A common single nucleotide polymorphism in endoplasmic reticulum aminopeptidase 2 induces a specificity switch that leads to altered antigen processing

Abstract

Endoplasmic reticulum aminopeptidases 1 and 2 (ERAP1 and ERAP2) cooperate to trim antigenic peptide precursors for loading onto MHC class I molecules and help regulate the adaptive immune response. Common coding single nucleotide polymorphisms in ERAP1 and ERAP2 have been linked with predisposition to human diseases ranging from viral and bacterial infections to autoimmunity and cancer. It has been hypothesized that altered Ag processing by these enzymes is a causal link to disease etiology, but the molecular mechanisms are obscure. We report in this article that the common ERAP2 single nucleotide polymorphism rs2549782 that codes for amino acid variation N392K leads to alterations in both the activity and the specificity of the enzyme. Specifically, the 392N allele excises hydrophobic N-terminal residues from epitope precursors up to 165-fold faster compared with the 392K allele, although both alleles are very similar in excising positively charged N-terminal amino acids. These effects are primarily due to changes in the catalytic turnover rate (k(cat)) and not in the affinity for the substrate. X-ray crystallographic analysis of the ERAP2 392K allele suggests that the polymorphism interferes with the stabilization of the N terminus of the peptide both directly and indirectly through interactions with key residues participating in catalysis. This specificity switch allows the 392N allele of ERAP2 to supplement ERAP1 activity for the removal of hydrophobic N-terminal residues. Our results provide mechanistic insight to the association of this ERAP2 polymorphism with disease and support the idea that polymorphic variation in Ag processing enzymes constitutes a component of immune response variability in humans.

Figure 1

Panel A Michaelis-Menten analysis of R-AMC hydrolysis by ERAP2^N (filled circles) and ERAP2^K (open circles) variants. Panel B: Double-reciprocal plot of the data.

Figure 2

Panel A, RP-HPLC chromatograms of products of the enzymatic digestion of peptide LSRHHAFSFR (substrate) by the two ERAP2 variants. The produced epitope SRHHAFSFR (product) is indicated. Two characteristic reaction conditions are shown, using 200ng (14.7nM, molar ratio peptide to enzyme=1361) of enzyme for 15min or 1hr and 2μg (147nM, molar ratio peptide to enzyme=136) enzyme for 1hr. Panel B, epitope generation rate for each ERAP2 variant. Panel C, the rate of hydrolysis by each ERAP2 variant of the fluorigenic substrate R-AMC was plotted versus the concentration of competing substrate LSRHHAFSFR and data were fit to a simple competition model to allow the calculation for the affinity for the competing substrate (57). Panel D, turnover number (k_cat) and specificity constant (k_cat/K_M) for the production of the SRHHAFSFR epitope by the two ERAP2 variants.

Figure 3

Panel A, RP-HPLC chromatograms of production of the SLYNTVATL epitope from a single residue extended precursor carrying a positively charged amino acid (KSLYNTVATL). Panel B, trimming rate of peptide KSLYNTVATL by the two ERAP2 variants. Panel C, R-AMC competition titrations for the KSLYNTVATL peptide as in Figure 2 reveal similar affinity for both ERAP2 variants.

Figure 4

Trimming rates of model epitope precursors by ERAP2^N (black bars) and ERAP2^K (white bars). 20μM of each indicated epitope precursor were incubated with an appropriate amount of ERAP2 to produce a limited amount of mature epitope as described in the methods section. Error bars correspond to standard deviation calculated from 3 separate experiments.

Figure 5

Panel A, specificity index (defined as the trimming rate for a peptide with a positively charged N-terminus divided by the trimming rate for the same peptide carrying a hydrophobic N-terminus) for three different antigenic epitope templates using the two ERAP2 variants. Y-axis is broken to enhance visibility of large numerical differences. Panel B, RP-HPLC chromatograms of trimming of the precursor RSRYWAIRTR by ERAP2^N and ERAP2^K. Notice the similar amount of epitope generation. Panel C, RP-HPLC chromatograms of trimming of the precursor LSRYWAIRTR by ERAP2^N and ERAP2^K. Notice the absence of a visible epitope peak produced by ERAP2^K.

Figure 6

Panel A, , Epitope generation versus epitope destruction: Trimming rates of epitope precursor LSRHHAFSFR (gray bars) and mature epitope SRHHAFSFR (white bars) by ERAP1, ERAP2^N and ERAP2^K. Panel B, ERAP2^N can trim the precursor LSRHHAFSFR with a rate corresponding to ~16% of that of ERAP1, whereas ERAP2^K is much less efficient. Bars represent the ratio of ERAP2/ERAP1 trimming rates for each ERAP2 variant. Panel C, HeLa cells, stably expressing HLA-B27, as well as TAP1 ER-transporter blocker (ICP47), were transfected with an ERAP2 variant and an ER-targeted miniprotein that after signal sequence cleavage gives rise to an HLAB27–specific peptide precursor with the sequence ASRHHAFSFR. HLAB27 cell-surface translocation was followed by flow cytometry using an MHC specific antibody (MIF=Mean Intensity of Fluorescence). Panel D, in vitro trimming rate of the ASRHHAFSFR by ERAP2 alleles.

Figure 7

Inhibition of ERAP2^N and ERAP2^K by a transition-state analog (DG001, panel A) and a substrate analog (Amastatin, panel B). The rate of hydrolysis of the fluorigenic substrate R-AMC was followed in the presence of increasing concentrations of inhibitor for each enzyme. Experimental data were fit to a simple binding model accounting for full or partial inhibition (see materials and methods). The calculated constants of inhibition (K_i) are shown.

Figure 8

Crystal structure of ERAP2^K allele suggests the structural basis for changes in activity. Panel A, sequence alignment of ERAP2 and homologous aminopeptidases showing conservation of the polymorphic residue 392 (in bold); the adjacent aminopeptidase motif (HELAH) is also indicated. Panel B, key catalytic residues in the ERAP2^K structure are shown in stick representation and superimposed with equivalent residues in ERAP2^N (3SE6). Note the relative positioning of Lys392 with respect to the catalytic Glu residues as well as the N-terminus of the bound ligand (Lys_a). Electron density 2|F_o|−|F_c| at 2σ is indicated around Lys392. Panel C, pairwise electrostatic interaction energies (ΔE_inter in Kcal/mol) calculated between the polymorphic residue 392 and the Lys_a ligand. The reported values are the average of the 20 runs with a standard deviation of less than 1%, which represents the uncertainty due to the granularity of the grid. Panel D, superimposition of ERAP2 residues that cap the S1 specificity pocket of the enzyme. |F_o|−|F_c| electron density (at 2.5σ) calculated in the absence of the ligand is shown around Lys_a. 2|F_o|−|F_c| electron density at 2σ is indicated around Glu177.