific
binding of the antigen to the PVDF filter.
The antibody-bound GAD65 or PC2 were separated from free antigen by addition of 50 ~1 of
50% protein A-Sepharose (Zymed, San Francisco, CA, USA) in each well and incubation for
45 min at 4C on a plate shaker.
Sepharose beads were washed nine times with
200 ~1 washing buffer using a vacuum manifold
(Millipore). The multiwell filters were dried under lamp for 15 min, transferred to 7 ml glass
scintillation vials using a Multiscreen multiple
punch system (Millipore), and scintillation fluid
(Ultima Gold, Packard Instrument Company) (4
ml/vial) then added. The Sepharose-bound radioactivity was determined in a liquid scintillation
counter.
In two experiments, 20000 cpm of TCA precipitable 3H-GAD65 was immunoprecipitated,
in
quadruplicate, with scalar dilutions (l/25-1/250)
of the JDF positive and one healthy negative
(l/25) stand ar d serum, following the procedures
described above.
2.5. Expression of GAD65 and PC2 antibody levels
and determination of an upper level of normal
values
To control for inter-assay variation, GAD65
antibody levels were expressed as a relative index
(GAD65 index) using a positive and three negative control sera in each assay. The positive control serum was the Juvenile Diabetes Foundation
world standard (JDF standard) for islet cell antibodies (ICA). This serum has been previously
shown to be positive for GAD65Ab (Grubin et
al., 1994; Hagopian et al., 1993a). The sera from
three healthy Swedish individuals served as negative controls. A GAD65Ab index (cpm of the

A. Falomi et al. /Journal

of Immunological Methods 186 (1995) 89-99

unknown sample - average cpm of three negative

standards)/(cpm
of the positive standard average cpm of three negative standards) was
calculated.
The effects of serum titration on the final
GAD65 index was evaluated in the eight positive
standard sera. From this analysis it was found
that sera with a GAD65 index above 0.75 had to
be reanalysed at a dilution of l/250. The final
GAD65 index was evaluated using similar serum
dilutions of the positive and the three negative
standard sera. The upper level of normal of the
assay was calculated using the average +3 SD of
the healthy Swedish individuals.
Similarly, PC2 antibody levels were expressed
as a relative index (PC2 index) using the specific
rabbit antiserum as a positive and three sera from
healthy individuals as negative controls. The upper level of normal for the assay was calculated
using the average +3 SD of a population of 50
healthy Swedish individuals.
2.6. Radioimmunologicat determination of the concentration of GAD65 in the in vitro translation
reaction and in human serum
The concentration of recombinant GAD65 at
the end of the in vitro translation reaction, and in
15 sera from healthy individuals was evaluated.
Non radiolabelled in vitro translated recombinant
GAD65 was prepared, in three different experiments, in the presence of cold 1 mM methionine
(Sigma).
In the GAD65 RIA, in vitro translated 35SGAD65 served as tracer and the serum from a
GAD65Ab-positive patient with so-called StiffMan Syndrome (a kind gift of Tiinamaija Tuomi,
Helsinki, Finland) served as antiserum. As a standard, recombinant human GAD65, expressed in
and purified from a bacculovirus system (a kind
gift from John Robertson, Synectics Biotechnology, Stockholm, Sweden) was used. The purity of
the standard GAD65, as evaluated by SDS-polyacrylamide gel electrophoresis,
was more than
98% and the concentration was determined by
total aminoacid composition.
In the GAD65 RIA, 12000 cpm of TCA pre-

93

cipitable 35S-GAD65, in triplicate, was diluted in

51 ~1 of IMP buffer. Before addition of 3 ~1 of a
l/5000 dilution of the Stiff-Man serum (final
serum dilution l/100000), 6 ~1 of a scalar dilution of standard GAD65 in IMP buffer (corresponding to an original concentration
ranging
from 0 to 130 ng/ml) or 6 ~1 of a l/100 or a
l/200 dilution in IMP buffer of the cold in vitro
translated GAD65 were added. The serum of the
patient with Stiff-Man Syndrome was negative for
GAD67Ab when tested at a l/100000 dilution.
To determine the concentration of GAD65 in
human serum, the standard GAD65 was diluted
in GAD65-depleted
human serum (by pre-adsorption on a column of GAD6 monoclonal antibody) instead of IMP buffer. 6 ~1 of human
serum, in triplicate, were used in the GAD65
RIA.
After overnight incubation at 4C, on a rotating platform, the immunoprecipitated
radioactivity was evaluated by separating the antibody
bound 35S-GAD65 using the procedures
described above for the GAD65Ab RIA.
The final GAD65 concentration in both the in
vitro translation reaction and in human serum
was estimated by constructing a semi-logarithmic
standard curve using the immunoprecipitated
radioactivity in the presence of scalar dilutions of
the standard GAD65.
2.7. Statistical analysis
The intra- and inter-assay coefficients of variation of the GAD65Ab and the PC2Ab assay and
the percent 3S-GAD65 or 35S-PC2 immunoprecipitated were calculated for the positive and the
three negative control sera.
Difference in GAD65Ab or PC2Ab levels between IDDM patients and healthy controls was
evaluated using the non-parametric Mann-Whitney test.
In the GAD65 RIA, the standard curve was
tested by linear regression analysis and the intraassay and the inter-assay coefficients of variation
calculated for five dilutions of the standard
GAD65 and 3 analyses of the in vitro translated
GAD65, respectively.

94

A. Falorni et al. /Journal

of Immunological Methods 186 (1995) 89-99

3. Results

100,

P
%

iI

3.1. In vitro translation of radiolabelled recombinant human GAD65 and PC2

The efficiency of translation was evaluated by

measuring the TCA precipitated
radioactivity
(cpm/pl of translated GAD65 or PC2). A comparison between the two-step and the coupled
transcription/translation
procedures, as well as a
comparison between the two plasmid preparation
techniques was carried out for 35S-GAD65 as
shown in Table 1. Large-scale preparation of
plasmid DNA/QIAGEN
purification and use of
the coupled in vitro transcription/translation
procedure resulted associated with the highest
efficiency of in vitro translation of 35S-GAD65.
The coupled transcription/translation
was also
evaluated using [ 3H]leucine. The percentage incorporation
of [3H]leucine in the translated
GAD65 was similar to that observed using
[ 35S]methionine (Table 1).
In vitro translation of 35S-PC2 resulted in
154600 + 23 400 cpm of TCA precipitable protein/pl of reaction, corresponding to 13.4 k 1.6
percentage incorporation of [ 35Slmethionine.
3.2. Validation of the GAD65 and PC2 antibody
assays

The immunoreactivity of the in vitro translated

recombinant human 35S-GAD65 and 35S-PC2 was
tested using the specific rabbit antisera. 35SGAD65 was immunoprecipitated
by the GAD65specific R7309 and R8100 antisera but neither by

35S-GAD65

%-PC2

Fig. 1. Percentage immunoprecipitation of recombinant human 35S-GAD65 and 35S-PC2 with GAD65-, GAD67- and
PCZ-specific rabbit antisera. As a negative control, the percentage immunoprecipitation obtained with healthy human
serum is shown.

the GAD67-specific R10266 nor the PCZpeptide

4 specific rabbit antisera (Fig. 1). Similarly, 35SPC2 was immunoprecipitated by the PC2-peptide
specific rabbit antiserum but neither by the
GAD65-specific
R7309 and R8100 nor the
GAD67-specific R10266 antisera (Fig. 1).
In 84 healthy Swedish individuals, the GAD65
index was - 0.0065 + 0.0105 (mean f SD) and was
normally distributed (Fig. 2). The upper level of
normal (mean + 3 SD) was estimated to be 0.025.
In one healthy serum, the GAD65 index (0.825)
was significantly higher than the upper level of
normal. This sample was not considered in the
above analysis.

Table 1
Efficiency of the in vitro translation of radiolabelled GAD65

Coupled transcription
translation of 35S-GAD65
Two steps transcription
translation of 35S-GAD65
Coupled transcription
translation of 3H-GAD65
Values are mean f SEM cpms/pl
ND: not determined.

Small-scale DNA preparation

PCIAA purification

Large-scale DNA preparation

QIAGEN purification

45670f3300
(4.8 f 0.9%)
95 180 f 7 740
(9.9 * 1.4%)
ND

567 250 f 46 620

(48.9 f 3.9%)
109 875 f 6 890
(11.4 f 1.2%)
180 750 f 7 330
(33.3 + 1.4%)

of translated GAD65 (percentage incorporation) of ten reactions.

A. Falomi et al. /Journal

of Immunological Methods 186 (1995) 89-99

The average percentage 35S-GAD65 immunoprecipitation in five assays was 53% with the
positive and 1.9%, 2.1%, and 2.7%, respectively
with the three negative standard sera. The ratio
between the immunoprecipitation
of the positive
and the average of the three negative standard
sera ranged between 2O:l and 25:l.
The intra-assay and inter-assay coefficients of
variation were 3% and 9% respectively for the
positive, and ll-17% and lo-14% respectively
for the negative standards.
When the GAD65 index was plotted as a function of the serum dilution in the eight GAD65Ab
positive IDDM standard sera linear dilution
curves were obtained for low level antibody positive sera and saturated binding for the high level
sera (Fig. 3). Using the JDF standard serum as
the GAD65Ab positive standard we detected
GAD65Ab in all of the seven positive standard
sera tested (GAD65 index 0.05-1.08). At increasing serum dilutions, the frequency of GAD65Ab
decreased as the IDDM sera gradually overlapped within the range of the negative healthy
controls. The GAD65 index was found to be a
linear function of the serum dilution in five IDDM
sera with an index between 0.05 and 0.62, but not
in two sera with an index above 1.00.
In 50 healthy Swedish individuals, the PC2Ab
index was -0.005 k 0.008 (mean 5 SD) and the

95

1.2

0.8

11100011250

11100

l/50

SERUM

1132

l/25

DILUTION

Fig. 3. Dilution curves of eight GAD65Ab

sera. Values are mean of triplicates.

positive

standard

upper level of normal (mean + 3 SD) was estimated to be 0.020.

The average 35S-PC2 immunoprecipitated with
the rabbit serum in five assays was 65% with the
positive standard sera and 1.3%, 1.2%, and 1.4%,
respectively with the three negative standard sera.
The ratio between the immunoprecipitation
of
the positive and the average of the three negative
standard sera ranged between 35:l and 45:l.
The intra-assay and inter-assay coefficients of
variation were 2% and 10% respectively for the
positive, and 12-15% and 9-15% respectively for
the negative standards. The sensitivity of the
PC2Ab RIA was tested with serial dilutions of
the rabbit PCZpeptide 4 antiserum. The endpoint dilution titre of this antiserum, as evaluated
in the RIA, was l/20000.
3.3. GAD65Ab and PC2Ab in IDDM patients and
healthy controls

Fig. 2. Frequency
distribution
healthy control subjects.

of the

GAD65

index

in 84

The levels of GAD65Ab, expressed as the

GAD65 index, were significantly higher in IDDM
patients (median 0.297; range - 0.020-2.402) than
in healthy individuals (median -0.007; range
-0.035-0.825)
(p < 0.005) (Fig. 4). GAD65Ab
were found in 75% (116/155) IDDM patients
and in 1.2% (l/85) healthy individuals. In IDDM

96

A. Falomi et ai. /Journal

of Imnwnofogicaf

Methods 186 (1995) 89-99

2.5

r=0.99

4,500

4,000

i
1

0.025

6
GAD65

IDDM

32.5

65

130

CONTROLS

Fig. 4. GAD65 index in IDDM patients and healthy controls.

The dotted line shows the upper level of normal of the
GAD65Ab assay.

patients, GAD65Ab occurred more frequently in

female (81.7%, 58/71) than in male (67.9%,
57/84) subjects (x2 = 3.86; p = 0.049). Furthermore, the median GAD65 index was significantly
higher (p = 0.048) in GAD65Ab positive female
(median 0.405, range 0.028-2.289) than in male
(median 0.223, range 0.026-2.402) IDDM patients. The occurrence of GAD65Ab in male
IDDM patients was related to the age at onset
(Table 2), being higher among those patients more
than 12 years old than among those less than 12
years old.
The levels of PC2Ab, expressed as the PC2
index, did not differ between IDDM patients
(median:
- 0.004; range - 0.025-0.016)
and
healthy control subjects (median -0.005; range
-0.027-0.015).
None of the sera from the 50

Table 2
Occurrence of GAD65 autoantibodies
relation to age and sex

16

@g/ml)

in IDDM patients in

Age

Males

Females

O-5.9 years
6-11.9 years
12-18 years

lo/17 (58.8%) a
24/39 (62.5%) a
23/28 (82.1%) b

12/15 (80%)
32/40 (80%)
14/16 (87.5%)

Numbers are frequencies of GAD65Ab-positive patients.

a p = 0.03 vs. females of similar age, x2 = 4.66.
bp < 0.05 vs. 0-11.9-year-old males, x2 = 3.93.

Fig. 5. Representative standard curve for GAD65 RIA (r =

0.99; p < 0.001). Note that the concentration of GAD65 is
expressed logarithmically on the x axis.

IDDM patients tested had a PC2 index above the

upper level of normal.
3.4. GAD65 radioimmunoassay
A representative
standard curve from the
GAD65 RIA is shown in Fig. 5. The lower detection limit of the RIA was 2 ng/ml (31 fmol/ml)
and the intra-assay and inter-assay coefficients of
variation were 9.2% and ll.l%, respectively.
Analysis of the in vitro translated cold GAD65,
at dilution of l/100 and l/50, resulted in a
GAD65 concentration of 7.5 k 0.9 ,ug/ml. This
concentration
was similar to that (9.2 f 1.1
pgg/ml) predicted from the percentage incorporation (49%) of [35S]methionine into the in vitro
translated GAD65, the concentration (1.1 f 0.1
pg/ml) and specific activity (1000 Ci/mmol) of
the [35S]methionine used, and the tota number
of methionine residues in the human GAD65
(n = 25).
The GAD65 concentration in healthy human
sera varied between 2 and 6 ng/ml in ten samples, 12.5 ng/ml in one sample and undetectable
(concentration < 2 ng/ml) in four samples.
3.5. Immunoreactiuity of 3H-GAD65
Immunoreactivity
immunoprecipitation

of 3H-GAD65 was tested by

with the JDF positive stan-

A. Falorni et al. /Journal

of Immunological Methods 186 (19951 89-99

Table 3
Immunoprecipitation
of H-GAD65
and 35S-GAD65 with the
JDF positive and one healthy negative standard serum
Serum (dilution)

ImmunoprecipiImmunoprecipitation of
tation of
3H-GAD65 (%Io) 3SS-GAD65(%)

Healthy negative (l/251

JDF positive (l/251
JDF positive (l/32)
JDF positive (l/501
JDF positive (l/100)
JDF positive (l/250)

3.Ok 0.1
50.7 + 1.5
45.0+ 1.0
32.7k2.1
l&6+0.9
9.5~05

Values

2.1 kO.2
53.2kO.9
49.3 f 1.2
35.5 k 2.3
19.6i 1.9
10.1 rf: 1.0

are mean + SD of quadruplicates.

dard serum at different dilutions, and with a

negative standard serum. As shown by the comparison with the values obtained when immunoprecipitating 35S-GAD65 with the same sera (Table 3), 3H-GAD65 was immunoreactive and potentially useful for the detection of GAD65Ab.

4. Discussion
In the present paper we describe in detail a
simple and semiautomated radioimmunoassay to
detect GAD65 autoantibodies and to determine
GAD65 concentration in a large number of samples. Additionally, we have developed a RIA to
determine the levels of autoantibodies
to the
converting enzyme PC2 and we demonstrate that
no such autoantibodies could be detected in 50
IDDM sera.
A large body of evidence indicates that GAD65
autoantibodies are strongly associated with IDDM
and might be potentially useful for the diagnosis
and identification of subjects at risk of developing
IDDM.
Several methods have been proposed for the
detection of GAD65 autoantibodies
in human
serum. Our group has previously suggested
(Grubin et al., 1992,1994) an alternative strategy
based on the in vitro transcription and translation
of recombinant human GAD65 radiolabelled with
[35S]methionine. The efficiency of this approach
was recently confirmed (Petersen et al., 1994;
Falorni et al., 1994). All these studies demonstrated that in vitro translated recombinant

97

GAD65 is immunoreactive and indistinguishable

from the native protein. Moreover, the fact that
radiolabelled GAD65 can be produced in an in
vitro translation system permits less highly specialized laboratories to repeat the procedure
without the need to express and purify GAD65
from eukaryotic cells.
In our RIA we found GAD65Ab in 75% IDDM
patients (sensitivity of the assay) and in only 1.2%
healthy controls (specificity of the assay 98.8%).
The adaptation of the assay to a multiwell system
(Millipore) represents a major improvement over
our previous radiobinding assay and allowed a
single person to analyze more than 400 samples
per week. The average work time for a single
assay (one multiwell plate: 28 samples plus 4
standards) was 2 h. A major improvement was
achieved by optimizing the purification of plasmid DNA which was critical for the efficiency of
the in vitro translation. The combination of a
highly pure plasmid DNA preparation and of a
coupled transcription/translation
procedure increased 5-lo-fold the efficiency of the translation
(evaluated as percentage incorporation of the
tracer). Moreover, we had previously demonstrated that the use of our cDNA clone, containing an appropriate 5 untranslated sequence was
associated with high quality of in vitro translated
GAD65, as evaluated by SDS-PAGE analysis
(Grubin et al., 1994; Falorni et al., 1994). Contamination products of a lower molecular weight
were minimal in our system but represent a major
problem in other similar assays (Petersen et al.,
1994). As a consequence, purification of the in
vitro translated GAD65 was not required in our
system. The adaptation of the procedure to a
coupled transcription/translation
system represents a major improvement because of the improved purity of the autoantigen and the reduction in work time with no need to handle mRNA.
Furthermore, we demonstrated that labelling
GAD65 with r3Hlleucine generated similar results thereby confirming the efficiency of our
transcription/translation
system and opening up
interesting possibilities of combining different autoantigens in a single assay.
Little information is available on the concentration of GAD65 in biological fluids and in vitro

98

A. Falorni et al. /Journal

of Immunological Methods I86 (1995) 89-99

cell systems. In the present paper we describe a

sensitive RIA to determine GAD65 in biological
samples and we have used this to determine the
concentration of GAD65 produced in our in vitro
translation system. Subsequently, we focused our
attention on the concentration of GAD65 in human serum. However, from the preliminary results we obtained we can conclude that the sensitivity of our RIA was probably not adequate and
only in a few samples could the concentration of
GAD65 be estimated.
Using coupled in vitro transcription/translation we were able to produce radiolabelled recombinant human PC2 and to develop a PC2Ab
RIA. Interestingly, none of the tested diabetic
sera proved positive for PC2Ab suggesting that
the autoimmune
reaction
against the islet
molecules is restricted to specific targets, such as
GAD65.
In conclusion we have developed a sensitive
and semiautomated RIA to detect GAD65 autoantibodies. In particular, we have optimized
our previosuly described procedure (Grubin et
al., 1994; Falorni et al., 19941, both increasing the
capacity of the assay and reducing the costs and
the work time/sample.
The resulting semiautomated and standardized GAD65Ab RIA with in
vitro translated 35S-GAD65 is highly sensitive,
highly specific, has a high capacitance, is easy to
perform, and is economical. The large scale application of assays similar to that described will
make possible the screening of the general population in order to identify subjects at risk of
autoimmune disease.

Acknowledgements

We thank Annemette
Borch and Tiziana
DAndrea for their superb technical assistance in
the GAD65Ab analysis, Susanne Sandberg for
her help in the adaptation of the Millipore Multiscreen System to our needs. Albert0 Falomi was
supported by a Juvenile Diabetes Foundation International Fellowship. This work was supported
by the Swedish Medical Research Council (104081,
the National Institutes of Health (DK 42654), the
Novo Nordisk A/S, the Nordisk Insulin Fund,

the Swedish Hoechst Foundation, the Swedish

Diabetes Association, the Tore Nilsons Foundation, the Petrus and Augusta Hedlunds Foundation and Associazione Italiana Lions per il Diabete.

References
Atkinson, M.A., Maclaren, N.K., Scharp, D.W., Lacy, P.E.
and Riley, W.J. (1990) 64,000 M, autoantibodies as predictors of insulin-dependent diabetes. Lancet 335, 1357-1360.
Baekkeskov, S., Nielsen, J.H., Marner, B., Bilde, T., Ludvigsson, J. and Lernmark, A. (1982) Autoantibodies in newly
diagnosed diabetic children immunoprecipitate
human
pancreatic islet cell proteins. Nature 298, 167-169.
Baekkeskov, S., Aanstoot, H.-J., Christgau, S., Reetz, A.,
Solimena, M., Cascalho, M., Folli, F., Richter-Olesen, H.
and De Camilli, P. (1990) Identification of the 64 K
autoantigen in insulin-dependent diabetes as the GABAsynthesizing enzyme glutamic acid decarboxylase. Nature
347, 151-156.
Bingley, P.J., Bonifacio, E. and Gale, E.A.M (1993) Can we
really predict IDDM? Diabetes 42, 213-20.
Bonifacio, E., Lemmark, A. and Dawkins, R.L. (1988) Serum
exchange and use of dilutions have improved precision of
measurement of islet cell antibodies. J. Immunol. Methods
106, 83-88.
Bosi, E., Bonifacio, E. and Bottazzo, G.F. (1994) Autoantigens in IDDM. Diabetes Rev. 1, 204-214.
Bottazzo, G.F., Florin-Christensen, A. and Dotiiach, D. (1974)
Islet-cell antibodies in diabetes mellitus with autoimmune
polyendocrine deficiencies. Lancet 2, 1279-1283.
Birmeier, H., McCulloch, D.K., Neifing, J.L., OWarnock, G.,
Rajotte, R.V., Palmer, J.P. and Lernmark, A. (1991) Risk
for developing type 1 (insulin-dependent) diabetes mellitus
and the presence of islet 64 K antibodies. Diabetologia 34,
727-733.
Falorni, A., Grubin, C.E., Takei, I., Shimada, A., Kasuga, A.,
Maruyama, T., Ozawa, Y., Kasatani, T., Saruta, T., Li, L.
and Lernmark, A. (1994) Radioimmunoassay detects the
frequent occurrence of autoantibodies to the M, 65,000
isoform of glutamic acid decarboxylase in Japanese insulin-dependent diabetes. Autoimmunity 19, 113-125.
Greenbaum, C.J., Palmer, J.P., Kuglin, B., Kolb, H. and
participating laboratories (1992) Insulin autoantibodies
measured by RIA methodology are more related to IDDM
than those measured by ELISA. J. Clin. Endocrinol.
Metab. 74, 1040-1044.
Grubin, C.E., Daniels, T., Karlsen, A.E., Boel, E., Hagopian,
W.A. and Lernmark, k (1992) The cDNA-directed, in
vitro-synthesized nascent peptide of glutamic acid decarboxylase (GAD21 is the autoantigen in insulin-dependent
diabetes. Clin. Res. 40, 299A [abstract].
Grubin, C.E., Daniels, T., Toivola, B., Landin-Olsson, M.,

A. Falorni et al. /Journal

of Immunological Methods 186 (1995) 89-99

Hagopian, W.A., Li,OL.,Karlsen, A.E., Boel, E., Michelsen,

B. and Lernmark, A. (1994) A novel radioligand binding
assay to determine diagnostic accuracy of isoform-specific
glutamic acid decarboxylase antibodies in childhood
IDDM. Diabetologia 37, 344-350.
Hagopian, W.A., Karlsen, A.E., Gottslter, A., Landin-Olsson,
M., Grubin, C.E., Sundkvist, G., Petersen, J.S., Boel, E.,
Dyrberg, T. and Lernmark, A. (1993a) Quantitative assay
using recombinant human islet glutamic acid decarboxylase (GAD65) shows that 64K autoantibody positivity at
onset predicts diabetes type. J. Clin. Invest. 91, 368-374.
Hagopian, W.A., Michelsen, B., Karlsen, A.E., Larsen, F.,
Moody, A., Grubin, C.E., Rowe, R., Petersen, J., McEvoy,
R. and Lemmark, A. (1993b) Autoantibodies in insulindependent diabetes primarily recognize the M, 65,000
rather than the M, 67,000 isoform of glutamic acid decarboxylase. Diabetes 42, 631-636.
Karlsen, A.E., Hagopian, W.A., Grubin, C.E., Dube, S., Disteche, C.M., Adler, D.A., Barmeier, H,., Mathewes, S.,
Grant, F.J., Foster, D. and Lernmark, A. (1991) Cloning
and primary structure of a human islet isoform of glutamic
acid decarboxylase from chromosome 10. Proc. Natl. Acad.
Sci. USA 89, 2115-2119.
Karlsen, A.E., Hagopian, W.A., Petersen, J.S., Boel, E., Dyrberg, T., Grubin, C.E., Michelsen, B.K., Madsen, O.D.,
Lernmark, A. (1992a) Recombinant glutamic acid decarboxylase (representing the single isoform expressed in
human islets) detects IDDM-associated 64,000-M, autoantibodies. Diabetes 41, 1355-1359.
Karlsen, A.E., Michelsen, B.K., Petersen, J.K., Hagopian,
W.A., Grubin, C;E., Boel, E., Madsen, O., Stenger, D.P.
and Lernmark, A. (1992b) Glutamic acid decarboxylase:
an autoantigen in insulin-dependent diabetes mellitus. Diabetes Nutr. Metab. 5 (suppl. I), 151-154.
Labarca, C. and Paigen, K. (1980) A simple, rapid and sensitive DNA assay procedure. Anal. Biochem. 102, 344-352.
Li, L., Jiang, J.. Hagopian, W.A, Karlsen, A.E., Skelly, M.,
Baskin, D.G. and Lernmark, A. (1995) Differential detection of rat islet and brain glutamic acid decarboxylase
(GAD) isoforms with sequence-specific peptide antibodies. J. Histochem. Cytochem. 43, 53-59.
Michelsen, B.K., Petersen, J.S., Boel, E., Moldrup, A., Dyrberg, T. and Madsen, O.D. (1991) Cloning, characteriza-

99

tion and autoimmune recognition of rat islet glutamic acid

decarboxylase in insulin-dependent diabetes mellitus. Proc.
Natl. Acad. Sci. USA 88, 8724-8758.
Palmer, J.P. and Lernmark, A. (1990) Pathophysiology of
Type 1 (insulin-dependent) diabetes. In: H. Rifkin and D.
Porte (Eds.), Diabetes Mellitus, Theory and Practice. Elsevier, Amsterdam, pp. 414-435.
Palmer, J.P., Asplin, C.M., Clemens, P., Lyen, K., Tatpati, O.,
Raghu, P.K. and Paquette, T.L. (1983) Insulin antibodies
in insulin-dependent
diabetes before insulin treatment.
Science 222, 1337-1339.
Petersen, J.S., Hejnaes, K.R., Moody, A., Karlsen, A.E., Marshall, M.O., Hoier-Madsen, M., Boel, E., Michelsen, B.K.
and Dyrberg, T. (1994) Detection of GAD65 antibodies in
diabetes and other autoimmune diseases using a simple
radioligand assay. Diabetes 43, 459-467.
Sambrook, J., Fritsch, E.F. and Maniatis, T. (1990) In: Molecular Cloning. A Laboratory Manual, 2nd edn. Cold Spring
Harbor Laboratory Press, Cold Spring Harbor, NY, pp.
1.25-1.28.
Schmidli. R.S., Colman, P.G., Bonifacio, E., Bottazzo, G.F.,
Harrison, L.C., and participating laboratories (1994) High
level of concordance between assays for glutamic acid
decarboxylase antibodies: the first International Glutamic
Acid Decarboxylase Antibody Workshop. Diabetes 43,
1005-1009.
Seissler, J., Amann, J., Mauch, L., Haubruck, H., Wolfahrt, S.,
Bieg, S., Richter, W., Hall, R., Heinze, E., Northemann,
W. and Scherbaum, W.A. (1993) Prevalence of autoantibodies to the 65- and 67-kD isoforms of glutamate decarboxylase in insulin-dependent diabetes mellitus. J. Clin.
Invest. 92, 1394-1399.
Smeekens, S.P., Steiner, D.F. (1990). Identification of a human insulinoma cDNA encoding a novel mammalian protein structurally related to the yeast dibasic processing
protease KEX2. J. Biol. Chem. 265, 2997-3000.
Velloso, L.A., KImpe, O., Hallberg, A., Christmanson, L.,
Betsholtz, C. and Karlsson, F.A. (1993) Demonstration of
GAD-65 as the main immunogenic isoform of glutamate
decarboxylase in type I diabetes and determination of
autoantibodies using a radioligand produced by eukaryotic
expression. J. Clin. Invest. 91, 2084-2090.