Characteristics of carbohydrate antigen binding to the presentation protein HLA-DR

Abstract

Zwitterionic polysaccharide antigens (ZPSs) were recently shown to activate T cells in a class II major histocompatibility complex (MHCII)-dependent fashion requiring antigen presenting cell (APC)-mediated oxidative processing although little is known about the mechanism or affinity of carbohydrate presentation (Cobb BA, Wang Q, Tzianabos AO, Kasper DL. 2004. Polysaccharide processing and presentation by the MHCII pathway. Cell. 117:677-687). A recent study showed that the helical conformation of ZPSs (Wang Y, Kalka-Moll WM, Roehrl MH, Kasper DL. 2000. Structural basis of the abscess-modulating polysaccharide A2 from Bacteroides fragilis. Proc Natl Acad Sci USA. 97:13478-13483; Choi YH, Roehrl MH, Kasper DL, Wang JY. 2002. A unique structural pattern shared by T-cell-activating and abscess-regulating zwitterionic polysaccharides. Biochemistry. 41:15144-15151) is closely linked with immunogenic activity (Tzianabos AO, Onderdonk AB, Rosner B, Cisneros RL, Kasper DL. 1993. Structural features of polysaccharides that induce intra-abdominal abscesses. Science. 262:416-419) and is stabilized by a zwitterionic charge motif (Kreisman LS, Friedman JH, Neaga A, Cobb BA. 2007. Structure and function relations with a T-cell-activating polysaccharide antigen using circular dichroism. Glycobiology. 17:46-55), suggesting a strong carbohydrate structure-function relationship. In this study, we show that PSA, the ZPS from Bacteroides fragilis, associates with MHCII at high affinity and 1:1 stoichiometry through a mechanism mirroring peptide presentation. Interestingly, PSA binding was mutually exclusive with common MHCII antigens and showed significant allelic differences in binding affinity. The antigen exchange factor HLA-DM that catalyzes peptide antigen association with MHCII also increased the rate of ZPS association and was required for APC presentation and ZPS-mediated T cell activation. Finally, the zwitterionic nature of these antigens was required only for MHCII binding, and not endocytosis, processing, or vesicular trafficking to MHCII-containing vesicles. This report is the first quantitative analysis of the binding mechanism of carbohydrate antigens with MHCII and leads to a novel model for nontraditional MHCII antigen presentation during bacterial infections.

Fig. 1

Chemical structure of PSA from B. fragilis. (A) Labeling reactions with hydrazide-linked biotin and AlexaFluor594 were targeted to the periodate-activated side chain galactofuranose. Radiolabeling was accomplished by tritiated borohydride reduction of periodate-created aldehydes. (B) Positive free amines were neutralized through acetylation by acetic anhydride. (C) Negative carboxylates were neutralized by carbodiimide and borohydride reduction.

Fig. 2

Analysis of MHCII antigen binding. (A) The plate capture efficiency for immobilization of DR protein was measured spectroscopically (n = 9) to facilitate the calculation of binding stoichiometries. 59.4% of the DR incubated in the Immulon 2 HB wells became immobilized, with 40.6% of the protein remaining in the supernatant as measured by absorbance at 280 nm. In subsequent binding assays, 0.1 μg DR construct was coated into wells (0.0594 μg actually immobilized) at 37°C and pH 5.0, except SEA (at pH 7.3). The degree of binding () is the moles ligand bound per mole of MHCII; thus, the B_max represents the stoichiometry. All data points represent the average of at least four replicates. PSA saturably binds to DR1 (panel B; K_d = 1.9 ± 0.4 μM; B_max = 0.8 ± 0.08), DR2 (panel C; K_d = 0.31 ± 0.05 μM; B_max = 1.0 ± 0.04), and DR4 (panel D; K_d = 1.0 ± 0.3 μM; B_max = 1.0 ± 0.1). Control binding with SEA (E) and MBPp (F) with DR2 show K_d values of 4.7 ± 2.0 μM and 4.3 ± 2.6 μM with B_max values of 1.5 ± 0.5 and 1.2 ± 0.2, respectively.

Fig. 3

Salt effects on in vitro MHCII binding. 0.1 μg of DR was used in all of the assays. (A) 4 μM MBPp and 4 μM SEA binding to DR2 with zero (open) or 1 M NaCl (shaded) added to the reaction buffer, normalized to the no salt added data with each data point representing the average of three replicates. Only very small effects are observed. (B) 1 μM PSA binding to DR2 with 0, 0.5, or 1.0 M NaCl added with each data point representing the average of eight replicates, showing significant loss of binding as a result of high ionic strength conditions. (C) Size exclusion elution profiles of PSA ligand used in these assays in high and low ionic strength buffer using a Superose 12 column, confirming that changes in binding are not due to changes in size or aggregation state under the specified ionic conditions.

Fig. 4

pH effects on in vitro MHCII binding. (A) Peptide and superantigen binding to DR2 at pH 5.0 (open) compared to pH 7.3 (shaded), normalized to pH 7.3 data, demonstrating modest alterations in binding (2-fold with peptide). (B) PSA binding to DR2 at pH 5.0 and 7.3, showing a profound increase in binding at acidic pH (5-fold), showing that binding of PSA to MHCII is strongly dependent upon an acid environment. (C) Size exclusion elution profiles of PSA ligand used in these assays at neutral and acidic pH using a Superose 12 column, confirming that changes in binding are not due to changes in size or aggregation state at the specified pH value.

Fig. 5

MHCII competitive binding assays (all normalized to no competitor), with each data point representing at least three replicates. (A) Titration of 1 μM PSA binding competition with a TTp competitor in DR2 binding at pH 5.0, showing complete inhibition of PSA binding at high peptide concentrations. (B) 4 μM MBPp binding (open) is competed with 20 μM PSA (shaded) using DR2 and DR1 at pH 7.3. (C) 4 μM SEA binding to DR1 competes with 20 μM PSA but is enhanced by 20 μM peptide at pH 7.3. (D) 4 μM SEA binding to DR2 with varying concentrations of PSA competitor, showing dose-dependence. (E) MBPp from 0.2 μg of MBPp-saturated DR2 and DR1 was displaced with the addition of a buffer alone (open) but more efficiently with 50 μM PSA (shaded).

Fig. 6

DM effects on binding. (A) 1.0 μM PSA binding time course with 0.1 μg DR2 in the presence (filled) or absence (open) of equimolar DM. Time to reach half completion (dotted line, T_½) in the absence of DM is 107.6 ± 16.4 min, but the rate of binding is increased 3-fold to T_½ = 33.3 ± 5.0 min with DM. Each data point represents the average of three replicates. (B) PSA co-IP with an anti-MHCII antibody from wild-type and DM^−/− splenocytes. DM^−/− cells fail to present significant quantities of PSA in MHCII. Each data point represents the average of two independent experiments with 12 mice in each group per experiment. (C) PSA-induced abscess formation (a measure of in vivo T cell activation by PSA) in either wild type or DM^−/− animals, showing a 2-fold decrease in abscess formation in the animals lacking DM. The numbers above each bar represent the number of mice with abscesses compared to total.

Fig. 7

Charge effects on cellular localization of PSA antigens (red) and MHCII (green) in murine MΦs and Raji cells by confocal microscopy. Surface colocalized PSA and MHCII (yellow) are indicated by arrow heads and internal colocalization (yellow) by arrows. (A) The native PSA images, shown as separate channels to illustrate the weakly stained MHCII positive vesicles obscured in the overlay (right) by the strong red signal of the polysaccharide. (B) NAc-PSA. (C) Carbo-PSA. (D) NAc-Carbo-PSA. In agreement with the in vitro binding experiments, none of the mod-PSA samples are presented by APCs at the cell surface, though all can enter vesicular compartments that contain MHCII protein (arrows).

Fig. 8

Charge effects on carbohydrate processing. Superose 12 analysis of (A) native PSA, (B) NAc-PSA, and (C) group B streptococcal (GBS) polysaccharide from APC endosomal compartments (open) compared to untreated controls (filled). PSA is reduced to a low molecular weight product (arrow) that is known to be presented by MHCII. Likewise mod-PSA and nonzwitterionic GBS polysaccharide also enter APCs and are processed (arrow). (D) Overlay plot of the control PSA elution profiles from panels A and B, showing no measurable difference in size or aggregation state of the native polysaccharide (filled) compared to charge-modified PSA (open) prior to endocytosis and processing.

Fig. 9

Charge effects on carbohydrate binding. (A) Native PSA (filled circles), NAc-PSA (open circles), Carbo-PSA (filled squares), and NAc-Carbo-PSA (open squares) binding to DR2 at 37°C and pH 5.0. All mod-PSA antigens failed to bind MHCII. (B) Native and NAc-PSA co-IP with anti-MHCII antibody from wt splenocytes, showing that mod-PSA antigens fail to be presented by APCs over an isotype background control.

Fig. 10

Schematic of ZPS Antigen Presentation by MHCII. In step 1, carbohydrate antigens enter the vesicular traffic of APCs and are quickly processed to low molecular weight forms. These same molecules cannot bind to surface localized MHCII molecules due to the requirement of low pH and HLA-DM, nor can the processed carbohydrates bind to vesicular Ii-bound MHCII since peptides must first be removed from the binding groove in order for binding to occur. In step 2, the pH drops and cathepsin S cleaves Ii from CLIP. In step 3, carbohydrate antigens cannot bind MHC-CLIP efficiently enough within the cell because HLA-DM is not present. By step 4, the pH has reached 4.5, enabling HLA-DM to catalyze the exchange of CLIP for ZPS molecules—but not singly charged or neutral carbohydrates. In step 5, high affinity electrostatic interactions are formed between PSA and MHCII in an allelic selective manner. In step 6, HLA-DM dissociates from the complex, enabling step 7, where the MHCII-ZPS complex is shuttled to the cell surface for T cell recognition.