of October 13, 2018.This information is current as

Capture-Independent FashionTranscription-Dependent,Periphery Display a Proinsulin Epitope in a Dendritic Cells in Human Thymus and

Camillo Ricordi, Douglas Hanahan and Alberto PuglieseDogra, Armando Mendez, Eliot Rosenkranz, Ulf Dahl,Alexander Cao, Gloria Allende, Markus Zeller, Rajpreet Carlos A. Garcia, Kamalaveni R. Prabakar, Juan Diez, Zhu

http://www.jimmunol.org/content/175/4/2111doi: 10.4049/jimmunol.175.4.2111

2005; 175:2111-2122; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/175/4/2111.full#ref-list-1

, 27 of which you can access for free at: cites 57 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2005 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Dendritic Cells in Human Thymus and Periphery Display aProinsulin Epitope in a Transcription-Dependent,Capture-Independent Fashion1

Carlos A. Garcia,2* Kamalaveni R. Prabakar,2* Juan Diez,2* Zhu Alexander Cao,�

Gloria Allende,* Markus Zeller,* Rajpreet Dogra,* Armando Mendez,*† Eliot Rosenkranz,§¶

Ulf Dahl,* Camillo Ricordi,*†§ Douglas Hanahan,� and Alberto Pugliese3*†‡

The natural expression of tissue-specific genes in the thymus, e.g., insulin, is critical for self-tolerance. The transcription oftissue-specific genes is ascribed to peripheral Ag-expressing (PAE) cells, which discordant studies identified as thymic epithelialcells (TEC) or CD11c� dendritic cells (DC). We hypothesized that, consistent with APC function, PAE-DC should constitutivelydisplay multiple self-epitopes on their surface. If recognized by Abs, such epitopes could help identify PAE cells to further definetheir distribution, nature, and function. We report that selected Abs reacted with self-epitopes, including a proinsulin epitope, onthe surface of CD11c� cells. We find that Proins�CD11c� PAE cells exist in human thymus, spleen, and also circulate in blood.Human thymic Proins� cells appear as mature DC but express CD8�, CD20, CD123, and CD14; peripheral Proins� cells appearas immature DC. However, DC derived in vitro from human peripheral blood monocytes include Proins� cells that uniquelydifferentiate and mature into thymic-like PAE-DC. Critically, we demonstrate that human Proins�CD11c� cells transcribe theinsulin gene in thymus, spleen, and blood. Likewise, we show that mouse thymic and peripheral CD11c� cells transcribe the insulingene and display the proinsulin epitope; moreover, by using knockout mice, we show that the display of this epitope depends uponinsulin gene transcription and is independent of Ag capturing. Thus, we propose that PAE cells include functionally distinct DCdisplaying self-epitopes through a novel, transcription-dependent mechanism. These cells might play a role in promoting self-tolerance, not only in the thymus but also in the periphery. The Journal of Immunology, 2005, 175: 2111–2122.

S elf-tolerance is established and maintained through central(thymic) and peripheral mechanisms. Thymic selection iscrucial for shaping a self-tolerant T cell repertoire in early

life during the maturation of the immune system (1). Peripheraltolerance mechanisms (deletion, anergy, ignorance, and regulatorycells) are active in peripheral lymphoid tissues (PLT)4 throughoutlife and are considered critical for self-molecules with tissue-re-stricted expression or peripheral proteins that are not expressed inthe thymus. However, many peripheral proteins are “ectopically”expressed in the thymus (2–7), and their thymic expression con-

tributes to the induction of self-tolerance (8–11). Proinsulin/insu-lin is a prototypical peripheral protein almost exclusively producedby pancreatic � cells (12). In humans, allelic variation and parentof origin effects correlate with levels of insulin gene (INS) tran-scription in the thymus and with the risk of type 1 diabetes (T1D),a disease in which autoimmunity to proinsulin/insulin plays animportant role (3, 4). Mouse models in which proinsulin/insulinexpression was genetically perturbed demonstrate that thymic pro-insulin/insulin expression dramatically affects the development ofself-tolerance (13–16). Mice have two nonallelic insulin genes,Ins1 and Ins2 (17), encoding slightly different proteins that areboth functional and secreted by � cells. However, only Ins2 isexpressed in the thymus of wild-type mice (14). The lack of pro-insulin II expression resulting from the disruption of Ins2 (18)causes accelerated diabetes development, heightened disease inci-dence, and more severe autoimmune responses to proinsulin/insu-lin in NOD mice (15, 16), a well-known model of autoimmunediabetes (19).

A subset of cells in the thymus expresses ostensible tissue-spe-cific proteins. However, data about their distribution, phenotype,and function are limited and discordant (20). Such cells were firstdescribed in the mouse and named peripheral Ag-expressing(PAE) cells (9). Initial studies in human thymus could not assignself-molecule expression to a single cell type but found evidencefor ectopic transcription of self-molecule genes in cell fractionsenriched for thymic epithelial cells (TEC) and bone marrow (BM)-derived APC (5). Later studies identified PAE cells as dendriticcells (DC)/macrophages (21, 22) or, alternatively, as medullaryTEC (mTEC) in both mouse (6, 21, 23) and human thymus (7, 22)by demonstrating peripheral protein gene transcripts in sorted cellpopulations. In some studies, these transcripts were not found in

*Immunogenetics Program and Cell Transplant Center, Diabetes Research Institute,†Department of Medicine, ‡Department of Immunology and Microbiology, §Depart-ment of Surgery, and ¶Division of Cardiothoracic Surgery, Miller School of Medi-cine, University of Miami, Miami, FL 33136; and �Department of Biochemistry andBiophysics and Diabetes Center, University of California, San Francisco Compre-hensive Cancer Center, San Francisco, CA 94143

Received for publication March 7, 2005. Accepted for publication May 30, 2005.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work is supported by National Institutes of health Grants AI-44456, HooperTraining Grant T32CA09043-25 (to Z.A.C.), American Diabetes Association Grant7-02-RA-72, and the Diabetes Research Institute Foundation.2 C.A.G., K.R.P., and J.D. contributed equally to this work.3 Address correspondence and reprint requests to Dr. Alberto Pugliese, Diabetes Re-search Institute, Miller School of Medicine, University of Miami, 1450 Northwest10th Avenue, Miami, FL 33136. E-mail address: apuglies@med.miami.edu4 Abbreviations used in this paper: PLT, peripheral lymphoid tissue; T1D, type 1diabetes; PAE, peripheral Ag expressing; TEC, thymic epithelial cell; BM, bonemarrow; DC, dendritic cell; mTEC, medullary TEC; GAD, glutamic acid decarbox-ylase; T-LDC, thymus light density cell; MBP, myelin basic protein; TPO, thyroidperoxidase; HEL, hen egg lysozyme; mHEL, membrane-bound HEL; CD, clusterdifferentiation; rh, recombinant human; SP-M, spleen monocyte; PB-M, peripheralblood monocyte.

The Journal of Immunology

Copyright © 2005 by The American Association of Immunologists, Inc. 0022-1767/05/$02.00

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

sorted DC populations. In our earlier studies, we used immuno-histochemistry to demonstrate cells stained for proinsulin, glu-tamic acid decarboxylase (GAD) and IA-2, islet cell moleculeswith tissue-restricted expression that can become target autoanti-gens in T1D (22), in the human thymus. These cells expressedmarkers of BM-derived APC by double immunofluorescence, asdefined by the expression of CD11c, CD14, CD40, CD80, CD86,CD83, CD8�, and HLA class II (22). Moreover, we found similarcells in human spleen and lymph nodes, tissues that also transcribegenes encoding for tissue-specific proteins, including INS (22, 24).These findings suggest but do not prove that human DC directlytranscribe peripheral protein genes.

We hypothesized that, consistent with APC function, PAE cellsshould constitutively display multiple self-epitopes on their sur-face. If recognized by Abs, such epitopes could help identify PAEcells and allow further studies to more precisely define their nature,distribution, and function. We report that selected Abs react withself-epitopes, including a proinsulin epitope, on the surface ofCD11c� cells. By using this proinsulin epitope as a selecting el-ement, we demonstrate that PAE cells include DC characterized bya distinct phenotype. These cells display multiple self-epitopes ontheir surface and, critically, such display is achieved through au-tonomous gene transcription.

Materials and MethodsHuman subjects and tissue specimens

Blood samples were obtained from healthy subjects (age range, 23–53years; six males, two females). Spleen specimens were obtained from ca-daver organ transplant donors (22–57 years old; four females, two males).Thymus was discarded material from children undergoing heart surgery (1wk to 3 mo old; three females, three males). Specimens were obtainedthrough informed consent and/or institutional approval.

Preparation of human thymus light density cells (T-LDC)

Thymus cells were processed to T-LDC, enriched in DC, according to theprotocol of Vandenabeele et al. (25). Thymus tissue was minced intopieces, digested for 20 min at room temperature, with 1 mg/ml grade IIbovine pancreatic DNase and 1 mg/ml type II collagenase in HBSS. EDTA(10 mM) was added to the cell suspension, which was filtered through anylon mesh and washed with HBSS, 10 mM EDTA, and 2% FBS. Cellswere centrifuged at 300 � g at 4°C, and the pellet was resuspended in 10ml of cold Nycodenz medium (d-1.069 g/ml, Accudenz, A.G.; AccurateChemical and Scientific). Cell suspensions (5 ml) were layered over 5 mlof d-1.069 g/ml cold Nycondenz medium. Two milliliters of RPMI 1640,10 mM EDTA, and 2% FBS were layered over the cell suspension. Sam-ples were centrifuged at 1700 � g at 4°C, and the LDC fraction recoveredfrom the upper zones.

Preparation of human splenocytes

Spleen specimens were cut into small pieces that were forced through themesh of a sterile sieve using the plunger of a syringe. Cell aggregates weredissociated by adding 10 mM EDTA and shaking. Mononuclear cells wereisolated by centrifugation over Ficoll-Paque Plus, d-1.077 g/ml (AmershamBiosciences).

Abs to self-molecules

We used two mAbs against human proinsulin, clone M32337 and 3A1,both specific for the B-chain/C-peptide junction; mAb M32337 has nocross-reactivity with human insulin, mAb 3A1 has no cross-reactivity withhuman, bovine, porcine insulin, or human C-peptide. When used in flowcytometry, both Abs were initially tested using the indirect method in com-bination with a secondary Ab against mouse IgG, Fc-specific F(ab�)2, con-jugated with FITC (Sigma-Aldrich). Once staining was demonstrated, weconjugated mAb M32337 with PE to facilitate staining for multiple mark-ers. We also used mAbs reacting with both human insulin and proinsulin:clone K36aC10 (26) and clone D6C4. For GAD65, we used a mAb(GAD65-AS) and a rabbit serum (lot no. 18080605) reacting with GAD65and GAD67. Abs to human IA-2 included mAbs 103/2C4, 97/4D9, 98/4H6, and rabbit sera against the intracellular (lot no. 8959) and extracel-lular (lot no. 9218H) domains, kindly provided by S. Gleason at Bayer(Pittsburgh, PA). Abs to human myelin basic protein (MBP) were mAbclone P12 and rabbit serum ab2404. Abs to human thyroid peroxidase(TPO) were mAb clone 6H7 and rabbit serum lot no. L3012864. Table Ireports Ab species, available data on epitope specificity, cross-reactivity,and sources. The rat mAb D8H21 (27) was used in the experiments in-volving transgenic mice. This Ab reacts with the MHC/Ag complex con-sisting of the I-Ak molecule and the 116-129 hen egg lysozyme (HEL)peptide. It is reported to stain mouse spleen cells of CBA and B10.A(4R)mice (both expressing I-Ak) when pulsed with this HEL peptide or wholeHEL (27).

Blocking experiments

We tested the specificity of the proinsulin M32337 mAb and theGAD65-AS mAb with synthetic peptides derived from the sequence ofhuman proinsulin (B24-C33, B24-C32, and B24-C36; a gift from L. C.Harrison, Walter and Eliza Hall Institute, Melbourne, Australia) and theGAD peptide used to generate the mAb (aa residues 4–25). Peptides wereadded in various concentrations to the dilution of Ab normally used forstaining and gently mixed for 30 min at room temperature before the so-lution was used for staining. In some experiments, blocking was done bydirect competition.

Abs to human cluster differentiation (CD) surface markers

FITC-conjugates included mAbs to human CD3, CD8�, CD14, CD15,CD16, CD20, CD40, CD80, CD83, CD86, HLA-DR, CD95 (BD Pharm-ingen), and CD178 (FasL; Calbiochem). CyChrome 5.5 conjugates in-cluded Abs against CD11c, HLA-DR, and CD123 (BD Pharmingen).

Table I. Abs to self-molecules tested for surface staining experiments

Clone/Lot Species Specificity Stained PAE Cells Source

M32337 Mouse Human proinsulin, C-peptide/B-chain junction Yes Fitzgerald Industries3A1 Mouse Human proinsulin, C-peptide/B-chain junction Yes Research DiagnosticsK36aC10 Mouse Human insulin/proinsulin No Sigma-AldrichD6C4 Mouse Human insulin/proinsulin (A19, B21) No Abcam LimitedGAD65-AS Mouse Rat GAD65 aa 4–25, N terminus Yes Alpha DiagnosticsAB1511 Rabbit Rat GAD65/67 aa 572–685, C terminus Yes Chemicon International103/2C4 Mouse Human IA-2, aa 771–979, intracellular No Labor fur Autoimmundiagnostik97/4D9 Mouse Human IA-2, aa 605–772, intracellular No Labor fur Autoimmundiagnostik98/4H6 Mouse Human IA-2, aa 618–628, transmembrane No Labor fur Autoimmundiagnostik8959 Rabbit Human IA-2, aa 601–979, intracellular Yes Gift from S. Gleason, Bayer9218H Rabbit Human IA-2, aa 389–576, extracellular Yes Gift from S. Gleason, BayerP12 Mouse Human MBP, aa 84–89 Yes Research Diagnostics2404 Rabbit Human MBP No Abcam Limited6H7 Mouse Human TPO Yes Research DiagnosticsL3012864 Rabbit Human TPO No United States Biologicals

2112 CAPTURE-INDEPENDENT SELF-Ag EXPRESSION BY DC

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Immunophenotyping and sorting of human cells

T-LDC and spleen mononuclear cells were adjusted to a 107 cells/ml con-centration in PBS, 2% FBS, 10 mM EDTA, and 0.02% sodium azide(washing solution) before staining. We stained unfixed, unpermeabilized,T-LDC, spleen cells, and whole blood samples (red cells were removedwith 1� FACS RBC lysis solution). Incubation with normal mouse serumwas used to block nonspecific binding. Cells were stained with 10–20 �lof direct conjugates and gently mixed at 4°C for 30 min. Samples werewashed and centrifuged at 300 � g for 5 min at 4°C. The pellet wasresuspended in 100 �l of washing solution. If indirect staining was used,this was performed first and revealed with FITC-conjugated polyclonalgoat anti-mouse IgG, F(ab�)2. After a final wash, cells were fixed in PBScontaining 1% paraformaldehyde. Monocyte-gated (whole blood andspleen cells), and ungated T-LDC were analyzed using a FACSCalibur(BD Biosciences). Triple staining was analyzed by gating cells positive forone marker in histogram analysis followed by double staining dot plotanalysis. Negative controls included unstained cells, omitting the primaryAb, and the appropriate isotype control Abs. CD4� and CD8� lympho-cytes were used as a lineage control populations. For cell sorting, cellswere incubated with lineage-positive Abs (CD3, CD7, CD15, CD19, andanti-glycophorin A) to eliminate T cells, B cells, granulocytes, and RBC.The remaining cells were stained with the proinsulin mAb M32337 andmAbs to several differentiation markers and then purified by FACS.

In vitro differentiation and maturation of human monocyte-derived DC

Peripheral blood samples or buffy coats were obtained from nine healthydonors (age range, 23–50 years; three males, four females, gender wasunknown for two donors). Mononuclear cells were isolated through Ficoll-Paque gradient centrifugation, suspended in medium (RPMI 1640, 1% glu-tamine, 1% penicillin/streptomycin, 25 mM HEPES, and 10% heat-inac-tivated FBS), plated in 6-well plates (3 � 106 cells/ml), and left adheringfor 3 h at 37°C. Nonadherent cells were removed and adherent cells cul-tured in medium alone, and medium was supplemented with recombinanthuman (rh)GM-CSF (50 ng/ml) and rhIL-4 (25 ng/ml) for 7 days. Freshcytokines were added on days 4 and 7 by removing half of the medium andadding back an equal volume of medium with cytokines. To induce mat-uration, the medium was supplemented with rhTNF-� (50 ng/ml) andPGE2 (2 �g/ml) on day 7. Cultured cells were harvested on day 8 andwashed with PBS. Viability was assessed by trypan blue exclusion. On day8, cells were harvested, stained with Abs, and analyzed by flow cytometry.

C-peptide and proinsulin levels measurements

Supernatants from the cell cultures were tested for the presence of C-peptide using a radioimmunoassay with a lower limit of detection of 0.10ng/ml (Diagnostic Products). Proinsulin levels were measured by ELISA(Mercodia); the limit of detection is �0.5 pmol/L and cross-reactivity withinsulin and C-peptide is �0.03 and �0.06%, respectively.

RNA isolation and RT-PCR from human cells

We used the Microto-Midi Total RNA Purification kit (Invitrogen LifeTechnologies) or the MicroPoly(A) Purist mRNA Purification kit (Am-bion) to purify total or mRNA from human tissues (whole thymus, spleenand pancreatic islets, positive controls), peripheral blood lymphocytes, andcells sorted from thymus, spleen, and blood. RNA integrity was assessedby inspection of 28S and 18S band intensities after agarose gel electro-phoresis. DNase-treated total RNA (1–4 �g) or mRNA (0.5–2.5 �g) wasreverse transcribed into cDNA with Oligo(dT)20 Primer and Superscript IIReverse Transcriptase using the SuperScript First-Strand Synthesis System(Invitrogen Life Technologies). Ten percent volume of the cDNA sampleswas amplified by PCR (a single denaturing step at 94°C for 3 min followedby 35 cycles of 94°C for 30 s, 60°C for 30 s, 72°C for 1 min, and a finalextension at 72°C for 10 min). Primers for the INS mRNA were reportedpreviously (3). Primers for a control transcript, GAPDH, were as follows:forward, 5�-CCATGGAGAAGGCTGGGGGCTC-3�, and reverse, 5�-CCTTCTTGATGTCATCATATTTGGCA-3�. Negative controls includedreactions lacking template or reverse transcriptase. Products were electro-phoresed in 2% agarose gels.

Mice

We studied tissues and cells from several mouse strains. These includedC57BL6, C3H, BALB/c, FVBN, CBA, and NOD mice. We also studiedmice in which Ins1 and Ins2 were disrupted (18). In Ins2 knockout mice,the LacZ gene replaces Ins2, and thus, �-galactosidase is expressed whereIns2 would be whereas Ins2 is not expressed. Breeding pairs (Ins1�/

�Ins2�/� and Ins1�/�Ins2�/�) were mated to obtain double knockout micelacking the expression of both insulin genes (Ins1�/�Ins2�/�). Double ho-mozygous mutant pups are smaller than heterozygous littermates and diewithin 48 h in diabetic ketoacidosis (18). The genotype of the mice wasconfirmed using a PCR-based typing method developed by M. Nakayama(Barbara Davis Center for Childhood Diabetes, University of ColoradoHealth Sciences Center, Aurora, CO). We studied mice lacking expressionof the pancreas duodenum homeobox gene PDX1 (or insulin promoterfactor 1), a transcription factor critical for pancreatic development, whichsuffer from pancreatic agenesis (28). We also studied transgenic mice ex-pressing membrane-bound HEL (mHEL) under the control of a tissue-specific promoter, the crystalline promoter (mHEL mice). These mice wereoriginated in the FVBN background and bred with B10.BR mice so thatthey express the I-Ak molecule (29, 30). Mice were maintained accordingto institutional guidelines.

Preparation of cells from mouse BM, thymus, and spleen

BM cells from Ins2 knockout mice were obtained from femurs and tibiaeby flushing cells out of the diaphysis with RPMI 1640 using a 23-gaugeneedle. Cells were filtered through a nylon mesh and counted. Viabilitywas assessed with trypan blue exclusion. BM cells were cultured in AIM-Vcomplete medium (Invitrogen Life Technologies) in the presence of GM-CSF (1–5 ng/ml) for 5 days before RNA extraction to assess LacZ expres-sion by RT-PCR. Thymus and spleen from various mouse strains weredissected and submerged in RPMI 1640 and 2% FBS. Depending on theexperiments, tissues were used to extract RNA or to obtain cells. For cellpreparation, tissues were pressed gently with the plunger of a syringe.Thymus tissues were digested with type II collagenase (1 mg/ml) and gradeII bovine pancreatic DNase I (1 mg/ml) in HBSS without calcium chloride,magnesium chloride, and magnesium sulfate for 15–20 min at room tem-perature. EDTA was added to a final concentration of 0.01 M to disruptDC-T cell rosettes. Both thymus and spleen cell suspensions were layeredover Ficoll-Paque (1077g/cm3), centrifuged, and then the low-density frac-tion was recovered.

Immunophenotyping and sorting of mouse cells

Cells were washed with FACS buffer (PBS, 2% FBS, 0.02% sodium azide,and 0.01 M EDTA), and 0.5–1 million cells were used for surface stainingand analyzed by flow cytometry. Before staining, cells were incubated withanti-CD16/CD32 to block nonspecific binding. Cells were then stainedwith Abs to mouse CD11c (HL3), MHC class II (10-3.6; BD Biosciences)and CD4 (L3T4; eBioscience). In some experiments, CD11c� cells weresorted by FACS, and RNA was extracted for RT-PCR expression studies.Similar to human cells, mouse thymus and spleen cells were also stainedwith the Abs to proinsulin (clone m32337) and polyclonal serum againstGAD65/67 (Table I) to assess the surface display of self-epitopes.

RT-PCR in mouse tissues and cells

RNA was extracted using the Microto-Midi Total RNA Purification kit(Invitrogen Life Technologies). RNA samples from mouse BM cells, thy-mus, and spleen cells and sorted populations (CD11c�, CD11c�, CD4�)were used in RT-PCR experiments to assess the expression of the Ins2,glucagon, and LacZ genes, which were performed using nested primers.Forward and reverse primers were as follows: Ins2 first pair, 5�-CACCCAGGCTTTTGTCAAGCAG-3�, and 5�-TGGTGGGTCTAGTTGCAGTAGT-3�; Ins2 second pair, 5�-GGGAGCGTGGCTTCTTCTACAC-3�,and 5�-GCTGGTGCAGCACTGATCTACA-3�; glucagon first pair, 5�-TGTGGCTGGATTGCTTATAATG-3�, and 5�-GCCCTCCAAGTAAGAACTCACA-3�; glucagon second pair, 5�-GCACGCCCTTCAAGACACAGAG-3�, and 5�-CGGTTCCTCTTGGTGTTCATCA-3�; LacZ firstpair, 5�-GTGCGGTGGTTGAACTGCAC-3�, and 5�-GCGCGTCGTGATTAGCGCCG-3�; and LacZ second pair, 5�-ACGGCACGCTGATTGAAGCA-3�, and 5�-CCAGCGACCAGATGATCACA-3�. The �2-micro-globulin transcript was amplified as control.

Statistical analysis

Means were compared using the Student’s t test for unpaired samples; thepaired samples test was used for the analysis shown in Fig. 4B. Onlytwo-tailed p values are reported.

2113The Journal of Immunology

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

ResultsAn epitope encompassing the B-chain/C-peptide junction ofproinsulin is detected on the surface of human CD11c� cells inthymus, spleen, and blood

Our initial phenotypic analysis of PAE cells in human thymus andPLT demonstrated that these cells express markers of BM-derivedAPC (including CD11c, CD14, and HLA class II) (22). We hy-pothesized that PAE-CD11c� cells could display self-epitopes ontheir surface, in keeping with their APC phenotype, and that self-epitopes might be accessible to Ab recognition. This approachcould also provide a selecting tool to study PAE cells. Using pro-insulin as a representative peripheral protein expressed by PAEcells, we tested two mAbs specific for human proinsulin (Table I),both recognizing the B-chain/C-peptide junction but not reactingwith mature insulin. Both proinsulin mAbs demonstrated similarlevels of surface reactivity in preliminary experiments. Two othermAbs (Table I) reacting with insulin and proinsulin did not stainthe cells’ surface. This may reflect the absence of intact protein onthe cell surface, as at least one of these Abs targets a conforma-tional epitope (26, 31). We then selected the anti-proinsulin mAbM32337 for further studies. As shown in Fig. 1, we found cellsstained by this Ab or Proins� cells in live, unpermeabilized cellpopulations from the thymus (T-LDC enriched in DC), and spleenmonocytes (SP-M); we also tested whole blood samples and dem-onstrated Proins� cells in the circulation when gating on peripheralblood monocytes (PB-M). Variable proportions of T-LDC(mean � 3.18 � SD 1.3%, n � 6), SP-M (mean � 66.2 � 19.4%,n � 6), and PB-M (mean � 62.9 � SD 11.4%, n � 8) were stainedon their surface by the proinsulin mAb M32337, with spleen andblood harboring higher proportions of Proins� cells than thymus( p � 0.0004, T-LDC vs SP-M; p � 0.000001, T-LDC vs PB-M).Importantly, the vast majority of the Proins� cells coexpressedCD11c (Fig. 1), which was the single marker most consistentlycoexpressed by Proins� cells in human T-LDC (69%), SP-M(93%), and PB-M (94%) populations. Proinsulin staining was notdetected in CD4� or CD8� lymphocytes (data not shown).

To further test the specificity of mAb M32337 and gain insightinto the nature of the Ags recognized on the surface of CD11c�

cells, we performed blocking experiments with overlapping pep-tides spanning the proinsulin B-chain/C-peptide junction (B24-C32, B24-C33, and B24-C36). All peptides prevented staining,e.g., preincubation with the B24-C32 peptide prevented staining(Fig. 2A), and staining inhibition with the proinsulin peptide wasdose dependent (Fig. 2B). This demonstrates that the mAbM32337 reacts with peptide Ags. Staining for CD11c andHLA-DR was unaffected by the proinsulin peptide (data notshown). Staining with the proinsulin mAb was also unaffected bypreincubation with a GAD65 peptide (aa residues 4–25) (Fig. 2C).Overall, the reactivity pattern observed with the Abs to insulin/proinsulin and proinsulin is consistent with the surface display ofa peptide epitope spanning the proinsulin B-chain/C-peptide junc-tion with no evidence for the presence of conformation-dependentepitopes that would require the presence of intact proinsulin orlarger fragments.

Human Proins�CD11c� cells correspond to PAE cells of theDC class

Fig. 3 shows an example of Proins� cells in T-LDC, SP-M, andPB-M populations coexpressing various differentiation markers to-gether with CD11c. Fig. 4A compares the expression of thesemarkers in these populations. Proins�CD11c� cells often coex-press CD14 and HLA-DR in all three populations. Other APCmarkers were differentially expressed: Proins�CD11c� T-LDCmore frequently express CD1a, CD20, CD40, CD80, CD83, andCD178 compared with Proins�CD11c� cells in SP-M and PB-Mpopulations. CD123� cells appear less frequent in SP-M as com-pared with T-LDC and PB-M populations. Approximately 40% ofthe Proins�CD11c� cells in T-LDC populations also expressCD8�, while this marker is essentially not expressed byProins�CD11c� cells in the periphery. Moreover,CD11c�CD8�� T-LDC are enriched in Proins� cells comparedwith CD11c�CD8�� cells, on average 4-fold (mean � 49.3 � SD

FIGURE 1. Surface staining with a proinsulin Ab inCD11c� cells from thymus, spleen, and blood. Flowcytometry analysis demonstrating surface staining withthe proinsulin mAb M32337 in T-LDC, spleen, andblood samples. The scatter plots show the populationsgated for analysis, which for spleen and blood sampleswere monocyte populations. In the histograms, theblack line represents the isotype control while the grayarea represents specific staining. The dot plots show thatmost Proins� cells coexpress CD11c. In the experimentshown, 85.7, 85.5, and 100% of Proins� cells in T-LDC, spleen, and PB-M populations, respectively, co-expressed CD11c.

2114 CAPTURE-INDEPENDENT SELF-Ag EXPRESSION BY DC

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

14.1 vs 14.1 � 12.4, n � 6, p � 0.004; Fig. 4B). The results areconsistent with our earlier characterization of proinsulin-positivePAE cells in human thymus and spleen tissue sections, where wealso detected the coexpression of DC markers including CD8�(22). Therefore, the surface expression of a proinsulin epitopeidentifies PAE-DC. The smaller proportions of peripheral (�5%),

and thymic Proins�CD11c� cells (�30%) appear to include cellsat various stages of differentiation and maturation, as indicated bythe heterogeneous expression of phenotypic markers in the differ-ent compartments. Approximately 50% of the Proins�CD11c�

cells in the thymus expressed and HLA-DR, with populations be-ing CD123high and HLA-DRhigh (Fig. 3); 23% of the thymicProins�CD11c� were CD14�. Based on the purification of lightdensity cells and gate selection, it is plausible that only a few of theProins�CD11c� cells may be of epithelial origin.

In vitro-derived DC include Proins� cells, which mature into athymic-like PAE-DC phenotype

To further explore self-epitope expression by DC, also in relationto the maturation state, we derived DC from human PB-M byinducing their differentiation with GM-CSF/IL-4 and their matu-ration with TNF-�/PGE2. Remarkably, �30% of the cultured cellswere Proins�. Although the proportion of Proins� cells remainedstable in response to the cytokines in the culture medium, we ob-served phenotypic changes typical of DC differentiation and mat-uration in the cultured cells. This was also confirmed by analyzingProins� and Proins� DC separately. Both populations respondedto differentiation and maturation stimuli by expressing typical DCmarkers (HLA-DR, CD83, CD1a, CD40, CD80, and CD86), asshown in Fig. 5 for CD83 and CD80. However, unlike Proins�

cells, Proins� DC maintained CD14 expression and up-regulatedCD8�, CD20, and CD123, both in terms of percentage of positivecells (Fig. 5) and mean fluorescence intensity (data not shown).Thus, Proins� cells have unique functional characteristics becausethey mature in vitro into cells having similar phenotype to thymicProins� PAE cells in their native microenvironment. C-peptideand proinsulin were not detectable in the supernatants of thesecultures, indicating that the protein was not secreted (data notshown).

Human Proins�CD11c� cells display multiple self-epitopes

We then investigated whether other self-epitopes are detectable onthe surface of CD11c� cells and Proins� cells. Using the Abslisted in Table I, we detected surface staining for selected GADand IA-2 epitopes in T-LDC, SP-M, and PB-M CD11c� popula-tions. This was noted with both a mAb and a polyclonal to GADand two polyclonal sera to IA-2, one reacting with the intracellulardomain. Three other mAbs to IA-2 did not stain (Table I). Signif-icant proportions of cells costained for GAD, IA-2, and proinsulin(Fig. 6A). In further blocking experiments, staining with theGAD65 mAb was virtually abolished by preincubating the Ab withthe peptide used for its generation (aa 4–25) (Fig. 6B). Staininginhibition was related to the amount of peptide used; 60 �g ofpeptide resulted in almost complete blocking (from 53 to 5%)when analyzing the monocytes-gated cells from peripheral blood.The GAD peptide did not affect proinsulin staining (Fig. 2C), andconversely, the proinsulin peptide did not affect GAD or IA-2staining (data not shown). We also tested mAbs and rabbit sera toMBP, a major constituent of the myelin sheath and an autoantigenin multiple sclerosis (32), and TPO, an enzyme expressed at theapical surface of thyrocytes and an autoantigen in autoimmunethyroid disease (33). There are no known cell surface receptors forthese molecules. Both MBP and TPO have been found to be ex-pressed in the human and mouse thymus (5–7). The MBP gene isalso expressed in mouse spleen and BM cells, and the expressionof a hemopoietic MBP mRNA/protein, encoded by exon 1 of theMBP gene or “Golli-MBP,” is reported in mouse T and B lym-phocytes, DC, and macrophages (34). The majority of Proins�

cells also stained for MBP and TPO (Fig. 6C). The MBP mAb isspecific for aa residues 84–89, encoded by exon 1 of the MBP

FIGURE 2. Blocking experiments. A, Preincubation of the proinsulinmAb M32337 with a synthetic peptide corresponding to the B24-C32 res-idues of proinsulin prevents staining on the surface of peripheral bloodcells. B, Surface staining was inhibited in a dose-dependent fashion bypreincubating the proinsulin mAb M32337 with a synthetic peptide corre-sponding to the B24-C32 residues of proinsulin. 0, 5, 10, 15, and 20 �gwere used; 20 �g of peptide completely blocked staining. The proinsulinAb has an affinity constant of 7 � 10�8 L/M, and 20 �g of proinsulinpeptide B24-C32 correspond to 1.47 � 10�8 M, which is in the same 10�8

M range of the Ab. Data from the whole scatter—not just the monocytegate—were included in this experiment to ensure that blocking was abso-lute, hence the lower proportion of Proins� cells (�20%) compared withthe average peripheral blood frequency usually observed when gating onmonocytes (�60%). C, Preincubation of the proinsulin mAb M32337 witha synthetic peptide corresponding to the aa 4–25 (N terminus) of GAD65did not affect proinsulin staining on the surface of peripheral blood cells.

2115The Journal of Immunology

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

gene, suggesting that an epitope of Golli-MBP is expressed on thesurface of CD11c� PAE cells. Overall, these data show that reac-tivity for multiple self-epitopes is detected with selected Abs onthe surface of Proins�CD11c� cells.

Human Proins�CD11c� cells transcribe INS

Transcription of self-molecule genes has been reported previouslyin human thymus, spleen, and lymph nodes (in which PAE-DCwere shown by immunohistochemistry/double immunofluores-cence) (22, 24), in gradient-purified mouse (9) and human (5) thy-mic cells, and in cells sorted for the expression of DC (mouse) (21)or mTEC markers (mouse and human) (6, 7, 23). With the excep-tion of one study in the mouse (21), these transcripts were notdetected in thymic DC (6, 7, 23). Such discrepancies might beexplained by the inability to enrich sufficiently for DC expressingself-Ags. Thus, we used proinsulin as a prototype Ag to test thehypothesis that the observed expression of self-epitopes is associ-ated with gene transcription in PAE cells of the DC class. Wesorted CD11c� cells from human thymus, spleen, and peripheralblood based on the surface expression of the proinsulin epitope.We extracted RNA from sorted Proins� and Proins� cells andassessed the presence/absence of full-length INS transcripts usinga well-validated RT-PCR assay (3). INS mRNA was detected inProins� but not in Proins� cells (Fig. 7). The identity of the RT-PCR product was confirmed by sequencing or restriction enzymedigestion (data not shown). Thus, the surface display of the pro-insulin epitope cosegregates with INS transcription.

Mouse BM-derived CD11c� cells transcribe Ins2, exist in mousePLT, and display multiple self-epitopes

Earlier studies did not find peripheral protein gene transcripts inmouse peripheral DC (6, 21). In light of the above data demon-strating transcription of an organ-specific gene such as insulin inhuman DC (Fig. 7) and human PLT (22), we further investigatedinsulin gene expression in mouse DC and PLT. As expected, wedetected Ins2, as well as glucagon transcripts, in mouse thymus butalso in spleen and lymph nodes (Fig. 8A). Ins2 transcripts werefound in CD11c� cells sorted from mouse thymus and spleen (Fig.8B) but not in spleen CD11c� cells and CD4� T cells. These data

are consistent with the expression of peripheral protein genes inhuman PLT (22, 24) and demonstrate it in mouse PLT and periph-eral CD11c� cells. To substantiate this observation, we examinedmice in which the LacZ gene replaces one copy of the codingregion in the Ins2 gene, serving as a marker of transcription (18).�-Galactosidase (LacZ) transcripts were detected in thymus andinguinal lymph nodes (Fig. 8C) and in both freshly isolated andcultured BM cells (Fig. 8D). This indicates that Ins2-expressingCD11c� PAE cells are BM derived. We then investigated whetherIns2 transcription in mouse CD11c� PAE cells correlates with cellsurface reactivity for the proinsulin epitope, similar to human cells.We stained thymus and spleen cells from several mouse strains(C57BL6, BALB/c, and NOD) and detected surface staining usingboth the proinsulin and GAD Abs (Fig. 9). Proins� cells from �5wk-old mice coexpressed CD11c, as well as other DC markers,and similar to human cells, they expressed Ins2 mRNA (data notshown).

Proinsulin surface staining depends on insulin genetranscription and is independent of Ag capturing

Having shown that a proinsulin epitope is detected on the surfaceof both human and mouse CD11c� cells, which transcribe theinsulin gene, we tested whether the display of this epitope dependson insulin gene transcription. Therefore, we asked whetherProins� cells could be detected in Ins1�/�Ins2�/� mice. We bredIns1�/�Ins2�/� with Ins1�/�Ins2�/� mice to obtain Ins1�/

�Ins2�/� mice. We evaluated newborn mice because these micedie shortly after birth with severe diabetic ketoacidosis (18). Al-though newborn mice have few CD11c� cells and mature DC donot fully develop until 5 wk of age (35), thymus and spleen cellsfrom newborn wild-type and Ins1�/�Ins2�/� mice stained withthe polyclonal Ab to GAD. Thus, PAE cells were present at birth,although possibly not fully matured. Proins� cells were detected inthe thymus and spleen of wild-type mice but were virtually absentin Ins1�/�Ins2�/� mice (Fig. 9). Thus, surface staining with theproinsulin Ab is dependent on the presence of intact insulin genesand inferentially on insulin gene transcription. This experimentdefinitively excludes any cross-reactivity of the proinsulin mAb

FIGURE 3. Phenotypic analysis of Proins�CD11c�

cells. Representative examples of Proins� cells in T-LDC, SP-M, and PB-M populations characterized forthe expression of the surface markers shown in conjunc-tion with CD11c, highlighting the phenotypic differ-ences among thymic and peripheral populations de-scribed in the main text.

2116 CAPTURE-INDEPENDENT SELF-Ag EXPRESSION BY DC

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

M32337 with unknown molecules and provides a critical valida-tion for our approach to identify PAE cells through the expressionof this proinsulin epitope in both human and mouse cells.

DC are known to help maintain peripheral tolerance by captur-ing, processing, and presenting self-molecules to T cells (36, 37).Thus, we investigated whether the surface expression of the pro-

insulin epitope by PAE cells of the DC class is alternatively ex-plained by the capture of proinsulin expressed in pancreatic �cells. We studied PDX1 knockout mice (28) because PDX1 is atranscription factor required for the normal development of thepancreas. PDX1�/� mice suffer from complete pancreatic agene-sis, lack ectopic sites of insulin or amylase production, and dieshortly after birth (28). Thus, PDX1�/� mice have no proinsulin/insulin in the circulation and lack pancreatic � cells that couldprovide pancreatic proinsulin for capturing by DC. We found thatthymus and spleen of newborn PDX1�/� mice contained Proins�

(and GAD�) cells similarly to wild-type mice (Fig. 9). Consistentwith the expectation that thymic and peripheral PAE cells auton-omously produce proinsulin through Ins2 transcription, we de-tected Ins2 expression in the thymus and spleen of PDX1�/� mice(data not shown). Thus, Ins2 transcription in these tissues and byinference in thymic and splenic CD11c� PAE cells is not affectedby the lack of PDX1. Collectively, the results in the insulin andPDX1 knockout mice show that the expression of the proinsulinepitope on the surface of PAE cells is not dependent on the pres-ence of the target organ as a source of captured Ag, clearly im-plicating gene transcription in PAE-DC with the surface display ofcognate self-epitopes.

Proins� cells from HEL transgenic mice display an HELepitope in the context of MHC class II

To investigate whether self-epitopes expressed by PAE cells areassociated with MHC molecules, we studied mice that express theHEL under the control of a tissue-specific promoter, the crystallinepromoter. The pattern of expression of HEL in mHEL transgenicmice mimics the expression of Ins2, with expression in a specifictissue (the eye), as well as in the thymus. As with Ins2, thymicexpression of the HEL gene has been shown to promote toleranceto the HEL protein (29, 30), which is a self-molecule in the trans-genic mice. Based on the finding that Proins� cells can expressmultiple self-epitopes, we tested the hypothesis that at least someProins� PAE cells from mHEL mice should also express HELepitopes on their surface. Using HEL as surrogate example of nat-ural self-epitopes expressed by Proins� PAE cells, we tested thehypothesis that HEL epitopes could be associated with MHC mol-ecules. As shown in Fig. 10, we demonstrate that Proins� cellsfrom mHEL transgenic mice, but not from nontransgenic controlmice, display both the proinsulin epitope and a HEL epitope ontheir surface. As expected, these cells expressed the transcripts forIns2 and the HEL transgene (data not shown). Because the anti-HEL mAb used is known to recognize a MHC class II/HEL Agcomplex when cells are pulsed with HEL (27), this experimentprovides initial evidence that PAE cells can display self-epitopesin association with MHC molecules.

DiscussionWe tested the hypothesis that PAE cells of the DC class shouldfunction as APC and constitutively display multiple self-epitopeson their surface. We also hypothesized that the display of suchepitopes could be demonstrated with Abs. Although Abs are notroutinely used to detect self-epitopes on the cell’s surface, the TCRand Ig (the B cell receptor) share many structural features allowingAbs to mimic TCR. Moreover, some Abs have been shown torecognize HLA/peptide complexes (27, 38, 39). Our data show forthe first time that a subset of CD11c� cells constitutively displayepitopes derived from self-molecules, in particular an epitope ofproinsulin. We extrapolate the loss of proinsulin reactivity on thesurface of thymus and spleen cells from insulin knockout mice tothe human system as indicating that reactivity with the proinsulin

FIGURE 4. Phenotypic analysis of Proins�CD11c� cells. A, Percentageof Proins�CD11c� cells expressing various surface markers in T-LDC,SP-M, and PB-M populations. Asterisks denote values that are statisticallydifferent. Proins�CD11c� T-LDC more frequently express CD1a (p �0.0002 vs SP-M and PB-M), CD8� (p � 0.00001 vs SP-M and PB-M),CD20 (p � 0.00008 vs SP-M and p � 0.00001 vs PB-M), CD40 (p �0.001 vs SP-M and p � 0.0001 vs PB-M), CD80 (p � 0.0002 vs SP-M andp � 0.00001 vs PB-M), CD83 (p � 0.00001 vs SP-M and PB-M), andCD178 (FasL, p � 0.04 vs SP-M and p � 0.002 vs PB-M) compared withProins�CD11c� cells in SP-M and PB-M populations. CD123� cells ap-pear less frequent in SP-M as compared with T-LDC and PB-M popula-tions (p � 0.002 vs PB-M, p � 0.0005 vs T-LDC). B, Flow cytometryanalysis of thymic CD11c� cells in relation to the expression of CD8�.Proins� cells were enriched in CD11c�CD8�� T-LDC compared withCD11c�CD8�� T-LDC.

2117The Journal of Immunology

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

mAb M32337 is not spurious but rather depends on gene tran-scription. The blocking experiments demonstrate that the proinsu-lin Ab reacts with a peptide Ag, and the overall pattern of reac-tivity observed with Abs to proinsulin and insulin/proinsulin isconsistent with the lack of conformational epitopes requiring thefull molecule on the surface of the cells. The data from the mHELtransgenic mice further show that PAE cells displaying the proin-sulin epitope also display an HEL epitope, the latter in the contextof MHC class II using an Ab known to react with MHC class II(I-Ak) and HEL Ag complexes (27). Although extensive studiesare needed to characterize the nature of the self-epitopes displayedby PAE cells and to fully determine their association with Ag-presenting molecules, our data suggest that the display of self-epitopes by PAE cells is relevant to Ag presentation.

Our data also show that the CD11c� cells expressing self-epitopes belong to the DC class of PAE cells, which have beenassociated with the expression of self-molecules in the thymus andself-tolerance. The data provide critical information on PAE cellsbecause their nature, phenotype, and distribution has been contro-versial. In fact, studies have identified thymic PAE cells as TEC (6,7, 23) or BM-derived APC (21, 22). Using double immunofluo-rescence, we previously identified CD11c� PAE-DC expressingproinsulin and other T1D autoantigens in human thymus and PLT(22), tissues that transcribe the genes coding for these molecules(22, 24). Other studies in mouse and humans found no evidencethat PAE cells include BM-derived APC or that PAE cells existoutside the thymus (6, 7, 21). Our data show that PAE cells existin human peripheral blood and provide definitive evidence that, in

FIGURE 6. Proins� cells display multiple self-epitopes. A, Analysis of monocyte gated peripheralblood cells stained with Abs against proinsulin(M32337), GAD (GAD65-AS), and IA-2 (serum 8959,against the intracellular domain). Surface staining isdemonstrated with Abs to GAD65 and IA-2 and ispresent on significant proportions of Proins� cells. B,Blocking experiment demonstrating the specificity ofthe GAD65 mAb. Staining with this mAb was virtuallyblocked by preincubation with the relevant GAD65 pep-tide. C, Analysis of monocyte-gated peripheral bloodcells stained with Abs against proinsulin (M32337),MBP, and TPO. Surface staining is demonstrated withAbs to MBP and TPO and is present on a significantproportion of Proins� cells.

FIGURE 5. In vitro-derivedProins� DC express CD8�, CD20,and CD123. Percentage of cells ex-pressing CD8�, CD20, CD123,CD14, CD80, and CD83 by flowanalysis of in vitro-derived DC, as awhole or as Proins� and Proins� cellpopulations. The frequency of cellsexpressing these markers was com-pared in Proins� and Proins� cells inboth the immature (CD8�, p � 0.03;CD20, p � 0.02; CD123, p � 0.15;CD14, p � 0.03; CD83, p � 0.01)and mature states (CD8�, p � 0.006;CD20, p � 0.01; CD123, p � 0.01;CD14, p � 0.01); n � 9 except n �4 for CD123. Asterisks denote valuesthat are statistically different.

2118 CAPTURE-INDEPENDENT SELF-Ag EXPRESSION BY DC

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

addition to mTEC, PAE cells do include BM-derived CD11c� DCin both thymus and PLT in humans and mice. This is shown by thecoexpression of CD11c and other APC markers in human Proins�

cells and by the selective presence of INS transcripts in sortedProins�CD11c� cells from thymus, spleen, and blood. Mouse BMprecursor cells also express Ins2 and CD11c� cells maintain suchexpression in thymus and PLT.

The relative contribution to self-tolerance provided by mTECand DC remains unclear (10). In NOD mice, diabetes developsdespite the expression of Ins2 in the thymus, although this is lowerthan in control strains (40). NOD.Ins2�/� mice lacking Ins2 ex-pression (in both mTEC and DC) develop diabetes at an acceler-ated rate with higher incidence in both sexes (15, 16). Conversely,the transgenic expression of Ins2 in MHC class II-positive cells,including in the thymus, prevents diabetes in NOD mice (13);moreover, syngeneic transplantation of transgenic BM expressingIns2 in APC and the transfer of CD11c� DC transgenically ex-pressing Ins2 prevent diabetes in NOD mice (41, 42). Thus, thereis evidence from genetically manipulated mouse models that theexpression of self-Ags in the thymus and by adoptively transferredBM-derived APC can influence self-tolerance. In the AIRE knock-out mouse (43, 44), a model of the autoimmune polyendocrinopa-thy (APE)-candidiasis (C)-ectodermal dystrophy (ED) syndrome,or APECED (45), the lack of the transcription factor AIRE (auto-immune regulator) causes mTEC to lose the ability to express self-molecule genes, including insulin (44). However, AIRE knockoutmice do not develop autoimmune diabetes or insulitis, even whenAIRE knockout mice transgenically express HEL in pancreatic �cells and a HEL-specific TCR (46). Lack of NOD susceptibilitygenes, anergy, or ignorance of the target Ag could explain thesefindings (47). An additional explanation is the persistence of self-molecules expression by PAE-DC. It is unclear whether the lack ofAIRE impairs self-molecules expression in these cells, but in ourearlier studies, we did not observe colocalization of AIRE withproinsulin expression in human thymus (22). Our finding in hu-mans and nontransgenic mouse strains that PAE cells of the DCclass normally display self-epitopes in both thymus and peripherysuggests that these cells might physiologically contribute to boththymic and peripheral tolerance.

Continuous exposure of the immune system to self is indeedthought to be critical for self-tolerance, but peripheral presentationof tissue-specific self-molecules is considered dependent on thecapturing of self-molecules by APC, in particular immature DC(48). The analysis of DC isolated from PDX1 knockout mice dem-onstrates that the surface display of a proinsulin epitope does notrequire proinsulin expression from pancreatic � cells and thus isnot readily explainable by Ag capture from peripheral sources. Thecombined data from the insulin gene double knockout mice (whichlack Proins� PAE cells) and the PDX1 knockout mice (which have

such cells) show that the display of a proinsulin epitope byPAE-DC in both thymus and periphery depends on autonomousgene transcription. Thus, it appears that at least three mechanismscan explain the expression of self-molecules in the thymus andlymphoid tissues: expression by mTEC, expression by DC throughcapturing from peripheral tissues (48) or from thymic mTEC (29,49), and expression by DC through the transcription-dependentmechanism described here. The existence of redundant and possi-bly complementary mechanisms may provide improved capacityto establish self-tolerance.

The possibility that Proins� PAE-DC might contribute to self-tolerance is further supported by the phenotypic analysis of humanProins� cells, which correspond to the PAE-DC we defined in thenative state (22). Proins� cells in T-LDC and monocyte popula-tions from spleen and blood often coexpress CD11c, CD14, and

FIGURE 7. Proins� cells transcribe INS. RT-PCR analysis to assess theexpression of the INS transcript (437 bp, full length) in Proins� andProins� cells sorted from thymus, spleen, and peripheral blood. Datashown are representative of several experiments (thymus n � 3, spleen n �2, peripheral blood n � 4). Pancreatic islets and negative controls are alsoshown (omitting the reverse transcriptase (�RT) or the template (NT)).RT-PCR products were loaded on 2% agarose gels, of which the negativeimage is shown. GAPDH was used as a housekeeping gene.

FIGURE 8. Ins2 is expressed in mouse PLT, thymic, and peripheralCD11c� cells. A, Nested RT-PCR demonstrates Ins2 and glucagon tran-scripts in thymus, spleen, inguinal lymph nodes but not liver of C3H mice(�RT lanes, RT � reverse transcriptase). Control PCR in which the re-verse transcriptase was omitted are also shown (�RT). �2-microglobulinwas used as housekeeping gene. B, Thymic and splenic CD11c� cells butnot splenic CD11� cells and CD4� T cells from CH3 mice expressed Ins2and glucagon transcripts. C, RNA from pancreas and thymus of Ins2�,LacZ�� mice were used in nested RT-PCR experiments to assess the ex-pression of �-galactosidase. Spleen and splenic CD11c� cells from wild-type C3H mice were also tested. The �-galactosidase gene was found ex-pressed in pancreas, thymus, and inguinal lymph nodes of Ins2�, LacZmice but not in splenic CD11c� cells from C3H mice. D, BM cells wereisolated from Ins2�, LacZ�� mice. BM cells were also cultured in thepresence of GM-CSF. Nested RT-PCR detected �-galactosidase gene ex-pression in both freshly isolated and cultured BM cells.

2119The Journal of Immunology

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

other APC markers. PAE-DC from thymus more frequently ex-press CD1a, CD83, CD40, and CD80 compared with their coun-terparts in the periphery, suggesting that these cells include matureDC. Thymic Proins�CD11c� cells also expressed CD8�, a markerof “lymphoid” DC, which have tolerogenic function in the mouse(48, 50) but are not well characterized in humans. We have pre-viously presented evidence for the existence of CD8�� DC ex-pressing proinsulin in the human thymus using double immuno-fluorescence on tissue sections (22). We substantiate and extendthose results by immunophenotyping and flow cytometry analysisof DC isolated from human thymus and produced in vitro fromPB-M. Myeloid and lymphoid DC originate from a common my-eloid progenitor, and the expression of CD8� likely marks thefunctional state rather than origin (51). In the mouse, CD8�� DCinduce T cell deletion through the FasR (52). Consistent with thesefindings, Proins� cells express FasL, and we previously demon-strated apoptosis in the lymphocytes surrounding Proins� cells inhuman thymus and spleen, suggesting that human PAE-DC cancontribute to self-tolerance through deletion (22). Moreover,Proins� PAE-DC in the thymus frequently express CD20 andCD123, markers that are associated with regulatory function insome studies (53). In contrast, the predominant phenotype of pe-ripheral PAE-DC resembled those of immature DC (infrequentexpression of CD83, CD40, and CD80). Immature DC have alsobeen reported to be tolerogenic (48). Thus, CD11c� PAE-DC inthymus and periphery differ in their maturation state and expresspartially distinct phenotypic markers.

Remarkably, we discovered that �30% of the DC derived invitro from PB-M using a standard protocol are Proins� cells.

These cells maintained CD14 expression, and their frequency wasnot affected by GM-CSF/IL-4, suggesting that these cells may besomewhat unresponsive to these cytokines. However, the Proins�

cells in the cultures responded to the differentiation (GM-CSF/IL-4) and maturation (PGE2/TNF-�) stimuli by up-regulatingCD8�, CD20, and CD123. Thus, Proins� cells acquired a pheno-type similar to that of thymic PAE-DC in their native state. Thisobservation suggests that the surface expression of a proinsulinepitope identifies a subset of DC with unique functional and mat-uration characteristics. The ability to generate DC in vitro thatinclude Proins� cells uniquely maturing into a thymic DC-like andpotentially tolerogenic phenotype may be relevant for developingimproved DC-based strategies against autoimmunity. The epitopesexpressed by PAE-DC may participate in thymic positive and neg-ative selection processes and thus might be more likely to interactproductively with the T cell repertoire, perhaps stimulating regu-latory cells in the periphery, than those derived from pulsing DCwith Ags (54). It should also be noted that the B-chain/C-peptidejunction of proinsulin is a known target epitope in patients withT1D and NOD mice (55–57). Genetic factors are critical modifiersof the expression of self-molecules by PAE cells and can influencesusceptibility to autoimmune responses against many of the self-Ags expressed by PAE cells (10); it remains to be determinedwhether PAE cells might activate autoreactive T cells against self-epitopes, perhaps in the presence of a predisposing genetic back-ground and a proinflammatory microenvironment that modifiestheir functional state.

In conclusion, we demonstrate that PAE cells include a class ofDC, which are present in the thymus and periphery. This subset of

FIGURE 9. Analysis of thymic and splenic Proins� cellsin wild-type, Ins1�/�Ins2�/�, and PDX1�/� knockout mice.Thymus and spleen cells from newborn C57BL/6 (wild-type), insulin double knockout (Ins1�/�Ins2�/�), andPDX1�/� mice were stained with the proinsulin mAbM32337 or the polyclonal serum to GAD65/67 (Table I).Cells displaying surface staining for GAD65/67 are present inwild-type and knockout mice while proinsulin staining is de-tected in wild-type mice and PDX1�/� mice but not inthe Ins1�/�Ins2�/� mice. Data shown are represen-tative of experiments replicated in three independentsets of littermates.

2120 CAPTURE-INDEPENDENT SELF-Ag EXPRESSION BY DC

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

DC displays self-epitopes through a novel mechanism based onautonomous gene transcription rather than Ag capturing. The pres-ence of PAE-DC in the circulation and the ability to purify thesecells by selective staining for self-epitopes should facilitate furtherstudies to better characterize the self-epitopes expressed, the mo-lecular mechanisms regulating such expression, the functionalproperties of these cells in both normal and autoimmune states,and, ultimately, their potential for therapeutic applications.

AcknowledgmentsDr. Ken Shortman kindly shared his protocol for T-LDC preparation. Dr.Len Harrison donated proinsulin peptides for the blocking experiments. Wethank him and Dr. Michael Clare-Salzler for critical discussions of the data.Dr. Constantin Polychronakos provided breeding pairs to generate insulin dou-ble knockout mice. We thank Dr. Maki Nakayama for sharing her unpublishedgenotyping protocol and primer sequences. Ingela Berglund-Dahl helped withmouse husbandry, and Helena Edlund provided PDX1 knockout mice. Dr. IgalGery provided mHEL mice and Dr. Ronald Germain the mAb D8H21. DavidStenger gave many helpful suggestions for the manuscript. We acknowledgethe support of Jim Phillips and the Flow Cytometry Core of the University ofMiami.

DisclosuresThe authors have no financial conflict of interest.

References1. Sprent, J., and H. Kishimoto. 2002. The thymus and negative selection. Immunol.

Rev. 185: 126–135.2. Jolicouer, C., D. Hanahan, and K. M. Smith. 1994. T cell tolerance toward a

transgenic � cell antigen and transcription of endogenous pancreatic genes inthymus. Proc. Natl. Acad. Sci. USA 91: 6707–6711.

3. Pugliese, A., M. Zeller, A. J. Fernandez, L. J. Zalcberg, R. J. Bartlett, C. Ricordi,M. Pietropaolo, G. S. Eisenbarth, S. T. Bennett, and D. D. Patel. 1997. Theinsulin gene is transcribed in the human thymus and transcription levels corre-lated with allelic variation at the INS VNTR-IDDM2 susceptibility locus for type1 diabetes. Nat. Genet. 15: 293–297.

4. Vafiadis, P., S. T. Bennett, J. A. Todd, J. Nadeau, R. Grabs, C. G. Goodyer,S. Wickramasinghe, E. Colle, and C. Polychronakos. 1997. Insulin expression inhuman thymus is modulated by INS VNTR alleles at the IDDM2 locus. Nat.Genet. 15: 289–292.

5. Sospedra, M., X. Ferrer-Francesch, O. Dominguez, M. Juan, M. Foz-Sala, andR. Pujol-Borrell. 1998. Transcription of a broad range of self-antigens in humanthymus suggests a role for central mechanisms in tolerance toward peripheralantigens. J. Immunol. 161: 5918–5929.

6. Derbinski, J., A. Schulte, B. Kyewski, and L. Klein. 2001. Promiscuous geneexpression in medullary thymic epithelial cells mirrors the peripheral self. Nat.Immunol. 2: 1032–1039.

7. Gotter, J., B. Brors, M. Hergenhahn, and B. Kyewski. 2004. Medullary epithelialcells of the human thymus express a highly diverse selection of tissue-specificgenes colocalized in chromosomal clusters. J. Exp. Med. 199: 155–166.

8. Antonia, S. J., T. Geiger, J. Miller, and R. A. Flavell. 1995. Mechanisms ofimmune tolerance induction through the thymic expression of a peripheral tissue-specific protein. Int. Immunol. 7: 715–725.

9. Smith, K. M., D. C. Olson, R. Hirose, and D. Hanahan. 1997. Pancreatic geneexpression in rare cells of thymic medulla: evidence for functional contributionto T cell tolerance. Int. Immunol. 9: 1355–1365.

10. Pugliese, A. 2004. Central and peripheral autoantigen presentation in immunetolerance. Immunology 111: 138–146.

11. Mathis, D., and C. Benoist. 2004. Back to central tolerance. Immunity 20:509–516.

12. Pugliese, A., and D. Miceli. 2002. The insulin gene in diabetes. Diabetes Metab.Res. Rev. 18: 13–25.

13. French, M. B., J. Allison, D. S. Cram, H. E. Thomas, M. Dempsey-Collier,A. Silva, H. M. Georgiou, T. W. Kay, L. C. Harrison, and A. M. Lew. 1997.Transgenic expression of mouse proinsulin II prevents diabetes in nonobese di-abetic mice. Diabetes 46: 34–39.

14. Chentoufi, A. A., and C. Polychronakos. 2002. Insulin expression levels in thethymus modulate insulin-specific autoreactive T cell tolerance: the mechanism bywhich the IDDM2 locus may predispose to diabetes. Diabetes 51: 1383–1390.

15. Thebault-Baumont, K., D. Dubois-Laforgue, P. Krief, J. P. Briand, P. Halbout,K. Vallon-Geoffroy, J. Morin, V. Laloux, A. Lehuen, J. C. Carel, et al. 2003.Acceleration of type 1 diabetes mellitus in proinsulin 2-deficient NOD mice.J. Clin. Invest. 111: 851–857.

16. Moriyama, H., N. Abiru, J. Paronen, K. Sikora, E. Liu, D. Miao, D. Devendra,J. Beilke, R. Gianani, R. G. Gill, and G. S. Eisenbarth. 2003. Evidence for aprimary islet autoantigen (preproinsulin 1) for insulitis and diabetes in the nono-bese diabetic mouse. Proc. Natl. Acad. Sci. USA 100: 10376–10381.

17. Wentworth, B. M., I. M. Schaefer, L. Villa-Komaroff, and J. M. Chirgwin. 1986.Characterization of the two nonallelic genes encoding mouse preproinsulin.J. Mol. Evol. 23: 305–312.

18. Duvillie, B., N. Cordonnier, L. Deltour, F. Dandoy-Dron, J. M. Itier,E. Monthioux, J. Jami, R. L. Joshi, and D. Bucchini. 1997. Phenotypic alterationsin insulin-deficient mutant mice. Proc. Natl. Acad. Sci. USA 94: 5137–5140.

19. Makino, S., D. Kunimoto, Y. Muraoka, Y. Mizushima, K. Katagiri, andY. Tochino. 1980. Breeding of a non-obese, diabetic strain of mice. JikkenDobutsu 29: 1–13.

20. Pugliese, A., and J. Diez. 2002. Lymphoid organs contain diverse cells expressingself-molecules. Nat. Immunol. 3: 335–336.

FIGURE 10. A, SP-M from 5-wk-old mHEL miceand 5-wk-old CBA mice were stained with the rat mAbD8H21, which reacts with the MHC/Ag complex con-sisting of the I-Ak molecule and the 116-129 HEL pep-tide. It is reported to stain mouse spleen cells of CBAand B10.A(4R) mice (both expressing I-Ak) whenpulsed with this HEL peptide or whole HEL (27). ThemAb D8H21 stained the surface of SP-M from mHELtransgenic mice without Ag pulsing. This is consistentwith the ability of SP-M from mHEL transgenic mice tospontaneously display a HEL epitope in associationwith the I-Ak MHC molecule. CBA mice express thesame I-Ak molecule as mHEL mice but lack the expres-sion of HEL. Some reactivity with the D8H21 Ab isseen in CBA mice but at much lower levels. This isexpected because this Ab is known to react with MHC/peptide complexes in which the peptide may derivefrom serum proteins (27). B, SP-M from 5-wk-oldmHEL transgenic mice were stained with the rat mAbD8H21 to HEL and the mouse mAb M32337 to proin-sulin directly conjugated with PE (left panel). A pro-portion of cells from these mice coexpresses HEL andproinsulin epitopes, and indeed, most Proins� cells co-express the HEL epitope. As expected, no colocaliza-tion between HEL and proinsulin was observed in CBAmice (right panel).

2121The Journal of Immunology

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

21. Throsby, M., F. Homo-Delarche, D. Chevenne, R. Goya, M. Dardenne, andJ. M. Pleau. 1998. Pancreatic hormone expression in the murine thymus: local-ization in dendritic cells and macrophages. Endocrinology 139: 2399–2406.

22. Pugliese, A., D. Brown, D. Garza, D. Murchison, M. Zeller, M. J. Redondo,J. Diez, G. S. Eisenbarth, D. D. Patel, and C. Ricordi. 2001. Self-antigen pre-senting cells expressing IDDM-associated autoantigens exist in both thymus andperipheral lymphoid organs in humans. J. Clin. Invest. 107: 585–593.

23. Chentoufi, A. A., M. Palumbo, and C. Polychronakos. 2004. Proinsulin expres-sion by Hassall’s corpuscles in the mouse thymus. Diabetes 53: 354–359.

24. Diez, J., Y. Park, M. Zeller, D. Brown, D. Garza, C. Ricordi, J. Hutton,G. S. Eisenbarth, and A. Pugliese. 2001. Differential splicing of the IA-2 mRNAin pancreas and lymphoid organs as a permissive genetic mechanism for auto-immunity against the IA-2 type 1 diabetes autoantigen. Diabetes 50: 895–900.

25. Vandenabeele, S., H. Hochrein, N. Mavaddat, K. Winkel, and K. Shortman. 2001.Human thymus contains 2 distinct dendritic cell populations. Blood 97:1733–1741.

26. Madsen, O. 1987. Proinsulin-specific monoclonal antibodies: Immunocytochem-ical application as � cell markers and as probes for conversion. Diabetes 36:1203–1211.

27. Zhong, G., C. Reis e Sousa, and R. N. Germain. 1997. Production, specificity, andfunctionality of monoclonal antibodies to specific peptide-major histocompati-bility complex class II complexes formed by processing of exogenous protein.Proc. Natl. Acad. Sci. USA 94: 13856–13861.

28. Jonsson, J., L. Carlsson, T. Edlund, and H. Edlund. 1994. Insulin-promoter-factor1 is required for pancreas development in mice. Nature 371: 606–609.

29. Zhang, M., M. S. Vacchio, B. P. Vistica, S. Lesage, C. E. Egwuagu, C. R. Yu,M. P. Gelderman, M. C. Kennedy, E. F. Wawrousek, and I. Gery. 2003. T celltolerance to a neo-self antigen expressed by thymic epithelial cells: the solubleform is more effective than the membrane-bound form. J. Immunol. 170:3954–3962.

30. Lai, J. C., A. Fukushima, E. F. Wawrousek, M. C. Lobanoff,P. Charukamnoetkanok, S. J. Smith-Gill, B. P. Vistica, R. S. Lee, C. E. Egwuagu,S. M. Whitcup, and I. Gery. 1998. Immunotolerance against a foreign antigentransgenically expressed in the lens. Invest. Ophthalmol. Vis. Sci. 39: 2049–2057.

31. Keilacker, H., H. Dietz, S. Witt, K. P. Woltanski, R. Berling, and M. Ziegler.1986. Kinetic properties of monoclonal insulin antibodies. Biomed. Biochim. Acta45: 1093–1102.

32. Hemmer, B., J. J. Archelos, and H. P. Hartung. 2002. New concepts in the im-munopathogenesis of multiple sclerosis. Nat. Rev. Neurosci. 3: 291–301.

33. McLachlan, S. M., and B. Rapoport. 2000. Autoimmune response to the thyroidin humans: thyroid peroxidase—the common autoantigenic denominator. Int.Rev. Immunol. 19: 587–618.

34. Marty, M. C., F. Alliot, J. Rutin, R. Fritz, D. Trisler, and B. Pessac. 2002. Themyelin basic protein gene is expressed in differentiated blood cell lineages and inhemopoietic progenitors. Proc. Natl. Acad. Sci. USA 99: 8856–8861.

35. Dakic, A., Q. X. Shao, A. D’Amico, M. O’Keeffe, W. F. Chen, K. Shortman, andL. Wu. 2004. Development of the dendritic cell system during mouse ontogeny.J. Immunol. 172: 1018–1027.

36. Steinman, R. M. 2003. The control of immunity and tolerance by dendritic cell.Pathol. Biol. (Paris) 51: 59–60.

37. Mougneau, E., S. Hugues, and N. Glaichenhaus. 2002. Antigen presentation bydendritic cells in vivo. J. Exp. Med. 196: 1013–1016.

38. Polakova, K., D. Plaksin, D. H. Chung, I. M. Belyakov, J. A. Berzofsky, andD. H. Margulies. 2000. Antibodies directed against the MHC-I molecule H-2Dd

complexed with an antigenic peptide: similarities to a T cell receptor with thesame specificity. J. Immunol. 165: 5703–5712.

39. Porgador, A., J. W. Yewdell, Y. Deng, J. R. Bennink, and R. N. Germain. 1997.Localization, quantitation, and in situ detection of specific peptide-MHC class Icomplexes using a monoclonal antibody. Immunity 6: 715–726.

40. Brimnes, M. K., T. Jensen, T. N. Jorgensen, B. K. Michelsen, J. Troelsen, andO. Werdelin. 2002. Low expression of insulin in the thymus of non-obese dia-betic mice. J. Autoimmun. 19: 203–213.

41. Steptoe, R. J., J. M. Ritchie, and L. C. Harrison. 2003. Transfer of hematopoieticstem cells encoding autoantigen prevents autoimmune diabetes. J. Clin. Invest.111: 1357–1363.

42. Steptoe, R. J., J. M. Ritchie, L. K. Jones, and L. C. Harrison. 2005. Autoimmunediabetes is suppressed by transfer of proinsulin-encoding gr-1� myeloid progen-itor cells that differentiate in vivo into resting dendritic cells. Diabetes 54:434–442.

43. Ramsey, C., O. Winqvist, L. Puhakka, M. Halonen, A. Moro, O. Kampe,P. Eskelin, M. Pelto-Huikko, and L. Peltonen. 2002. Aire deficient mice developmultiple features of APECED phenotype and show altered immune response.Hum. Mol. Genet. 11: 397–409.

44. Anderson, M. S., E. S. Venanzi, L. Klein, Z. Chen, S. P. Berzins, S. J. Turley,H. von Boehmer, R. Bronson, A. Dierich, C. Benoist, and D. Mathis. 2002.Projection of an immunological self shadow within the thymus by the aire pro-tein. Science 298: 1395–1401.

45. Pitkanen, J., and P. Peterson. 2003. Autoimmune regulator: from loss of functionto autoimmunity. Genes Immun. 4: 12–21.

46. Liston, A., S. Lesage, J. Wilson, L. Peltonen, and C. C. Goodnow. 2003. Aireregulates negative selection of organ-specific T cells. Nat. Immunol. 4: 350–354.

47. Sprent, J., and C. D. Surh. 2003. Knowing one’s self: central tolerance revisited.Nat. Immunol. 4: 303–304.

48. Steinman, R. M., D. Hawiger, and M. C. Nussenzweig. 2003. Tolerogenic den-dritic cells. Annu. Rev. Immunol. 21: 685–711.

49. Gallegos, A. M., and M. J. Bevan. 2004. Central tolerance to tissue-specificantigens mediated by direct and indirect antigen presentation. J. Exp. Med. 200:1039–1049.

50. Vremec, D., and K. Shortman. 1997. Dendritic cell subtypes in mouse lymphoidorgans: cross-correlation of surface markers, changes with incubation, and dif-ferences among thymus, spleen, and lymph nodes. J. Immunol. 159: 565–573.

51. Traver, D., K. Akashi, M. Manz, M. Merad, T. Miyamoto, E. G. Engleman, andI. L. Weissman. 2000. Development of CD8�-positive dendritic cells from acommon myeloid progenitor. Science 290: 2152–2154.

52. Belz, G. T., G. M. Behrens, C. M. Smith, J. F. Miller, C. Jones, K. Lejon,C. G. Fathman, S. N. Mueller, K. Shortman, F. R. Carbone, and W. R. Heath.2002. The CD8�� dendritic cell is responsible for inducing peripheral self-tol-erance to tissue-associated antigens. J. Exp. Med. 196: 1099–1104.

53. Nakano, H., M. Yanagita, and M. D. Gunn. 2001. CD11c�B220�Gr-1� cells inmouse lymph nodes and spleen display characteristics of plasmacytoid dendriticcells. J. Exp. Med. 194: 1171–1178.

54. Tarbell, K. V., S. Yamazaki, K. Olson, P. Toy, and R. M. Steinman. 2004.CD25�CD4� T cells, expanded with dendritic cells presenting a single autoan-tigenic peptide, suppress autoimmune diabetes. J. Exp. Med. 199: 1467–1477.

55. Rudy, G., N. Stone, L. C. Harrison, P. G. Colman, P. McNair, V. Brusic,M. B. French, M. C. Honeyman, B. Tait, and A. M. Lew. 1995. Similar peptidesfrom two � cell autoantigens, proinsulin and glutamic acid decarboxylase, stim-ulate T cells of individuals at risk for insulin-dependent diabetes. Mol. Med. 1:625–633.

56. Narendran, P., A. J. Williams, K. Elsegood, N. J. Leech, and C. M. Dayan. 2003.Humoral and cellular immune responses to proinsulin in adults with newly di-agnosed type 1 diabetes. Diabetes Metab. Res. Rev. 19: 52–59.

57. Chen, W., I. Bergerot, J. F. Elliott, L. C. Harrison, N. Abiru, G. S. Eisenbarth, andT. L. Delovitch. 2001. Evidence that a peptide spanning the B-C junction ofproinsulin is an early autoantigen epitope in the pathogenesis of type 1 diabetes.J. Immunol. 167: 4926–4935.

2122 CAPTURE-INDEPENDENT SELF-Ag EXPRESSION BY DC

by guest on October 13, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from