Translational Mini-Review Series on Type 1 Diabetes: Systematic analysis of T cell epitopes in autoimmune diabetes

Abstract

T cell epitopes represent the molecular code words through which the adaptive immune system communicates. In the context of a T cell-mediated autoimmune disease such as type 1 diabetes, CD4 and CD8 T cell recognition of islet autoantigenic epitopes is a key step in the autoimmune cascade. Epitope recognition takes place during the generation of tolerance, during its loss as the disease process is initiated, and during epitope spreading as islet cell damage is perpetuated. Epitope recognition is also a potentially critical element in therapeutic interventions such as antigen-specific immunotherapy. T cell epitope discovery, therefore, is an important component of type 1 diabetes research, in both human and murine models. With this in mind, in this review we present a comprehensive guide to epitopes that have been identified as T cell targets in autoimmune diabetes. Targets of both CD4 and CD8 T cells are listed for human type 1 diabetes, for humanized [human leucocyte antigen (HLA)-transgenic] mouse models, and for the major spontaneous disease model, the non-obese diabetic (NOD) mouse. Importantly, for each epitope we provide an analysis of the relative stringency with which it has been identified, including whether recognition is spontaneous or induced and whether there is evidence that the epitope is generated from the native protein by natural antigen processing. This analysis provides an important resource for investigating diabetes pathogenesis, for developing antigen-specific therapies, and for developing strategies for T cell monitoring during disease development and therapeutic intervention.

T cell reactivity in type 1 diabetes

It is now widely accepted that type 1 diabetes is an autoimmune disease associated with the activation of CD4 and CD8 T cells recognizing islet autoantigens [1,2]. It is considered likely by many that these autoreactive T cells are the mediators of islet β cell damage, although direct evidence for this is compelling only in rodent models of the disease in which adoptive transfer experiments are feasible ethically and technically. Epitopes represent the molecular code words through which the adaptive immune system generates cellular immunity with specificity and memory; islet autoantigen epitopes represent the targets of regulatory and effector T cells that preside over the fate of the β cell. Because type 1 diabetes is a relatively common disorder with several representative animal models, and because some of the major autoantigens in humans and rodents have been well characterized, spontaneous diabetes has come to be considered as the prototypic cell-mediated autoimmune disease.

The considerable advances in autoimmune serology in type 1 diabetes during the period 1975–91 led to the unequivocal identification of three major islet autoantibody specificities, targeted against insulin, glutamic acid decarboxylase (GAD) and the islet tyrosine phosphatase, IA-2 [3–5]. Not surprisingly, most T cell epitope studies have focused on these three autoantigens; the existence of class-switched IgG autoantibodies against these particular islet proteins strongly implies the influence of T cell help. From the 1990s onwards, increasing numbers of reports noted the detection of T cells directed against these three proteins in the peripheral blood, leading to an increasing emphasis on epitope discovery. To date, even a relatively partial unravelling of the epitope code in type 1 diabetes has enabled the characterization of T cell responses and some key research advances. These include the demonstration that human type 1 diabetes may be Th1-dominated [6,7]; identification of a previously unrecognized population of naturally arising CD4 T cells producing interleukin (IL)-10 and associated with late disease onset [6]; evidence for T cell cross-reactivity between β cell antigens and common pathogens [8]; identification of CD8 T cells associated with recurrent autoimmunity after islet transplantation [9]; the first steps towards developing T cell assays for standardization and use in trial monitoring [10]; and the development of epitope-based intervention strategies [11].

In light of these advances, this review is designed to take stock of the large portfolio of epitopes identified to date, and examine their cellular source and major histocompatibility complex (MHC) restriction. Our review highlights the question of what truly constitutes a T cell epitope, and seeks to provide some indication as to the likelihood that particular epitopes may represent useful research and therapeutic tools, by categorizing the evidence that led to their identification.

Epitope discovery

The starting-points for Tables 1–4 were previous listings of diabetes-relevant T cell epitopes compiled by our groups [12,13]. These previous compilations have been supplemented here by the results of web-based searches of PubMed (the US National Library of Medicine's database of biomedical citations and abstracts) conducted in April 2006 using the search terms ‘diabetes, T cell, epitope, antigen’ or ‘diabetes, T cell, peptide, antigen’. Peptides listed as T cell epitopes for mouse β cell antigens in Tables 2 and 4 are those reported to be recognized by T cells from non-obese diabetic (NOD) mice either spontaneously or as a consequence of peptide or protein immunization. Peptides listed as T cell epitopes for human β cell antigens in Tables 1 and 3 are those shown to be recognized by T cells from type 1 diabetes patients and/or at-risk individuals, or those recognized by human leucocyte antigen (HLA)-restricted T cells in HLA-transgenic mice (not necessarily of the NOD background). For epitopes reported multiple times by the same group, only a single reference is provided in Tables 1–4. However, for epitopes reported by more than one group, all are credited. In this way, it can be seen readily which epitopes have been verified by multiple groups. While the epitope listings are intended to be comprehensive, the authors will welcome notification of inadvertent omissions.

Table 1.

CD4 T cell epitopes for human β cell antigens

			Type of T cell response
				Human
Aga	Position	Sequence	MHC	Clone or line	PBMC	Controlsd	HLA Tg miceb	Level of evidencec	Comments	Ref.
GAD65	88–99	NYAFLHATDLLP	DR1	Yes		3		C		[43]
	101–115	CDGERPTLAFLQDVM	DQ8		Yes	2	Prot-imm	A		[44]
	115–127	MNILLQYVVKSFD	DR4				Pept-imm	C		[45]
	115–130	MNILLQYVVKSFDRST	DR4				Prot-imm	C		[46]
	116–127	NILLQYVVKSFD	DR53	Yes		1		C		[47]
	121–140	YVVKSFDRSTKVIDFHYPNE	DQ8				Prot-imm	D		[48]
	126–140	FDRSTKVIDFHYPNE	DQ8		Yes	2	Prot-imm	A		[44]
	146–165	NWELADQPQNLEEILMHCQT	DR2	Yes		3		C		[49]
	173–187	TGHPRYFNQLSTGLD	DQ8		Yes	3		D		[50]
	174–185	GHPRYFNQLSTG	DR2	Yes		3		C		[49]
	201–220	NTNMFTYEIAPVFVLLEYVT	DQ8				Prot-imm	D		[48]
	202–216	TNMFTYEIAPVFVLL	DR8 or DR9	Yes		1		D		[47]
	206–220	TYEIAPVFVLLEYVT	DQ8		Yes	2	Prot-imm	A		[44]
	206–225	TYEIAPVFVLLFYVTLKKMR	DR2	Yes		3		C		[49]
	231–250	PGGSGDGIFSPGGAISNMYA	DQ8				Prot-imm	D		[48]
	247–266	NMYAMMIARFKMFPEVKEKG	DQ and/or DR		Yes	1, 2		D		[51,52]
	247–266	NMYAMMIARFKMFPEVKEKG	DR3	Yes	Yes	3		D		[53]
	248–257	MYAMMIARFK	DR51	Yes		3		C		[43]
	251–270	MMIARFKMFPEVKEKGMAAL	DR12	Yes	Yes	3		D		[53]
	260–279	PEVKEKGMAALPRLIAFTSE	DQ and/or DR		Yes	1, 2		D		[51,52]
	261–280	EVKEKGMAALPRLIAFTSEH	DQ8	Yes	Yes	3		D		[53]
	270–283	LPRLIAFTSEHSHF	DR4	Yes		3		C		[49]
	271–285	PRLIAFTSEHSHFSL	DR4				Prot-imm	C		[46]
	274–286	IAFTSEHSHFSLK	DR4				Pept-imm	C		[45]
	339–352	TVYGAFDPLLAVAD	DR3	Yes	Yes	3		A		[54]
	356–370	KYKIWMHVDAAWGGG	DR4				Prot-imm	C		[46]
	370–386	GLLMSRKHKWKLSGVER	DP2	Yes		1		D		[47]
	376–390	KHKWKLSGVERANSV	DR4				Prot-imm	C		[46]
	417–429	NCNQMHASYLFQQ	DR1	Yes		1		C		[47]
	431–445	KHYDLSYDTGDKALQ	DQ8		Yes	2	Prot-imm	A		[44]
	461–475	AKGTTGFEAHVDKCL	DQ8		Yes	2	Prot-imm	A		[44]
	471–490	VDKCLELAEYLYNIIKNREG	DQ8				Prot-imm	D		[48]
	481–495	LYNIIKNREGYEMVF	DR4				Prot-imm	C		[46]
	491–510	YEMVFDGKPQHTNVCFWYIP	DR3 and DQ5	Yes	Yes	3		D		[53]
	493–507	MVFDGKPQHTNVCFW	DQ8		Yes	3		D		[50]
	501–520	HTNVCFWYIPPSLRTLEDNE	DR1 and DR4	Yes	Yes	3		D		[53]
	505–519	CFWYIPPSLRTLEDN	DQ1 or DR1	Yes	Yes	2		D		[55]
	505–519	CFWYIPPSLRTLEDN	DQ8	Yes	Yes	3	Pept-imm	D		[50]
	506–518	FWYIPPSLRTLED	DQ and/or DR		Yes	1		D		[56]
	511–525	PSLRTLEDNEERMSR	DR4				Prot-imm	C		[46]
	511–530	PSLRTLEDNEERMSRLSKVA	DR3	Yes	Yes	3		D		[53]
	521–535	ERMSRLSKVAPVIKA	DQ8	Yes	Yes	3	Pept-imm	D	Greater response in diabetic compared to non-diabetic twins	[50]
	521–535	ERMSRLSKVAPVIKA	DQ8 or DR4	Yes	Yes	1		D		[55]
	533–547	IKARMMEYGTTMVSY	DQ8		Yes	3		D		[50]
	536–550	RMMEYGTTMVSYQPL	DQ8		Yes	2	Prot-imm	A		[44]
	546–560	SYQPLGDKVNFFRMV	DR4				Prot-imm	C		[46]
	551–565	GDKVNFFRMVISNPA	DR4				Prot-imm	C		[46]
	555–567	NFFRMVISNPAAT	DR4	Yes	Yes	1		B	Eluted from MHC	[57,58]
	556–570	FFRMVISNPAATHQD	DR4				Prot-imm	C		[46]
	563–575	NPAATHQDIDFLI	DR53	Yes		3		C		[59]
	566–580	ATHQDIDFLIEEIER	DR4				Prot-imm	C		[46]
HSP60	31–50	KFGADARALMLQGVDLLADA	?		Yes	3		D		[60]
	136–155	NPVEIRRGVMLAVDAVIAEL	?		Yes	3		D		[60]
	255–275	QSIVPALEIANAHRKPLVIIA	?		Yes	3		D		[60]
	286–305	LVLNRLKVGLQVVAVKAPGF	?		Yes	3		D		[60]
	436–455	IVLGGGCALLRCIPALDSLT	?		Yes	3		D		[60]
	437–460	VLGGGCALLRCIPALDSLTPANED	?		Yes	1		D		[60]
	466–485	EIIKRTLKIPAMTIAKNAGV	?		Yes	1		D		[60]
	511–530	VNMVEKGIIDPTKVVRTALL	?		Yes	3		D		[60]
HSP70	1–20	MAKAAAVGIDLGTTYSCVGV	?		Yes	3		D		[61]
	166–185	GLNVLRIINEPTAAAIAYGL	?		Yes	3		D		[61]
	210–229	TIDDGIFEVKATAGDTHLGG	?		Yes	3		D		[61]
	225–244	THLGGEDFDNRLVNHFVEEF	?		Yes	3		D		[61]
	271–290	KRTLSSSTQASLEIDSLFEG	?		Yes	3		D		[61]
	391–410	LLLLDVAPLSLGLETAGGVM	?		Yes	3		D		[61]
	421–440	PTKQTQIFTTYSDNQPGVLI	?		Yes	3		D		[61]
	496–515	KANKITITNDKGRLSKEEIE	?		Yes	3		D		[61]
	511–530	KEEIERMVQEAEKYKAEDEV	?		Yes	3		D		[61]
IA-2	502–514	GSFINISVVGPAL	?		Yes	2		D		[62]
	575–587	RSVLLTLVALAGV	?		Yes	2		D		[62]
	601–618	RQHARQQDKERLAALGPE	DQ8				Pept-imm	D		[63]
	608–620	DKERLAALGPEGA	?		Yes	3		D		[64]
	616–633	GPEGAHGDTTFEYQDLCR	DQ8				Pept-imm	D		[63]
	646–663	EGPPEPSRVSSVSSQFSD	DQ8				Pept-imm	D		[63]
	654–674	VSSVSSQFSDAAQASPSSHSS	DR4		Yes	1		B	Eluted from MHC	[7]
	661–678	FSDAAQASPSSHSSTPSW	DQ8				Pept-imm	D		[63]
	685–700	ANMDISTGHMILAYME	DR4–DQ8		Yes	3		D		[65]
	709–732	LAKEWQALCAYQAEPNTCATAQGE	DR4		Yes	1		B	Eluted from MHC	[7]
	713–728	WQALCAYQAEPNTCAT	DR4–DQ8		Yes	3		D		[65]
	721–738	AEPNTCATAQGEGNIKKN	DQ8				Pept-imm	D		[63]
	745–760	PYDHARIKLKVESSPS	DR4–DQ8		Yes	3		D		[65]
	751–770	IKLKVESSPSRSDYINASPI	DR4	Yes		3		C		[66]
	752–775	KLKVESSPSRSDYINASPIIEHDP	DR4		Yes	1		B	Eluted from MHC	[6]
	766–783	NASPIIEHDPRMPAYIAT	DQ8				Pept-imm	D		[63]
	787–802	LSHTIADFWQMVWESG	DR4–DQ8		Yes	3		D		[65]
	793–808	DFWQMVWESGCTVIVM	DR4–DQ8		Yes	3		D		[65]
	797–809	MVWESGCTVIVML	?		Yes	3		D		[64]
	797–817	MVWESGCTVIVMLTPLVEDGV	DR4		Yes	1		B	Eluted from MHC	[7]
	799–814	WESGCTVIVMLTPLVE	DR3–DQ2; DR4–DQ8		Yes	3		D		[65]
	804–816	TVIVMLTPLVEDG	?		Yes	3		D		[64]
	805–820	VIVMLTPLVEDGVKQC	DR3–DQ2; DR4–DQ8		Yes	3		D		[65]
	826–843	DEGASLYHVYEVNLVSEH	DQ8				Pept-imm	D		[63]
	830–842	SLYHVYEVNLVSE	?		Yes	2		D		[62]
	831–850	LYHVYEVNLVSEHIWCEDFL	DPB4·1	Yes	Yes	3		A		[66]
	841–856	SEHIWCEDFLVRSFYL	DR3–DQ2; DR4–DQ8		Yes	3		D		[65]
	841–860	SEHIWCEDFLVRSFYLKNVQ	DPB4·1	Yes	Yes	3		A		[66]
	845–860	WCEDFLVRSFYLKNVQ	DR4–DQ8		Yes	3		D		[65]
	847–862	EDFLVRSFYLKNVQTQ	DR3–DQ2; DR4–DQ8		Yes	3		D		[65]
	854–866	FYLKNVQTQETRT	?		Yes	3		D		[64]
	854–872	FYLKNVQTQETRTLTQFHF	DR4		Yes	1		B	Eluted from MHC	[7]
	889–904	DFRRKVNKCYRGRSCP	DR4–DQ8		Yes	3		D		[65]
	918–930	TYILIDMVLNRMA	?		Yes	2		D		[62]
	919–934	YILIDMVLNRMAKGVK	DR3–DQ2; DR4–DQ8		Yes	3		D		[65]
	931–948	KGVKEIDIAATLEHVRDQ	DQ8				Pept-imm	D		[63]
	933–945	VKEIDIAATLEHV	?		Yes	3		D		[64]
	955–975	SKDQFEFALTAVAEEVNAILK	DR4		Yes	1		B	Eluted from MHC	[7]
	959–974	FEFALTAVAEEVNAIL	DR4–DQ8		Yes	3		D		[65]
	961–979	FALTAVAEEVNAILKALPQ	DQ8				Pept-imm	D		[63]
ICA69	36–47	AFIKATGKKEDE	?		Yes	1		D		[67]
IGRP	13–25	QHLQKDYRAYYTF	DR3	Yes		2		D		[68]
	23–35	YTFLNFMSNVGDP	DR4	Yes		2		D		[68]
	226–238	RVLNIDLLWSVPI	DR3	Yes		2		D		[68]
	247–259	DWIHIDTTPFAGL	DR4	Yes		2		D		[68]
Insulin	L1–16	MALWMRLLPLLALLAL	?		Yes	3		D		[69]
	L1–24	MALWMRLLPLLALLALWGPDPAAA	DQ8				Pept-imm	D		[70]
	L5–20	MRLLPLLALLALWGPD	?		Yes	3		D		[69]
	L9–24	PLLALLALWGPDPAAA	?		Yes	3		D		[69]
	L11–B2	LALLALWGPDPAAAFV	DR4				Prot-imm	C		[71]
	L13–B4	LLALWGPDPAAAFVNQ	?		Yes	3		D		[69]
	L17–B8	WGPDPAAAFVNQHLCG	?		Yes	3		D		[69]
	L21–B12	PAAAFVNQHLCGSHLV	DR4				Prot-imm	C		[71]
	L21–B12	PAAAFVNQHLCGSHLV	?		Yes	3		D		[69]
	B1–17	FVNQHLCGSHLVEALYL	?		Yes	1		D		[72]
	B6–22	LCGSHLVEALYLVCGER	?		Yes	3		D		[69]
	B9–23	SHLVEALYLVCGERG	DQ8		Yes	1		D		[73]
	B10–25	HLVEALYLVCGERGFF	?	Yes		3		D		[74]
	B11–27	LVEALYLVCGERGFFYT	DR16	Yes		3		C		[75]
	B11–27	LVEALYLVCGERGFFYT	?		Yes	1		D		[72]
	B14–30	ALYLVCGERGFFYTPKT	?		Yes	3		D		[76]
	B16–32	YLVCGERGFFYTPKTRR	?		Yes	3		D		[69]
	B20–C4	GERGFFYTPKTRREAED	?		Yes	3, 1		D		[53,72]
	B20–C7	GERGFFYTPKTRREAEDLQV	DQ8				Pept-imm	D		[70]
	B21–C5	ERGFFYTPKTRREAEDL	?		Yes	3		D		[76]
	B24–C4	FFYTPKTRREAED	DQ and/or DR		Yes	1		D		[56]
	B25–C8	FYTPKTRREAEDLQVG	?	Yes		3		D		[74]
	B25–C9	FYTPKTRREAEDLQVGQ	?		Yes	3		D		[69]
	B30–C14	TRREAEDLQVGQVELGG	?		Yes	1		D		[72]
	C3–18	EDLQVGQVELGGGPGA	DR	Yes		3		D		[74]
	C3–19	EDLQVGQVELGGGPGAG	?		Yes	3		D		[69]
	C8–24	GQVELGGGPGAGSLQPL	?		Yes	1		D		[72]
	C13–29	GGGPGAGSLQPLALEGS	?		Yes	3		D		[69]
	C13–32	GGGPGAGSLQPLALEGSLQK	DR4		Yes	1		B	Eluted from MHC	[6]
	C17–A1	GAGSLQPLALEGSLQKRG	DR4		Yes	3	Prot-imm	A		[71]
	C18–A1	AGSLQPLALEGSLQKRG	?		Yes	1		D		[72]
	C19–A3	GSLQPLALEGSLQKRGIV	DR4		Yes	1		B	Eluted from MHC	[6]
	C22–A5	QPLALEGSLQKRGIVEQ	DR4		Yes	1		B	Eluted from MHC	[6]
	C23–A6	PLALEGSLQKRGIVEQC	?		Yes	3		D		[69]
	C24–A7	LALEGSLQKRGIVEQCC	?		Yes	3		D		[76]
	C28–A11	GSLQKRGIVEQCCTSIC	?		Yes	1		D		[72]
	C29–A12	SLQKRGIVEQCCTSICS	DR4				Prot-imm	C		[71]
	A1–13	GIVEQCCTSICSL	DR4	Yes		1		C		[77]
	A1–15	GIVEQCCTSICSLYQ	DR4	Yes		3		D	T cells from pancreatic lymph nodes	[35]
	A1–15	GIVEQCCTSICSLYQ	?	Yes		3		D		[74]
	A1–16	GIVEQCCTSICSLYQL	?		Yes	3		D		[69]
	A6–21	CCTSICSLYQLENYCN	?		Yes	1		D		[72]

^aGAD, glutamic acid decarboxylase; HSP, heat shock protein; IA-2, insulinoma-associated protein 2; ICA, islet cell autoantigen; IGRP, islet-specific glucose-6-phosphatase catalytic subunit-related protein.

^bPept-imm, peptide-immunized; Prot-imm, protein-immunized.

^cA, T cell clone, line, or hybridoma [human or human leucocyte antigen (HLA) Tg mouse] exists that responds to peptide and whole protein; human peripheral blood mononuclear cells (PBMCs) respond. B, other evidence of epitope presentation from whole protein (e.g. elution); human PBMCs respond. C, T cell clone, line, or hybridoma (human or HLA Tg mouse) exists that responds to peptide and whole protein; no positive human PBMC data. D, T cell clone, line, or hybridoma (human or HLA Tg mouse) or human PBMCs or T cells from HLA Tg mice respond to peptide; no evidence of epitope presentation from whole protein.

^d1: Responses from cases were increased compared to normal controls or differed qualitatively from control responses. 2: Cases and normal controls responded similarly. 3: No normal controls were examined, or differences between cases and controls were unclear. MHC: major histocompatibility complex.

Table 4.

CD8 T cell epitopes for mouse β cell antigens

				Type of T cell responseb
Aga	Position	Sequence	MHC	Clone, line or hybridoma	Islet T cells	Spleen or lymph nodes	Level of evidencec	Comments	Ref.
DMK	138–146	FQDENYLYL	Db	Spont	Spont		A		[98]
GAD65	206–214	TYEIAPVFV	Kd			Spont	A		[99]
	507–516	WFVPPSLRTL	Kd	Pept-imm		Pept-imm	C		[100]
	546–554	SYQPLGDKV	Kd			Spont	A		[99]
GAD67	515–524	WYIPQSLRGV	Kd	Pept-imm		Pept-imm	D		[101]
IGRP	21–29	TYYGFLNFM	Kd		Spont		B		[24]
	206–214	VYLKTNVFL	Kd	Spont	Spont		A	Eluted from MHC	[27]
	225–233	LRLFGIDLL	Db		Spont		B		[24]
	241–249	KWCANPDWI	Db		Spont		B		[24]
	324–332	SFCKSASIP	Kd		Spont		B		[24]
Insulin 1/2	B15–23	LYLVCGERG	Kd	Spont	Spont		A	Ag identified by expression cloning	[102]
Insulin 2	B25–C2	FYTPMSRREV	Kd			Pept-imm	D		[103]

^aDMK, dystrophia myotonica kinase; GAD, glutamic acid decarboxylase; IGRP, islet-specific glucose-6-phosphatase catalytic subunit-related protein.

^bPept-imm, peptide-immunized; Spont, spontaneous.

^cA: Evidence of epitope presentation from whole protein (e.g. T cell clone, line, or hybridoma exists that responds to peptide and whole protein; elution); spontaneous T cell response. B: No evidence of epitope presentation from whole protein; spontaneous T cell response. C: Evidence of epitope presentation from whole protein; no positive spontaneous T cell response data. D: No evidence of epitope presentation from whole protein; no positive spontaneous T cell response data. MHC: major histocompatibility complex.

Table 2.

CD4 T cell epitopes for mouse β cell antigens

				Type of T cell responseb
Aga	Position	Sequence	MHC	Clone, line or hybridoma	Islet T cells	Spleen or lymph nodes	Level of evidencec	Ref.
GAD65	202–221	TNMFTYEIAPVFVLLEYVTL	Ag7			Spont; prot-imm; pept-imm	B	[78]
	206–220	TYEIAPVFVLLEYVT	Ag7	Spont; prot-imm		Pept-imm	A	[79,80]
	217–236	EYVTLKKMREIIGWPGGSGD	Ag7			Spont; prot-imm; pept-imm	B	[78]
	221–235	LKKMREIIGWPGGSG	Ag7	Prot-imm			C	[80]
	247–266	NMYAMLIARYKMSPEVKEKG	Ag7			Spont	B	[81]
	286–300	KKGAAALGIGTDSVI	Ag7	Spont; prot-imm			A	[80]
	401–415	PLQCSALLVREEGLM	Ag7	Prot-imm			C	[80]
	509–528	VPPSLRTLEDNEERMSRLSK	Ag7			Spont	B	[81]
	524–543	SRLSKVAPVIKARMMEYGTT	Ag7	Prot-imm		Spont; pept-imm	A	[30,79,81]
	561–575	ISNPAATHQDIDFLI	Ag7	Prot-imm			C	[80]
	571–585	IDFLIEEIERLGQDL	Ag7	Prot-imm			C	[82]
GAD67	29–48	DTWCGVAHGCTRKLGLKICG	Ag7			Spont; pept-imm	B	[78]
	44–62	LKICGFLQRTNSLEEKSRL	Ag7			Spont; pept-imm	B	[78]
HSP60	76–95	DGVTVAKSIDLKDKYKNIGA	Ag7			Pept-imm	D	[83]
	166–185	EEIAQVATISANGDKDIGNI	Ag7			Pept-imm	D	[83]
	195–214	RKGVITVKDGKTLNDELEII	Ag7			Pept-imm	D	[83]
	361–380	KGDKAHIEKRIQEITEQLDI	Ag7			Pept-imm	D	[83]
	437–460	VLGGGCALLRCIPALDSLKPANED	Ag7	Prot-imm		Spont	A	[84]
	526–545	RTALLDAAGVASLLTTAEAV	Ag7			Pept-imm	D	[83]
	541–560	TAEAVVTEIPKEEKDPGMGA	Ag7			Pept-imm	D	[83]
IA-2	676–693	PSWCEEPAQANMDISTGH	Ag7			Pept-imm	D	[63]
	691–708	TGHMILAYMEDHLRNRDR	Ag7			Pept-imm	D	[63]
	706–723	RDRLAKEWQALCAYQAEP	Ag7			Pept-imm	D	[63]
	751–768	IKLKVESSPSRSDYINAS	Ag7			Pept-imm	D	[63]
	766–783	NASPIIEHDPRMPAYIAT	Ag7			Pept-imm	D	[63]
	781–798	IATQGPLSHTIADFWQMV	Ag7			Pept-imm	D	[63]
	961–979	FALTAVAEEVNAILKALPQ	Ag7			Pept-imm	D	[63]
IA-2β	640–659	KLSGLGADPSADATEAYQEL	Ag7	Prot-imm		Pept-imm	C	[85]
	755–777	QREENAPKNRSLAVLTYDHASRI	Ag7	Prot-imm		Spont; pept-imm	A	[85]
ICA69	36–47	AFIKATGKKEDE	Ag7			Spont; prot-imm; pept-imm	B	[86]
IGRP	4–22	LHRSGVLIIHHLQEDYRTY	Ag7			Spont; pept-imm	B	[87]
	123–145	WYVMVTAALSYTISRMEESSVTL	Ag7			Spont; pept-imm	B	[87]
	128–145	TAALSYTISRMEESSVTL	Ag7			Spont; pept-imm	B	[87]
	195–214	HTPGVHMASLSVYLKTNVFL	Ag7			Spont; pept-imm	B	[87]
Insulin 1/2	A7–21	CTSICSLYQLENYCN	Ag7	Spont	Spont		A	[88]
Insulin 1	L7–23	FLPLLALLALWEPKPTQ	Ag7			Pept-imm	D	[89]
	L20–B11	KPTQAFVKQHLCGPHL	Ag7			Pept-imm	D	[89]
	B9–23	PHLVEALYLVCGERG	Ag7	Spont	Spont	Pept-imm	A	[89,90]
	C15–30	SPGDLQTLALEVARQK	Ag7	Spont	Spont	Pept-imm	B	[89]
	C21–A5	TLALEVARQKRGIVDQ	Ag7			Pept-imm	D	[89]
Insulin 2	L14–B6	LFLWESHPTQAFVKQHL	Ag7			Pept-imm	D	[89]
	L20–B11	HPTQAFVKQHLCGSHL	Ag7			Pept-imm	D	[91]
	B2–17	VKQHLCGSHLVEALYL	Ag7	Spont	Spont		B	[89]
	B9–23	SHLVEALYLVCGERG	Ag7	Spont	Spont	Pept-imm	A	[89,90]
	B24–C1	FFYTPMSRRE	Ag7			Spont; prot-imm	B	[92]
	C15–32	GPGAGDLQTLALEVAQQK	Ag7			Pept-imm	D	[89]

^aGAD, glutamic acid decarboxylase; HSP, heat shock protein; IA-2, insulinoma-associated protein 2; ICA, islet cell autoantigen; IGRP, islet-specific glucose-6-phosphatase catalytic subunit-related protein.

^bPept-imm, peptide-immunized; Prot-imm, protein-immunized; Spont, spontaneous.

^cA: Evidence of epitope presentation from whole protein (e.g. T cell clone, line, or hybridoma exists that responds to peptide and whole protein; elution); spontaneous T cell response. B: No evidence of epitope presentation from whole protein; spontaneous T cell response. C: Evidence of epitope presentation from whole protein; no positive spontaneous T cell response data. D: No evidence of epitope presentation from whole protein; no positive spontaneous T cell response data. MHC: major histocompatibility complex.

Table 3.

CD8 T cell epitopes for human β cell antigens

				Type of T cell response
				Human
Aga	Position	Sequence	MHC	Clone or line	PBMC	Controlsd	HLA Tg miceb	Level of evidencec	Comments	Ref.
GAD65	114–123	VMNILLQYVV	A2	Yes		1		C		[93]
IA-2	797–805	MVWESGCTV	A2	Yes		2		D		[94]
IAPP (prepro)	5–13	KLQVFLIVL	A2		Yes	1		D		[95]
IGRP	265–273	VLFGLGFAI	A2				Spont	D		[96]
Insulin	B9–18	SHLVEALYLV	A2				Pept-imm	C		[14]
	B10–18	HLVEALYLV	A2				Pept-imm	C		[14]
	B10–18	HLVEALYLV	A2		Yes	1		B	Generated by proteasome	[15]
	B10–18	HLVEALYLV	A2	Yes	Yes	3		B	Generated by proteasome; translocated by TAP; response correlated with islet graft failure	[9]
	B14–22	ALYLVCGER	A3, A11		Yes	3		B	Precursor generated by proteasome	[15]
	B15–23	LYLVCGERG	A24		Yes	1		D		[97]
	B15–24	LYLVCGERGF	A24		Yes	3		B	Precursor generated by proteasome	[15]
	B17–26	LVCGERGFFY	A1, A3, A11		Yes	1		B	Precursor generated by proteasome	[15]
	B18–27	VCGERGFFYT	A1, A2, B8, B18		Yes	1		B	Precursor generated by proteasome	[15]
	B20–27	GERGFFYT	A1, B8		Yes	1		B	Generated by proteasome	[15]
	B21–29	ERGFFYTPK	A3		Yes	3		B	Precursor generated by proteasome	[15]
	B25–C1	FYTPKTRRE	B8		Yes	1		B	Generated by proteasome	[15]
	B27–C5	TPKTRREAEDL	B8		Yes	2		B	Precursor generated by proteasome	[15]
	C20–28	SLQPLALEG	A2				Pept-imm	C		[14]
	C25–33	ALEGSLQKR	A2				Pept-imm	D		[14]
	C29–A5	SLQKRGIVEQ	A2				Pept-imm	C		[14]
	A1–10	GIVEQCCTSI	A2				Pept-imm	C		[14]
	A12–20	SLYQLENYC	A2				Pept-imm	C		[14]

^aGAD, glutamic acid decarboxylase; IA-2, insulinoma-associated protein 2; IAPP, islet amyloid polypeptide; IGRP, islet-specific glucose-6-phosphatase catalytic subunit-relatedprotein.

^bPept-imm, peptide-immunized; Spont, spontaneous.

^cA: T cell clone, line, or hybridoma [human or human leucocyte antigen (HLA) Tg mouse] exists that responds to peptide and whole protein; human peripheral blood mononuclear cells (PBMCs) respond. B: Other evidence of epitope presentation from whole protein (e.g. elution); human PBMCs respond. C: T cell clone, line, or hybridoma (human or HLA Tg mouse) exists that responds to peptide and whole protein; no positive human PBMC data. D: T cell clone, line, or hybridoma (human or HLA Tg mouse) or human PBMCs or T cells from HLA Tg mice respond to peptide; no evidence of epitope presentation from whole protein.

^d1: Responses from cases were increased compared to normal controls or differed qualitatively from control responses. 2: Cases and normal controls responded similarly. 3: No normal controls were examined, or differences between cases and controls were unclear. MHC: major histocompatibility complex.

Our registry also provides information that allows the epitopes to be evaluated according to criteria that indicate the probability that a given epitope is important in the pathogenesis of type 1 diabetes. In the case of epitopes identified using standard or HLA-transgenic NOD mice, those shown to be recognized by islet-infiltrating T cells during the development of spontaneous type 1 diabetes in the mice are most likely to be disease-relevant, while reactivities detected only in response to peptide immunization may be less so. For the epitopes of human β cell antigens, peptides recognized preferentially or differentially in type 1 diabetes patients and/or at-risk individuals compared to normal controls are the best candidates for disease-relevant epitopes that could ultimately be exploited to monitor autoimmune activity in at-risk individuals or patients undergoing intervention therapies. In addition, for both the murine and human systems, evidence that a given epitope is naturally processed and presented (e.g. elution from MHC, proteasome processing or the existence of a T cell clone that recognizes both the peptide and whole protein) further increases the probability that the epitope is a relevant one. This is the first time that islet cell epitopes have been documented with this level of stringency, and we believe that this approach also represents a first among the organ-specific autoimmune diseases.

CD4 T cell epitopes in mouse and man

Tables 1 and 2 show the CD4 T cell epitopes identified for human and mouse islet cell autoantigens. The relative contribution of the different islet cell autoantigens to the epitope lists is depicted graphically in Fig. 1a. It is noteworthy that for CD4 epitopes, the relative contributions of GAD65, proinsulin and IA-2 are similar in man and mouse, although over three times as many epitopes have been reported in man. It is possible that the greater number of epitopes identified in man represents a more diverse MHC class II background compared with the single MHC class II molecule present in the NOD mouse.

Fig. 1.

Pie charts showing the antigenic distribution of (a) CD4 and (b) CD8 T cell epitopes in autoimmune diabetes in humans and mice. Data used are from Tables 1–4. For the sake of clarity, the single epitope of ICA69 is not shown in the pie charts in (a).

CD8 T cell epitopes in mouse and man

In contrast with CD4, for CD8 T cells the major contributor to epitopes recognized in the mouse is islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP), and in man is proinsulin/insulin (Tables 3 and 4; Fig. 1b). The latter phenomenon most probably reflects the influence of two recent publications in which large numbers of potential proinsulin/insulin epitopes were identified using approaches that lend themselves to large-scale epitope identification (proteasomal digestion of whole protein and scanning proteins in silico using binding algorithms), as well as the development and exploitation of HLA-A2 transgenic mice [14,15]. It is also noteworthy that there is a distinct lack of identification of CD8 T cell epitopes of IA-2, despite the fact that, in man at least, there is a high prevalence of CD4 T cell responses to this autoantigen (Figs 1a and 2), as well as the known importance of IA-2 autoantibodies in heralding the imminent development of the disease. It is likely that in the coming months and years these apparent species-based differences will balance out, and the pie charts for CD8 will come to resemble the harmonized appearance seen for CD4 epitopes. For example, there is now a considerable effort being expended on the search for IGRP epitopes in man, given the demonstration of their importance in NOD mice. This is also a good example of the use of animal models in instructing research into human disease.

Fig. 2.

Representation of linear sequence and location of CD4 T cell epitopes of the major islet autoantigens preproinsulin, GAD65, and IA-2 in human type 1 diabetes. GAD65 and IA-2 are drawn to the same scale, whereas preproinsulin is enlarged three-fold for clarity. Epitopes shown are from Table 1. Where data exist that show an epitope to be unambiguously restricted by a particular HLA class II molecule, this is shown with shading, where black = HLA-DQ8; diagonal stripes = HLA-DR3 (DRB1*0301), and grey = HLA-DR4 (DRB1*0401). All other epitopes are shown as white boxes. TM = transmembrane.

Evaluating the importance of specific antigens in the pathogenesis of type 1 diabetes

An autoantigen could reasonably be termed important, and even essential, if it could be shown that genetic ablation of its expression led to protection from disease. Fulfilment of this very stringent criterion has been achieved for murine preproinsulin 1 [16]. More recently, the presence of the CD4 T cell epitope InsB₉₋₂₃ (which also contains a CD8 T cell epitope) in particular has been reported to be required for the development of islet autoimmunity in NOD mice [17]. In contrast, NOD mice deficient in expression of IA-2 [18,19], IA-2β[19,20], ICA69 [21] or GAD65 [22,23] are not protected from disease, indicating that these antigens are not essential for disease development. We would argue, however, that such results do not necessarily imply that T cell responses to these antigens play no role in β cell elimination. Rather, in their absence, other specificities may more efficiently compete (e.g. at the level of the antigen-presenting cell) and fill the niche occupied normally by T cells specific for the now-ablated antigens. For example, when CD8 T cells specific for IGRP_206−214/H-2K^d (which, as discussed below, is a disease-relevant epitope by several criteria) were completely depleted by high-dose treatment of NOD mice with an altered peptide ligand, disease occurred none the less, and the expansion of clonotypes specific for subdominant epitopes was observed [24].

A number of other strategies have been employed to address the issue of the importance of particular β cell autoantigens in the pathogenesis of type 1 diabetes in NOD mice. The ability of adoptively transferred T cell clones to accelerate diabetes in young NOD mice or to cause disease in NOD-scid mice provides evidence for the importance of the corresponding antigenic specificity. For example, adoptive transfer studies have demonstrated the pathogenicity of the CD8 T cell clone 8·3 [25,26], which is specific for IGRP_206−214/H-2K^d[27]. Acceleration of disease in T cell receptor (TCR) transgenic mice is another indication of the pathogenicity of a particular specificity. Consistent with the adoptive transfer studies, NOD mice transgenic for the 8·3 TCR show accelerated disease [28]. Further evidence for the importance of IGRP in the development of diabetes in NOD mice comes from the finding that quantification of IGRP-reactive T cells in the peripheral blood can be used to predict which individual NOD mice will go on to develop disease [29].

For GAD65, adoptive transfer and TCR transgenic mouse studies are in apparent disagreement. A GAD65-reactive CD4 T cell line was capable of inducing diabetes in NOD-scid mice [30]. However, mice transgenic for a class II MHC-restricted TCR specific for either GAD65_286−300 or GAD65_206−220 were protected from disease [31,32], suggesting that at least certain GAD65-reactive CD4 T cell clonotypes may play a beneficial regulatory role. It should be noted here that two InsB₉₋₂₃-reactive TCR transgenic mice have been described, one that develops disease [33] and one that does not [34]. These results support the notion that both pathogenic and non-pathogenic GAD65-reactive T cell clonotypes also exist.

For the most part, the types of experiments used to evaluate the importance of specific antigens in the development of diabetes in NOD mice cannot be applied to patients. In this case, quantitative or qualitative differences in T cell responses between patients or at-risk individuals and normal controls are the only indicator of disease-relevant antigens. This information is provided in Tables 1 and 3. However, it must be acknowledged that peripheral blood is the only T cell source that has been monitored routinely in humans, and the argument could always be made that T cells in the islets or the pancreatic lymph nodes might represent more accurately truly disease-relevant specificities. To this end, CD4 T cells from the pancreatic lymph nodes of type 1 diabetes patients have been examined recently [35]. Studies in NOD mice suggest the possibility of antigen-specific imaging of islet-infiltrating T cells [36], which may one day allow non-invasive examination of islet-infiltrating T cells in humans and a novel strategy to assess the disease relevance of particular β cell antigens.

Avoiding ‘tunnel vision’

It is, of course, entirely conceivable that β cell proteins other than those appearing in Tables 1–4 serve as targets in the autoimmune destruction in type 1 diabetes. Such candidates include imogen38, fetal antigen 1 (or Pref-1), and as yet unidentified components of β cells and their secretory granules [37–39]. By no means should the present compilation be regarded as comprehensive or fully representative. It merely reflects the current knowledge of proven and published epitopes as well as, most probably, a biased focusing on proinsulin, IA-2 and GAD65. It is also possible that the current selection is biased to those peptides displaying high binding affinity to HLA class I or II restriction elements. Although it remains to be demonstrated that binding affinity correlates with T cell recognition and possible autoimmune disease, it is remarkable that the majority of CD4 epitopes described to be recognized by T1D patients display high binding affinity to HLA [6,7]. In the case of HLA class I, epitope discovery has often been led by high binding affinity to the respective HLA class I molecules and as such could reflect a bias.

The exploitation of epitope discovery

The epitope lists contained in this review reflect a conjunction of human and animal model-based research, and as such have several potential utilities. In particular we anticipate that peptides based on these sequences could be used in the design of assays to detect T cell autoreactivity. Such assays are increasingly being deployed to provide secondary measures of clinical efficacy [40], as well as mechanistic insights, during intervention and prevention trials for type 1 diabetes. Some assays, of course, are entirely reliant upon the discovery of robust T cell epitopes, for example MHC-multimer-based technologies, as well as those that function optimally when peptides rather than whole proteins are used as stimuli. Recent mouse and human studies indicate that peptides targeted by CD8 T cells in new-onset and graft-recurrent type 1 diabetes overlap [15,41]. Thus, assays to monitor autoimmune activity in at-risk individuals and patients might also be applicable to the monitoring of islet graft rejection. A further use for epitope discovery may be in the design of novel intervention strategies. It has long been argued that antigen-specific immunotherapies for autoimmune diseases offer the best hope for the development or re-establishment of tolerance to islet autoantigens, and some of these may be peptide-based strategies [42]. Examples of the successful use of peptide immunotherapy in the prevention or treatment of type 1 diabetes in NOD mice are presented in Table 5.

Table 5.

Examples of the use of peptide immunotherapy in the prevention or treatment of type 1 diabetes in NOD mice

Aga	Position	Sequence	MHC	Treatment	Adjuvant	Outcome	Ref.
GAD65	247–266	NMYAMLIARYKMSPEVKEKG	Ag7	50 µg of each peptide	None	Mice followed to 52 weeks	[104]
	509–528	VPPSLRTLEDNEERMSRLSK	Ag7	in a mixture administered intranasally		of age; decreased incidence
	524–543	SRLSKVAPVIKARMMEYGTT	Ag7	at 2–3 weeks of age (single dose)		of disease with treatment
	539–558	EYGTTMVSYQPLGDKVNFFR	Ag7
HSP60	437–460	VLGGGCALLRCIPALDSLKPANED	Ag7	50 µg administered subcutaneously after the appearance of hyperglycaemia (single dose)	Incomplete Freund's	Mice followed to 40 weeks of age; increased survival with treatment	[105]
IGRP	206–214	KYNKANAFLb	Kd	100 µg administered intraperitoneally every 2–3 weeks beginning at 3 weeks of age	None	Mice followed to 32 weeks of age; decreased incidence of disease with treatment	[106]
Insulin 2	B9–23	SHLVEALYLVCGERG	Ag7	40 µg administered intranasally for 3 consecutive days every 4–5 weeks beginning at 4 weeks of age	None	Mice followed to 32 weeks of age; decreased incidence of disease with treatment	[88]

^aGAD, glutamic acid decarboxylase; HSP, heat shock protein; IGRP, islet-specific glucose-6-phosphatase catalytic subunit-related protein.

^bMimotope peptide; changes from natural peptide (VYLKTNVFL) are denoted by underlining. MHC: major histocompatibility complex.

It is our hope that this annotated T cell epitope compilation will prevent duplication of effort, while at the same time encouraging verification, validation and wider use of the most promising epitopes identified to date. Antigen discovery efforts ongoing in our laboratories and others will probably reveal additional candidate peptides to be evaluated in the future.

Note added in proof

After submission of our article, glial fibrillary acidic protein (GFAP) 143–151 (NLAQDLATV), GFAP 192–200 (SLEEEIRFL), GFAP 214–222 (QLARQQVHV), IAPP 5–13 (KLQVFLIVL), IAPP 9–17 (FLIVLSVAL), IGRP 152–160 (FLWSVFWLI), IGRP 215–223 (FLFAVGFYL), and IGRP 293–301 (RLLCALTSL) were reported to be recognized in the context of HLA-A2 by PBMC from T1D patients and/or at-risk individuals (Standifer NE, Ouyang Q, Panagiotopoulos C, Verchere CB, Tan R, Greenbaum CJ, Pihoker C, Nepom GT. Identification of novel HLA-A*0201-restricted epitopes in recent-onset type 1 diabetic subjects and antibody-positive relatives. Diabetes 2006; 55:3061–7), as were IA-2 172–180 (SLSPLQAEL), IA-2 482–490 (SLAAGVKLL), IAPP 5–13 (KLQVFLIVL), insulin L2–10 (ALWMRLLPL), and insulin B10–18 (HLVEALYLV) (Ouyang Q, Standifer NE, Qin H, Gottlieb P, Verchere CB, Nepom GT, Tan R, Panagiotopoulos C. Recognition of HLA class I-restricted β-cell epitopes in type 1 diabetes. Diabetes 2006; 55:3068–74).