| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Experimental Studies |
Barbara Davis Center for Childhood Diabetes, University of Colorado Health Sciences Center (E.K., L.Y., R.G., C.F.V., S.B., G.S.E.), Denver, Colorado 80262; and Department of Medicine I, Istituto Scientifico San Raffaele, University of Milan (E.B.), Milan, Italy 20132
Address all correspondence and requests for reprints to: George S. Eisenbarth, M.D., Ph.D., Barbara Davis Center for Childhood Diabetes, University of Colorado Health Sciences Center, Box B-140, 4200 East 9th Avenue, Denver, Colorado 80262. E-mail: George.Eisenbarth{at}UCHSC.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Amongst patients with new-onset diabetes and prediabetic relatives, using RIAs for autoantibodies reacting with ICA512/IA-2 (256–979), insulin, and glutamic acid decarboxylase 65, 98% expressed one or more of these autoantibodies and 78% expressed two or more, whereas no control (n = 208) expressed more than a single autoantibody. A combined ICA512/IA-2 (256–979), glutamic acid decarboxylase 65 autoantibody RIA with differential autoantigen labeling (35S-methionine, 3H-leucine) has been developed that uses 96-well plate membrane filtration and Top Counter ß counting. Concordance between results of dual and single RIAs was greater than 90%. This simple combined autoantibody assay should facilitate large-scale autoantibody screening.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Rabin and co-workers (9, 10) reported an enzyme-linked immunosorbent assay for anti-ICA512 (amino acids 389–937) autoantibodies, and we recently have described an RIA for anti-ICA512 (amino acids 389–948) antibodies (11). Lan and co-workers (12) reported the isolation of a molecule termed IA-2 containing sequence virtually identical to ICA512 but with an open reading frame extended at the 5' end and a longer predicted carboxyterminal amino acid sequence. The aim of this study was to determine whether the carboxy- and aminoterminal IA-2 extensions of ICA512 are important for autoantibody binding and to develop an assay with high sensitivity and rapid throughput for multiple autoantibody screening (ICA512/IA-2 and GAD65). For this purpose, we examined serum reactivity to products from constructs of the full-length IA-2 complementary DNA (cDNA) and fragments of ICA512/IA-2 and compared them with that against our initial ICA512 RIA. To improve efficiency of autoantibody screening, we developed a combined RIA for GAA and ICA512/IA-2 autoantibodies.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Sera were obtained from 45 (21 male and 24 female) prediabetic relatives (relatives followed to the onset of overt diabetes) of patients with type 1 diabetes (from the Joslin Diabetes Center and the Barbara Davis Center), 46 (23 male and 23 female) new-onset patients with type 1 diabetes (tested within 7 days of diagnosis), and 208 healthy control subjects with no family history of diabetes. The median age of the prediabetic relatives at the time of obtaining the sera and patients with new-onset diabetes was 10.5 yr (range: 2.5–66.4 yr) and 10.9 yr (range: 2.2–27.2 yr), respectively, and that of the controls was 15.4 yr (range: 0.4–67.5 yr). The median age at diagnosis of diabetes in the prediabetic group was 12.6 yr (range: 3.9–69.0 yr). Seventy-three new-onset patients and prediabetic relatives followed to diabetes were under 15 yr of age at sampling. The diagnosis of type 1 diabetes was according to Adult National Diabetes Data Group criteria, with either fasting hyperglycemia or oral glucose tolerance testing. All relatives and patients gave informed consent to being studied, and the protocol was approved by the Institutional Review Boards of the University of Colorado and the Joslin Diabetes Center. Sera were stored at -20 C until use. For testing of the combined ICA512/IA-2 (256–979), GAD65 autoantibody RIA, 137 sera (46 new-onset patients with type 1 diabetes, 45 prediabetic relatives, and 46 healthy controls randomly selected from 208 control subjects) were analyzed with single assays (GAD65 or ICA512/IA-2 autoantibodies) and the combined GAD65 and ICA512/IA-2 autoantibody assay.
cDNA cloning
As illustrated in Fig. 1
, five constructs of the
ICA512/IA-2 molecule were used in this study. We had screened a human
islet lambda gt11 cDNA library with ICA-positive first-degree relatives
of type 1 diabetes, and a positive clone (clone HB-1) that contained
the partial sequence (nucleotide 837-3613) of the receptor-type protein
tyrosine phosphatase, ICA512/IA-2, was obtained (11). The coding region
of initial ICA512 [ICA512/IA-2 (amino acid 389–948 of IA-2)] cDNA
was amplified by PCR as previously described (11).
|
. PCR-amplified products were subcloned directly into
the pCRII vector (Invitrogen, San Diego, CA). Previously described
full-length IA-2 [ICA512/IA-2 (1–979)] cDNA (13) subcloned into the
EcoRI site of the SP64PolyA vector (Promega, Madison, MA),
amplified cDNA corresponding to nucleotides 2122–3097 of IA-2 sequence
[ICA512/IA-2 (687–979)] and ligated into the pGEM-T vector
(Promega), was kindly provided by Ezio Bonifacio. Nucleotide sequence
was determined in both directions by automated sequencing using a
sequencer (ABI 373A; Applied Biosystems Inc., Foster City, CA).
|
The constructs illustrated in Fig. 1
were transcribed and
translated in vitro in the presence of
35S-methionine (Amersham International, Amersham, Bucks,
UK; > 1000 Ci/mmol) using the TNT-coupled rabbit reticulocyte system
(Promega). The ICA512/IA-2 RIAs were performed in duplicate using a
96-well plate format similar to a recombinant GAD65 autoantibody RIA,
as described previously (14), with some modifications. In brief,
in vitro translated 35S-labeled protein [20,000
cpm trichloroacetic acid (TCA)] precipitable protein) was incubated
with 5 µL patients’ serum at a 1:25 dilution overnight at 4 C in
Tris-buffered saline/Tween 20 (TBST; 20 mmol/L Tris-HCl, pH 7.4, 150
mmol/L NaCl, 0.1% BSA, 0.15% Tween 20) containing 0.1% aprotinin and
10 mmol/L benzamidine. Twenty-five microliters of 50% Protein
A-sepharose (Pharmacia, Uppsala, Sweden) in TBST was added to the
reaction in a MultiScreen-DV 96-well filtration plate (Millipore,
Burlington, MA) and incubated for 45 min at 4 C and washed nine times
with cold TBST using the Millipore vacuum-operated 96-well plate washer
(Millipore). After washing, 40 µL scintillation liquid
(Microscint-20; Packard, Meriden, CT) was added to each well and
radioactivity was determined directly in the 96-well plate with Top
Count 96-well plate ß counter (Packard). For autoantibodies to
ICA512/IA-2 constructs, except for ICA512/IA-2 (1–577), the positive
control sera and negative control sera were included in every assay,
and the antibody levels were expressed as an index defined as: (cpm in
the unknown sample - negative control)/(positive control -
negative control). Positive for these assays was based on the 99th
percentile of sera from 208 healthy control subjects. For
autoantibodies to ICA512/IA-2 (1–577) construct, SD score
was calculated for each serum based on the mean value and
SD of cpm precipitated with 23 normal control serum.
Positive was defined as 
3 SD score.
Competition studies
Unlabeled recombinant ICA512/IA-2 (1–577) and ICA512/IA-2 (687–979) were prepared by in vitro transcription/translation system as detailed above, except that amino acid mixture contained unlabeled methionine rather than 35S-methionine. For the competition studies, 10 µL unlabeled ICA512/IA-2 protein (50- to 75-fold excess amount of [35S)-protein) was incubated with patients’ sera for 3 h at room temperature followed by the addition of in vitro translated [35S]-labeled ICA512/IA-2 (20,000 cpm TCA precipitable protein) in TBST, and samples were processed in the ICA512/IA-2 autoantibody RIAs described above.
IAA and GAA assays
The IAA assay was carried out by a fluid-phase RIA using competition with cold insulin and precipitation with polyethylene glycol (15). The 99th percentile of the normal range for IAA is 42 nU/mL. The interassay coefficient of variation (CV) was 10.3% (n = 7).
GAA assay used in vitro transcribed and translated human islet 35S-labeled GAD65 and precipitation with Protein A-sepharose (14). The levels of GAA were expressed as an index calculated with the same formula used in the ICA512/IA-2AA autoantibody RIAs. The 99th percentile of levels in the normal controls for this assay was an index of 0.032. The interassay CV and intraassay CV were 6.5% (n = 10) and 5.0% (n = 6), respectively.
Combined GAA and ICA512/IA-2 autoantibody RIA
Recombinant GAD65 and ICA512/IA-2 (256–979) proteins were translated in vitro with differential labeling [3H-leucine for GAD65 and 35S-methionine for ICA512/IA-2 (256–979)], and the combined GAA and ICA512/IA-2 (256–979) autoantibody RIA was performed using the same format as ICA512/IA-2 autoantibody RIAs described above. In vitro transcribed 3H-labeled GAD65 (20,000 cpm TCA precipitable protein) and 35S-labeled ICA512/IA-2 (256–979) (20,000 cpm TCA precipitable protein) were used as antigens. The positive control sera and negative control sera for GAA and ICA512/IA-2 (256–979) autoantibodies were included in every assay. The levels of both antibodies were expressed as an index calculated with the same formula used in the ICA512/IA-2 autoantibodies or GAA RIAs with channel windows set for each radionucleotide and mathematical correction for 35S counts in the 3H channel and 3H counts in the 35S channel.
Statistical analysis
Unpaired data were analyzed by the 
-square test and the
Mann-Whitney U test. A SD score was calculated
for each serum: SD score = (antibody index of test
serum - mean index of healthy control sera)/SD of the
indices of 208 healthy control sera. The SD score provides
a measure of the certainty with which individual serum can be deemed
different from normal. The statistical significance of paired
differences of SD score among RIAs for a series of
ICA512/IA-2 constructs was determined by the paired t test.
Agreement between the single and combined RIA methods was assessed with
the Kappa statistic, which is a measure of the strength of agreement on
a scale from 0–1. A P < 0.05 was considered
statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Sera were considered ICA512/IA-2 autoantibody positive if their indices were more than 0.160 for ICA512/IA-2 (389–948) autoantibodies, 0.048 for ICA512/IA-2 (256–979) autoantibodies, 0.284 for ICA512/IA-2 (1–979) autoantibodies, and 0.039 for ICA512/IA-2 (687–979) autoantibodies, which were based on the 99th percentile of healthy control subjects. For 208 healthy control subjects, the mean indices of autoantibodies to ICA512/IA-2 (389–948), ICA512/IA-2 (256–979), ICA512/IA-2 (1–979), and ICA512/IA-2 (687–979) were -0.031 ± 0.089 (mean ± SD), 0.001 ± 0.012, 0.038 ± 0.072, and -0.001 ± 0.011, respectively. The interassay CV and intraassay CV were 9.6% (n = 12) and 6.0% (n = 7) for ICA512/IA-2 (389–948) autoantibodies, 11.7% (n = 9) and 10.0% (n = 10) for ICA512/IA-2 (256–979) autoantibodies, 6.5% (n = 9) and 10.1% (n = 10) for ICA512/IA-2 (1–979) autoantibodies, and 9.9% (n = 9) and 10.0% (n = 10) for ICA512/IA-2 (687–979) autoantibodies, respectively.
Evaluation of ICA512/IA-2 autoantibodies in prediabetic relatives and new-onset patients with type 1 diabetes
Forty percent (18 of 45) of prediabetic relatives and 41% (19 of
46) of new-onset patients with type 1 diabetes had ICA512/IA-2
(389–948) autoantibody levels exceeding the 99th percentile of the
healthy control group (Table 2 and Fig. 2). In contrast, using ICA512/IA-2 constructs containing
the complete IA-2 carboxyterminus (to amino acid 979), between 61 and
69% of prediabetic relatives and new-onset patients were positive
(Table 2). The sensitivities of these three autoantibodies in
prediabetic patients were significantly higher than that of ICA512/IA-2
(389–948) autoantibodies (P < 0.05). For all assays,
only 1% (2 of 208) of healthy control subjects had an index defined as
positive (> the 99th percentile).
|
|
shows the concordance among the 4 ICA512/IA-2
autoantibody assays. All prediabetic relatives or new-onset patients
positive for ICA512/IA-2 (389–948) autoantibodies also were positive
for ICA512/IA-2 (256–979) autoantibodies, ICA512/IA-2 (1–979)
autoantibodies, and ICA512/IA-2 (687–979) autoantibodies. Two percent
(1 of 45) of prediabetic relatives and 6% (3 of 46) new-onset patients
were ICA512/IA-2 (256–979) autoantibody negative but ICA512/IA-2
(1–979) autoantibody positive. A high level of positivity for
ICA512/IA-2 (256–979) autoantibodies was associated with positivity on
the ICA512/IA-2 (389–948) autoantibody assay, but many sera were
strongly ICA512/IA-2 (256–979) autoantibody positive (e.g.
index of 1.6) but ICA512/IA-2 (389–948) autoantibody negative (Fig. 2). To examine the serum reactivity against the extracellular
aminoterminal domain of IA-2, autoantibodies to an ICA512/IA-2 (1–577)
construct was also determined in 20 new-onset patients and 18
prediabetic relatives. Only 2 of 38 sera reacted with this construct,
and the bindings of these 2 sera with ICA512/IA-2 (1–577) construct
were blocked by preincubation with unlabeled ICA512/IA-2 (1–577). All
of 4 sera that were ICA512/IA-2 (1–979) autoantibody positive but
ICA512/IA-2 (256–979) autoantibody negative did not react with this
construct. In competition studies, the bindings of these 4 sera with
ICA512/IA-2 (1–979) were not inhibited by unlabeled ICA512/IA-2
(1–577) or ICA512/IA-2 (687–979) protein. However, the binding of
ICA512/IA-2 (256–979) autoantibodies in ICA512/IA-2 (389–948)
autoantibody-negative/ICA512/IA-2 (256–979) autoantibody-positive sera
were completely inhibited by preincubation with unlabeled ICA512/IA-2
(687–979).
|
The levels of ICA512/IA-2 (256–979) autoantibodies were correlated
with those of ICA512/IA-2 (1–979) autoantibodies (r = 0.96,
P < 0.0001) and ICA512/IA-2 (687–979) autoantibodies
(r = 0.98, P < 0.0001) (Fig. 3).
The prevalence of ICA512/IA-2 autoantibodies in patients under 14 yr
old were higher than those in the 15 yr group (Table 4).
|
|
illustrates the mean SD score
((index of test sera - mean index of normal
controls)/SD of normal controls) for individual sera for
the ICA512/IA-2 (256–979) autoantibodies as compared with other
ICA512/IA-2 autoantibody RIAs. The mean SD score obtained
using the ICA512/IA-2 (256–979) construct was significantly higher
than those obtained when other ICA512/IA-2 constructs were used
(P < 0.0001).
|
Table 5 summarizes the sensitivity of individual
assays for ICA512/IA-2 (256–979) autoantibodies, GAA, and IAA. Amongst
patients with new-onset diabetes and prediabetics, 98% (89 of 91)
expressed one or more autoantibodies and 78% (71 of 91) expressed two
or more of these autoantibodies. Of note, with the number of healthy
control subjects evaluated to date (n = 208) with positivity
defined at the 99th percentiles, none of the controls expressed more
than a single autoantibody.
|
One-hundred thirty-seven serum samples were analyzed for GAA and
ICA512/IA-2 (256–979) autoantibodies with a combined RIA (described in
Materials and Methods) and compared in parallel with
individual RIAs for GAA and ICA512/IA-2 (256–979) autoantibodies using
in vitro translated 35S-labeled products.
Figure 5 shows the correlation between the levels of GAA
or ICA512/IA-2 (256–979) autoantibodies determined with the individual
assays (35S-GAD65, 35S-ICA512/IA-2), and
the combined assay (3H-GAD65,
35S-ICA512/IA-2). A close correlation was observed
between the assays (r = 0.92 for GAA and r = 0.96 for
ICA512/IA-2 (256–979) autoantibodies). Considering the assignment of
sera as positive or negative, the agreement between the single assay
and combined assay was 90% (123/137) for the GAA RIA (Kappa
statistic = 0.81) and 100% (137/137) for ICA512/IA-2 (256–979)
autoantibody RIA (Kappa statistic = 1.0). The assays disagreed
only for 14/274 determinations at the lower detection limit for the
assays, suggesting random differences at the cutoff levels (Fig. 5).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We previously have reported the development of an RIA for a shortened ICA512 [ICA512/IA-2 (389–948)] autoantibodies (11). In this current study, we developed RIAs for ICA512/IA-2 autoantibodies using a series of carboxyterminal or aminoterminal IA-2 extensions of ICA512 and assessed the sensitivity of these assays compared with our initial ICA512 autoantibody RIA. With positivity defined above the 99th percentile of control subjects, autoantibodies to ICA512/IA-2 (256–979), ICA512/IA-2 (1–979), and ICA512/IA-2 (687–979) were present in 60–70% of prediabetic relatives and new-onset patients with type 1 diabetes, which is similar to that previously reported for IA-2 [ICA512/IA-2 (1–979)] autoantibodies detected by a radiobinding assay (13). This compares with 40% of prediabetic relatives and 41% of new-onset patients positive for ICA512/IA-2 (389–948) autoantibodies. These results indicate that carboxyterminus extension of ICA512 enhances the sensitivity of the ICA512/IA-2 autoantibody RIA while preserving specificity.
In this study, at least three ICA512/IA-2 antibody epitopes were identified. In prediabetic relatives and new-onset patients, all sera positive for ICA512/IA-2 (389–948) autoantibodies were positive also for autoantibodies to ICA512/IA-2 (256–979), ICA512/IA-2 (1–979), and ICA512/IA-2 (687–979), and there was a high correlation among the latter three antibodies. These results suggest that the major epitope recognized by sera from type 1 diabetic patients is located within residues 687–979 of the intracellular domain of the IA-2. A few sera, however, were positive for ICA512/IA-2 (1–979) autoantibodies but were ICA512/IA-2 (256–979) autoantibody negative, thereby having distinct antibody specificity including the aminoterminal residues 1 to 255 of IA-2. Furthermore, the finding that all of 4 sera positive for ICA512/IA-2 (1–979) autoantibodies without ICA512/IA-2 (256–979) autoantibodies did not react with the ICA512/IA-2 (1–577) construct and that the reactivity against ICA512/IA-2 (1–979) in these sera were not inhibited by preincubation with unlabeled ICA512/IA-2 (687–979) protein suggests that these sera might recognize a conformational epitope of the molecule.
The mean SD score ((antibody index of test serum -
mean index of control sera)/SD of the indexes of controls),
a measure of certainty that an individual sera is different from normal
controls, was highest for ICA512/IA-2 (256–979) autoantibodies amongst
the autoantibody RIAs studied (Fig. 4). The higher SD score
of ICA512/IA-2 (256–979) autoantibodies in contrast to the ICA512/IA-2
(1–979) construct results primarily from a lower background count for
the ICA512/IA-2 (256–979) construct amongst normal control sera.
In 91 patients with new-onset type 1 diabetes and prediabetic relatives, 98% expressed 1 or more autoantibodies against ICA512/IA-2 (256–979), GAD65, and insulin. This shows that combined analysis of these 3 sets of autoantibodies has high predictive and diagnostic potential for type 1 diabetes. It has been shown recently that the number of these 3 sets of autoantibodies is highly predictive of diabetes risk in first-degree relatives of type 1 diabetes, independent of ICA status (16).
Our assay format for detecting ICA512/IA-2 (256–979) autoantibodies is similar to that used for the detection of GAA. We therefore developed a combined ICA512/IA-2 (256–979), GAD65 autoantibody RIA. In contrast to a combined assay described by Bonifacio and coworkers (13), we used 3H-labeled GAD65 and 35S-labeled ICA512/IA-2 (256–979), which allowed simultaneous detection and discrimination of both autoantibody specificities. We used the ICA512/IA-2 (256–979) constructs for combined autoantibody assay because this construct gave a maximum SD score and high sensitivity. The 4 of 91 sera positive for ICA512/IA-2 (1–979) autoantibodies and negative for ICA512 IA-2 (256–979) did not affect the overall sensitivity when multiple autoantibodies were examined, in that these 4 sera all had GAA (data not shown). The combined assay gave essentially identical results to those obtained in the single RIAs and detected either antibody in more than 90% of new-onset and prediabetic relatives. The simplicity of this assay with dual determination of the two antibodies using 5 µL of sera for duplicate (7 µL for triplicate), 96-well membrane separation of autoantibody bound labeled autoantigen, and 96-well ß counting should facilitate large-scale population screening.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 A Research Fellow of the Japan
Society for the Promotion of Science.
Received June 18, 1996.
Revised August 2, 1996.
Accepted October 2, 1996.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  R. P. Sahu, A. Aggarwal, G. Zaidi, A. Shah, K. Modi, S. Kongara, S. Aggarwal, S. Talwar, S. Chu, V. Bhatia, et al. Etiology of Early-Onset Type 2 Diabetes in Indians: Islet Autoimmunity and Mutations in Hepatocyte Nuclear Factor 1{alpha} and Mitochondrial Gene J. Clin. Endocrinol. Metab., July 1, 2007; 92(7): 2462 - 2467. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Gianani, A. Putnam, T. Still, L. Yu, D. Miao, R. G. Gill, J. Beilke, P. Supon, A. Valentine, A. Iveson, et al. Initial Results of Screening of Nondiabetic Organ Donors for Expression of Islet Autoantibodies J. Clin. Endocrinol. Metab., May 1, 2006; 91(5): 1855 - 1861. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Pihoker, L. K. Gilliam, C. S. Hampe, and A. Lernmark Autoantibodies in Diabetes Diabetes, December 1, 2005; 54(suppl_2): S52 - S61. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Takeda, E. Kawasaki, I. Shimizu, E. Konoue, M. Fujiyama, S. Murao, K. Tanaka, K. Mori, Y. Tarumi, I. Seto, et al. Clinical, Autoimmune, and Genetic Characteristics of Adult-Onset Diabetic Patients With GAD Autoantibodies in Japan (Ehime Study) Diabetes Care, June 1, 2002; 25(6): 995 - 1001. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. S. Hampe, L. P. Hammerle, L. Bekris, E. Örtqvist, I. Kockum, O. Rolandsson, M. Landin-Olsson, C. Törn, B. Persson, and A. Lernmark Recognition of Glutamic Acid Decarboxylase (GAD) by Autoantibodies from Different GAD Antibody-Positive Phenotypes J. Clin. Endocrinol. Metab., December 1, 2000; 85(12): 4671 - 4679. [Abstract] [Full Text]
|
|
|
|
 |
|
 M. J Redondo, M. Rewers, L. Yu, S. Garg, C. C Pilcher, R. B Elliott, and G. S Eisenbarth Genetic determination of islet cell autoimmunity in monozygotic twin, dizygotic twin, and non-twin siblings of patients with type 1 diabetes: prospective twin study BMJ, March 13, 1999; 318(7185): 698 - 702. [Abstract] [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
, prediabetic
relatives; 
, new-onset patients with type 1 diabetes. ——, 99th
percentile of 208 control subjects for the 2 assays.
