identification of glutamic acid decarboxylase autoantibody - diabetes

Identification of Glutamic Acid DecarboxylaseAutoantibody Heterogeneity and Epitope Regionsin Type I DiabetesNoriko Ujihara, Kendra Daw, Roberto Gianani, Esper Boel, Liping Yu, and Alvin C. Powers

Glutamic acid decarboxylase (GAD) is an autoantigen ofthe islet cell antibodies (ICAs) present in type I diabetes.GAD autoantibodies are also found in patients with stiff-man syndrome and in certain ICA-positive individualswho rarely develop diabetes on long-term follow-up. Thislatter subset of ICA has been termed restricted or p-cell-specific ICA because the antibodies react with only theP-cells of the islet. By immunoprecipitation of recombi-nant GAD65 and GAD67 protein and protein fragments,83% of sera from individuals with new-onset diabetes orprediabetes (ji = 30) had GAD65 autoantibodies, but only26% had GAD67 autoantibodies. In contrast, all restrictedICA sera (ji = 6) had both GAD65 and GAD67 autoanti-bodies. In both types of sera, the binding of GAD67autoantibodies could be blocked by preincubation of theserum with GAD65 and GAD67, but the binding of GAD65autoantibodies could not be blocked by preincubationwith GAD67. The titer of GAD65 autoantibodies was muchhigher in the restricted ICA sera (titer > 1:1,000) than inthe sera from individuals with new-onset diabetes orprediabetes (titer < 1:100) and was reflected by thegreater amount of GAD65 protein immunoprecipitated byrestricted ICA sera (2.61 ± 1.39 U) compared with serafrom individuals with new-onset diabetes (0.51 ± 0.34 U).The restricted ICA sera immunoprecipitated equimolaramounts of GAD65 protein fragments, suggesting a non-conformational or linear epitope; epitope mapping local-ized the major epitope region to amino acids 361-442 anda second minor epitope region to amino acids 1-195. Forthe monoclonal antibody GAD6, the GAD65 epitope regionlocalized to amino acids 529-585. In contrast, the serafrom individuals with new-onset diabetes or prediabetesimmunoprecipitated equal amounts of full-length GAD65protein and a GAD65 fragment containing amino acids188-585, but did not immunoprecipitate smaller GAD65protein fragments, which suggests an epitope(s) dependenton protein conformation. These results suggest that subsetsof GAD autoantibodies exist and indicate a heterogeneity inthe immune response to GAD. Diabetes 43:968-975,1994

From the Division of Endocrinology (N.U., K.D., A.C.P.), Department of Medicine,Vanderbilt University, Nashville, TN; the Department of Veterans Affairs MedicalCenter (A.C.P.)> Nashville, TN; the Barbara Davis Center for Childhood Diabetes (R.G.,L.Y.), University of Colorado Health Sciences Center, Denver, CO; and Bioscience(E.B.), Diabetes Care Division, Novo Nordisk A/S, Bagsvaerd, Denmark.

N.U. and K.D. contributed equally to this article.Address correspondence and reprint requests to Dr. Alvin C. Powers, Division of

Endocrinology, AA 4206, MCN, Vanderbilt University, Nashville, TN 37232.Received for publication 23 August 1993 and accepted in revised form 14 April

1994.ICA, islet cell antibody; GAD, glutamic acid decarboxylase; PCR, polymerase

chain reaction; AMV, avian myeloblastosis virus; MMLV, Moloney murine leukemiavirus; bp, base pair; TCA, trichloroacetic acid; TBST, Tris-buffered saline/0.05%Tween 20; PAS, protein A-sepharose; SDS-PAGE, sodium dodecyl sulfate-polyac-rylamide gel electrophoresis; MICA, monoclonal islet cell antibody.

Type I diabetes is an autoimmune disease directedagainst the pancreatic p-cell. The development oftype I diabetes is heralded and accompanied byislet cell autoantibodies (ICAs) that are detected

immunocytochemically with sections of human or rat pan-creas (1). The immunocytochemical ICAs typically react withall cells of the pancreatic islet, are present in the serum ofindividuals at the onset of clinical diabetes, and have beenuseful in predicting the development of type I diabetes (1-3).The immunocytochemical ICA is likely a composite of anti-bodies directed at several pancreatic islet molecules; one ofthe target antigens is glutamic acid decarboxylase (GAD)(4,5).

GAD is the biosynthetic enzyme for the inhibitory neuro-transmitter 7-amino butyric acid (6). Two isoforms, GAD65

(65 kDa) and GAD67 (67 kDa), are the products of twoseparate genes, are highly homologous, and differ mostly inthe amino terminal-third of the protein (6-8). The respectivecDNAs for GAD65 and GAD67 are identical in the brain andislet (9-14). In the human, GAD65 is the predominant formexpressed in the islet, whereas both GAD65 and GAD67 arefound in rat islets (15,16). Mouse islets contain little, if any,GAD65 (15,16).

Antibodies against GAD, originally detected by Baekke-skov et al. (4), are found in the majority of individuals withpreclinical and new-onset type I diabetes and have beendetected utilizing recombinant GAD and porcine brain GAD(17-24). Some ICA sera react with GAD65 and GAD67, butGAD65 is the predominant autoantigen (17,25). An initialreport mapped the reactivity of four ICA sera to the middleand COOH-terminal of GAD65 (26). A recent study using aseries of GAD-directed monoclonal antibodies from a singlepatient with type I diabetes has detected at least two distinctepitopes, one that is linear in nature and one that is depen-dent on protein conformation (27,28).

GAD autoantibodies are also found in patients with stiff-man syndrome, patients with polyendocrine failure, andcertain ICA-positive individuals who rarely develop diabeteson long-term follow-up (29-31). This latter subset of ICA hasbeen termed restricted ICA or p-cell-specific ICA because itreacts with only the (3-cells of the islet, and their reactivitycan be blocked by preincubation with brain GAD (30,31).

To investigate the role of GAD in the autoimmune processof type I diabetes, we have analyzed GAD autoantibodies inrestricted ICA sera and sera from individuals with new-onset

968 DIABETES, VOL. 43, AUGUST 1994

Dow

nloaded from http://diabetesjournals.org/diabetes/article-pdf/43/8/968/360734/43-8-968.pdf by guest on 28 D

ecember 2021

N. UJIHARA AND ASSOCIATES

diabetes or prediabetes and have mapped GAD proteinepitope regions.

RESEARCH DESIGN AND METHODSIsolation of GAD65 and GAD67 cDNAs. Total RNA from a humanpancreatic insulinoma and rat/human cerebellum were used to isolatethe GAD65 and GAD07 cDNAs. Total RNA was reverse transcribed intocDNA, and subsequent polymerase chain reaction (PCR) amplificationwas performed using specific oligonucleotide primers corresponding topublished rat and human GAD cDNA sequences (10,11,32). First-strandcDNA synthesis was performed using either avian myeloblastosis virus(AMV) or Moloney murine leukemia virus (MMLV) reverse transcriptase;briefly, 1-10 |xg of total RNA samples were incubated in a mixture ofRNase inhibitor, oligo(dt), and deoxynucleotides for 30-60 min at 42°C.Subsequent PCR amplification with first-strand cDNA as template werecarried out in a thermal cycler (Perkin-Elmer/Cetus) for 30 cycles (1 min94°C denaturing, 2 min 60°C annealing, and 3 min 72°C extension) in a100-|xl volume containing buffer (10 niM Tris, pH 8.3, 50 mM KC1,1.5 mMMgCl2, 0.001% wt/vol gelatin), 200 yM of each dNTP, 2.5 U Taq DNApolymerase (Promega), and 1 (xM of each oligonucleotide primer. Re-actions were covered with 50-60 (xl of mineral oil before thermalcycling. Following the amplification, the PCR products were analyzed on1% agarose gels with 0.5 |xg/ml of ethidium bromide and purified using aGeneclean II kit (Bio 101).

Oligonucleotide primers for the PCR amplification of GAD cDNAwere complementary to the published GAD cDNA sequence (10,11,32),with the incorporation of unique restriction enzyme sites to facilitatesubcloning. For the amplification of rat GAD05) primer 1 (TGC TCT AGAATG GCA TCT CCG GGC TCT GGC TTT) corresponds to the 5' endcoding strand of the rat GAD65 cDNA at base pairs (bp) 75-98 withincorporation of a Xba I site. Primer 2 (CTA GTC GAC TTA CAA ATCTTG TCC CAG GCG TTC) corresponds to the 3' end of the noncodingstrand of the rat GAD05 cDNA at bp 1809-1832 with incorporation of aSal, I site. For the amplification of human GAD67, primer 1 (ACT GGATCC GAG CAA ACT GTG CAG TTC TTA CTG) corresponds to the 5' endcoding strand of the rat GADG7 cDNA (20) at bp 510 (of human GAD67sequence) with incorporation of a Bam HI site. Primer 2 (ACT GAA TTCTTA CAG ATC CTG GCC CAG TCT) corresponds to the 3' endnoncoding strand of the human GADG5 cDNA at bp 1764-1784 withincorporation of an Eco RI site. A full-length GADG5 cDNA and a partialGAD67 cDNA (predicted coding region for amino acids 110-593 of GAD67protein) were isolated by this method. A full-length GAD67 cDNA wasisolated by extending the previously reported partial human GADG7cDNA (14) by rapid amplification of cDNA ends synthesis protocols (33).

The full-length rat GAD6i5 cDNA, the partial human GAD67 cDNA(predicted coding region for amino acids 110-593 of GAD67 protein), andthe full-length human GADG7 cDNA (predicted coding region for aminoacids 1-593) were ligated into the polylinker region of an expressionvector containing an upstream T7 promotor for transcriptional direc-tion. Nucleotide sequences were determined by the dideoxy-chaintermination method (Sequenase Version 2.0 Kit, USB Corp). The GADcDNAs were analyzed by restriction enzyme digestion and DNA se-quencing and agreed with the published sequences for the rat GAD65cDNA and human GAD07 cDNAs (12,32).Preparation of GAD protein. For RNA preparation, 30 |xg of GADplasmid cDNA was linearized on the 3' side of the insert and treated withproteinase-K followed by phenol-chloroform extraction and ethanolprecipitation. A typical transcription reaction contained 1 |xg of templatecDNA, 40 U of T7 RNA polymerase, and 7.5 mM of each rNTP intranscription buffer. Procedures used for the in vitro transcription andrecovery of mRNA were carried out following the protocols of Ambion'sMegascript Kit.

Recombinant rat GAD05 and human GADG7 proteins were prepared byin vitro translation in a reticulocyte lysate (Promega) in the presence of[35S]methionine (1,200 Ci/mmol). The only methionine present in thein vitro translation reaction was [35S]methionine. Trichloroacetic acid(TCA) precipitation of the translation products was utilized to determinethe percent incorporation of the labeled amino acid in the GAD protein.For the TCA precipitation, an aliquot of the translation product wasincubated with 1 M NaOH/2% H2O2. The GAD protein was precipitatedwith 25% TCA/2% casamino acids and filtered on a GF/A glass fiber filter(Whatman). The filter was placed in 5 ml of Ecolite® (ICN) and countedin a p-scintillation counter. Percent incorporation was calculated as(radioactivity in TCA-precipitated protein/total radioactivity in the reac-tion) x 100 and usually ranged from 20-30%.

The rat GAD65 protein consisted of the full-length protein (as deducedfrom its DNA sequence) and reacted with a sheep polyclonal GADantiserum and two monoclonal antibodies, GAD, and GADC (34,35). TheGAD, monoclonal antibody was a gift of Dr. David Gottlieb (Washing-ton, MO); the GADG monoclonal antibody was obtained from theDevelopmental Studies Hybridoma Bank at the University of Iowa(NICHD contract no. N01-HD-2-3144); and the polyclonal GAD anti-serum was a gift of Dr. Enrico Mugnaini (University of Connecticut) andVirginia Weise (Laboratory of Clinical Science at NIMH). The partial-length human GADG7 protein consisted of amino acids 110-593 (asdeduced from its DNA sequence) and reacted with the sheep polyclonalGAD antiserum, but not with the monoclonal antibodies, GAD, andGADG. Unless stated otherwise, GADG5 refers to full-length rat GAD0Bprotein, and GADG7 refers to human GADG7 protein of amino acids110-593. Full-length human GADG7 protein was also prepared and gaveidentical results to the partial-length human GADG7 protein containingamino acids 110-593.

Immunoprecipitation of GAD protein. Thirty microliters of type Idiabetic or control subject serum was incubated with 30 (xl of GADantigen containing 50,000 cpm of labeled, TCA-precipitable GAD proteinfor 12-15 h at 4°C. All dilutions were made with Tris-buffered saline/0.05% Tween 20 (TBST). Protein A-sepharose (PAS, Sigma) was washedand resuspended in TBST to a 2% solution. Two microliters of PAS (finalbed volume) was added to the reaction and mixed for 1 h at 4°C withgentle rotation. The PAS pellets were washed three times with 500 \x,\ ofTBST. The immunoprecipitated GAD protein was removed from the PASwith sodium dodecyl sulfate and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) or by scintillation quan-titation. SDS-PAGE was performed under nonreducing conditions with10% polyacrylamide gels that were analyzed by fluorography. For thequantitative GAD assay, the PAS pellet was resuspended in 100 \x\ of 1%SDS, and the supernate was quantitated by ^-scintillation counting withan aqueous-based scintillant. All samples in the quantitative GAD assaywere assayed in duplicate. The amount of GADer, protein immunopre-cipitated was expressed as relative GAD binding ([cpm in sample - cpmin a normal control serum]/[cpm in a positive control serum - cpm in anormal control serum]). The same positive control serum and negativecontrol serum were used for each assay. Each serum was assayed induplicate, and the binding of a serum to different GAD fragments isshown as the mean ± SE of at least three experiments. Comparisons ofthe relative GAD binding of different groups are reported as the mean ±SD of each group. Statistical comparisons of the two groups used theStudent's t test or Fisher's exact test. Our laboratory participated in theGAD antibody workshop sponsored by the Immunology of DiabetesWorkshop (Orlando, FL, April 1993) and our assay achieved a sensitivityof 75% and a specificity of 100% with a mean difference betweenduplicates of 16.7%.Blocking of binding to GAD65 and GADG7 with unlabeled GAD05 orGAD67. Unlabeled GADG5 and GADG7 were prepared in the reticulocytelysate system as detailed above except that the amino acid mixturecontained unlabeled methionine rather than [UBS] methionine. For theblocking studies, a portion of the unlabeled GAD protein (representing~25% of an in vitro translation with 1 |xg of GAD RNA) was incubatedwith a dilution of the ICA serum for 4 h at room temperature. Thismixture of serum and unlabeled GAD was then added to 50,000 cpm of35S-labeled GAD protein (representing ~10% of an in vitro translationwith 1 (xg of GAD RNA) and incubated for 12-15 h at 4°C. Theimmunoprecipitates were analyzed by gel electrophoresis or by thequantitative GAD assay as described above. Each sera was assayed induplicate, and the mean and SE of three experiments was determined.Preparation of GADG5 protein fragments. GAD65 protein fragmentscontaining amino acids 1-195, 1-361, 1-442, and 1-529 were preparedfrom restriction enzyme fragments of GADG5 cDNA after linearizationwith Hinc II, Nsi I, Stu I, and Nar I enzymes, respectively. Digestion withthe restriction enzyme Nsi I (two internal Nsi I sites present) andsubsequent religation resulted in a GADG5 protein that does not containthe amino acids 361-421. PCR was used to create a GADGr> proteinfragment containing amino acids 188-585. A fragment containing aminoacids 361-585 was created with a unique Bgl II site in GADG5. Aschematic summary of the GAD65 fragments is shown in Fig. 5. TheGAD65 protein fragments were prepared by in vitro transcription andtranslation as described in METHODS.

The relative amount of different GAD fragments immunoprecipitatedwas compared by expressing the immunoprecipitates as molar ratios.Because the in vitro translation reaction contains only labeled methio-nine, the number of cpm in a GAD protein fragment correlates with the

DIABETES, VOL. 43, AUGUST 1994 969

Dow


ecember 2021

DIFFERENCES IN GAD AUTOANTIBODIES

TABLE 1Individuals studied for GAD antibodies

Restricted ICA123456

Individuals with new-onset diabetes7891011121314151617181920212223242526272829303132

Individuals with prediabetes33343536

Sex

FFFMFM

FMMMFMFMMMFMMFMFFMMMMMMFMM

FFFF

IDDM

NNYNNN

YYYYYYYYYYYYYYYYYYYYYYYYYY

YYYY

Age atonset(years)

NANA68NANANA

587

1025' 916302321

714304230141211304525212911311845

18101420

ICA(JDF U)

640160POS128016040

POS160NEGNEG320POSNEG320NEG160160320NEGNEG80160NEGNEGNEGNEGNEGPOSNDNEG160NEG

160POS640160

GAD65

1:20,0001:10,0001:1,0001:1,0001:1,0001:1,000

POS*POSPOSPOSPOSPOSPOS*POSNEGNEGPOSPOSPOSPOSPOS*POSPOSPOSPOSPOSPOSPOSPOSNEGNEGNEG

POS*POSPOSPOS

GAD67

1:10,0001:1,0001:1001:1001:1001:100

NEGPOSNEGNEGPOSNEGPOSNEGNEGNEGNEGNEGNEGNEGPOSPOSNDPOSNEGNEGNEGNEGNDNEGNEGNEG

POSNEGPOSNEG

For ICA, titer of ICA is reported in Juvenile Diabetes Foundation units or positive if titer not available. GAD65 and GAD67 autoantibodieswere determined as described in the METHODS. GAD65 and GAD67 titers in new-onset diabetes were < 1:100 unless marked (*). NA, notapplicable; ND, not done; NK, not known; Y, yes; N, no; POS, positive; NEG, negative.

number of methionine residues in that protein fragment. The molar ratioof a GAD protein fragment was calculated from the amount of cpm inthe immunoprecipitate and the number of methionine residues in thatGAD protein fragment ([cpm of GAD fragment immunoprecipitated bysample serum — cpm of GAD fragment immunoprecipitated by normalcontrol serum]/number of predicted methionine residues in that GADprotein fragment). Each assay was performed in duplicate, and theresults of at least three assays are expressed as the mean ± SE.Sera population. Serum was obtained from 30 individuals with new-onset type I diabetes or prediabetes and from 6 ICA-positive individualswith the restricted or P-cell-specific ICA pattern of reactivity (31).Information regarding individuals with restricted ICA, individuals withnew-onset diabetes, or individuals with prediabetes and the reactivity ofeach serum to GAD are shown in Table 1. The sera from individuals withnew-onset diabetes or prediabetes were selected from individuals withthe clinical features of type I diabetes. Prior studies by several groups ofinvestigators have shown that such sera react with all cells of thepancreatic islet (5,30,31,36-40). We have confirmed this prior observa-tion by demonstrating that eight of the sera from individuals withnew-onset diabetes or prediabetes react with all islet cells (data notshown). The immunocytochemical staining of rat pancreatic sections byrestricted ICA sera was blocked by preincubating the serum withporcine brain GAD isolated with GADG monoclonal antibody affinitychromatography as described previously (31). The restricted ICA seraalso reacted with brain GAD by immunoblotting and, unlike the ICAfrom individuals with new-onset diabetes, stained rat islet but not mouseislets (31). Only one of the six individuals with a restricted ICA pattern

had diabetes, and she developed it at 68 years of age (Table 1). Theimmunocytochemical ICA (Juvenile Diabetes Foundation units) wasdetermined as described previously at the Joslin Diabetes Center(Boston, MA) by Drs. Richard Jackson and George Eisenbarth (1,31).Sera from healthy, nondiabetic individuals were included as negativecontrols.

RESULTSThis study examines the humoral response to GAD inICA-positive sera from individuals with and without diabetes.GAD antibodies in individuals with new-onset diabetes orprediabetes were compared with GAD antibodies in individ-uals with a restricted ICA pattern (only one individual in thisgroup had diabetes). The features of the individuals exam-ined are summarized in Table 1. Comparative assays of GADantibody titer and epitope recognition indicate that GADautoantibodies in restricted ICA sera differ from those in thesera of individuals with new-onset diabetes or prediabetes.ICA sera reactivity with GAD65 and GAD67. Immunopre-cipitation of recombinant GAD65 and GAD67 protein wasperformed with restricted ICA sera (n = 6) and with serafrom individuals with new-onset diabetes or prediabetes (n= 30). GAD65 is immunoprecipitated by restricted ICA sera


Dow


ecember 2021


B lOOOn

kDa

69

46

GAD 65

GAD 67

In VitroTranslation Products

' 120 -

100 -

-y 8 0 •

r*.

Immunoprecipitations

Q<

60 H

40 -

20 -

\///////A r^-^^ V//////ABUFFER GAD 65 (1-593)

GAD 67(110-593)

GAD 67

and by sera from individuals with new-onset diabetes, butnot by normal sera (Fig. L4, Table 1). GAD67 is immunopre-cipitated by all restricted ICA sera but by only some serafrom individuals with new-onset diabetes (Fig. \A and B,Table 1). The reactivity with full-length GAD67 and a GAD67

fragment consisting of amino acids 110-593 were identical,and subsequent experiments used either full-length GAD67 orthe GAD67 fragment (Fig. IB and C). Of the sera fromindividuals with new-onset diabetes or prediabetes, 83% (25of 30) had GAD65 autoantibodies, but only 26% (8 of 30) hadGAD67 autoantibodies (Table 1). In contrast, all restrictedICA had both GAD65 and GAD67 autoantibodies. With Fish-er's exact test, the frequency of GAD65 antibodies in new-onset diabetes or prediabetes was not different from therestricted ICA group (P = 0.56), but the frequency of GAD67

antibodies was statistically different between the two groups(P = 0.002). The presence of both GAD65 and GAD67 anti-bodies was not dependent on the titer of GAD65 antibodiesbecause individuals with comparable GAD65 antibody titersmay or may not also have GAD67 antibodies. (See blockingresults below.) No restricted ICA sera or sera from individ-uals with new-onset diabetes or prediabetes had only GAD67

antibodies.

Blocking of binding to GAD65 and GAD67 with unlabeledGAD65 or GAD67. Preincubation of the ICA sera withunlabeled GAD65 or GAD67 was used to determine if sera thatreact with both GAD65 and GAD67 contain GAD65-specificand GAD67-specific antibodies or a single GAD antibody thatreacts with both GAD65 and GAD67 proteins. For the blockingstudies, the restricted ICA sera were diluted 1:1,000-1:10,000

T

FIG. 1. ICA-positive sera immunoprecipitate GAD6S. A: left panel:GADe5 and GADr>7 in vitro translation products, prepared as described inMETHODS. Metabolically labeled GADfi5 and GADr>7 were incubated withserum and the resulting immunoprecipitate analyzed by SDS-PAGE andautoradiography. Right two panels: the results with either labeled GAD05(upper) or GAD67 (lower). Full-length GAD05 is immunoprecipitated bytwo sera from individuals with new-onset diabetes, but not by normalsera. GAD67 is immunoprecipitated by only one of two sera fromindividuals with new-onset diabetes. B: immunoprecipitation wasperformed with either full-length GADf>7 ( H^ ) or with a GAD07 fragmentconsisting of amino acids 110-593 ( • ) , and the resultingimmunoprecipitate analyzed quantitatively as described in METHODS. Eachexperiment represents a duplicate measurement, and the mean and SE ofthree experiments is shown for a restricted ICA serum (serum 1) andtwo sera from individuals with new-onset diabetes (sera 2 and 3). Theerror bars are too small to be seen on some data points. C: a restrictedICA sera was preincubated with either buffer, unlabeled GAD05,unlabeled GAD67 protein (amino acids 1-593), or unlabeled GADr>7fragment (amino acids 110-593), and the mixture was added to 50,000cpm of labeled, full-length GAD07 protein (1-593). Theimmunprecipitation was analyzed quantitatively as described in METHODS.Each experiment represents a duplicate measurement, and the mean ±SE of experiments is shown.

for GAD65 antibodies and 1:100-1:1,000 for GAD67 antibodies,and the sera from individuals with new-onset diabetes werediluted from 1:10 to 1:50. These dilutions were chosen so thatreactivity with the respective GAD proteins could be easilydetected, and the choices reflect the observation that GAD65

reactivity is usually detected at a 10-fold higher dilution thanis GAD67 reactivity. The results with one serum is shown inFig. 2A. Using restricted ICA sera and sera from individualswith new-onset diabetes or prediabetes (both those reactingwith GAD65 only and those reacting with both GAD65 andGAD67), GAD65 reactivity could be blocked with unlabeledGAD65 protein but not with unlabeled GAD67 protein (Fig. 2Aand B). Similar results were obtained with nine other sera.GAD67 reactivity of the restricted ICA sera and the sera fromindividuals with new-onset diabetes reacting with GAD07 wasblocked with either GAD65, full-length GAD67, or a GAD67

fragment with the amino terminal 109 amino acids deleted(Figs. 1C and 2A and C). Results were identical with eitherfull-length GAD67 protein or the GAD67 fragment consistingof amino acids 110-593 (Fig. IB and C, and data not shown).

These results suggest that at least two distinct sets ofantibodies are present in ICA-positive sera that react withboth GAD isoforms: one that reacts with GAD65 only, andone that reacts with both GAD65 and GAD67. The fact that


Dow


ecember 2021


APreabsorbedwtth:

GAD 65

4 B

GAD 67

0Sera Preincubated with: Buffer GAD 65 GAD 67

GAD65 reactivity can be blocked only with GAD65 and notwith GAD67 indicates that this antibody is specific for GAD65.The ability to block the GAD67 reactivity of both restrictedICA sera and sera from individuals with new-onset diabeteswith either GAD65 or GAD67 protein indicates that the GAD67

antibody reacts with a region that is highly homologousbetween the two proteins.Titer of GAD antibodies. Immunoprecipitation of recom-binant GAD65 and GAD67 protein was performed with variousdilutions of serum. All restricted ICA sera immunoprecipi-tated GAD65 using a 1:1,000 serum dilution or greater (Table1 shows the GAD65 titer of each sera). In contrast, only 13%(4 of 30) of the sera from individuals with new-onset diabetesor prediabetes immunoprecipitated GAD65 at a dilutiongreater than 1:100 (Table 1).

The amount of labeled GAD65 protein immunoprecipitatedby a 1:1 dilution of serum (mean ± SD of each group) wasmuch higher in the sera from individuals with new-onsetdiabetes (relative GAD binding = 0.51 ± 0.34 U) than in thenormal sera (relative GAD binding < 0.02 U) (P < 0.005)(Fig. 3). The relative GAD65 binding was much higher in therestricted ICA sera (relative GAD binding = 2.61 ± 1.39 U)than in the sera from individuals with new-onset diabetes(relative GAD binding = 0.51 ± 0.34 U) (P < 0.005) (Fig. 3).GAD epitope region recognition. Quantitative assays ofthe immunoprecipitated GAD65 protein showed that re-stricted ICA sera (dilution of 1:10) immunoprecipitatedequimolar amounts of the following GAD65 fragments: GAD65

(1-442), GAD65 (361-585), and GAD65 (amino acids 361^121deleted). However, the restricted ICA sera immunoprecipi-tated noticeably less of GAD65 fragments containing aminoacids 1-195 or 1-361. Figure 4A shows the results with onerestricted ICA serum, and similar results were obtained withthree other sera. Restricted ICA sera reactivity to full-lengthGAD65 and the GAD65 (188-585) fragment was detectable ata 10-fold greater dilution of serum than the reactivity to the

Sera Preincubated with: BUFFER GAD 65 GAD 67

FIG. 2. Blocking of GAD65 and GADG7 binding. A: a restricted ICA serawas preincubated with either buffer, unlabeled GAD65, or unlabeledGAD67, and then the mixture was added to 50,000 cpm of either labeledGAD65 protein or labeled GADf>7 protein as described in METHODS. Theimmunoprecipitation was analyzed by SDS-PAGE and autoradiography.B: a restricted ICA sera ( CH ), a sera from an individual with new-onsetdiabetes (GA065 positive) ( iHi ), and a sera from an individual withnew-onset diabetes (GAD65 and GAD67 positive) ( • ) werepreincubated with either buffer, unlabeled GAD65, or unlabeled GAD67,and then the mixture was added to 50,000 cpm of labeled GADG6 proteinas described in METHODS. The percentage of GAD proteinimmunoprecipitated following the preincubation is plotted. Each serumwas assayed in duplicate in each experiment and the mean ± SE of threeor four experiments is shown. C: a composite figure of the blocking offive sera (two restricted and three from individuals with new-onsetdiabetes). Restricted ICA sera and sera from an individuals withnew-onset diabetes (GAD65 and GAD67 positive) were preincubated witheither buffer, unlabeled GAD65, or a GAD67 fragment consisting of aminoacids 110-593. The mixture was added to the 50,000 cpm of labeledGAD67 protein (amino acids 110-593) as described in METHODS. Thedilution of the restricted ICA serum was 1:10,000 for GAD65 and 1:1,000for GAD67. The sera from individuals with new-onset diabetes wasdiluted from 1:10. Each sera was assayed in duplicate in one to fourexperiments, and the mean ± SE of all five sera is shown. Similar resultswere obtained with two additional sera using the autoradiographicapproach shown in Fig. 1A.

GAD65 fragments GAD65 (1-442), GAD65 (361-585), andGAD65 (amino acids 361-421 deleted) (data not shown).

In contrast, sera from individuals with new-onset diabetesor prediabetes (dilution 1:10) immunoprecipitated equimolaramounts of full-length GAD65 and GAD65 (188-585), but notthe GAD fragments GAD65 (361-585) and GAD65 (aminoacids 361-421 deleted). In addition, the sera from individualswith new-onset diabetes did not react with GAD65 fragmentscontaining amino acids 1-195, 1-361, and 1-442. Figure 4Bshows the results with one serum, and similar results wereobtained with five other sera.

The GAD6 monoclonal antibody immunoprecipitatedequimolar amounts of full-length GAD65 and fragments GAD65

5 j

45 ••

4 - •

3.5 ••

3 • •

2 . 5 ••

2 -•

1 .5 ••

0 5 •

0 -

NORMAL NEW ONSET RESTRICTED ICA

Type of Sera

FIG. 3. Quantitation of the amount of GAD6S immunoprecipitated.Relative GAD binding (calculated as described in METHODS) is shown fornormal sera (n = 10), sera from individuals with new-onset diabetes(n = 18), and restricted ICA sera (n = 6). Each point represents aduplicate determination. The mean ± SD is shown for each sera type.


Dow


ecember 2021


111

? 1000-Q.

O 800-

t0- 600 -o2 400 -5

Q 200-

n -

I

' . . . •

T

T l

585

Restrictedew Onset ICA GAD-6

B p

Q.

o.\-Q.Q.

Z

§Q<

450 -

400 -

350 -

300 -

250-

200 -

150 -

100 -

50 -

0 -

00in0000

i n

O

O

I

•

00inob00

i n

Q

i n

Q

O

42)

^-• ^

i n<£>Q

O

QCM

i^ ^(OCO

in

Q

O

CM

(OM̂ J

in

Q

m00inT_CDCO

mIDQ

O

in00i n

(OCO

intoO

i ntoQ

O

95)

i

mtoQ

O

<D

i n(DQ

O

|

61)

CO

mtoQ

O

o O oFIG. 4. GAD0B epitope region recognition by GAD-positive sera.GAD-positive sera were incubated with fragments of the GAD65 protein(amino acid boundaries of each fragment are shown). GAD65 (361-421D) refers to a GAD protein in which amino acids 361-421 were deletedby use of internal Nsi I sites in the cDNA. A: a restricted ICA serum; B:a serum from an individual with new-onset diabetes. Each serum wasdiluted 1:10, and immunoprecipitation was performed as described inMETHODS. The amount of protein precipitated/number of methionineresidues in that protein fragment is calculated as described in METHODS.Note the difference in the scale of the y axis of the two panels. Eachserum was analyzed in duplicate in each experiment, and the mean ± SEof three experiments is shown.

(188-585) and GAD65 (361^21 deleted), but it did not reactwith fragments GAD65 (1-442) and GAD65 (1-529) (data notshown). The GAD6 antibody is known to react with denaturedGAD65 protein by immunoblotting (4,34).

Figure 5 summarizes the epitope region mapping for therestricted ICA sera, sera from individuals with new-onsetdiabetes, and the GAD6 monoclonal antibody.

442

+++4

+++4

+++•»

FIG. 5. Summary of GAD65 epitope regions. Left: the amino acidboundaries of the GAD65 protein fragments; right: the reactivity of thesera or GAD6 with each GAD protein fragment. +, Plus SD range fromnormal mean; ++, plus 2 SD range from normal mean; and + + ++,exceeding plus 3 SD from normal mean.

DISCUSSIONGAD autoantibodies are an important immune marker intype I diabetes. Our study has found that GAD antibodies inICA sera are heterogenous and differ in titer, GAD isoformspecificity, and epitope region reactivity. At least threepatterns of GAD autoantibodies in ICA sera exist: 1) a set ofGAD65-speciflc antibodies is found in most sera from individ-uals with new-onset diabetes or prediabetes; 2) a set ofGAD65-specific antibodies and a second set of antibodies thatreact with both GAD65 and GAD67 are found in a minority ofindividuals with new-onset diabetes; and 3) both GAD65 andGAD67 antibodies in high titer are found in restricted ICAsera. Each pattern of GAD autoantibodies has distinctivereactivity, which suggests that the GAD antibody populationsare different in each ICA subset.

Our results indicate that GAD65 is the predominant GADautoantigen in type I diabetes and that the majority of serafrom individuals with new-onset diabetes or prediabetescontain a set of GAD65-specific antibodies. The GAD65-specific antibodies are likely a set of antibodies that havecommon features rather than a single antibody. This findingis in agreement with recent reports (17,25). However, somesera from individuals with new-onset diabetes or prediabetes(24%) also possess an antibody that reacts with both GAD65

and GAD67. This GAD67 antibody reacts to a region that ishighly homologous between the two GAD isoforms becauseGAD67 binding can be blocked by preincubating the serumwith GAD65. No sera from individuals with new-onset diabe-tes had antibodies only to GAD67. Our results differ withthose reported by Kaufman et al. (26), who found that 15 of23 (65%) individuals with type I diabetes or preclinicaldiabetes reacted with both GAD65 and GAD67, but agree withthe results of Hagopian et al. (25) and Velloso et al. (17). Apossible reason for these different results is that Kaufman etal. (26) used GAD prepared in a prokaryotic expressionsystem, but the other studies used GAD prepared in aeukaryotic system (17,25). There are currently no cleardifferences in the clinical characteristics of individuals withboth GAD65 and GAD67 antibodies compared with individualswho possess only GAD65 antibodies, but a systematic exam-ination of clinical characteristics of individuals based ontheir GAD65/GAD67 reactivity is lacking. Our results alsoshow that some individuals with new-onset diabetes arepositive for only GAD antibodies or ICA. A likely explanationis that the sera from individuals with new-onset diabetes


Dow


ecember 2021


possess several autoantibodies (directed at GAD, insulin,gangliosides, and other islet molecules) and that some indi-viduals have more of one autoantibody type compared withother individuals, in whom another autoantibody type pre-dominates. GAD is likely only a component of the immuno-cytochemical ICA, as suggested by Richter et al. (38).

Restricted ICA sera differ in the titer of GAD antibodiesand reactivity to GAD65 protein fragments. Both the titer ofGAD65 and GAD67 antibodies and the relative GAD bindingwere greater in the restricted ICA sera than in individualswith new-onset diabetes. The greater amount of GAD immu-noprecipitated by the restricted ICA may reflect higher titerantibodies in restricted ICA sera or the ability of the re-stricted GAD antibodies to react with GAD protein present ina non-native state that is not recognized by the sera fromindividuals with new-onset diabetes. This non-native GADmay consist of partially denatured protein, protein dimers,and GAD protein fragments resulting from less than full-length translation. Sera from individuals with new-onsetdiabetes do not immunoblot GAD protein or react withdenatured GAD protein (4) and likely do not react with suchnon-native GAD. An alternative explanation for some of thedifferences between the restrictive ICA and sera from indi-viduals with new-onset diabetes is that the antibody affinitymay be much higher in the restricted ICA sera.

Restricted ICA sera immunoprecipitated equimolaramounts of the GAD65 fragments GAD65 (1-442), GAD65

(361-585), and GAD65 (amino acids 361-421 deleted), but thesera from individuals with new-onset diabetes reacted poorlywith these GAD65 fragments. This indicates that some of theGAD autoantibodies in the restricted ICA detect a noncon-formational or linear epitope as further suggested by theability of these GAD antibodies to Western blot the GADprotein (R.G., unpublished observation). Epitope mappingwith GAD65 protein fragments indicates that at least twononconformational or linear epitope regions exist for therestricted ICA. The major epitope region resides betweenamino acids 361^442, and a second minor epitope regionresides between amino acids 1-195. Restricted ICA sera mayalso react with a third region of GAD65 near the COOH-terminal of the protein (K.D., A.C.P., unpublished observa-tion). These epitope regions can likely be further mapped indetail with synthetic GAD peptides because these GADantibodies react with GAD protein fragments. Restricted ICAsera also contain a set of antibodies that react only withfull-length GAD65 protein, not GAD65 fragments, because theantibody titer is 10-fold greater to full-length GAD65 proteinthan to the GAD fragments. This antibody has featuressimilar to the GAD65 antibody found in the sera of individualswith new-onset diabetes or prediabetes. Thus, the restrictedICA sera contain at least four distinctive sets of antibodiesthat react to the following GAD forms: full-length GAD65,GAD65 amino acids 361-442, GAD65 fragment 1-195, andGAD67.

In contrast, sera from individuals with new-onset diabetesor prediabetes react less well with GAD fragments comparedwith full-length GAD65 protein, which suggests that theseGAD antibodies recognize a conformational epitope that islost when fragments of the GAD protein are used. The GAD65

fragment (188-585) reacts equally well as the full-lengthprotein and is capable of blocking the binding of the full-length GAD65 protein, which suggests that this GAD fragmenthas sufficient protein conformation. This epitope cannot be

further localized using GAD fragments or synthetic peptidesand will require techniques that maintain the overall confor-mation of the protein.

The GAD65 epitope regions for the restricted ICA, for thesera from individuals with new-onset diabetes, and for theGAD6 monoclonal antibody reside in the COOH-terminal ofthe GAD65 protein. This is somewhat surprising, because thisregion has the highest degree of homology with GAD67 andyet these GAD antibodies are specific for GAD65. The regioncontaining the epitope for the sera from individuals withnew-onset diabetes or prediabetes, but not the major re-stricted ICA epitope region, includes the region of GAD65

with homology to the P2-C protein of the Coxsackie virus B4noted by Kaufman et al. (26) (amino acids 250-273 ofGAD65). The GAD epitope region described in our reportdoes not correspond to an epitope recently detected withEscherichia cota-prepared GAD, and the source of the GADprotein may account for this difference (41). Our results arein agreement with recent reports by Richter et al. (27,28),who used human monoclonal islet cell antibodies (MICA)derived from a single patient with type I diabetes to examineGAD reactivity. These authors found that the amino terminalof GAD is not necessary for antibody binding and thatconformational and linear epitopes exist. The sera from alarge number of individuals with new-onset diabetes paral-lels the behavior of the monoclonal antibodies, MICA 1 andMICA 3, in that both recognize a conformational epitopelocated in the COOH-terminal of the GAD65 (27,28). TheMICA 2 antibody also recognized a linear epitope, but it isfound very near the COOH-terminal of the GAD65 protein(27) and appears to be similar to the epitope detected with E.cott-prepared GAD (41). We did not detect antibodies resem-bling MICA 2 in the sera from individuals with new-onsetdiabetes, which is not surprising because these sera areknown not to immunoblot GAD protein. The presence ofseveral different monoclonal antibodies in the serum of oneindividual with new-onset diabetes likely results from theimmortalization of very infrequent monoclonal antibodyclones producing GAD antibodies or clones producing lowtiter GAD antibodies. Our results suggest that the majorantibody present in the sera from individuals with new-onsetdiabetes resembles MICA 1 and MICA 3 and that otherantibodies, if present in other individuals with new-onsetdiabetes, are at a much lower titer.

Our study demonstrates that the humoral response to GADis quite heterogenous. In addition, there are major differ-ences between the GAD autoantibodies found in restrictedICA sera and those found in the sera from individuals withnew-onset diabetes or prediabetes. These distinctive featuresinclude: 1) all restricted ICA sera reacted with both GAD65

and GAD67, but a minority of sera from individuals withnew-onset diabetes have antibodies to both GAD65 andGAD67; 2) the GAD antibody titers are much higher (at least10-fold) in the restricted ICA sera than in the sera fromindividuals with new-onset diabetes or prediabetes; and 3)restricted ICA sera contain at least four distinct GAD anti-bodies, but most sera from individuals with new-onset dia-betes or prediabetes have only one detectable set of GADantibodies. The identification of subsets of GAD autoantibod-ies that differ in their GAD65 and GAD67 reactivity and bytheir GAD65 protein epitope may help distinguish GADantibodies that are associated with the development ofdiabetes from GAD antibodies found in individuals with a


Dow


ecember 2021


low incidence of diabetes. In addition, the presence of suchsubsets requires that GAD antibodies be carefully measuredusing GAD in its native protein conformation before corre-lating GAD antibody status with clinical and immunologicfindings.

ACKNOWLEDGMENTSThese studies were supported by National Institutes ofHealth Grant R01-DK-43736, Career Development Awardsfrom the Department of Veterans Affairs Research Serviceand the American Diabetes Association (A.C.P), a MeritReview Award from the Veterans Affairs Research Service,the Diabetes Research and Training Center at VanderbiltUniversity (NIH DK-20593), a University Grant from Vander-bilt University, a BRSG grant administered by VanderbiltUniversity (BRSG RR-05424), and a grant from the DiabetesResearch and Education Foundation.

We wish to thank Dr. Richard Jackson, Joslin DiabetesCenter, and Dr. George Eisenbarth, Barbara Davis Center forChildhood Diabetes for measurement of ICA, for provision ofpatient sera and for helpful discussions; Dr. James Thomas,Vanderbilt University, for provision of patient sera and forreading the manuscript; and Dr. Mark Atkinson, University ofFlorida, for provision of patient sera and for helpful discus-sions.

REFERENCES1. Castano L, Eisenbarth GS: Type I diabetes: a chronic autoimmune disease

of human, mouse, and rat. Annu Rev Immunol 8:647-679, 19902. Riley WJ, Atkinson MA, Schatz DA, Maclaren NK: Comparison of islet

autoantibodies in "pre-diabetes" and recommendations for screening. JAutoimmun 3:47-51, 1990

3. Palmer JP, McCulloch DK: Perspectives in Diabetes: prediction andprevention of IDDM: 1991. Diabetes 40:943-947, 1991

4. Baekkeskov S, Aanstoot HJ, Christgau S, Reetz A, Solimena M, CascalhoM, Folli F, Richter-Olesen H, DeCamilli P: Identification of the 64Kautoantigen in insulin-dependent diabetes as the GABA-synthesizingenzyme glutamic acid decarboxylase. Nature 347:151-156, 1990

5. Atkinson MA, Maclaren NK: Islet cell autoantigens in insulin-dependentdiabetes. J Clin Invest 92:1608-1616, 1993

6. Erlander MG, Tobin AJ: The structural and functional heterogeneity ofglutamic acid decarboxylase: a review. Neurochem Res 16:215-226, 1991

7. Bu DF, Erlander MG, Hitz BC, Tillakaratne NJ, Kaufman DL, Wagner-McPherson CB, Evans GA, Tobin AJ: Two human glutamate decarbox-ylases, 65-kDa GAD and 67-kDa GAD, are encoded by a single gene. ProcNad Acad Sci USA 89:2115-2119, 1992

8. Christgau S, Schierbeck H, Aanstoot HJ, Aagaard L, Begley K, Kofod H,Ilejnaes K, Baekkeskov S: Pancreatic 3-cells express two autoantigenicforms of glutamic acid decarboxylase, a 65-kDa hydrophilic form and a64-kDa amphiphilic form which can be both membrane-bound andsoluble. J Biol Chem 266:21257-21269, 1991

9. Clare-Salzler MJ, Tobin AJ, Kaufman DL: Glutamate decarboxylase: anautoantigen in IDDM. Diabetes Care 15:132-135, 1992

10. Giorda R, Peakman M, Tan KC, Vergani D, Trucco M: Glutamic aciddecarboxylase expression in islets and brain. Lancet 338:1469-1470, 1991

11. Kelly C, Carter ND, Johnstone AP, Nussey SS: Cloning of large isofonn ofhuman brain glutamic acid decarboxylase. Lancet 338:1468-1469, 1991

12. Karlsen AE, Hagopian WA, Grubin CE, Dube S, Disteche CM, Adler DA,Barmeier H, Mathewes S, Grant FJ, Foster D, Lernmark A: Cloning andprimary structure of a human islet isoform of glutamic acid decarboxyl-ase from chromosome 10. Proc Nad Acad Sci USA 88:8337-8341, 1991

13. Cram DS, Barnett LD, Joseph JL, Harrison LC: Cloning and partialnucleotide sequence of human glutamic acid decarboxylase cDNA frombrain and pancreatic islets. Biochem Biophys Res Commun 176:1239—1244, 1991

14. Michelsen BK, Petersen JS, Boel E, Moldrup A, Dyrberg T, Madsen OD:Cloning, characterization, and autoimmune recognition of rat islet glu-tamic acid decarboxylase in insulin-dependent diabetes mellitus. ProcNad Acad Sci USA 88:8754-8758, 1991

15. Velloso LA, Kampe O, Eizirik DL, Hallberg A, Andersson A, Karlsson FA:Human autoantibodies react with glutamic acid decarboxylase. Diabeto-logia 36:39-46, 1993

16. Kim J, Richter W, Aanstoot HJ, Shi Y, Fu Q, Rajotte R, Warnock G,Baekkeskov S: Differential expresssion of GAD(i5 and GAD07 in human,rat, and mouse pancreatic islets. Diabetes 42:1799-1808, 1993

17. Velloso LA, Kampe 0, Hallberg A, Christmanson L, Betsholtz C, Karlsson

FA: Demonstration of GAD0B as the main immunogenic isoform ofglutamate decarboxylase in type I diabetes and determination of autoan-tibodies using a radioligand produced by eukaryotic expression. J ClinInvest 91:2084-2090, 1993

18. Hagopian WA, Karlsen AE, Gottsater A, Landin-Olsson M, Grubin CE,Sundkvist G, Petersen JS, Boel E, Dyrberg T, Lernmark A: Quantitativeassay using recombinant human islet glutamic acid decarboxylase(GADG5) shows that 64K autoantibody positivity at onset predicts diabe-tes type. J Clin Invest 91:368-374, 1993

19. Karlsen AE, Hagopian WA, Petersen JS, Boel E, Dyrberg T, Grubin CE,Michelsen BK, Madsen OD, Lernmark A: Recombinant glutamic aciddecarboxylase (representing the single isofonn expressed in humanislets) detects IDDM-associated 64,000-Mr autoantibodies. Diabetes 41:1355-1359, 1992

20. Christie MR, Brown TJ, Cassidy D: Binding of antibodies in sera fromtype 1 (insulin-dependent) diabetic patients to glutamate decarboxylasefrom rat tissues: evidence for antigenic and non-antigenic forms of theenzyme. Diabetologia 35:380-384, 1992

21. Rowley MJ, Mackay IR, Chen QY, Knowles WJ, Zimmet PZ: Antibodies toglutamic acid decarboxylase discriminate major types of diabetes melli-tus. Diabetes 41:548-551, 1992

22. Clare-Salzler MJ, Tobin AJ, Kaufman D: Glutamate decarboxylase: anautoantigen in IDDM. Diabetes Care 15:132-135, 1992

23. DeAizpurua HJ, Harrison LC, Cram DS: An ELISA for antibodies torecombinant glutamic acid decarboxylase in IDDM. Diabetes 41:1182—1187, 1992

24. DeAizpurua HJ, Wilson YM, Harrison LC: Glutamic acid decarboxylaseautoantibodies in preclinical insulin-dependent diabetes. Proc Nad Acad,Sci USA 89:9841-9845, 1992

25. Hagopian WA, Michelsen B, Karlsen AE, Larsen F, Moody A, Grubin CE,Rowe R, Petersen J, McEvoy R, Lernmark A: Autoantibodies in IDDMprimarily recognize the 65,000-Mr rather than the 67,000-Mr isoform ofglutamic acid decarboxylase. Diabetes 42:631-636, 1993

26. Kaufman DL, Erlander MG, Clare-Salzler M, Atkinson MA, Maclaren NK,Tobin AJ: Autoimmunity to two forms of glutamate decarboxylase ininsulin-dependent diabetes mellitus. J Clin Invest. 89:283-292, 1992

27. Richter W, Shi Y, Baekkeskov S: Autoreactive epitopes defined bydiabetes-associated human monoclonal antibodies are localized in themiddle and C-terminal domains of the smaller form of glutamate decar-boxylase. Proc Natl Acad Sci USA 90:2832-2836, 1993

28. Richter W, Endl J, Eiermann TH, Brandt M, Kientsch-Engel R, Thivolet C,Jungfer H, Scherbaum WA: Human monoclonal islet cell antibodies froma patient with insulin-dependent diabetes mellitus reveal glutamatedecarboxylase as the target antigen. Pwc Natl Acad Sci USA 89:8467-8471, 1992

29. Solimena M, Folli F, Aparisi R, Pozza G, DeCamilli P: Autoantibodies toGABA-ergic neurons and pancreatic B-cells in stiff-man syndrome. NEnglJMed 322:1555-1560, 1990

30. Genovese S, Bonifacio E, McNally JM, Dean BM, Wagner R, Bosi E, GaleEA, Bottazzo GF: Distinct cytoplasmic islet cell antibodies with differentrisks for type 1 (insulin-dependent) diabetes mellitus. Diabetologia,35:385-388, 1992

31. Gianani R, Pugliese A, Bonner-Weir S, Shiffrin AJ, Soeldner JS, Erlich H,Awdeh Z, Alper CA, Jackson RA, Eisenbarth GS: Prognostically signifi-cant heterogeneity of cytoplasmic islet cell antibodies in relatives ofpatients with type I diabetes. Diabetes 41:347-353, 1992

32. Julien JF, Samama P, Mallet J: Rat brain glutamic acid decarboxylasesequence deduced from a cloned cDNA. J Neunchem 54:703-705, 1990

33. Frohman MA: Rapid amplification of complementary DNA ends forgeneration of full-length cDNAs: thermal RACE. Methods Enzymol218:340-356, 1993

34. Chang YC, Gottlieb DL: Characterization of the proteins purified withmonoclonal antibodies to glutamic acid decarboxylase. J Neuwsci8:2123-2130, 1988

35. Mugnaini E, Dahl AL: Zinc-aldehyde fixation for light-microscopic imnui-nocytochemistry of nervous tissue. J Histochem Cylocliem 31:1435-1438,1983

36. Bottazzo GF, Doniach D: Islet cell antibodies (ICA) in diabetes mellitus:evidence of an autoantigen common to all the cells in the islet; ofLangerhans. Ric Clin Lab 8:29-38, 1978

37. Lipton R, LaPorte RE: Epidemiology of islet cell antibodies. EpidemiolRev 11:182-203, 1989

38. Richter W, Seissler J, Northemann W, Wolfhart S, Meinck HM,Scherbaum WA: Cytoplasmic islet cell antibodies recognize distinct isletantigens in IDDM but not in stiff-man syndrome. Diabetes 42:1642-1648,1993

39. Vardi P, Dibella EE, Pasquarello TJ, Srikanta S: Islet cell autoantibodies:pathobiology and clinical applications. Diabetes Care 10:645-656, 1987

40. Bottazzo GF, Genovese S, Bosi E, Dean BM, Christie MR, Bonifacio E:Novel considerations on the antibody/autoantigen system in type I(insulin-dependent) diabetes mellitus. Ann Med 23:453-461, 1991

41. Mauch L, Abney CC, Berg H, Scherbaum WA, Liedvogel B, NorthemannW: Characterization of a linear epitope within the human pancreatic64-kDa glutamic acid decarboxylase and its autoimmune recognition.Euro J Biochem 212:597-603, 1993


Dow


ecember 2021

identification of glutamic acid decarboxylase autoantibody - diabetes

Documents