HLA-DM captures partially empty HLA-DR molecules for catalyzed removal of peptide

Abstract

The mechanisms of HLA-DM-catalyzed peptide exchange remain uncertain. Here we found that all stages of the interaction of HLA-DM with HLA-DR were dependent on the occupancy state of the peptide-binding groove. High-affinity peptides were protected from removal by HLA-DM through two mechanisms: peptide binding induced the dissociation of a long-lived complex of empty HLA-DR and HLA-DM, and high-affinity HLA-DR-peptide complexes bound HLA-DM only very slowly. Nonbinding covalent HLA-DR-peptide complexes were converted into efficient HLA-DM binders after truncation of an N-terminal peptide segment that emptied the P1 pocket and disrupted conserved hydrogen bonds to HLA-DR. HLA-DM thus binds only to HLA-DR conformers in which a critical part of the binding site is already vacant because of spontaneous peptide motion.

Figure 1. Peptide disrupts long-lived complex between empty DR and DM

(a,b) Specificity of SPR assay. (a) Activity of DM WT, Mut1 (DMα R98A) and Mut2 (DMα R98A/R194A) in accelerating binding of Alexa-488 labeled MBP_85-99 peptide (30 nM) to DR2 (150 nM), measured by fluorescence polarization (FP). (b) Binding of DR2-CLIP complexes to DM WT and mutants by SPR. DR2-CLIP (2 μM) was run over DM WT and mutant surfaces (5 min, pH 5.35, 25 μl/min, 30°C) (stage 1), followed by buffer (stage 2) and 1 μM MBP_85-99 (stage 3). Readings from streptavidin flow cell were subtracted from DM WT and mutant flow cells. (c,d) DR2 mutant without DM interaction. (c) DM-accelerated binding (+/− 25 nM DM) of labeled MBP_85-99 to DR2 WT and DRα S53D mutant (150 nM) was measured as in (a). (d) DM binding of DR2 WT and mutant. WT or mutant DR2-CLIP (1 μM) were injected (5 min), followed by buffer and 10 μM CLIP_87-101. (e) Dissociation of DM-DR2 complex by high-affinity peptide. DR2-CLIP complexes (2 μM) were injected (5 min), followed by buffer and MBP_85-99 or MBP P4D analog (1 μM). Data are representative of two (a, c, d) or more than three (b, e) independent experiments.

Figure 2. Rate of DM-DR complex dissociation is determined by peptide affinity

(a,b) Dissociation of DM-DR complex by CLIP and CLIP mutants. DR2-CLIP (5 μM) was injected (5 min, pH 5.35, 15 μl/min, 30 °C), followed by buffer and CLIP or CLIP mutants (10 μM). (c) Relationship between peptide affinity and DM-DR dissociation rate. DR2 (100 nM) binding of peptides (3-fold dilutions, 10 μM to 14 nM) was examined in competition assay against Alexa-488 labeled MBP_85-99 (10 nM). IC₅₀ values were plotted against rates of peptide-induced DM-DR dissociation (a,b). (d) Titration of CLIP mutants. CLIP and two CLIP mutants were injected in stage 3 (0.5 to 100 μM). Rates of DR dissociation from DM were plotted against peptide concentration. (e) Testing of DM- DR dissociation by MBP_85-99 and mutants. Indicated MBP peptides were injected in stage 3 as in (a). (f) Relationship between MBP peptide affinity and DM-DR dissociation rate. IC₅₀ values determined in competition assay (as in c) were plotted against rate of DR dissociation from DM (e). Data are representative of two independent experiments as shown (a-f) and more than three independent experiments under similar conditions (a–c, e, f). Peptide competition assays were performed in triplicates.

Figure 3. High-affinity DR-peptide complexes interact only slowly with DM

(a, b) Absence of detectable DM binding by DR molecules with covalent peptides. (a) DR2-CLIP complexes with or without thrombin cleavage of the peptide linker (5μM) were injected for 5 min, followed by buffer (stage 2) and 20 μM CLIP_87-101 (stage 3). (b) 3C cut or uncut DR1-CLIP_low complexes (5 μM) were injected (2 min), followed by buffer and HA_306-318 (50 μM). (c,e,g) Slow DM binding by DR molecules with bound high-affinity peptides. DR complexes preloaded with indicated peptides (2 μM for DR2 and DR4; 10 μM for DR1) were injected for 5-7 min, followed by buffer and 10 μM CLIP_87-101 (DR2) or 50 μM HA_306-318 (DR4 and DR1). (d,f) Concentration-dependent binding of high-affinity DR-peptide complexes under optimal conditions (37°C, long injection time, high DR-peptide concentrations). DR-peptide complexes preloaded with the indicated peptides were injected for 10 min, followed by buffer (stage 2) and 5 μM MBP_85-99 (DR2) and 50 μM HA_306-318 (DR1) (stage 3). (a-g) SPR assays were carried out at 25°C (a, b) or 37°C (c-g) (pH 5.35, 15 μl/min). Data are representative of three (a, c, e) and of two (b, d, f, g) independent experiments.

Figure 4. DM binds with fast kinetics to DR-peptide complexes without engaged peptide N-terminus

(a, b) The hydrogen bonding network between DR1 and full length, covalently linked HA_306-318 peptide (a) is compared to a linked HA mutant peptide lacking three N-terminal residues (b). Peptide residues are numbered from P_-2 to P₁₁. DRα Val65 was mutated to cysteine to enable disulfide bond formation to HA peptides with a cysteine at the P6 position (HA₆ peptides). (c) Space filling model of the empty DR1 groove showing residues contacted by two N-terminal peptide residues (red) and the P1 anchor (green). (d, f) DM binding was enabled by loss of two N-terminal residues and the P1 anchor. DR1 complexes with linked HA₆ peptides (5 μM) were injected for 2 minutes (pH 5.35, 15 μl/min 25 °C), followed by injection of buffer. (e) C-terminal truncation of HA₆ peptide did not result in substantial DM binding. Complexes linked at the P₆ position were tested as in (d). Data are representative of at least three (d, f) and of two (e) independent experiments.

Figure 5. Truncated peptides covalently linked through the P3 peptide position also bind DM

(a, b) DRα Gly58 was mutated to a cysteine to permit disulfide bond formation with HA peptides carrying a cysteine at the P3 position (HA₃ peptides). DM binding of DR molecules disulfide-linked to the indicated HA₃ peptides (5 μM) was examined as in Fig. 4. Data are representative of three (a) and of two (b) independent experiments.

Figure 6. DRα Trp43 is important for interaction with DM

(a) DRα Trp43 is surface accessible and interacts with the P1 tyrosine of HA_306-318. Hydrophobic residues in the DRα40-54 segment interacting with P1 anchor (Trp43, Ala52 and Phe54) are indicated, as well as residues implicated in DM interaction (Glu40, Phe51 and Ser53). HA_306-318 P-2 to P2 segment is shown as a stick model. (b-d) Conservative DRα W43F mutation reduced DM susceptibility (b), DM binding (c) and intrinsic stability of DR2-peptide complexes (d). (b) Dissociation rate of Alexa-488 labeled MBP_85-99 from DR2 WT or W43F was measured by FP over a range of DM concentrations. (c) DR2 WT or W43F with bound CLIP (1 μM) were injected over DM WT and DM Mut2 flow cells (pH 5.35, 30 °C, 15 μl/min) followed by buffer and 5 μM CLIP_87-101. (d) Dissociation of Alexa-488 labeled MBP_85-99 from preloaded DR2 WT or W43F complexes (100 nM) was measured by FP in presence of 10 μM unlabeled MBP_{85- 99} competitor peptide (pH 5.4, 25°C). (e,f) Comparison of a panel of four DRα chain mutants in FP (e) and SPR (f) assays, using conditions described in b and c, respectively. Data are representative of two independent experiments.

Figure 7. Large energy barrier for DR-peptide binding to DM

(a,b) Temperature profiles for DM binding by DR with full length or linked truncated peptide (HA₆ P₂-P₁₁). (a,b) DR1-CLIP_low (1 μM, a) and DR1-HA₆ P₂-P₁₁ (1 μM, b) complexes were run at indicated temperatures (15 μl/min, pH 5.35), followed by buffer and 50 μM HA_306-318. (c) Relative binding at time 250-270 sec of the association phase in (a,b and Supplementary Fig. 6a) was plotted against corresponding temperature. (d) DM affinity for DR1 with covalently linked HA₆ P₂-P₁₁ peptide. DR1-HA₆ P₂-P₁₁ complexes were injected at indicated concentrations as in (a). Data were fitted with a 1:1 Langmuir binding model using BIAevaluation software. (e-g) Comparison of pH profiles for binding by DR molecules with full-length peptides (e) and covalent truncated peptide (f). (e,f) DR1-CLIP_low complexes (5μM, e) and DR1-HA₆ P₂-P₁₁ complexes (5 μM, f) were injected as in (a) for 2 min at the indicated pH (25 °C). (g) Relative binding at time 91-110 seconds of the association phase in (e,f and DR2-CLIP) was plotted against pH, with the binding at pH 5.0 defined as 1.0. Data are representative of two (a, e) and of three (b, d, f) independent experiments.