Specificity of peptide selection by antigen-presenting cells homozygous or heterozygous for expression of class II MHC molecules: The lack of competition

Abstract

We isolated and identified naturally processed peptides selected by antigen-presenting cells homozygous for expression of I-A^g7 or I-A^d class II MHC molecules, or from heterozygous antigen-presenting cells that express I-A^g7 along with I-A^g7PD or I-A^d. Identification of large numbers of peptides demonstrated that despite being closely related on a structural level, each class II MHC molecule selected for very unique peptides. The large data sets allowed us to definitively establish the preferred peptide-binding motifs critical for selection of peptides by I-A^g7, I-A^g7PD, and I-A^d. Finally, extensive analyses of peptide families reveals that there was little competition among class II MHC alleles for display of peptides and that presence of one allele had minimal impact on the repertoire of peptides selected by another.

The onset of many organ-specific autoimmune disorders is linked to the expression of certain class II MHC alleles. The class II MHC molecule I-A^g7 is expressed by the nonobese diabetic (NOD) mouse that spontaneously develops autoimmune type 1 diabetes mellitus (T1DM) (1). In particular, both I-A^g7 and the human T1DM-susceptible class II MHC alleles (such as HLA DQ2 and DQ8) share biochemical and structural features, most prominent among which is the presence of a non-Asp residue at position 57 of the β chain (2–4). According to structural studies, the P9 pocket of I-A^g7, composed of the crucial β57 residue (Ser), is wider and more open toward the C terminus than in other class II MHC alleles that express the conserved Asp at β57 (5, 6). Similar features were also described for the human T1DM-susceptible class II MHC, HLA-DQ8 (7).

A recent report from our group identified a large number of naturally processed peptides selected by antigen-presenting cells (APCs) expressing either wild type I-A^g7 or the mutant I-A^g7PD (wherein the β57 Ser was changed to the conserved Asp). Two important findings emerged: first, the nature of the P9 pocket was central in determining the specificity of peptides selected by I-A^g7, and second, peptides selected by I-A^g7 showed unique sequence motifs (8). In fact, based on the analysis of the naturally selected peptides we argue that there is a high degree of specificity, an issue that has been obscured by binding analyses with synthetic or artificial peptide libraries.

In the NOD model of autoimmune diabetes, numerous past reports have indicated that MHC heterozygosity results in protection from disease (9–14). In fact, the presence of any other β57 Asp-expressing class II MHC, in addition to I-A^g7, results in markedly decreased incidence of diabetes; for example is the result of F1 mice derived from mating of wild type NOD (I-A^g7) to NOD.GD (I-A^d), which are protected from diabetes (15). Moreover, in the case of the human disease where MHC heterozygosity is widely prevalent, the presence of certain class II MHC alleles (such as DQB1*0602) among DQ8-expressing susceptible individuals imparts protection from onset of diabetes (16–18). The primary mechanism(s) underlying this protective effect of MHC heterozygosity are not known, although three primary possibilities have been proposed: (i) the presence of additional class II MHC alleles may alter the thymic environment for selection of CD4⁺ T cells such that self-reactive T cells are efficiently deleted (19–21); (ii) additional class II MHC alleles may select for regulatory T cells that may impact on islet-reactive T cells (10, 22, 23); and (iii) other class II MHC molecules may alter the repertoire of peptides selected by I-A^g7, referred to as “determinant capture” (24, 25).

For our current studies, we examined whether the presence of other class II MHC alleles affects the repertoire of peptides sampled by I-A^g7, as well the degree of specificity of peptide selection exhibited by closely related class II MHC alleles. We generated heterozygous APC lines expressing I-A^g7 and I-A^d, or I-A^g7 and I-A^g7PD, and compared the class II MHC-associated peptides from them to those isolated from APCs expressing a single MHC allele. Results from these experiments indicate that the presence of additional class II MHC molecules that express the conserved β57 Asp residue has little or no impact on the repertoire of peptides selected by I-A^g7, and vice versa. Moreover, we find that peptides bound to each of these different class II molecules are unique.

Materials and Methods

Cell Lines and Antibodies.

Lipopolysaccharide-activated spleen cells from NOD mice (I-A^g7+), NOD.PD mice (I-A^g7+ and I-A^g7PD+), and NODxNOD.GD (H-2^g2: K^d+, D^b+, I-A^d+, I-E^null) F₁ mice were fused to the M12.C3 B cell lymphoma line to generate NOD.C3 (I-A^g7+), NOD.PD.C3 (I-A^g7+ and I-A^g7PD+), NOD.GD.C3 (I-A^g7+ and I-A^d+), and NOD.GD.C3-D (I-A^d+) cell lines, respectively (Table 1). For simplicity, we refer to them as g7+, g7+PD+, g7+d+, and d+, respectively. The APC lines were cultured in DMEM supplemented with 10% calf serum. Monoclonal antibodies (mAbs) AG2.42.7 (anti-I-A^g7), AG2.27.5 (anti-I-A^g7 and I-A^g7PD), and MKD6 (anti-I-A^d) were used to immunoprecipitate peptide–MHC complexes.

Table 1.

Class II MHC molecules expressed by the different APC lines

Cell line	Class II MHC	α chain	β chain	β56	β57
NOD.C3 (g7+)	I-Ag7	d	g7	His	Ser
NOD.PD.C3 (g7+PD+)	I-Ag7	d	g7	His	Ser
	I-Ag7PD	d	g7PD	Pro	Asp
NOD.GD.C3 (g7+d+)	I-Ag7	d	g7	His	Ser
	I-Ad	d	d	Pro	Asp
NOD.GD.C3-D (d+)	I-Ad	d	d	Pro	Asp
C3G7PD (PD+)	I-Ag7PD	d	g7PD	Pro	Asp

Purification of Naturally Processed Peptides.

The g7+, g7+PD+, g7+d+ and d+ lymphoma cells (5–12 × 10⁹) were lysed in PBS containing 40 mM MEGA 8 and MEGA 9 detergents, in the presence of protease inhibitors (1 mM PMSF/10 mM iodoacetamide/20 μg/ml leupeptin). After 1 h at 4°C, the lysate was spun in a Sorvall RC5C centrifuge at 10,000 × g for 30 min and the class II MHC molecules were purified by using AG2.42.7 mAb (for I-A^g7), AG2.275.5 (for I-A^g7PD) or MKD6 mAb (for I-A^d) coupled to cyanogen bromide-activated Sepharose beads (Sigma). (In the case of the NOD.PD.C3 cell line, the lysate was passed through AG2.42.7 beads three times to isolate and preclear I-A^g7–peptide complexes after which the eluent was subsequently incubated with AG2.27.5 beads to precipitate I-A^g7PD–peptide complexes.) Sepharose beads were loaded into disposable chromatography columns (Bio-Rad) and washed first with 40 ml of wash buffer 1 (10 mM MEGA 8 and MEGA 9 in PBS) and 40 ml of wash buffer 2 (2.5 mM MEGA 8 and MEGA 9 in PBS), and then extensively with PBS (100 ml) and milliQ H₂O (100 ml). I-A^g7–, I-A^g7PD–, or I-A^d–peptide complexes were eluted with 15 ml of 0.1% trifluoroacetic acid (pH 1.9). The eluted peptides were separated from MHC molecules by using centriprep YM-10 (MWCO, 10 kDa) membrane separation. Peptide extracts were then analyzed by RP-HPLC coupled with tandem MS (MS/MS) to identify the amino acid sequences of the naturally processed peptides isolated from I-A^g7, I-A^g7PD, and I-A^d.

RP-HPLC and MS/MS Analysis.

Peptide preparations were desalted and further purified by using three ZipTip pipette tips containing C18 reverse-phase media (Millipore, Bedford, MA). The samples from each tip were eluted by using 40 μl of 50:50 acetonitrile/H₂O, in 0.1% trifluoroacetic acid. The final 120-μl volume was dried, and the pellet was resuspended in 45 μl of solvent A (3% acetonitrile/97% H₂O/0.1% formic acid). Samples were analyzed by on-line RP-HPLC-MS and MS/MS. Sample (6 μl) was loaded onto a Delta-Pak (Waters, Bedford, MA) C18 0.075 × 100 mm, 5-μm, 300-Å capillary column (New Objective, Woburn, MA) by using a Waters CapLC. The gradient was from 0% solvent B (97% acetonitrile/3% H₂O/0.1% formic acid) to 5% solvent B over 3 min, then to 50% solvent B over 70 min. Eluent flow was ≈6 μl/min and split before the column at a 1:30 ratio to maintain a flow rate at the PicoFrit tip of 280 nl/min. Flow was directed into the entrance of the heated capillary of a LCQ-Deca quadrupole ion-trap mass spectrometer (Thermo-Finnigan, San Jose, CA) equipped with a custom-built nanospray source.

MS and dependent-scan mode MS/MS were performed by using the spectrometer and the control XCALIBUR 1.1 software. For MS, the scan range was m/z 600–1,400 in the profile mode. Every three “microscans” were averaged to give one scan. Acquisition was started 15 min after the beginning of the LC run. For MS/MS, a scan range of m/z 600-1,400 in centroid mode was used for the MS, and the MS/MS range was from 30% of the m/z of the parent ion to m/z 2,000. The isotope cluster of parent ions were dynamically selected, and isolated with a 2.2 m/z window. The collision energy was set to 28% of the maximum, which is ≈5 eV. For detection of peptides, two major strategies were used. In the first, the spectrometer selected peptides based on their signal intensity. In such experiments, the first and second most abundant ions were analyzed in the first run, after which the third and fourth most abundant ions were analyzed in another subsequent run. These peptides, which were submitted to collision-induced dissociation for sequencing, were often found associated with the same MHC molecule from different cell lines (see Tables 4–6, which are published as supporting information on the PNAS web site, www.pnas.org).

The second approach was designed to give more control to the selection of the precursor peptide ions and complemented the first approach. The peptides present in one cell line were searched according to the m/z of the doubly charged molecular ions and their precise retention time(s) among the peptides isolated from another cell line. This approach overcame the possibility of missing one peptide from one line; for example, a peptide from the IL-21 receptor (351–366) was sequenced among I-A^g7 peptides from g7+ cell line but not from g7+PD+ or g7+d+ cell lines. However, searching for this peptide's presence by the m/z of its [M + 2H]⁺ and corresponding retention time in the chromatograms of I-A^g7 peptides from g7+PD+ or g7+d+ revealed this family in both samples (Table 4).

Product-ion spectra were analyzed, and peptide sequences were determined automatically by using sequest software provided by the instrument manufacturer. All of the automatically determined sequences were manually verified against the experimental product-ion spectra. The relative abundance for each peptide was estimated as described (8).

Results

Cell Lines.

Table 1 summarizes the APC lines and the MHC molecules expressed by them. Each line was derived from the same source so presumably should contain the same proteins for processing. Three lines express a single MHC allele: I-A^g7, I-A^g7PD, or I-A^d. Two lines express both I-A^g7 and I-A^g7PD, and I-A^g7 and I-A^d. It is important to note that each MHC set in the double expressing lines is represented at about the same level (Fig. 1). We previously analyzed peptides from I-A^g7 and I-A^g7PD and found that peptides selection was highly specific for each (8). The comparison made here between I-A^g7 and I-A^d is particularly important for as establishing the degree of specificity of the selected peptides. Both haplotypes share similar features: the same αd chain, and, from past structural analysis, similar P4 and P6 binding sites.

Figure 1.

Class II MHC expression among heterozygous APC lines. (A) NOD.GD.C3 APC line (I-A^g7 and I-A^d) stained with AG2.42.7 (anti-I-A^g7 mAb) and MKD6 (anti-I-A^d mAb). (B) NOD.C3 (I-A^g7 only) and NOD.PD.C3 (I-A^g7 and I-A^g7PD) APCs stained with AG2.42.7 (anti-I-A^g7, solid line) and AG2.27.5 (anti-I-A^g7 and I-A^g7PD, dotted line). (Note that NOD.PD.C3 stains 2-fold more with AG2.27.5 mAb, which recognizes both I-A^g7 and I-A^g7PD.) Isotype-matched antibodies were used as controls.

Peptides Isolated from I-A^g7 or I-A^d molecules.

We isolated and identified peptides from each of our APC lines. As described in previous studies, many of the peptides emerge as families in which all of the members share the same core sequence, along with varying lengths of flanking residues on the amino and carboxy termini (8, 26–29). (We define as “core” the 9-aa stretch from the P1 to the P9 binding sites or pockets. The variable flanking residues contribute to binding independent of the nature of the amino acid.) Peptides were derived from proteins in the extracellular fluids, membrane-bound surface proteins or cytosolic proteins.

We examined 74, 129, and 59 I-A^g7-bound peptides from 45, 57, and 24 different families from g7+, g7+d+, and g7+PD+ cell lines, respectively. (Table 4 indicates all of the peptide families that were identified from I-A^g7, and Fig. 2 shows the frequency of residues at each position in the core sequence, comparing it to those found from I-A^g7PD and I-A^d.) Peptides selected by I-A^g7 showed sequence features that distinguished them from peptides selected from I-A^g7PD or I-A^d (Fig. 2). The features of the I-A^g7 selected peptides were described before, and as expected, similar features were found in the same sets of peptides examined in the present study: some of the peptide families studied before were also found in the homozygous or double MHC-expressing lines. The most distinguishing feature of I-A^g7-bound peptides was the presence of acidic residues at the C terminus (8, 30). Interestingly, there are peptides that contained more than one acidic residue. In the previous report on a g7 line, we examined the binding of the peptides to purified I-A^g7 molecules and established that the acidic residues were interacting at the corresponding P9 pocket site (8). Concerning P4 and P6 residues, these were mainly represented by hydrophobic amino acids (such as Ile/Leu, Val, Ala), small-to-medium-sized polar residues (Ser, Thr, Asn, Gln), and Gly. The P1 residues included a variety of amino acids including charged (Arg/Lys or Asp/Glu), small hydrophobic (Ile/Leu, Val, Ala), and polar (Ser, Thr, Asn, Gln) (Fig. 2).

Figure 2.

Preferred anchor residues for I-A^g7, I-A^d, and I-A^g7PD. P1, P4, P6, and P9 residues were identified on the basis of the binding core of naturally processed peptide families, as shown in Tables 4–6.

Our recent report indicated that the peptides isolated from I-A^g7PD were distinct from and did not overlap with those isolated from I-A^g7 (8). A summary of their preferred residues in the nine-core segment is also indicated in Fig. 2. The P1 pocket of I-A^g7PD accommodated various amino acids, prominent among which were hydrophobic (Ile/Leu, Val, Ala, Gly) and polar (Gln/Asn) residues. The P4 and P6 pockets were the same as those in I-A^g7 and hence were dominated by similar anchor residues (i.e., mostly small hydrophobic amino acids such as Ile/Leu, Val, Ala, and Gly). The P9 pocket of I-A^g7PD contains β57 Asp that forms an ion pair with the opposing α76 Arg, and was mostly occupied by small hydrophobic (Ile/Leu, Val, Ala, Gly) or polar (Ser/Thr) residues. This, of course, is in sharp contrast to the acidic residue preference of the P9 pocket of wild-type I-A^g7 molecule (Fig. 2).

From I-A^d, we isolated and identified 86 and 88 peptides among 50 and 46 families from d+ and g7+d+ cell lines, respectively (Table 5). This number of peptides plus the six previously identified in an early study by Hunt et al. (29) provides us with a larger database to analyze the I-A^d binding motif. The I-A^d binding motif was proposed in the pioneering work of Gray, Sette, and colleagues, and by the subsequent structural analysis done by Wilson's group, using synthetic peptides (31–33). The binding studies identified a 6-mer-core peptide, now known from the structural analysis to encompass the amino acid residues from P4 to P9 (31, 34). The position of Gray's motif was established from the crystallographic analysis of I-A^d complexed to peptides derived either from ovalbumin (323–339) or influenza hemagglutinin (126–138) (33). We used this information to establish the core sequence of the isolated peptides. Fig. 2 indicates the preferred residues and indicates the differences from the peptides extracted from I-A^g7. Table 2 contains examples on how we determined the possible register for some of the naturally processed peptides.

Table 2.

Alignment of some naturally processed I-A^d peptides isolated from g7+d+ or d+ cell lines

						P1			P4		P6			P9
I-E α chain 52–67		A	S	F	E	A	Q	G	A	L	A	N	I	A	V	D	Q
Erp 29 169–181					E	F	I	K	A	S	S	I	E	A	R	Q	A
Chaperonin subunit-2 199–213			I	H	V	I	K	K	L	G	G	S	L	A	D	S	Y
α-d-mannosidase 310–326	N	Q	R	T	A	Q	F	G	I	S	V	Q	Y	A	T	L	N
β-Hexoaminidase β 131–148	Y	S	L	L	V	Q	E	P	V	A	V	L	K	A	N	S	V	W
α-1,2-mannosidase 122–135			G	V	I	E	A	F	L	H	A	W	K	G	Y	Q
α-1,2-mannosidase 350–365	E	T	Q	L	L	E	D	Y	V	K	A	I	E	G	I	K
Pre-prolegumain 109–124		D	V	T	P	E	N	F	L	A	V	L	R	G	D	A	E
Transferrin receptor 466–481	A	T	E	W	L	E	G	Y	L	S	S	L	H	L	K	A
Nucleobindin 258–272			V	L	D	E	Q	E	L	E	A	L	F	T	K	E	L
LAMP 35–47			L	F	I	E	H	V	V	E	V	A	R	G	K
CDEI-box binding protein 414–429			I	A	L	E	N	Y	L	A	A	L	Q	S	D	P	P	R

Naturally processed peptides presented by I-A^d, from g7+d+ and d+ APC lines, were aligned based on the motif predicted by previous crystal structure and binding studies (31–33). Note that the top five peptides agree well with the previously described motif; however, the bottom seven examples represent peptides that prefer an acidic amino acid at P1. The same set of peptides also can tolerate hydrophobic residues such as Gly or Leu at the P9 position, which in previous binding studies were shown to be unfavorable (32). 

Small hydrophobic residues, in agreement with the previous findings using synthetic peptides, almost exclusively represented the P4 and P6 residues (Table 2 and Fig. 2). The residue interacting at P4 appears to provide the major contribution to binding, fitting in well at the binding site of I-A^d. It is this residue that sets the binding register (33).

The results shown in Table 2 and Fig. 2 add to our understanding of the I-A^d binding-motif. First, a large fraction (>50%) of the peptides contained residues that were considered unfavorable to binding, on the basis of data from binding studies (32). For example, approximately half of all peptides have an “unfavorable” residue at the P9 position (notably Gly, Ile/Leu, Val, and Gln/Asn). As we indicated for I-A^k bound peptides, a single unfavorable or hindering residue can be tolerated, though it does lower the binding affinity (35). As noted here, some of these selected peptides are well represented, indicating that the putative unfavorable residues had little impact on their final display. Second, there was a definite preference for residues at P1, the largest pocket site of the I-A^d binding groove. Thus, ≈40% of the P1 residues were acidic and 25% were represented by Ile,Leu, Val, Ala, or Gly. From a structural perspective, the large size of the P1 pocket as well as the presence of a His residue at α24 may allow for favorable interactions with a negatively charged anchor residue.

Peptides Isolated from APC Expressing Two Different MHC Molecules.

Without exception, peptides that were expressed at intermediate or large levels on I-A^g7 or I-A^d were also found in the double MHC-expressing line. For those peptides expressed in low amounts, a few were found in one line but not in another. These peptides were identified at the lowest range of detection, at levels where it is difficult to be certain when examining a very complex mixture of peptides like that of a class II MHC extract. Table 3 indicates, for example, that of the total 90 peptide families isolated from I-A^g7, only seven were not found in the double- g7+d+ line. At the low levels of analysis, the absence of a peptide may not indicate lack of expression. Proof of this is the fact that of the combined 15 peptides not found in g7+d+ and g7+PD+, 13 were localized in either one or the other. The same findings hold true for the peptides isolated from I-A^d, where 11 were not found in the double expressing line g7+d+. Note also that some peptides were found in the double expressing lines and not in the homozygous.

Table 3.

Analyses of naturally processed peptide families isolated from homozygous and heterozygous APC lines

Peptides	n
I-Ag7 peptides (n = 90)
Peptides found in g7+ and absent in g7+d+	7
Peptides found in g7+ and absent in g7+PD+	8
Peptides found in g7+ and absent in both double lines	2
Peptides found in g7+PD+ and absent in g7+	1
Peptides found in g7+d+ and absent in g7+	1
I-Ad peptides (n = 85)
Peptides found in d+ and absent in g7+d+	11
Peptides found in g7+d+ and absent in d+	4

Note that the peptides marked as absent were present in their original source in low abundances; most were present in amounts that were <5% of the most abundant peptide. This made it unreliable for us to definitively establish their presence in a complex peptide extract (listed in detail in Tables 4 and 5). 

Discussion

Three main findings derive from these analyses. The first conclusion is that each haplotype selects for highly specific and defined repertoire of peptides. This issue started to become very evident from the initial examination of bulk peptides in the studies of Rammensee et al. (36). Although more difficult to appreciate among the class II peptides because of their size heterogeneity, the specific nature of peptide selection became very apparent with the identification of large numbers of naturally processed peptides by MS analysis and the mapping of the core segments of these peptides (26–29). Such specificity between peptides selected from I-A^g7 or I-A^d or I-A^g7PD may not have been expected based on the supposed lack of specificity or degeneracy of binding among haplotypes. Studies using synthetic peptides in in vitro binding assays indicate that class II MHC molecules can bind a largely overlapping repertoire of peptides with little or no specificity (37–39). That is to say, many peptides bind to different alleles. However, if one examines the natural selected peptides, a very different result becomes evident; the majority of the I-A^g7 peptides showed a very specific motif, distinctly different from those selected by I-A^d or I-A^g7PD. The same cell line could have provided autologous peptides to both, yet the selected population was highly specific for each MHC. Many peptides isolated from I-A^g7 were never found in I-A^g7PD (an issue confirmed here and extended for the I-A^d peptides). The important finding is that the peptides isolated from I-A^g7 and never found in I-A^g7PD could bind to each of the purified MHC molecules in vitro (8). Thus, the analysis of selected peptides and the analysis of binding of library of synthetic peptides give very different conclusions that need to be taken into perspective in immunogenic responses (8, 39).

A second contribution of these studies is to precise the peptide-binding motif for I-A^d favored for selection during protein processing. Both I-A^g7 and I-A^d share the same I-Aα^d chain and differ by 17 aa in their I-Aβ chain, which makes them very closely related. The structural features of I-A^g7 and I-A^d are alike especially with respect to the P4 and P6 pockets (5, 6, 33). Indeed, the analysis of naturally processed peptides indicates that the P4 and P6 pockets of both prefer small hydrophobic residues and disfavor charged or bulky residues. The P9 pocket of I-A^d also prefers small, hydrophobic residues including Gly, Val, and Ile/Leu, which previously were identified as unfavorable P9 anchor residues for I-A^d (32). On the other hand, the P9 position for I-A^g7 peptides is almost exclusively dominated by the presence of an acidic amino acid. Finally, the P1 pocket of I-A^d, which by earlier studies was deemed degenerate in specificity (33), is dominated by the presence of acidic amino acids and to a lesser extent by small hydrophobic residues.

As motifs for class II MHC molecules are determined by identification of naturally processed peptides (26–29), a prominent theme emerging revolves around the contribution of specific binding pockets in selection of epitopes. In our own studies, we have isolated and identified naturally processed peptides from three different murine class II MHC alleles, namely I-A^k, I-A^g7, and I-A^d. For each of these alleles, the major contribution for favorable binding derives from distinct allelic sites or pockets – in the case of I-A^k, an acidic residue at P1 contributes most of the binding strength (26, 40); for I-A^d, a small hydrophobic residue buried in the P4 pocket enhances binding (29, 31), where as for I-A^g7, an acidic in the P9 pocket as well at P10 or 11 favors binding (8, 41).

The third result is the finding of limited competition in peptides between two alleles expressed in the same APC. The concept of competition for epitope display among MHC alleles has evolved from past studies analyzing T cell responses toward model antigens, which suggested that MHC heterozygosity may influence the repertoire of peptides selected by class II MHC molecules (24, 25). On the other hand, other T cell studies have indicated that I-A^g7 and I-A^g7PD seem to select for distinct epitopes derived from the same protein (42, 43). Because these previous studies have all used T cells as the indirect readout, we took the most direct approach, which was to isolate and identify peptides from class II MHC molecules on the surface of an APC.

Our results demonstrate that the vast majority of peptides selected by I-A^g7 and I-A^d were not affected by the presence of additional class II MHC molecules and moreover, were derived from distinct proteins. There were very few proteins, such as BSA and LAMP, which donated distinct epitopes to both I-A^g7 and I-A^d (Tables 4 and 5). (For example, BSA 304–320, 559–573, and 568–580 were selected by I-A^g7, whereas BSA 226–241 was selected by I-A^d. Similarly, LAMP 35–47 and 213–227 peptide families were selected by I-A^d and I-A^g7, respectively.) From a physiological viewpoint, the conclusions stated here are relevant because the unmanipulated APCs are displaying the final products of peptide selection derived from endogenous antigen processing and presentation pathways. Taking into consideration the specificity of peptides selected by each MHC allelic form, the lack of competition was not unexpected.

On a slightly different note, in our previous analysis we also tested the other side of the coin, which is whether for a given MHC molecule there would be competition among different epitopes from the same protein. For these experiments, we chemically quantitated the amounts of a minor epitope of lysozyme selected in the presence or absence of the dominant major epitope. Results from those studies indicated that the dominant segment of lysozyme, which binds about 50 times stronger than the adjacent minor epitope, had no influence on the selection of the weaker minor epitope (44).

With respect to the pathogenesis of T1DM, MHC-heterozygosity may affect T cell selection or generation of regulatory T cells, which in turn afford protection from onset of autoimmunity (19–23). However, the idea that one allele competes with the other for display of islet β cell-derived antigens, which could affect activation of diabetogenic T cells is not supported by our results.

Abbreviations

APC

antigen-presenting cell

MS/MS

tandem MS

NOD

nonobese diabetic

T1DM

type 1 diabetes mellitus