© 1998 Nature Publishing Group http://www.nature.com/naturemedicine• ....................................................................................................................................................................................................................... ARTICLES
Tissue transglutaminase selectively modifies gliadin peptides
that are recognized by gut-derived T cells in celiac disease
0YVIND MOLBERG1, STEPHEN N. MCADAM1, ROMAN KORNER2, HANNE QUARSTEN1,
CHRISTEL KRIST!ANSEN1, LARS MADSEN3, LARS FUGGER3'4 , HELGE ScoTI'5, OVE NOREN6,
PETER ROEPSTORFF2, KNUT E.A. LUNDIN1, HANS SJ0STR6M6 & LUDVIG M. SOLLID1
'Institute of Transplantation Immunology and 'LIIPA T, Institute of Pathology,
Rikshospitalet, University of Oslo, N-0027 Oslo, Norway
2Department of Molecular Biology, Odense University, DK-5230 Odense M, Denmark
'Department of Clinical Immunology, Aarhus University Hospital, Skeiby Sygehus DK-8200 Aarhus N,Denmark
4Deptartment of Clinical Immunology, Rigshospitalet, University of Copenhagen,
6Department of Medical Biocehmistry and Genetics, Biochemistry Laboratory C, The Panum Institute,
University of Copenhagen, DK-2200 Copenhagen N, Denmark
Correspondence should be addressed to K.E.A.L.
present address: Department of Medicine, Ullevaal Hospital, Oslo, Norway
The action of tissue Transglutaminase1 (TGase) on specific pro-
tein-bound glutamine residues plays a critical role in numerous
biological processes2-s. Here we provide evidence for a new role
of this enzyme in the common, HLA-DQ2 (and DQ8) associated6
enteropathy, celiac disease (CD). The intestinal inflammation in
CD is precipitated by exposure to wheat gliadin in the diet' and
is associated with increased mucosa! activity of TGase8 • This en-
zyme has also been identified as the main target for CD-associ-
ated anti-endomysium autoantlbodies•, and is known to accept
gliadin as one of its few substrates 10. We have examined the pos-
sibility that TGase could be involved in modulating the reactivity
of gliadln specific T cells. This could establish a link between pre-
vious reports of the role of TGase in CD and the prevailing view
of CD as a T-cell mediated disorder1•11•12• We found a specific ef-
fect of TGase on T-cell recognition of gliadin. This effect was lim-
ited to gliadin-specific T cells isolated from intestinal CD lesions.
We demonstrate that TGase mediates its effect through an or-
dered and specific deamidation of gliadins. This deamidation
creates an epitope that binds efficiently to DQ2 and is recog-
nized by gut-derived T cells. Generation of epitopes by enzy-
matic modification is a new mechanism that may be relevant for
breaking of tolerance and initiation of autoimmune disease.
Immunofluorescence staining of small-intestine biopsy speci-
mens from control subjects and untreated CD patients demon-
strated expression of TGase mainly in the subepithelial region of
the mucosa. The CD patients had a higher expression level than
control subjects as well as some TGase expression in the brush
border. Notably, in the untreated CD patients, the subepithelial
region staining positive for TGase is heavily infiltrated with T
cells carrying the CD3 antigen and large cells expressing DQ
molecules (data not shown).
We tested a panel of T-cell lines and clones (from eight differ-
ent CD patients) isolated from duodenal biopsies challenged ex
vivo with pepsin -trypsin digested gliadin (PT-gliadin, an antigen
known to provoke disease in vivo13) and we found that addition
ofTGase to proliferative assays enhanced reactivity and sensitiv-
ity to PT-gliadin for all these gut-derived T cells (Fig. 1, left panel
and data not shown). None of the gut-derived T cells were reac-
tive to TGase alone. The effect of TGase was not dependent on
NATURE MEDICINE • VOLUME 4 • NUMBER 6 • JUNE 1998
type of antigen presenting cells (APC); both B lymphoblastoid
cell lines (B-LCL) and T-cell-depleted peripheral blood mononu-
clear cells (PBMC) had enhanced responses (10-fold and 30-fold
respectively). Further, the effect of TGase on T-cell recognition
appeared to be specific for gliadin. Two DR2 restricted, gut-de-
rived T-cell lines specific for astrovirus (one from an untreated
CD patient) and two DR3 restricted, PBMC-derived T-cell clones
specific for PPD (Purified Protein Derivative from Mycobacterium
tuberculosis) were not sensitive to TGase-mediated antigen modi-
fication (data not shown).
TGase catalyzes protein crosslinking through the formation of
isopeptide bonds between Gin and Lys residues1•3• Gliadins are
rich in Gin residues, but have very few Lys residues14•
Nonetheless, gliadin- gliadin complexes are known to form in
the presence of TGase10• Since particulate antigens are more effi-
ciently taken up by APC15, TGase could enhance gut-derived T-
eel! recognition of gliadin by catalyzing formation of gliadin
complexes. However, with the exception of one single DQ2 re-
stricted T-cell clone, no increase in PT-gliadin-specific prolifera-
tion of PBMC-derived T cells from six CD patients16 was detected
after addition of TGase (Fig. 1, right panel). This indicates that
increased uptake of gliadin mediated by TGase cannot account
for the enhanced proliferation of gut-derived T cells. These data
also indicate that gliadin-specific T cells that reside in peripheral
blood and gut mucosa may recognize distinct epitopes and, most
importantly, that TGase could be instrumental in the creation of
gliadin epitopes recognized by gut-derived T cells.
The PT-gliadin recognized by gut-derived T cells has been sub-
jected to digestion at low pH, a condition known to facilitate non-
enzymatic deamidation of Gin residues 11 . Interestingly, TGase can
catalyze substrates to react with water to be deamidated when
amine acceptors are in deficit' . Indeed, such Gln deamidations
have been demonstrated with gliadin as a substrate10• We tested
whether TGase-mediated deamidation of gliadin was important
for its recognition by gut-derived T cells by analyzing the effect of
TGase on gut-derived T-cell recognition of a water-soluble, chy-
motrypsin-digested gliadin. Digestion with chyrnotrypsin was
done at neutral pH, thereby avoiding spontaneous non-enzymatic
deamidation. Chymotrypsin-digested gliadin only weakly stimu-
lated gut-derived T cells from six CD patients (four DQZ+ and two
© 1998 Nature Publishing Group http://www.nature.com/naturemedicine
TCC restriction TCC restriction
412 R-2 002
+ TGasel!lil RNB.14 DR3 + TGase II
412 R-3 002 AK35 DR3
412 R-5 002 OHR 17 DR3
387 R-3 002 DHR41 DR3
387 R-16 002 DHR23 DR4
411 R-1 002 DHR26 DR4
380 E-27 002 AK16 002
282 1.12 DOB GVL 13 002
282 2.8 DOB AK44 002
282 2.27 DOB GVL40 008
282 2.35 DOB AK35 DP1
20 40 60 BO 100 20 60 100 140 180
cpmx 10·1 cpm x 10·'
Fig. 1 Comparison of the effect of TGase on PT-gliadin-specific T-cell
clones derived from the small intestine or from peripheral blood of celiac
disease patients. 100 µg/ml PT-gliadin was incubated for 2 h at 37 °C with
or without 100 µg/ml TGase and then incubated with APC (irradiated allo-
genic HLA homozygous B-LCL) for 16 h. Recognition of antigens by the T
cells was measured as 3H-thymidine incorporation 48-72 h after antigenic
stimulation. Data given as (cpm; Proliferation of [T cells + APC incubated
with antigen] - [T cells + APC incubated without gliadin]. Data for each T-
eel\ clone are from one representative experiment (of two or three).
DQ8·). However, it could be converted to an efficient antigen by
incubating it in acid (pH 1.8) for 1 hour at 98 °C. Notably, TGase
treatment of the chymotrypsin-digested gliadin generated a more
potent antigen for all of the gut-derived T cells than the acid/heat
treatment (Fig. 2 and data not shown).
For a detailed molecular analysis of the actions of TGase on
gliadin, we focused on a recently characterized gliadin epitope
recognized by gut-derived DQ2 restricted T cells from three dif-
ferent CD patients (H.S. et al., manuscript submitted). This epi-
tope was originally identified within a 20 amino acid fragment
isolated from a purified y-type gliadin. In its unmodified form
this fragment were determined to have the sequence
QQLPQPQQPQQSFPQQQRPF, a sequence that matched residues
134-153 of GDB2_WHEAT (SWISSPROT accession number
P08453). Acid/heat treatment of this fragment resulted in a com-
plex mixture of peptides with varying numbers of deamidated
Gin residues. Of these, the Glu at position 148 was essential for
recognition by gut-derived T cells, as without this deamidation
the 134-153 peptide was barely stimulatory. The minimal T-cell
epitope was mapped to residues 141-150 (H.S. et al, manuscript
T-cell lines from two patients (CD370 and CD412), as well as a
T-cell transfectant (60.6) expressing a chimeric T-cell receptor
containing variable regions cloned from the gliadin peptide-spe-
cific, gut-derived T-cell clone 4.32 (H.S. et al, manuscript submit-
ted), recognized the acid/heat treated peptide 134- 153. These T
cells were used to demonstrate that TGase treatment trans-
formed peptide 134-153 from a non-stimulatory to a potent T-
cell stimulatory peptide.
Ion exchange chromatography of peptide 134-153, treated
with TGase showed two distinct and dominant peaks (Fig. 3a).
Fractions collected from the chromatography were separated into
aliquotes. One aliquot was tested for recognition by the T-cell
tranfectant 60.6 and showed that the T-cell stimulatory capacity
of the peptide was retained in the peak corresponding to fraction
28. Fractions 15 and 28, corresponding to the two peaks were sub-
jected to methyl-esterification of acid residues (C-terminus and
Glu) followed by Matrix-Assisted Laser Desorption/Ionization-
Time-Of-Flight (MALDI-TOF) and ElectroSpray Ionization (ESI)
ion trap mass spectrometry to determine the number and posi-
tions of the deamidated glutamine residues. Both fractions had
one dominant peptide, with two and three deamidations respec-
tively (Fig. 3b). In the non-stimulatory fraction 15, Ql40 and
Q150 were identified as the deamidated glutamine residues. In
the T-cell stimulatory fraction 28 these residues were also deami-
dated, as well as a third residue, predominantly at Ql48.
Incubation of synthetic 1251-labeled peptides encoding T-cell epi-
topes (that is, peptides 134-153;£140 and 134-153;£140 E148,
E150) with TGase and subsequent analysis by SDS-PAGE indicated
that a fraction of each peptide became covalently coupled to the
enzyme (data not shown). This demonstrates that TGase and
gliadin peptides may form hapten-carrier-like complexes that can
be involved in production of TGase autoantibodies9 •
The proposed peptide-binding motif for DQ2 predicts a prefer-
ence for negatively charged residues in relative position 7 (refs.
18, 19) Testing of synthetic peptides in a cell-free DQ2 binding
assay demonstrated a IO-fold increase in binding affinity of pep-
tides with an E148 substitution (Fig. 4a and H.S. et al., manu-
script submitted), indicating that position 148 could be in DQ2
pocket 7. Although the binding affinity to DQ2 of acid/heat
treated peptide 138- 152; £148 the latter was a more potent anti-
gen for the T-cell transfectant (Fig. 4). This indicates that al-
though non-enzymatic deamidation generates a complex
mixture of deamidated variants, of which only a minority are ad-
vantageous for T-cell recognition, enzymatic deamidation by
TGase preferentially acts on residues favorable for T-cell recogni-
tion and binding to DQ2.
Our data indicate that the repertoire of antigen-specific HLA
class II restricted CD4·T cells can be shaped by modified antigens
and that organ-specific T cells in an inflammatory disease can
recognize epitopes generated by a specific enzyme activated by
the inflammation. This may go unreported as the standard use of
synthetic peptides for the characterization of T-cell epitopes
means that most in vivo modified epitopes would escape detec-
tion. Our data indicate that the fine specificity of peripheral
Fig. 2 Recognition of
gliadin by a DQ2-re-
stricted T-cell line from
the gut of an untreated
celiac disease patient
(CD4 l 2). Gliadin was in-
cubated as described in
Fig. 1. The acid heat-
treated gliadin was incu-
bated at pH 1.8, 98 °C
for 60 min, freeze-dried
and resuspended at pH
7.4. Allogenic, HLA-
PBMC (irradiated 25 Gy)
were used as APC. T cell
proliferation was mea-
-1:r- : Gladin alone
+ : Gliadin incubated with TGase
-0-; Acid/heat-treated glladin
-+-: Acidlheat-lreated glladin
incubated with TGase
10 0.1 0,01
Chymotrypsin-digested gliadin (mg/ml)
sured as in Fig. 1. Data are from one representative experiment (of three).
Similar data were obtained with T-cell lines from the small intestines of five
other CD patients.
NATURE M EDICINE • VOLUME 4 • NUMBER 6 • JUNE 1 998
© 1998 Nature Publishing Group http://www.nature.com/naturemedicine• ······················································································································································································································· ARTICLES
a 5 15 28 38
15 28 46
134· · . .. . . ... . ········· · --·· ·153
15 ·· .... .. E· ·" •• •· .... · ·E •••
28 .. . .. . . . E .. ..... ... E ... E ...
blood T cells and organ-specific T cells recognizing the same
complex antigen is different, emphasizes the danger of sampling
PBMC when studying the immunology of other organs.
The spatial relationship (revealed by immunohistochemistry),
between abundantly expressed extracellular TGase and DQ ex-
pressing cells just under the mucosa! epithelium of CD patients
suggests that TGase might be involved in processing gliadin be-
fore it binds to DQ2 molecules. There is a infiltrate of T cells in
the same region and ex vivo studies of gliadin-challenged biop-
sies have shown that some of these T cells are gliadin-specific20 •
Thus, the subepithelial region seems to be a microenvironment
for TGase-mediated modification, DQ2 binding and T cell recog-
nition of gliadin.
Detailed analysis of one T cell epitope (within residues
134-153 of a y-gliadin) recognized by DQ2 restricted gut-derived
T cells from three CD patients showed that TGase mediated an
ordered and specific deamidation of Gln residues which was im-
portant for DQ2 binding and critical for T cell recognition . Our
data indicate that peptide 134- 153 is not the only gliadin pep-
tide that can be transformed into an efficient T-cell epitope
through TGase modification. All the gut-derived T-cell clones
displayed in figure 2 showed enhanced reactivity to TGase mod-
ified gliadin, but none recognized TGase-treated peptide
134-153. Furthermore, analyses with different purified gliadins
have demonstrated that these T-cell clones are heterogeneous
with respect to fine-specificity (0.M., unpublished observa-
tions), which suggests that TGase-mediated deamidation poten-
tates several different gliadin epitopes.
Fig. 4 a, T cell recognition and DQ2 binding of the synthetic gliadin pep-
tide 138-152 and variants with Gin to Glu amino acid substitutions. Peptide
binding to DQ2 was measured in a cell-free assay with purified HLA-mole-
cules and binding affinity is given as IC,0, i.e. the amount of gliadin peptide
necessary for 50% inhibition of binding of a labeled indicator peptide. The
synthetic peptides were made with Gin to Glu substitutions in the same po-
sitions (140, 148 and 150) that are deamidated upon incubation of gliadin
peptide with TGase (Fig. 3b). Recognition of peptides by the DQ2 restricted
T-cell transfectant 60.6 was measured by IL-2 release in response to peptide
stimulation. b, Effect of acid/heat treatment of peptide 138-152; El 40,
El 48, El 50 on T cell recognition and DQ2 binding. Acid/heat treatment of
the peptide was done at pH 1.8, 98 °C for 60 min .
NATURE MEDICINE • VOLUME 4 • NUMBER 6 • IUNE 1998
Fig. 3 a, Kinetic analysis of TGase-mediated deamidation of gliadin pep-
tide 134-153. The peptide was incubated with TGase for various time pe-
riods and then analyzed by ion exchange chromatography. Depicted are
the chromatographs of the peptide after incubation with TGase for 0, 30,
60 and 180 min. The fractions of the 180 min chromatography were di-
vided. One aliquot was tested for recognition by the DQ2-restricted T-cell
transfectant 60.6. Fractions 15 and 28 from the other aliquot were methyl-
esterified and sequenced by ESI MS/ MS to determine the number and po-
sitions of deamidated glutamine residues. The peak corresponding to
fractions 38-46 in the chromatographs contains mainly impurities in the
Tris-HCI buffer. Recognition of each of the fractions by DQ2-restricted T-
cell transfectant 60.6 was measured by quantification of IL-2 release.
b, The upper lines show the amino acid sequence of gliadin peptide
134-153 prior to TGase treatment; lower lines, the number and positions
of deamidated glutamine residues in chromatography fractions 15 and 28
of peptide 134-153 after 180 min TGase treatment. The number and posi-
tions of deamidated glutamine residues in each fraction was assessed by
ESI MS/MS analysis.
The DQ2 binding epitopes recognized in CD are created by an
enzyme which is the main focus of the autoantibody response in
this disease9 • We have proposed that a prerequisite for produc-
tion of these disease-associated autoantibodies is that hapten-
carrier like complexes of gliadin and TGase are formed in vivo,
which would provide (gliadin-specific) T-cell help to TGase-spe-
cific B cells21 • Jndeed, our data indicate that, in addition to gen-
erating gliadin fragments recognized by gut-derived T cells,
TGase can also catalyze formation of complexes between gliadin
T-cell epitopes and TGase. Notably, in this model autoantibody
production to TGase does not involve breaking of T-cell toler-
ance to the abundantly expressed TGase protein.
lntroduction of negative charges at defined positions in pep-
tides probably affects the repertoire of HLA molecules able to
bind them. We observed an effect of TGase on both DQ2 and
DQ8 restricted T-cell epitopes. Notably, both DQ2 and DQ8 have
been predicted to preferentially bind peptides with negatively
charged residues' 8· ' 9•22 • The skewing towards DQ2 and DQ8 re-
striction of gliadin-specific gut-derived T cells may therefore re-
flect that TGase deamidation favors selective binding of gliadin
to these HLA molecules.
Although this report describes the enzymatic deamidation of a
foreign antigen, self antigens could also be substrates for similar
types of modifications. Mass spectrometric analysis of peptide
fragments eluted from HLA molecules have identified post-trans-
lationally-modified self proteins23 • Notably, autoreactive human
CDS· T-cells antigen specifically recognizing modified peptides
a Binding to T cell transfectant DQ2 (IC50)
Py.PfQPQQSFP.!;Q.!;RP 1.8 50 " -0
Py.P.!;QPQQSFP.!;Q.!;RP 1.8 150
Py.P.!;,QPQQSFP.!;Q.!;RP 1.2 50
0,4 2 10 50
© 1998 Nature Publishing Group http://www.nature.com/naturemedicine• ARTICLES
have been described23•24. In a mouse arthritis model it has also
been demonstrated that modified self-peptides are recognized by
the class JI-restricted autoreactive T cells2s. Future work should
determine whether enzymatic modification of antigens is a gen-
eral mechanism for the immune system to define epitopes or if it
is a process invariably associated with the kind of pathological
immune activation described in this study.
lmmunohistochemistry. Cryostat sections from healthy controls and un-
treated CD patients were stained using a indirect three-color immunofluo-
rescence technique20• Images were digitally analyzed using a Eclipse E800
microscope (Nikon) and a C5810 imaging system (Hamatsu, Shizouka,
Japan). The following mouse monoclonal antibodies were used; anti-TGase
(TG l 00, lgGl; Neomarkers, Fremont, California), anti-DQ (SPV-L3, lgG2a;
gift H. Spits, DNAX Research Institute, Palo Alto, California) and anti-CD3
(RIV9, lgG3; Sanbio, Am Uden, The Netherlands). Secondary goat anti
mouse antibodies (Southern Biotechnology, Birmingham, Alabama) were
biotinylated anti-lgG3, FITC-conjugated anti-lgG2a and cy.3-conjugated
anti-lgGl. The biotinylated anti-lgG3 was visualized by an AMCA conju-
gated streptavidin (Vector, Burlingame, California).
Preparation of gliadin antigens. Digestion of crude gliadin (Sigma; G-
3375) with pepsin (Sigma; P-7012) and trypsin (Sigma; T-7418) was done
as described". a-chymotrypsin (Sigma; C-3142) digestion of gliadin from
the wheat strain Kadett was performed at 200:l (w/w) in 0.1 M NH,HC03
with 2 M urea at 37 •c for 24 hand stopped by incubating in boiling water
for 5 min. Synthetic gliadin peptide quality-controlled by HPLC (> 80% pu-
rity) and MS were purchased from Research Genetics (Huntsville, Alabama).
Generation of antigen specific T cells. T-cell lines derived from gut and
specific for PT-gliadin were established from small intestinal biopsies as de-
scribed", except that isolation of some T cells from the gliadin-challenged
biopsy material was done without enrichment for CD25-expressing cells
(0.M. et al., manuscript submitted). Gut derived T cell lines of three DQ2 +
CD patients (CD370, CD380 and CD387) and two DQ8 + patients (CD282
and CD360) on a gluten-free diet and three untreated DQ2 + patients
(CD4 l 0, CD4 l l and CD4 l 2) were tested. In addition, DQ2 restricted T-cell
clones (370.E-l 4; 380.E-3, E-11 and E-27; 387.R-3,R-12 and R-16; 411 .R-1;
412.R-2, R-3 and R-5) (0.M. et al., manuscript submitted) made from these
T-cell lines were used. Generation of gut-derived astrovirus-specific T-cell
lines (ref. 26 and 0 .M . et al., manuscript submitted), gut-derived DQ8 re-
stricted T-cell clones", PPD-reactive T-cell clones (RN4.26 and 5.28 from
PBMC; E.A.L. et al., unpublished) and P/T-gliadin-specific T-cell clones from
PBMC'6 have been described elsewhere.
Generation of a DQ2 restricted T-cell transfectant specific for gliadin.
Va and VI} gene segments from the T-cell clone 4.32 derived from the gut"
were amplified from genomic DNA and cloned into the T-cell receptor ex-
pression vectors, p T acass and p T(}cass, respectively". These vectors allow
for functional expression of murine or human variable region T-cell receptor
spliced to murine constant regions (L.M., LS. and l.F., unpublished obser-
vations). Transfection of these plasmids with CD4:pBs and a Neo selection
marker into the T-cell hybridoma BW 58a-/(}- deficient for T-cell receptors29
allowed for the selection and cloning of a stable transfectant (60.6) express-
ing human CD4 and chimeric T-cell receptor.
T-cell assays. T cells (5 x 10') were added to 5 x 104 APC (either HLA-
matched allogenic B-LCL transformed by Epstein Barr virus and irradiated
with 80 Gy or HLA-matched allogenic PBMC irradiated with 25 Gy) which
had been incubated for 16-20 h with complex antigen or four h with pep-
tides in a volume of l O µI RPMI 1640 (Life Technologies) supplemented
with 15% pooled human serum. Assays were performed in 96-well U-bot-
tom plates (Nunc) and T-cell proliferation was measured as 3H-thymidine
incorporation 48-72 h after antigenic stimulation. Activation of the 60.6 T-
cell transfectant was measured as IL2 release following an 18 h incubation
of 2.5 x 1 o• transfectant cells with 5 x 1 o• DQ2 + B-LCL as APC. The con-
centration of IL2 was quantified by time-resolved fluorometry, using Delfia
reagents (Wallac, Finland) and murine anti-lL2 antibodies (JES6-1 A 12 and
JES6-5H4, PharMingen, California). Parallel assays using the original T-cell
clone and the transfectant confirmed that the chimeric T-cell receptor ex-
pressed by the transfectant maintained its specificity for MHC restriction
and peptide (data not shown).
Tissue transglutaminase assays. PT-gliadin (1-1000 µg/ml), chy-
motrypsin-digested gliadin (0.1-1000 µg/ ml) or gliadin peptides (0.2-100
µM) were incubated with 100 µg/ml guinea pig TGase (Sigma; T-5398) at
37 °c for 2 h in PBS with 0.8mM CaCI,. These concentrations of TGase and
CaCl,were found to give optimal T-cell stimulation and are comparable to
that described in other systems'·10• After 2 h incubation, gliadin/TGase mix-
tures were incubated with APC and tested in T-cell assays as above.
DQ2 peptide binding assay. A competitive peptide binding assay using
affinity purified HLA-DQ(al *0501,(}*0201) molecules, a 1251 -labeled indica-
tor peptide KPLLIIAEDVEGEY (Mycobacterium Bovis 65 kDa heat shock pro-
tein 243-255Y) and various unlabeled gliadin peptides was performed as
SOS-PAGE of gliadin peptides incubated with TGase. Peptides 134-153;
134-153, El 48; and 134-153, El 40; E148, El 50 were synthesized with an
extra N-terminal Y and radiolabeled with 1251 using the chloramine T
method'". TGase was 125I-labeled as control. Trace amounts of '"I-peptides
(20,000 cpm) were mixed with their corresponding unlabeled peptides in
various concentrations and incubated with 1 OOµg/ml unlabeled TGase at
37 °c for 2 h in PBS with 0.8 mM CaCl2• The samples were boiled in
Laemmli buffer with 2-mercaptoethanol for 5 min and separated by 15%
Ion exchange chromatography. Mono-Q chromatography in the SMART
system (Pharmacia) was run in 5 mM Tris-HCI buffer purified by treatment
with DAEA-Sephadex A50 (buffer A, pH 6.5). The column was eluted by a
gradient with a final concentration of 0.05 M NaCl in buffer A. Synthetic
peptide 134-153 (1 mg/ml) was dissolved in buffer A supplemented with
0.8 mM CaCI, and incubated with 1 mg/ml TGase.
Mass spectrometry. Fractions collected from ion exchange chromatogra-
phy of TGase treated peptides were subjected to methyl-esterification'0 of
acid residues (C-terminus and Glu) followed by MALDI-TOF mass spectrom-
etry" to determine the number of Glu residues. ESl32-ion trap" tandem
mass spectrometry30 was used to determine the position of Glu residues.
Peptide ions were isolated in the ion trap and fragmented by collisions with
He gas creating a series of N- and C-terminal fragments with differing pep-
tide chain lengths. By comparing the fragmentation patterns of the native
peptide with the TGase treated fractions, mass increments of 15 Da were as-
signed to methyl-esterified Glu residues in these fragment ions. MALDI-TOF
mass spectra were acquired on a Reflex mass spectrometer (Bruker-Franzen
Analytik GmbH, Bremen, Germany). ESI tandem mass spectra were
recorded on an ESQUIRE ion trap mass spectrometer (Bruker-Franzen
We thank E. Thorsby for review of the manuscript, E. Brundtland, T. Jensen and
D. Anthonsen for technical assistance, H. Gjertsen for providing T-ce/1 clones
and K. Kett for providing biopsy specimens. This work was supported with
grants from the Research Council of Norway, Medinnova, the Danish Medical
Research Council, the Danish Cancer Society, the Novo Nordisk Foundation
and the Lundbeck Foundation. 0. N. and H. S. are members of the Danish
Biomembrane Research Centre.
RECEIVED 10 FEBRUARY; ACCEPTED 30 MARCH 1998
1. Folk, 1-E. Mechanism and basis for specificity of transglutaminase-catalyzed e-(y -glu-
tamyl) lysine bond formation . Adv. Enzymol. Re/at. Areas Mo/. Biol. 54, 1- 56 (1983).
2. Kahlem, P., Terre, C., Green, H. & Djian, P. Peptides containing glutamine repeats as
substrates for transglutaminase-catalyzed cross-linking: relevance to diseases of the
nervous system Proc. Natl. Acad. Sci. USA 93, 14580-14585 (1996).
3. Aeschlimann, D. & Paulsson, M. Transglutaminases: protein cross-linking enzymes in
tissues and body fluids. Thromb. Haemost. 71, 402-415 (1 994).
NATURE MEDICINE • VOLUME 4 • NUMBER 6 • JUNE 1998
© 1998 Nature Publishing Group http://www.nature.com/naturemedicine•
4. Davies, P. J. et al. Transglutaminase is essential in receptor-mediated endocytosis of
a2-macroglobulin and polypeptide hormones. Nature 283, 162-167 (1980).
5. Teshigawara, K., Kannagi, R., Noro, N . & Masuda, T. Possible involvement of trans-
glutaminase in endocytosis and antigen presentation. Microbial. lmmunal. 29,
6. Sollid, L. M . & Thorsby, E. HLA susceptibility genes in celiac disease: genetic map-
ping and role in pathogenesis. Castroenterology 105, 910-922 (1993).
7. Trier, J. S. Celiac sprue. N. Engl./. Med. 325, 1709-1719 (1991 ).
8. Bruce, S. E., Bjarnason, I. & Peters, T.J . Human jejunal transglutaminase: demonstra-
tion of activity, enzyme kinetics and substrate specificity with special relation to
gliadin and coeliac disease. Clin. Sci. (Co/ch.) 68, 573-579 (1985).
9. Dieterich, W. et al. Identification of tissue transglutaminase as the autoantigen of
celiac disease. Nature Med. 3, 797-801 (7-1997).
10. Larre, C. et al. Gliadin modifications catalysed by guinea pig liver transglutaminase.
/. Food. Biochem 267-282 (1993).
11. Lundin, K. E.A. et al. Gliadin-specific, HLA-DQ(al •0501,p1 *0201) restricted T cells
isolated from the small intestinal mucosa of celiac disease patients. /. Exp. Med. 178,
12. MacDonald, T. T. In Coeliac Disease (ed. Marsh, M . N.) 283-304 (Blackwell Scientific,
13. Frazer, A. C. Gluten-induced enteropathy, the effect of partially digested gluten.
Lancet 252-255 (1959).
14. Wieser, H. In Caeliac disease (ed . Howdle, P. D.) 191-208 (Bailliere Tindall, London,
15. Watts, C. Capture and processing of exogenous antigens for presentation on MHC
molecules. Ann. Rev. lmmunol. 15, 821 - 850 (1997).
16. Gjertsen, H. A. et al. T cells from the peripheral blood of coeliac disease patients rec-
ognize gluten antigens when presented by HLA-DR, -DQ, or -DP molecules. Scand. / .
lmmunol. 39, 567- 574 (1994).
17. Hamada, J. S. Deamidation of food proteins to improve functionality. Crit. Rev. Food
Sci. Nutr. 34, 283-292 (1994).
18. Johansen, B. H. et al. Identification of a putative motif for binding of peptides to
HLA-DQ2. Int. lmmunol. 8, 1 77-182 (1996).
19. van de Wal, Y. et al. Peptide binding characteristics of the coeliac disease-associated
DQ(al •0501, Pl *0201) molecule. lmmunogenetics 44, 246-253 (1996).
NATURE MEDICINE • VOLUME 4 • NUMBER 6 • JUNE 1998
20. Halstensen, T. S. et al. In situ two- and three-color immunofluorescence staining of
mucosal T-cells in celiac disease. Increase of TCR -y/o+CD8- and TCR a/f3+CD45RO+
intraepithelial lymphocytes and IL2R+ TCR aJP+CD4+CD45RO+lamina propria lym-
phocytes. Prag. Histochem. Cytochem. 26, 201-210 (1992).
21 . Sollid, L. M ., Molberg, 0 ., McAdam, S. & Lundin, K.E.A. Autoantibodies in coeliac
disease, . Tissue transglutaminase - guilty by association . Cut 41, 851-852 (1997).
22. Kwok, W. W. et al. HlA-DQBl codon 57 is critical for peptide binding and recogni-
tion./. Exp. Med. 183, 1253-1258 (1996).
23. Meadows, L. et al. The HLA-A·0201-restricted H-Y antigen contains a posttransla-
tionally modified cysteine that significantly affects T cell recognition . Immunity 6,
24. Skipper, J. C. et al. An HLA-A2-restricted tyrosinase antigen on melanoma cells results
from posttranslational modification and suggests a novel pathway for processing of
membrane proteins. /. Exp. Med. 183, 527-534 (1996).
25. Michaelsson, E. et al. T cell recognition of carbohydrates on type II collagen . / . Exp.
Med. 180, 745- 749 (1994).
26. Molberg, 0. et al. CD4+ T cells with specific reactivity against astrovirus isolated
from normal human small intestine. Gastroenteralogy 114, 115-122 (1998).
27. Lundin, K. E. A. et al. T cells from the small intestinal mucosa of a DR4, DQ7/DR4,
DQ8 celiac disease patient preferentially recognize gliadin when presented by DQ8.
Hum. lmmunol. 41, 285-291 (1994).
28. Kouskoff, V., Signorelli, K., Benoist, C. & Mathis, D. Cassette vectors directing ex-
pression of T cell receptor genes in transgenic mice. /. lmmunol. Methods 180,
29. Letourneur, F. & Malissen, B. Derivation of a T cell hybridoma variant deprived of
functional T cell receptor a1p chain transcripts reveals a nonfunctional alpha-mRNA
of BW5147 origin. Eur. J. lmmunol. 19, 2269-2274 (1989) .
30. Hunt, D. F. et al. Protein sequencing by tandem mass spectrometry. Proc. Natl. Acad.
Sci. USA 83, 6233-6237 (1986).
31. Karas, M. & Hillenkamp, F. Laser desorption ionization of proteins with molecular
masses exceeding 10,000 daltons. Anal. Chem. 60, 2299-2301 (1988).
32. Fenn, J. B. et al. Electrospray ionization for mass spectrometry of large biomolecules.
Science 246, 64-71 (1989) .
33. Cooks, G. L., Glish, G.L., McLuskey, S.A. & Kaiser, R.E. Ion trap mass spectrometry.
Chem. Eng. News 25, 26-41 (1991 ).
- Tissue transglutaminase selectively modifies gliadin peptides that are recognized by gut-derived T cells in celiac disease