of May 18, 2011
This information is current as
Alemzadeh, Soumitra Ghosh and Martin J. Hessner
Xujing Wang, Shuang Jia, Rhonda Geoffrey, Ramin
Type 1 Diabetes Mellitus Using Serum and
Identification of a Molecular Signature in Human
Article cited in:
, 12 of which can be accessed free at:
cites 40 articles
is online at
The Journal of Immunology
Information about subscribing to
Submit copyright permission requests at
Receive free email-alerts when new articles cite this article. Sign up at
Print ISSN: 0022-1767 Online ISSN: 1550-6606.
Immunologists, Inc. All rights reserved.
by The American Association ofCopyright ©2008
9650 Rockville Pike, Bethesda, MD 20814-3994.
The American Association of Immunologists, Inc.,
is published twice each month by
The Journal of Immunology
on May 18, 2011
Identification of a Molecular Signature in Human Type 1
Diabetes Mellitus Using Serum and Functional Genomics1
Xujing Wang,*†Shuang Jia,* Rhonda Geoffrey,* Ramin Alemzadeh,‡Soumitra Ghosh,*†
and Martin J. Hessner2*†
Understanding active proinflammatory mechanisms at and before type 1 diabetes mellitus (T1DM) onset is hindered in humans,
given that the relevant tissues are inaccessible and pancreatic immune responses are difficult to measure in the periphery by
traditional approaches. Therefore, we investigated the use of a sensitive and comprehensive genomics strategy to investigate the
presence of proinflammatory factors in serum. The sera of recent onset diabetes patients (n ? 15, 12 possessing and 3 lacking islet
cell autoantibodies), long-standing diabetes patients (n ? 12), “at risk” siblings of diabetes patients (n ? 9), and healthy controls
(n ? 12) were used to induce gene expression in unrelated, healthy PBMC. After culture, gene expression was measured with
microarrays and normalized expression data were subjected to hierarchical clustering and multidimensional scaling. All recent
onset sera induced an expression signature (192 UniGenes; fold change: >1.5, p < 0.01; false discovery rate: <0.01) that included
IL-1 cytokine family members and chemokines involved in monocyte/macrophage and neutrophil chemotaxis, as well as numerous
receptors and signaling molecules. This molecular signature was not induced with the sera of healthy controls or long standing
diabetes patients, where longitudinal analysis of “at risk” siblings (n ? 3) before and after onset support the hypothesis that the
signature emerges years before onset. This study supports prior investigations of serum that reflect disease processes associated
with progression to T1DM. Identification of unique inflammatory mediators may improve disease prediction beyond current islet
autoantibodies. Furthermore, proinflammatory serum markers may be used as inclusion criteria or endpoint measures in clinical
trials aimed at preventing T1DM. The Journal of Immunology, 2008, 180: 1929–1937.
by the presence of islet cell autoantibodies that are useful for pre-
dicting T1DM (1–3). Risk estimates vary, however; a titer for a
single autoantibody among first-degree relatives imparts little risk,
whereas high titers for two or three Abs are predictive (?80%) of
disease within 5 years (4, 5). T1DM patients have an estimated
80–90% loss in ? cell mass at diagnosis, although functioning
mass still remains (6). Upon initiation of insulin therapy, 25–100%
of newly diagnosed patients experience a transient restoration of re-
maining ? cell function, termed the “honeymoon period”, that lasts
from months to years (7–10). This immunologically active time is
ype 1 diabetes mellitus (T1DM)3is a ? cell-specific au-
toimmune disease that results in life-long dependency on
daily insulin injections. Clinical onset is often preceded
clinically significant because it offers: 1) a window for potential ther-
apeutic intervention aimed at arresting ? cell destruction, thus pre-
serving endogenous ? cell mass and insulin secretion; and 2) a time
to potentially measure processes related to ? cell destruction.
The development of T1DM involves both an adaptive T cell
response as well as a significant cytokine-based arm that kills pan-
creatic ? cells (11–14). In recent onset T1DM populations, ele-
vated serum levels of IL-1?, IL-1?, IL-6, IL-8, IL-18, TNF-?,
CXCL10, and IFN-? have been reported (15–19); however, suffi-
ciently reliable detection methods to enable potentially mechanis-
tic or prognostic measurement remains challenging. Recently, a
genomics approach was used to investigate the presence of inflam-
matory factors in the sera of juvenile rheumatoid arthritis (JRA)
patients (20). Gene expression profiling was conducted on PBMCs
isolated from healthy individuals after culture with autologous,
patient, or allogeneic control serum. A proinflammatory signature
was induced by patient sera and provided a foundation for more
directed studies that ultimately led to successful treatment. Because
serum cytokine levels in T1DM are often too low to measure directly
but potentially sufficient to induce the expression of genes under their
influence, we applied this sensitive comprehensive approach to recent
onset diabetics, at risk siblings of probands, long-standing diabetics,
and healthy controls.
Materials and Methods
Subjects and subject characterization
Recent onset (RO) diabetes patients (after stabilization on exogenous in-
sulin but within 7 mo of diagnosis; n ? 15), long-standing (LS) diabetes
patients (?10 years after diagnosis; n ? 12) and “at risk” (AR) subjects
(Ab-positive siblings of probands; n ? 9) were recruited through Chil-
dren’s Hospital of Wisconsin. Diabetes was defined according to World
Health Organization criteria and included blood glucose levels of ?200
mg/dl with symptoms confirmed by a physician (21). Healthy control (HC;
n ? 12) criteria included fasting blood glucose of ?100 mg/dl, no familial
*Max McGee National Research Center for Juvenile Diabetes, Department of Pediatrics
at Medical College of Wisconsin and Children’s Research Institute of Children’s Hospital
of Wisconsin,†Human and Molecular Genetics Center, Medical College of Wisconsin,
and‡Children’s Hospital of Wisconsin Diabetes Center, Pediatric Endocrinology and
Metabolism, Medical College of Wisconsin, Milwaukee WI 53226
Received for publication August 16, 2007. Accepted for publication November 15, 2007.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
1This work was supported by National Institutes of Health, National Institute of
Biomedical Imaging and Bioengineering Grant EB001421, National Institute of Al-
lergy and Infectious Diseases Grant P01-AI-42380, General Clinical Research Cen-
ters Grant M01-RR00058, as well as Advancing a Healthier Wisconsin Initiative
Grant 5520065, and The Children’s Hospital of Wisconsin Foundation.
2Address correspondence and reprint requests to Dr. Martin J. Hessner, Department
of Pediatrics, The Medical College of Wisconsin, 8701 Watertown Plank Road, Mil-
waukee, WI 53226. E-mail address: firstname.lastname@example.org
3Abbreviations used in this paper: T1DM, type 1 diabetes mellitus; AR, at risk; FDR,
tyrosine phosphatase-2; JRA, juvenile rheumatoid arthritis; LS, long standing; MDS, mul-
tiple dimensional scaling; qRT-PCR, quantitative RT-PCR; RO, recent onset.
Copyright © 2008 by The American Association of Immunologists, Inc. 0022-1767/08/$2.00
The Journal of Immunology
on May 18, 2011
Table I. Subject characteristics
Relative to OnsetHgA1c (%)GAD IA2 IAAb
Recent onset T1DM
AR1a (visit 1)
AR1b (visit 2)
AR2a (visit 1)
AR2b (visit 2)
AR3a (visit 1)
AR3b (visit 2)
AR4a (visit 1)
AR4b (visit 2)
AR5a (visit 1)
AR5b (visit 2)
AR6a (visit 1)
AR6b (visit 2)
RO Ab-negative T1DM
SS1 (visit 1)
SS1 (visit 2)
SS1 (visit 3)
SS2 (visit 1)
SS2 (visit 2)
SS3 (visit 1)
SS3 (visit 2)
SS3 (visit 3)
SS3 (visit 4)
SS3 (visit 5)
SS3 (visit 6)
aMeasured as described (22) after the initiation of treatment with exogenous insulin. The sera of healthy controls was used to define 99th percentile threshold cutoffs for positive
assays. Threshold cutoffs for GAD, IA2, and IAA were ?0.07, ?0.007, and ?0.05, respectively. Sample values exceeding these respective cutoffs are shown in boldfaced font.
cAge at sample collection.
dAb-positive siblings of probands; longitudinal sample pairs.
eRelative to visit 1.
fPreonset to postonset longitudinal sample series.
1930IDENTIFICATION OF A T1DM MOLECULAR SIGNATURE
on May 18, 2011
history of any autoimmune disorder, and negativity for islet autoantibodies
at the 99th percentile (22). All study subjects were free of known infection
at the time of sample collection; subject characteristics are shown in Table
I. All blood samples were drawn by trained phlebotomists at Children’s
Hospital of Wisconsin and immediately processed. Under sterile condi-
tions, peripheral blood (acid citrate dextrose solution B, anti-coagulated)
was collected and components were separated by Ficoll-Histopaque den-
sity gradient centrifugation. PBMCs were viably frozen in RPMI 1640
medium at ?80oC supplemented with 20% FCS and 10% DMSO until
DNA or RNA was extracted. Serum was stored at ?80°C until use. Auto-
antibody titers for glutamic acid decarboxylase (GAD), protein tyrosine
phosphatase-2 (IA2), and insulin were determined as previously described
(22). HLA-DQB1 genotypes were determined with the SeCore DQB1 se-
quencing kit in accordance with the manufacturer’s instructions (Invitrogen
Life Technologies). Serum IL-1? levels were determined by ELISA (hu-
man IL-1?/IL-1F2 immunoassay, catalog no. DLB50; R&D Systems) in
accordance with the manufacturer’s instructions. The study was approved
by the Institutional Review Board (IRB) of the Children’s Hospital of
Wisconsin (IRB no. 01-15) and informed consent was obtained from par-
PBMC cultures, RNA extractions, and GeneChip analysis
Fresh PBMCs of healthy donors for use in cultures were isolated from
20–40 ml of blood by Ficoll-Histopaque density gradient centrifugation
and used immediately. As previously described (20), the induction of gene
expression was accomplished by culturing the PBMCs of healthy blood
donors for 6 h at 37°C in 5% CO2with 20% of autologous sera, HC sera,
RO sera, LS sera, or AR sera. Cultures were prepared in a Costar 24-well
plate (Corning) using 4–6 wells per condition (106cells/well in 800 ?l of
RPMI 1640 medium plus 200 ?l of sera). In this study a total of 62 subject
sera were analyzed, requiring 13 different blood donors. A given serum
sample was used to induce gene expression in cells of a single donor. The
donor cells used in the testing of each serum are shown in supplement B.4
After culture, the cells distributed over 4–6 wells for analysis of each
subject were pooled, and total RNA was extracted using TRIzol reagent
(Invitrogen Life Technologies), providing sufficient material for global gene
expression analysis and quantitative RT-PCR confirmation. The GeneChip
human genome U133 plus 2.0 array interrogates ?47,000 transcripts and thus
was selected for these studies for its overall comprehensive coverage. Purified
RNA (?50 ng) was amplified using an Affymetrix two-cycle cDNA synthesis
kit (catalog no. 900432), and cRNA was synthesized, labeled, fragmented, and
hybridized to the array in accordance to the Affymetrix GeneChip expression
analysis technical manual. RNA from each culture was independently ana-
lyzed. After hybridization, arrays were washed, stained with PE-conjugated
streptavidin (Molecular Probes), and scanned. Image data were analyzed with
Multichip Analysis (RMA; www.bioconductor.org/) to determine signal log
ratios. The statistical significance of differential gene expression was de-
rived through a Student’s t test and false discovery rates (FDR) were de-
termined with Significance Analysis of Microarrays (SAM) software as
described (23). Ontological pathway analysis was performed with Onto-
Express (http://vortex.cs.wayne.edu/ontoexpress) and Database for Anno-
gov/) (24, 25). These tools conduct overrepresentation analyses of the
functional gene categories detected relative to the total functional gene
categories assayed by the array using the Gene Ontology (GO) project
databases as references. Hierarchical clustering was conducted with
Genesis (26) and multidimensional scaling was performed with
Real-time quantitative RT-PCR (qRT-PCR)
Specific oligonucleotide primers for the following selected genes were de-
signed with Oligo 6.66 (Molecular Biology Insights): IL1B, IL1R1,
CXCL5, KCNJ15 (potassium inwardly rectifying channel, subfamily J,
member 15), SDC2 (syndecan 2), CREM (cAMP-responsive element mod-
ulator), CXCL1, and CCR1. Monoplex real-time qRT-PCR was performed
using a Rotor-Gene 3000 thermal cycler (Corbett Research), a Quan-
tumRNA 18S internal standards kit (Ambion), locus-specific primers
(Sigma Genosys), and QuantiTect SYBR Green PCR Master Mix (Qiagen)
according to the manufacturers’ instructions. Synthesis of first-strand
cDNA from 0.2–0.5 ?g unamplified RNA was accomplished with random
hexamers (Invitrogen Life Technologies) and SuperScript II (Invitrogen
Life Technologies) according to the manufacturer’s instructions. Duplicate
locus-specific and 18S PCRs were performed for each gene analyzed in
20-?l reactions that included 2 ?l of cDNA and 10 ?l of 2? SYBR Quan-
tiTect SYBR Green PCR Master Mix (Qiagen) possessing 1.2 ?l of locus-
specific (10 ?M) or 18S-specific competimers (used as a 3:7 ratio of primer/
competimer set; each stock is at 5 ?M) and 6.8 ?l of deionized water.
Reactions were typically cycled as follows: stage 1, 95°C for 900 s; stage
2, 50 cycles at 95°C for 30 s, 50–66°C for 30 s (locus specific), 72°C for
30 s, and fluorescence acquisition at 72–82°C for 20 s (locus specific); and
stage 3, melt curve at 60–95°C. The 18S reactions were cycled as follows:
stage 1, 95°C for 900 s; stage 2, 50 cycles at 95°C for 30 s, 55°C for 30 s,
72°C for 30 s, and 82°C for 15 s; and stage 3, melt curve at 60–95°C. A
pooled and concentrated sample of HC or RO cDNA was used for both the
4The online version of this article contains supplemental material.
Table II. qRT-PCR performance parameters and primer designs
Gene NameUniGene Slopea
Oligonucleotide Primer Sequencesd
Interleukin 1, ? (IL1B)Hs.126256
Interleukin 1 receptor, type I (IL1R1) Hs.557403
Chemokine (C-X-C motif) ligand 5
Chemokine (C-X-C motif) ligand 1
?3.49 0.99 1.94
Chemokine (C-C motif) receptor 1
K inwardly-rectifying channel,
subfamily J, member 15 (KCNJ15)
Syndecan 2 (SDC2) Hs.1501
cAMP responsive element modulator
aSlope of standard curve.
bLinear regression of standard curve.
cReaction efficiency, determined as previously described (?27?).
dTop sequence is forward primer, bottom is reverse.
1931The Journal of Immunology
on May 18, 2011
locus-specific and the 18S standard curves at undiluted, 1/5, 1/25, 1/125,
and 1/625 concentrations. At least two points from the standard curve were
used as positive controls in each assay. Specificity for all qRT-PCR was
verified by both melting curve analysis and 1.5% agarose gel detection of
a single product of the predicted size. The data were analyzed with the
Rotor-Gene 3000 software using the cycle threshold for quantification.
Relative gene expression data (fold change) between samples was accom-
plished using the mathematical model described by Pfaffl (27). Primer de-
signs and reaction performance parameters are provided in Table II.
Analysis of RO diabetes patients and HC
Similarly as in the prediabetic stage, RO T1DM subjects experi-
ence active killing of their pancreatic ? cells. To determine
whether proinflammatory serum factors related to the autoimmune
processes of T1DM could be detected through their ability to in-
duce gene expression, healthy PBMCs were cultured in the pres-
ence of autologous, allogeneic RO, or allogeneic HC sera. We
opted for this strategy vs directly assaying PBMCs of cases and
controls because: 1) immune responses are considered to be local
events and thus participating cells may not be highly repre-
sented in peripheral circulation; and 2) Pascual et al., (20) ob-
served that the gene expression differences between PBMCs
incubated with JRA vs control sera were more robust than those
observed between the direct gene expression profiling of case
and control PBMCs. To avoid inducing gene expression due to
factors related to hyperglycemia, samples from RO T1DM pa-
tients were collected after stabilization on exogenous insulin
(2–7 mo after diagnosis; Table I).
Gene expression profiling was conducted on healthy PBMCs
that were cultured with the allogeneic sera of either RO diabetes
patients (n ? 12) or healthy controls (n ? 12). The transcrip-
tional responses of healthy PBMCs to either healthy allogeneic
serum or recent T1DM onset serum were normalized with those
induced by autologous sera to account for gene expression in-
duced by isolating and placing the PBMCs in culture. First,
unsupervised hierarchical clustering was conducted. This anal-
ysis used any probe set that exhibited a minimum 2-fold change
(log2ratio of ?1.0 relative to the autologous induction; n ?
827) in any six of the 24 total samples (12 RO and 12 HC). This
analysis showed that the RO and HC samples cluster into two
groups (supplement A).
To focus the analysis, we first performed a permutation test with
SAM (23). We then identified the differentially expressed probe
sets (log2ratio ? 0.58 (1.5-fold); Student’s t test, p ? 0.01 of log
ratio, FDR ? 0.01) between the RO and HC groups (n ? 233
probe sets, 192 unique UniGenes). These are tabulated in supple-
ment B. Our power, determined through resampling, was calcu-
lated to be on average 70% (supplement B). Multiple probe sets
targeting the same UniGene were averaged and evaluated by mul-
tiple dimensional scaling (MDS) and hierarchical clustering (Fig.
1, A and B); with either approach the RO and HC groups distinctly
clustered. Although the expression profiles induced by RO and
HC sera are clearly resolved by MDS and hierarchical clustering,
cantly regulated (fold change of ?1.5; Student’s t test, p ? 0.01) after
healthy PBMCs are cultured with RO (n ? 12) vs HC (n ? 12) sera. All
data were normalized with that of the autologous induction to account for
gene expression induced by placing the PBMCs into culture. The related-
ness of the 24 profiles was examined by MDS and hierarchical clustering.
A, In MDS analysis, each sample is plotted in a three-dimensional space
where the similar samples are plotted in closer proximity compared with
the dissimilar ones. RO samples are shown as red circles and HC samples
are shown as black squares. B, Hierarchical clustering was conduced with
Genesis (26). With either algorithm, the samples cluster to form two dis-
tinct groups, each possessing a high degree of intragroup similarity among
RO and HC samples. The scale represents fold of change between the
serum tested relative to autologous serum (?4-fold to ?4-fold). (The sup-
porting data for these analyses are provided in supplement B).
Gene expression profiles of 192 unique UniGenes signifi-
Table III. Pathway analysis of genes differentially regulated by PBMCs
incubated with RO or HC seraa
Positive regulation of
Response to stimulus
3.2 ? 10?10
3.2 ? 10?10
1.2 ? 10?8
1.3 ? 10?8
2.4 ? 10?6
4.2 ? 10?6
5.7 ? 10?5
8.9 ? 10?5
1.4 ? 10?4
6.8 ? 10?4
1.1 ? 10?3
1.1 ? 10?3
3.3 ? 10?3
5.9 ? 10?3
aAnalysis restricted to Gene Ontology (GO) biological processes where more
than three UniGenes were detected per category and possessed FDR-corrected p val-
bUnique UniGene total (unique UniGenes detected).
cUnique reference UniGene total (unique UniGenes assayed on array).
1932 IDENTIFICATION OF A T1DM MOLECULAR SIGNATURE
on May 18, 2011
the molecular signature induced by RO sera is not an all or none
response as is reflected by Fig. 1B, where it is clear that a given
gene within a cohort is not always up- or down-regulated to the
same degree. Ontological analysis of this subset of 192 genes iden-
tified 14 significant Gene Ontology biological processes (p ?
0.01, after FDR correction) possessing at least four unique Uni-
Genes (Table III). These Gene Ontology biological processes in-
cluded chemotaxis, cell-cell signaling, signal transduction, im-
mune response, positive regulation of cell proliferation, cell
adhesion, and G protein-coupled receptor protein signaling. An-
notated within these 14 categories were 68 UniGenes up-regulated
by RO sera related to proinflammatory processes (Fig. 2). RO sera
induced transcription of numerous immune signaling molecules
and receptors including: 1) the IL-1 cytokine family members IL-
1?, IL-1R1, and IL-1R2; and 2) the chemokines CCL2 and CCL7,
involved in monocyte/macrophage chemotaxis, as well as IL-8,
CXCL1, CXCL3, and CXCL5, involved in neutrophil chemotaxis
(28). CCR1, a receptor associated with a number of autoinflam-
matory diseases, was also induced by RO sera (29, 30). Several
innate immunity receptors were also significantly up-regulated by
RO sera, including HEBGF (heparin-binding EGF-like growth fac-
tor) and the TREM1 (triggering receptor on myeloid cells 1). Not
using 68 well-annotated genes up-reg-
ulated by RO sera. Expression profiles
induced by PBMCs cultured with RO
(n ? 12) and HC (n ? 12) sera dis-
tinctly cluster. Indicated are the Uni-
Gene identifier, gene symbol, and
mean-induced fold of change induced
by incubation with RO and HC (rela-
tive to autologous induction) for each
gene (the detailed results of samples
for the 192 and 68 UniGene subsets
are provided in supplement B). The
scale represents fold of change be-
tween the serum tested relative to au-
tologous serum (?4-fold to ? 4-fold).
1933The Journal of Immunology
on May 18, 2011
listed are TLR2 and TLR4 (?1.6-fold) that were detected at a
significance of p ? 0.05. Genes involved with cell adhesion and/or
cell motility were up-regulated, including ICAM1, SDC2, and en-
dolyn (CD164). Genes associated with G protein-coupled receptor
signaling, such as EDG3 (endothelial differentiation, sphingolipid
G protein-coupled receptor 3) and RGS1 (regulator of G protein
signaling 1) were also up-regulated by RO sera.
Follow-up qRT-PCR and ELISA studies
In the array analysis, many of the genes differentially regulated
when healthy PBMCs are cultured with either RO or HC sera are
known to be functionally related to and/or influenced by IL-1?.
Confirmatory qRT-PCR was performed for eight such loci (Table
IV). The IL-1 family members IL-1? and IL-1R1 were determined
to be 3.3 ? 1.9- and 9.0 ? 1.7 -fold increased, respectively, when
comparing PBMCs incubated with RO vs HC sera. CXCL5,
CXCL1, and CCR1 were respectively determined to be 3.8 ? 2.0-,
3.8 ? 2.9-, and 4.9 ? 1.2 -fold increased when comparing PBMCs
incubated with RO vs HC sera. Lastly, KCNJ15, SDC2, and
CREM were determined to be 1.7 ? 1.6-, 4.6 ? 3.4-, and 3.6 ?
1.4-fold increased, respectively, in cultures incubated with RO vs
HC sera. Consistent with the array results, qRT-PCR analysis con-
firmed statistically significant overrepresentation of each of these
eight transcripts in PBMCs cultured with 20% RO sera.
The serum levels of IL-1? of RO and HC subjects were mea-
sured by a commercial ELISA possessing a sensitivity of detection
of ?1 pg/ml. Among the 12 RO and 12 HC serum samples ana-
lyzed in the microarray studies, sufficient serum was available for
10 of the RO subjects and all 12 HC subjects. Serum samples from
an additional 42 RO T1DM subjects (n ? 52 total) and an addi-
tional 41 HC subjects (n ? 53 total) were added to this analysis.
These additional subjects met the RO and HC criteria defined in
the methods section. Differences in serum IL-1? levels were not
observed between these two groups (RO, 0.93 ? 1.1 pg/ml; HC,
0.77 ? 1.0 pg/ml; p ? 0.5, Student’s t test).
Using qRT-PCR, we investigated whether PBMCs drawn di-
rectly from RO T1DM patients overexpressed these eight tran-
scripts relative to PBMCs collected from healthy controls. These
studies used PBMCs isolated (by density gradient centrifugation)
from the same blood draws (2–6 mo after diagnosis) that were
used in the serum culture experiments described above. Aside from
finding a marginally significant increase in the abundance of CCR1
transcript (2.5 ? 1.0-fold increase in RO PBMCs relative to HC
PBMCs; p ? 0.0415), we did not find statistically significant dif-
ferential expression (Table IV).
Overall, the qRT-PCRs on PBMCs cultured with RO and HC
sera confirm the differential expression of transcripts regulated
by or related to IL-1? that were detected in the array analysis.
The ELISA results fail to show differences between RO and HC
subjects in serum IL-1? levels, highlighting the difficulty mea-
suring events related to T1DM in the periphery and suggesting
that if IL-1? is present in the sera of the RO cohort, it is below
the ELISA detection limit. Alternatively, other cytokine(s) may
be responsible for the molecular signature induced by RO sera.
The qRT-PCR analysis of the RO and HC PBMCs indicate that
IL-1? is not being produced by leukocytes in the periphery.
Analysis of other subject groups
To better understand the nature of the RO signature, other sub-
ject groups were investigated. Twelve LS subjects were ana-
lyzed, all were ?10 years postonset and 11/12 possessed a mea-
surable titer for at least one islet cell autoantibody (Table I). LS
sera induced expression signatures that clustered with those in-
duced by HC sera (Fig. 3A and supplement B). This is consis-
tent with long-standing T1DM not being immunologically active
and supports the hypothesis that the molecular signature expressed by
PBMCs cultured with RO serum arises from factors related to active
We then investigated how the emergence of the molecular sig-
nature correlates with islet cell autoantibody status and disease
progression in 12 individuals. Six currently healthy autoantibody-
positive AR siblings of diabetics were each monitored on two
occasions with a minimum interval between visits of 1 year (Table
I and Fig. 3B). Sera of subjects AR1, AR4, AR5, and AR6 induced
profiles similar to those induced with HC sera when analyzing the
192 UniGenes differentially regulated between inductions with HC
and RO sera or the focused subset of 68 genes (Fig. 3B). The sera
of subjects AR2 and AR3 induced a signature more similar to and
clustered with the RO inductions. Sample pairs from a given subject
were always concordant for the presence or absence of the proinflam-
between the induced expression profile and the number of autoanti-
bodies, Ab specificity, or HLA-DQB1 genotype. Given the current
sample size, the power to detect such associations is limited. The
serum of three recent onset T1DM patients that lacked measurable
to induce expression profiles that clustered with those induced by RO
sera when using the complete set of 192 regulated UniGenes or the
focused set of 68 UniGenes (Fig. 3C), further associating the molec-
ular signature with active inflammatory processes vs autoantibody
status. However, the presence of autoantibodies targeting other ? cell
proteins cannot be excluded in these subjects. Lastly, three longitu-
dinal subject series collected from prospectively monitored siblings of
diabetics that developed T1DM were also evaluated. The first subject
Table IV. qRT-PCR of selected genes on cultured PBMCs vs PBMCs directly collected from subjects
Gene expression induced in cultured PBMCs incubated with RO, HC,
and autologous (Auto) serum
Gene expression measured directly
in PBMCs of RO and HC subjects
RO vs Auto
HC vs Auto
RO vs HC
RO vs HC
Student’s t testa
RO vs HC
RO vs HC
Student’s t testa
8.6 ? 15.3
15.7 ? 24.4
18.9 ? 33.6
69.5 ? 169.1
6.9 ? 7.8
4.9 ? 6.3
25.6 ? 24.9
11.7 ? 13.5
2.6 ? 1.5
1.8 ? 1.1
5.0 ? 4.8
18.2 ? 28.6
1.4 ? 0.6
2.9 ? 2.5
7.6 ? 4.6
3.2 ? 2.3
3.3 ? 1.9
9.0 ? 1.7
3.8 ? 2.0
3.8 ? 2.9
4.9 ? 1.2
1.7 ? 1.6
4.6 ? 3.4
3.6 ? 1.4
p ? 0.0003
p ? 0.0009
p ? 0.00001
p ? 0.0005
p ? 0.005
p ? 0.005
p ? 0.00001
p ? 0.00001
0.5 ? 1.5
1.0 ? 2.1
0.6 ? 1.7
0.2 ? 2.9
2.5 ? 1.0
1.7 ? 1.5
0.5 ? 1.6
1.6 ? 1.4
p ? 0.8
p ? 0.3
p ? 0.6
p ? 0.8
p ? 0.0415
p ? 0.8
p ? 0.1
p ? 0.09
aStudent’s t test performed on log ratio.
1934IDENTIFICATION OF A T1DM MOLECULAR SIGNATURE
on May 18, 2011
(SS1), diagnosed at the age of 9.0 years, was evaluated at ?18.0 mo
(?1 autoantibody), ?4.4 mo (?2 autoantibodies), and ? 1.7 mo (?3
autoantibodies) relative to onset. The second subject (SS2), diagnosed
at the age of 4.9 years, was evaluated at ?3.7 mo (?3 autoantibodies)
and ? 1.0 mo (?3 autoantibodies) relative to onset. In subjects
1 and 2, sera from all preonset time points induced the proin-
flammatory signature and clustered with the recent onset ex-
pression profile when using the complete set of 192 regulated
UniGenes or the focused set of 68 UniGenes. The third subject
(SS3), diagnosed at the age of 21.7 years, was the most com-
prehensively studied. Samples were collected at ?63.8 mo
(autoantibody negative), ?39.8 mo (autoantibody negative),
PBMCs incubated with the sera of
additional subject groups using the
68 annotated genes up-regulated by
RO sera. A, Hierarchical clustering
of expression profiles induced by
RO (n ? 12), HC (n ? 12), and LS
(n ? 12) sera. Profiles of PBMCs
incubated with sera collected from
LS diabetics (?10 years postonset)
cluster with those incubated with
HC sera, consistent with an absence
of active autoimmunity. B, The sera
of AR autoantibody-positive sib-
lings of diabetics (n ? 6) were an-
alyzed at two longitudinal time
points each (a vs b; Table I). The
sera of two subjects (AR2 and AR3
at both time points analyzed) in-
duced proinflammatory expression
profiles similar to those induced by
RO sera. C, The sera of three RO
Ab-negative T1DM patients (ROAN;
lacking titers for GAD, IA2, and insulin
autoantibodies) were analyzed and
found to induce signatures similar to
those of autoantibody-positive T1DM
serum. D, Longitudinal sample series 3
(SS3) is from an AR sibling of a dia-
betic that developed T1DM. Samples
were collected at ?63.8 mo (autoanti-
body negative), ?39.8 mo (autoantibody
negative), ?29.3 mo (?2 autoanti-
bodies), ?17.6 mo (?2 autoanti-
bodies), ?4.0 mo (?2 autoantibodies),
and ? 3.3 mo (?2 autoantibodies) rel-
ative to onset. Hierarchical clustering
was conducted using an average of the
RO (n ? 12) and HC (n ? 12) expres-
sion values to better view the develop-
ment of the proinflammatory profile.
All preonset sample profiles cluster
with the recent onset sera profile. The
scale represents fold of change between
the serum tested relative to autologous
serum (?4-fold to ? 4-fold).
Expression profiles of
1935 The Journal of Immunology
on May 18, 2011
?29.3 mo (?2 autoantibodies), ?17.6 mo (?2 autoantibodies),
?4.0 mo (?2 autoantibodies), and ?3.3 mo (?2 autoantibod-
ies) relative to onset. In SS3, the proinflammatory signature was
again observed in all preonset samples analyzed before the
emergence of islet cell autoantibodies (Fig. 3D).
In this article, we define a molecular signature that is induced by
factors present in the sera of RO diabetes patients. The sera of HC
and LS diabetes patients fail to induce the molecular signature.
These observations are consistent with an absence of inflammatory
factors in HC sera and an overall absence of autoimmunity. Likewise,
in LS sera these factors are absent due to significantly reduced or no
? cell mass and an absence of active autoimmunity. Together, these
results support the hypothesis that induction of the signature by RO
sera is reflective of active autoimmune processes.
Three longitudinal sample series were analyzed. Although the
number of cases is small, the proinflammatory signature was ob-
served before the clinical onset of T1DM in all cases. In SS3, the
signature was present 63.8 mo (?5 years) before onset and before
the emergence of islet cell autoantibodies, implying that this indi-
rect yet sensitive approach may be useful in predicting onset in AR
subjects. Among the six autoantibody positive subjects studied the
signature was induced by the sera of only two, raising two ques-
tions. First, is onset pending in signature-positive/Ab-positive in-
dividuals? Second, do signature-negative/Ab-positive individuals
represent those Ab-positive individuals that never progress to
T1DM? The prospective study of these and additional subjects will
clarify potential prognostic value of these strategies. It is not likely
that these studies have merely measured a general inflammatory
response, as the sera of individuals undergoing an atopic asthmatic
response to Parietaria judaica pollen induce a completely unique
profile consisting of different proinflammatory transcripts/path-
ways (M. J. Hessner, unpublished results).
Previous studies have clearly established that there exists a com-
plex cytokine milieu in T1DM patients (15–19); however, many of
the 68 UniGenes of the proinflammatory signature are known to be
influenced by IL-1 through either increased transcription or increased
mRNA half-life. These include IL-1?, IL-1R2, IL-8, CCR1, CCL2,
CCL7, CXCL1, CXCL3, CXCL5, SDC2, ICAM1, TIMP1, MT1X,
CTSL, APLP2, PLAUR, heparin-binding EGF-like growth factor
(HBEGF), CD86, SLC11A1, plasminogen activator inhibitor 2
(SERPINB2), and CREM. Furthermore, KCNJ15, which has been
linked to the lysosome-mediated control of IL-1? secretion (31), is
also up-regulated by RO sera. Although not reaching our significance
threshold, RO sera increased the expression of cyclooxygenase type 2
(PTGS2) and platelet activating factor (PAFAH1B1), consistent with
the presence of IL-1. The expression profiles induced by T1DM and
JRA serum (20) exhibit some parallels. Pascual et al. (20) reported on
46 genes up-regulated by JRA serum, and 22 of these are detected
among the 192 UniGenes significantly induced when comparing RO
to HC serum. It must be emphasized that these studies have evaluated
the induction of gene expression after culturing healthy PBMCs with
RO sera; the degree to which the transcription of these genes will
correlate to detection of their respective translated protein products
remains to be determined.
Recently, Kaizer et al. (32) reported the direct expression pro-
filing of PBMCs isolated from healthy controls, T1DM patients at
diagnosis, T1DM patients at 1 mo postdiagnosis, and T1DM pa-
tients at 4 mo postdiagnosis. In patient PBMCs at diagnosis, they
identified an expression signature that included IL1?, CCR1,
CXCL1, and TREM1, which are in common with the focused 68
UniGene signature induced in healthy PBMCs by RO serum. The
PBMC signature observed in PBMCs collected at diagnosis was
reported to resolve by 4 mo postonset despite the fact that pancre-
atic ? cell destruction is ongoing during this time, prompting the
authors to hypothesize that many of their observations may be a
direct or indirect consequence of hyperglycemia (32). Although
there is a mixed body of evidence showing that hyperglycemia can
affect IL-1? secretion by pancreatic ? cells and other cell types
(33–37), the molecular signature defined here is likely not a con-
sequence of hyperglycemia because it was detected years before
T1DM onset in the longitudinal case studies and all of the RO
samples evaluated were collected after stabilization on exogenous
insulin, were normoglycemic, and possessed histories of good gly-
cemic control based upon hemoglobin A1c values (group average
7.7 ? 1.1% SD). In contrast, the long-standing diabetes patients
had a slightly poorer history of glycemic control (average 8.0 ?
1.3 SD), yet the profiles induced by LS sera cluster with those of
induce by HC sera. The molecular signature induced by RO serum
also resolves sometime between 7 mo (maximum collection time
of our recent onset cohort) and 10 years (minimum time of our LS
cohort). Defining when the molecular signature resolves and how
this coincides with the end of the honeymoon period is the focus
of ongoing studies. In this study, we performed confirmatory qRT-
PCR for eight genes (IL-1?, IL-1R1, CXCL5, CXCL1, CCR1,
KCNJ15, SDC2, and CREM) that were significantly up-regulated
in healthy PBMCs cultured with RO sera. Parallel studies directly
examining PBMCs isolated from peripheral blood in general failed
to show overexpression of these same genes in RO vs HC PBMCs,
although the CCR1 transcript was 2.5-fold more abundant in RO
than in the HC PBMCs (p ? 0.045). These results, generated from
PBMCs collected 2–6 mo postdiagnosis, are consistent with those
of Kaizer et al., (32) in that our measurements were taken at a time
when differential gene expression in T1DM PBMCs is reported to
be largely diminished to levels similar to those of healthy controls.
Given that the IL-1 receptor antagonist has been identified as a
successful therapy for JRA (20, 38), other autoinflammatory diseases
(39), and most recently type II diabetes (40), questions are raised as to
whether such treatment of recent onset diabetics may prolong the
honeymoon period and whether treatment of Ab-positive/profile-pos-
itive AR subjects may delay or prevent T1DM onset. As a first step to
address this possibility and dissect active inflammatory pathways, ex-
periments to impair induction of the proinflammatory signature in
vitro by blocking the action of IL-1? (and IL-1?) with IL-1RA are
ongoing. Because cytokine milieu associated with T1DM is complex
(15–19), we are also exploring the effect of immunodepletion of mul-
tiple cytokines in an effort to define the major contributors to the
signature induced in healthy PBMCs by RO serum.
We have determined that sera of preonset and RO diabetics possess
is reflective of active autoimmunity. Although the overall sample size
is modest, we observe a dramatic result that is supported through the
comprehensive testing of appropriate additional subject cohorts. This
study lays the foundation for urgent further work to define the utility
of this approach in predicting onset in AR subjects and as a measure
in primary and secondary prevention trials.
We thank Marilyn Koppen and Joanna Kramer for their excellent sample
and database management. We also thank the physicians, nurses, and staff
of Children’s Hospital of Wisconsin and The Max McGee National Re-
search Center for Juvenile Diabetes who assisted in subject recruitment and
The authors have no financial conflict of interest.
1936 IDENTIFICATION OF A T1DM MOLECULAR SIGNATURE
on May 18, 2011
References Download full-text
1. Diabetes Prevention Trial Type 1 Diabetes Study Group. 2002. Effects of insulin
in relatives of patients with type 1 diabetes mellitus. N. Engl. J. Med. 346:
2. Bingley, P. J., E. Bonifacio, A. J. Williams, S. Genovese, G. F. Bottazzo, and
E. A. Gale. 1997. Prediction of IDDM in the general population: strategies based
on combinations of autoantibody markers. Diabetes 46: 1701–1710.
3. Nejentsev, S., M. Sjoroos, T. Soukka, M. Knip, O. Simell, T. Lovgren, and
J. Ilonen. 1999. Population-based genetic screening for the estimation of Type 1
diabetes mellitus risk in Finland: selective genotyping of markers in the HLA-
DQB1, HLA-DQA1 and HLA-DRB1 loci. Diabet. Med. 16: 985–992.
4. Lernmark, A., H. Kolb, and T. Mire-Sluis. 1999. Towards a World Health Or-
ganization (WHO) approved standard sample for islet cell antibodies, GAD65
and IA-2 autoantibodies. Diabetologia 42: 381–382.
5. Franke, B., T. S. Galloway, and T. J. Wilkin. 2005. Developments in the pre-
diction of type 1 diabetes mellitus, with special reference to insulin autoantibod-
ies. Diabetes Metab. Res. Rev. 21: 395–415.
6. Meier, J. J., J. C. Lin, A. E. Butler, R. Galasso, D. S. Martinez, and P. C. Butler.
2006. Direct evidence of attempted ? cell regeneration in an 89-year-old patient
with recent-onset type 1 diabetes. Diabetologia 49: 1838–1844.
7. Abdul-Rasoul, M., H. Habib, and M. Al-Khouly. 2006. “The honeymoon phase”
in children with type 1 diabetes mellitus: frequency, duration, and influential
factors. Pediatr. Diabetes 7: 101–107.
8. Bonfanti, R., E. Bazzigaluppi, G. Calori, M. C. Riva, M. Viscardi, E. Bognetti,
F. Meschi, E. Bosi, G. Chiumello, and E. Bonifacio. 1998. Parameters associated
with residual insulin secretion during the first year of disease in children and
adolescents with Type 1 diabetes mellitus. Diabet. Med 15: 844–850.
9. Bonfanti, R., E. Bognetti, F. Meschi, A. Brunelli, M. C. Riva, M. R. Pastore,
G. Calori, and G. Chiumello. 1998. Residual ?-cell function and spontaneous
clinical remission in type 1 diabetes mellitus: the role of puberty. Acta Diabetol.
10. Wallensteen, M., G. Dahlquist, B. Persson, M. Landin-Olsson, A. Lernmark,
G. Sundkvist, and B. Thalme. 1988. Factors influencing the magnitude, duration,
and rate of fall of B-cell function in type 1 (insulin-dependent) diabetic children
followed for two years from their clinical diagnosis. Diabetologia 31: 664–669.
11. Green, E. A., and R. A. Flavell. 2000. The temporal importance of TNF? ex-
pression in the development of diabetes. Immunity 12: 459–469.
12. Guerder, S., D. E. Picarella, P. S. Linsley, and R. A. Flavell. 1994. Costimulator
B7-1 confers antigen-presenting-cell function to parenchymal tissue and in con-
junction with tumor necrosis factor ? leads to autoimmunity in transgenic mice.
Proc. Natl. Acad. Sci. USA 91: 5138–5142.
13. Shimabukuro, M., K. Koyama, Y. Lee, and R. H. Unger. 1997. Leptin- or tro-
glitazone-induced lipopenia protects islets from interleukin 1? cytotoxicity.
J. Clin. Invest. 100: 1750–1754.
14. Hussain, M. J., M. Peakman, H. Gallati, S. S. Lo, M. Hawa, G. C. Viberti,
P. J. Watkins, R. D. Leslie, and D. Vergani. 1996. Elevated serum levels of
macrophage-derived cytokines precede and accompany the onset of IDDM. Dia-
betologia 39: 60–69.
15. Perez, F., A. Oyarzun, E. Carrasco, B. Angel, C. Albala, and J. L. Santos. 2004.
Plasma levels of interleukin-1?, interleukin-2 and interleukin-4 in recently diag-
nosed type 1 diabetic children and their association with ?-pancreatic autoanti-
bodies. Rev. Med. Chil. 132: 413–420.
16. Erbagci, A. B., M. Tarakcioglu, Y. Coskun, E. Sivasli, and E. Sibel Namiduru.
2001. Mediators of inflammation in children with type I diabetes mellitus: cyto-
kines in type I diabetic children. Clin. Biochem. 34: 645–650.
17. Nicoletti, F., I. Conget, R. Di Marco, A. M. Speciale, R. Morinigo, K. Bendtzen,
and R. Gomis. 2001. Serum levels of the interferon-?-inducing cytokine inter-
leukin-18 are increased in individuals at high risk of developing type I diabetes.
Diabetologia 44: 309–311.
18. Nicoletti, F., I. Conget, M. Di Mauro, R. Di Marco, M. C. Mazzarino,
K. Bendtzen, A. Messina, and R. Gomis. 2002. Serum concentrations of the
interferon-?-inducible chemokine IP-10/CXCL10 are augmented in both newly
diagnosed Type I diabetes mellitus patients and subjects at risk of developing the
disease. Diabetologia 45: 1107–1110.
19. Dogan, Y., S. Akarsu, B. Ustundag, E. Yilmaz, and M. K. Gurgoze. 2006. Serum
IL-1?, IL-2, and IL-6 in insulin-dependent diabetic children. Mediators Inflamm.
20. Pascual, V., F. Allantaz, E. Arce, M. Punaro, and J. Banchereau. 2005. Role of
interleukin-1 (IL-1) in the pathogenesis of systemic onset juvenile idiopathic
arthritis and clinical response to IL-1 blockade. J. Exp. Med. 201: 1479–1486.
21. Alberti, K. G., and P. Z. Zimmet. 1998. Definition, diagnosis and classification of
diabetes mellitus and its complications: Part 1. Diagnosis and classification of
diabetes mellitus provisional report of a WHO consultation. Diabet. Med. 15:
22. Woo, W., J. M. LaGasse, Z. Zhou, R. Patel, J. P. Palmer, H. Campus, and
W. A. Hagopian. 2000. A novel high-throughput method for accurate, rapid, and
economical measurement of multiple type 1 diabetes autoantibodies. J. Immunol.
Methods 244: 91–103.
23. Tusher, V. G., R. Tibshirani, and G. Chu. 2001. Significance analysis of microar-
rays applied to the ionizing radiation response. Proc. Natl. Acad. Sci. USA 98:
24. Khatri, P., P. Bhavsar, G. Bawa, and S. Draghici. 2004. Onto-tools: an ensemble
of web-accessible, ontology-based tools for the functional design and interpre-
tation of high-throughput gene expression experiments. Nucleic Acids Res. 32:
25. Hosack, D. A., G. Dennis, Jr., B. T. Sherman, H. C. Lane, and R. A. Lempicki.
2003. Identifying biological themes within lists of genes with EASE. Genome
Biol. 4: R70.
26. Sturn, A., J. Quackenbush, and Z. Trajanoski. 2002. Genesis: cluster analysis of
microarray data. Bioinformatics 18: 207–208.
27. Pfaffl, M. W. 2001. A new mathematical model for relative quantification in
real-time RT-PCR. Nucleic Acids Res. 29: e45.
28. Moser, B., M. Wolf, A. Walz, and P. Loetscher. 2004. Chemokines: multiple
levels of leukocyte migration control. Trends Immunol 25: 75–84.
29. Ribeiro, S., and R. Horuk. 2005. The clinical potential of chemokine receptor
antagonists. Pharmacol. Ther. 107: 44–58.
30. Saeki, T., and A. Naya. 2003. CCR1 chemokine receptor antagonist. Curr.
Pharm. Des. 9: 1201.
31. Andrei, C., P. Margiocco, A. Poggi, L. V. Lotti, M. R. Torrisi, and A. Rubartelli.
2004. Phospholipases C and A2 control lysosome-mediated IL-1 ? secretion:
implications for inflammatory processes. Proc. Natl. Acad. Sci. USA 101:
32. Kaizer, E. C., C. L. Glaser, D. Chaussabel, J. Banchereau, V. Pascual, and
P. C. White. 2007. Gene expression in peripheral blood mononuclear cells from
children with diabetes. J. Clin. Endocrinol. Metab. 92: 3705–3711.
33. Donath, M. Y., J. Storling, K. Maedler, and T. Mandrup-Poulsen. 2003. Inflam-
matory mediators and islet ?-cell failure: a link between type 1 and type 2 dia-
betes. J. Mol. Med. 81: 455–470.
34. Maedler, K., P. Sergeev, F. Ris, J. Oberholzer, H. I. Joller-Jemelka, G. A. Spinas,
N. Kaiser, P. A. Halban, and M. Y. Donath. 2002. Glucose-induced ? cell pro-
duction of IL-1? contributes to glucotoxicity in human pancreatic islets. J. Clin.
Invest. 110: 851–860.
35. Welsh, N., M. Cnop, I. Kharroubi, M. Bugliani, R. Lupi, P. Marchetti, and
D. L. Eizirik. 2005. Is there a role for locally produced interleukin-1 in the
deleterious effects of high glucose or the type 2 diabetes milieu to human pan-
creatic islets? Diabetes 54: 3238–3244.
36. Shanmugam, N., Y. S. Kim, L. Lanting, and R. Natarajan. 2003. Regulation of
cyclooxygenase-2 expression in monocytes by ligation of the receptor for ad-
vanced glycation end products. J. Biol. Chem. 278: 34834–34844.
37. Valencia, J. V., M. Mone, C. Koehne, J. Rediske, and T. E. Hughes. 2004.
Binding of receptor for advanced glycation end products (RAGE) ligands is not
sufficient to induce inflammatory signals: lack of activity of endotoxin-free al-
bumin-derived advanced glycation end products. Diabetologia 47: 844–852.
38. Verbsky, J. W., and A. J. White. 2004. Effective use of the recombinant inter-
leukin 1 receptor antagonist anakinra in therapy resistant systemic onset juvenile
rheumatoid arthritis. J. Rheumatol. 31: 2071–2075.
39. Dinarello, C. A. 2005. Interleukin-1?. Crit. Care Med. 33: S460–S462.
40. Larsen, C. M., M. Faulenbach, A. Vaag, A. Volund, J. A. Ehses, B. Seifert,
T. Mandrup-Poulsen, and M. Y. Donath. 2007. Interleukin-1-receptor antagonist
in type 2 diabetes mellitus. N. Engl. J. Med. 356: 1517–1526.
1937The Journal of Immunology
on May 18, 2011