B cell receptor (BCR)-mediated antigen processing is a mechanism that allows class II–restricted presentation of specific antigen by B cells at relatively low antigen concentrations. Although BCR-mediated antigen processing and class II peptide loading may occur within one or more endocytic compartments, the functions of these compartments and their relationships to endosomes and lysosomes remain uncertain. In murine B cells, at least one population of class II– containing endocytic vesicles (i.e., CIIV) has been identified and demonstrated to be distinct both physically and functionally from endosomes and lysosomes. We now demonstrate the delivery of BCR-internalized antigen to CIIV within the time frame during which BCR-mediated antigen processing and formation of peptide–class II complexes occurs. Only a fraction of the BCR-internalized antigen was delivered to CIIV, with the majority of internalized antigen being delivered to lysosomes that are largely class II negative. The extensive colocalization of BCR-internalized antigen and newly synthesized class II molecules in CIIV suggests that CIIV may represent a specialized subcellular compartment for BCR-mediated antigen processing. Additionally, we have identified a putative CIIV-marker protein, immunologically related to the Igα subunit of the BCR, which further illustrates the unique nature of these endocytic vesicles.
The recognition of MHC class II–restricted antigens by antigen-specific T cells requires the proteolytic processing of protein antigens to immunogenic peptides by class II–positive antigen-presenting cells (1, 2). The first step in antigen processing by B cells involves B cell receptor (BCR)1–mediated internalization of antigen (3–5). BCR-internalized antigen is then proteolytically processed and the resultant peptides preferentially loaded onto newly synthesized class II molecules (6–8) from which the class II– associated invariant chain has been removed by the concerted action of acid proteases and the protein HLA-DM/ H-2M (9). The resultant peptide–class II complexes are then transported to the surface of the B cell.
The intracellular compartments where antigen processing occurs have only recently been characterized and there is considerable variation in the intracellular localization of class II molecules among different cell types. Many cells, such as human lymphoblasts and macrophages, sequester much of their class II in lysosomes or lysosome-like structures referred to as the MHC class II–enriched compartment (MIIC; reference 10). Although delivery of BCR-internalized antigen to MIIC has been demonstrated (11), the fate of the antigen delivered to these structures (i.e., complete degradation versus processing and binding to class II molecules) remains unknown.
In other professional antigen-presenting cells such as many murine B cell lines, there is little accumulation of class II in lysosomes under normal conditions (12–14). Instead, class II is found in endosomes and endosome-related structures, at least one population of which (class II vesicles [CIIV]) can be purified and physically separated from conventional endocytic and secretory organelles by cell fractionation techniques (14).
Although many or all endocytic, class II–containing vesicle populations may host the loading of peptides onto class II molecules, there may be important qualitative differences regarding the subcellular compartments where antigenic peptides are generated and efficiently loaded onto class II molecules. Specifically, although BCR-mediated antigen presentation appears to involve binding of peptide to newly synthesized class II molecules (6–8), presentation of fluid phase proteins by B cells appears to be able to occur via both newly synthesized and recycling class II molecules (7, 8, 15, 16), possibly reflecting differences in the intracellular sites of peptide generation and class II loading.
Additionally, not all receptors are equivalent at mediating antigen processing and presentation. In murine B cells, antigen internalized via the transferrin receptor (while presented more efficiently than soluble antigen) is presented 10–100 times less efficiently than the same antigen internalized via the BCR (17). This result may reflect the fact that the transferrin receptor has far more restricted access to intracellular class II compartments in B cells than does the BCR (11). Even more dramatic is the demonstration that a single amino acid substitution in the transmembrane region of the human IgM BCR (huBCR) can completely abolish the ability of this receptor to mediate efficient antigen processing and presentation without affecting BCR-mediated antigen endocytosis and bulk antigen degradation (18, 19). Thus, antigen uptake and degradation is necessary, but not sufficient, for antigen processing and presentation.
Thus, it has become important to determine the intracellular compartments to which physiologically important receptors (e.g., the BCR) deliver antigens. In this paper, we demonstrate that, within the time frame during which the intracellular events of BCR-mediated antigen processing are known to occur, BCR molecules and BCR-internalized antigen have access not only to predominantly class II– negative endosome and lysosomes, but also to a novel population of endocytic vesicles that are highly enriched in newly synthesized class II molecules (i.e., CIIV). Moreover, CIIV contain a putative marker protein, immunologically related to the Igα subunit of the BCR, further illustrating the distinct nature of these endocytic vesicles.
Materials And Methods
Cell Culture and Fractionation.
A20μWT (i.e., A20 cells expressing a transfected, phosphorylcholine (PC)–specific human mIgM BCR (huBCR); reference 19) were cultured in αMEM, 5% FBS, 50 μM 2-mercaptoethanol, and 500 μg/ml G418. A20μWT cells were homogenized, fractionated by free flow electrophoresis (FFE), and the distribution of plasma membrane, lysosomes, and CIIV was determined as previously reported (14).
Distribution of huBCR-internalized Antigen in A20μWT FFE Fractions.
A20μWT cells (2 × 108 total cells) were incubated at 4 × 107 cells/ml for 30 min at 37°C in media containing 2 μg/ml PC-modified Fab fragments of rabbit γ globulin labeled with 125I (PC–RGG–125I [2μCi 125I/μg PC–RGG]), homogenized, fractionated by FFE, and the distribution of the plasma membrane and lysosomes was determined. The distribution of PC–RGG–125I was determined by counting each FFE fraction in a γ counter. Background counts (<100 cpm) were subtracted from the counts for each sample and the results normalized to a maximum value of 1.00.
Immuno-electron Microscopy Localization of huBCR-internalized Antigen and BCR Molecules to Isolated CIIV.
A20μWT cells were incubated in 400-nM PC-modified ovalbumin (PC–OVA) for 20 min at 37°C, homogenized, and fractionated by FFE. Isolated CIIV, as well as endosomes–lysosomes, were processed for immuno-electron microscopy (immunoEM) as previously described (14). Cryosections were stained with rabbit IgG specific for either murine class II (14), murine IgG (315-005-046; Jackson Immunologicals, West Grove, PA), human IgM (309-005-095; Jackson Immunologicals; 309-005-095), or ovalbumin (RaOVALBUMI; East Acres Biologicals, Southbridge, MA). Antibody to class II molecules, anti-ovalbumin, and both anti-Ig antibodies were visualized with either 1, 5, or 10 nm protein A–gold (14).
Steady-state Distribution of the PC-specific huBCR in A20μWT Cells.
A20μWT cells were homogenized and fractionated by FFE. 200 μl of each FFE fraction, along with 50 μl of homogenization buffer containing 5% Triton X-100 and 0.5 mg/ml BSA, was added to a PC–BSA-coated 96-well plate and the samples allowed to bind for 6 h at 4°C. The plates were washed and probed with rabbit anti–human IgM (1:1,000; 309-005-095; Jackson Immunologicals) followed by horseradish peroxidase (HRP)–labeled goat anti–rabbit IgG (1:1,000; 31462; Pierce Chemical Co., Rockford, IL). Bound goat anti–rabbit Ig–HRP was detected by addition of 200 μl of 0.5 mg/ml O-phenylenediamine and 0.015% H2O2 in borate buffer. After sufficient time, 50 μl of 1N HCl was added and the absorbance (OD 490 nm) measured. The absorbance above background is reported (background = ∼0.200 OD 490 nm).
Surface Labeling and Endocytosis.
A20μWT cells were collected by centrifugation and washed two times with PBS. Cells were labeled for 15 min at 108 viable cells/ml in PBS pH 7.5 containing 1 mg/ml sulfosuccinimidyl-6-(biotinamido) hexanoate (NHS– LC–biotin; Pierce Chemical Company; 21335). The labeling was quenched by addition of 5–10 vol of 10 mM lysine in PBS. The labeled cells were pelleted and then washed twice in PBS 0.1% BSA before incubation at 37°C in complete media containing 1 μM PC–OVA (Cells were >98% viable after labeling, washing, and incubation).
Detection of Biotin-labeled huBCR Molecules in A20μWT FFE Fractions.
FFE fractions (200 μl) from biotin-labeled/PC–OVA pulsed A20μWT cells were added to NeutrAvidin (50 μg/ml; 31000; Pierce)-coated plates along with 50 μl of homogenization buffer containing 5% Triton X-100 and 0.5 mg/ml BSA and allowed to bind for 6 h at 4°C. The plate was washed and probed with rabbit anti–human IgM (1:1,000, 309-005-095; Jackson Immunologicals) followed by HRP-labeled goat anti–rabbit Ig (1:1,000; 31462; Pierce). Bound goat anti–rabbit Ig–HRP was detected as described above. The absorbance above background is reported (background = ∼0.100 OD 490 nm).
Steady-state Distribution of Igα in A20μWT Cells.
Individual A20μWT FFE fractions were concentrated by centrifugation and analyzed by SDS-PAGE and Western blotting (14) with a rabbit antiserum raised against intact, full length, bacterially expressed murine Igα (rabbit anti-Igα, 1:5,000). Binding of rabbit anti-Igα was detected with HRP-labeled goat anti–rabbit Ig (1:5,000; 31462; Pierce) and enhanced chemiluminescence (ECL; 34080; Pierce).
Immunological Relationship Between Igα and p50Igα.
A20μWT low density membranes (LDM) were separated by preparative SDS-PAGE and electroblotted onto nitrocellulose. The blot was probed with rabbit anti-Igα (1:1,000) and then extensively washed. The regions of the blot containing either Iga and p50Igα were individually excised and bound antibodies eluted with 500 μl of 0.1 M glycine, pH 2.5. The eluted antibody was neutralized with 100 μl, 1.0 M Tris, pH 8.0, and 4.5 ml of blotting buffer. The affinity-purified antibody was used to probe Western blots of total A20μWT LDM and binding revealed with HRP-labeled goat anti–rabbit Ig (1:5,000; 31462; Pierce) and ECL (34080; Pierce).
Results And Discussion
BCR-mediated Delivery of Antigen to Endosome and CIIV.
The murine A20μWT cell line (19) was used as a model cell for these studies. A20μWT cells express a PC-specific huBCR as well as an endogenous murine IgG2a BCR (muBCR) and the ability of these cells to process and present antigens via both BCRs has been well characterized (4, 19). Importantly, A20μWT cells localize only a small fraction (<10%) of their total class II to intracellular membranes (13, 14), with little class II present in high density, hydrolase-rich lysosomes (14). Owing to the lack of class II molecules in the lysosomes of these cells, it is easy to distinguish, by FFE, a distinctive population of low density, novel endocytic vesicles (i.e., CIIV) that are enriched in newly synthesized class II molecules (14). Moreover, comparative analysis of A20 (14, 23) and A20μWT cells (Figs. 1, 2,A, and 4; data not shown) demonstrates that CIIV isolated from both cells possesses the same morphological, biophysical, biochemical, and immunological characteristics (e.g., class II–positive, lgp110-negative, β-Hexosaminidase– negative multivesicular membrane structures with a characteristic electrophoretic mobility). Importantly, in A20μWT cells, BCR-mediated antigen processing and peptide loading of class II molecules can occur exclusively in low density endocytic structures without the involvement of high density lysosomal structures (13, 14).
If CIIV are involved in BCR-mediated antigen processing, then BCR-internalized antigens should be delivered to CIIV within 30–60 min after BCR-mediated internalization, the time required for BCR-mediated antigen processing in A20μWT cells (18). To determine whether this was the case, A20μWT cells were incubated with PC–RGG–125I for 30 min under conditions where antigen internalization and processing occur exclusively via cell-surface huBCR molecules (18, 19). The antigen-pulsed A20μWT cells were then homogenized, fractionated by FFE, and the distribution of huBCR-internalized PC–RGG–125I, as well as that of markers for plasma membrane (PM), endosomes, lysosomes, and CIIV (14), was determined. As shown in Fig. 1, the vast majority of the huBCR-internalized antigen was found in FFE fractions that contained endosomes and lysosomes (fractions 45–53). However, a small but significant amount of labeled antigen was present in the anodally shifted CIIV-containing FFE fractions (fractions 54–58). Thus, BCR-bound antigen could be found in CIIV-containing FFE fractions within the time frame during which BCR-mediated antigen processing is occurring.
This biochemical analysis, along with our previous observations that CIIV-containing FFE fractions consist almost entirely of class II–positive vesicles, strongly suggested that BCR-internalized antigen can gain access to CIIV. To demonstrate this point directly, and rule out the possibility that the BCR-internalized antigen present in the CIIV-containing FFE fractions was contained exclusively in class II–negative, β-Hexosaminidase–positive lysosomes, CIIV-containing FFE fractions from antigen (i.e., PC–OVA)– pulsed A20μWT cells were examined by multiple label immunoEM (14). As shown in Fig. 2,A, huBCR-internalized antigen (arrow) was present in CIIV isolated from antigen-pulsed B cells. Additionally, immunoEM analysis of endosome/lysosome–enriched FFE fractions demonstrated that BCR-internalized antigen was present both in class II–negative endosomes and lysosomes as well as class II–positive endosomes (Fig. 2, B and C, respectively). Quantitation of these immunoEM samples (Table 1) revealed that the vast majority (71–85%) of CIIV were endocytic (i.e., accessible by BCR-internalized antigen) and that the bulk (58–90%) of the endocytic vesicles within the CIIV-containing FFE fractions were class II positive. On the contrary, only 34% of the endocytic vesicles in the endosome/lysosome–containing FFE fractions were class II–positive endosomes, with the majority of the antigen-containing vesicles being class II–negative endosomes and lysosomes.
CIIV Contain BCR Molecules Internalized from the Cell Surface.
The presence of BCR-internalized antigen in CIIV raised the question of whether antigen was delivered to this compartment while still bound to the BCR or after dissociation of BCR–antigen complexes. To determine whether antigen is delivered to CIIV while still bound to the BCR, we first determined whether BCR molecules could be found in CIIV at steady-state. To this end, A20μWT cells were homogenized, fractionated by FFE, and the distribution of PC-binding huBCR molecules was determined by an antigen-specific anti–human IgM ELISA. As shown in Fig. 3 A, most of the PC-binding huBCR was present in PM-containing FFE fractions (fractions 35–44) with lower but significant levels also detected in endosome and lysosome, as well as CIIV-containing FFE fractions (fractions 45–54 and 55–64, respectively). A similar steady-state distribution was observed for the endogenous muBCR as determined by Western blotting (data not shown).
To confirm that the huBCR molecules detected in the CIIV-containing FFE fractions were actually localized to CIIV, we performed multiple label immunoEM analysis of FFE-isolated CIIV for the presence of huBCR molecules. As shown in Fig. 4, CIIV do contain huBCR molecules (arrows) at steady-state. Similar analysis of endosome/lysosome–containing FFE fractions demonstrated that huBCR molecules could be found in both class II–positive as well as class II–negative vesicles in these fractions (data not shown).
Because we have previously demonstrated that newly synthesized BCR molecules do not traffic through CIIV before arrival at the cell surface (14), these results strongly suggest that the BCR molecules found in CIIV are derived from the PM by endocytosis and suggest that antigen is delivered to this compartment while bound to these internalized BCR molecules. To demonstrate directly that cell surface BCR molecules, internalized in the presence of polyvalent antigen, are delivered to CIIV, A20μWT were surface labeled with biotin, incubated for various times at 37°C in the presence of polyvalent antigen (i.e., PC–OVA), homogenized, and then fractionated by FFE. The level of biotin-labeled (i.e., internalized) huBCR molecules in each FFE fraction was then determined by a human IgM-specific, avidin-capture ELISA. As shown in Fig. 3,B, after 20 min of incubation, internalized huBCR molecules could be detected in CIIV-containing FFE fractions (fractions 53–60) as well as those enriched in endosomes and lysosomes (fractions 45–52). Surprisingly, even though we had found that, under similar conditions, a vast majority of the BCR-internalized antigen was ultimately delivered to endosomes and lysosomes (see Fig. 1), a significant fraction of the internalized huBCR molecules were found in CIIV-containing FFE fractions. Therefore, we suggest that a portion, and possibly all, of the huBCR molecules detected in CIIV by immunoEM (Fig. 4) were derived from the PM after endocytosis, further supporting the contention that antigen is delivered to CIIV while bound to the BCR. Interestingly, as suggested by the presence of huBCR molecules in CIIV isolated from nonantigen-pulsed A20μWT cells, preliminary analysis of the constitutive endocytosis and trafficking of the huBCR of A20μWT cells suggests that BCR endocytosis and delivery to CIIV can occur in the absence of antigen cross-linking (Drake, J.R., unpublished results).
Subcellular Distribution of BCR Subunits and Identification of a Putative CIIV Marker Protein Immunologically Related to Igα.
Although both BCR molecules and BCR-internalized antigen clearly gained access to CIIV, as well as endosomes and class II–negative lysosomes, the extent to which these molecules are, or are not, selectively targeted to CIIV remains unclear. To begin to address whether there is any selective targeting of BCR molecules or antigen–BCR complexes to CIIV, we first attempted to determine whether there is any difference in the subunit composition of the BCR molecules present in PM, endosome and lysosome, or CIIV-containing FFE fractions.
To this end, we examined the steady state distribution of the Igα subunit of the BCR. As shown in Fig. 5,A, the majority of the 32-kD Igα protein (arrow B) is present in PM-containing FFE fractions with lesser amounts detected in endosome–lysosome and CIIV-containing fractions. The same distribution was also found for the Igβ subunit of the BCR as well as the heavy and light chain subunits of both the huBCR and muBCR (data not shown), suggesting that the subunit composition of the BCRs in these compartments is similar. Surprisingly, the anti-Igα antiserum also recognized a second, 50-kD protein (Fig. 5 A, arrow A) that appears to be selectively enriched in CIIV-containing FFE fractions.
To determine whether this 50-kD putative CIIV-marker protein was immunologically related to Igα (as opposed to being recognized by antibodies of a second specificity present in the rabbit anti-Igα antiserum), we affinity-purified antibodies to both Igα and the 50-kD protein on Western blots of A20μWT LDM, and tested the specificity of these purified antibodies. As shown in Fig. 5 B, the unfractionated rabbit anti-Igα antiserum recognized both Igα as well as the 50-kD protein present in unfractionated A20μWT LDM. Although affinity-purified anti-Igα antibody failed to recognize the 50-kD protein (occasionally, reactivity of the affinity-purified anti-Igα toward the 50-kD protein was observed although the results were variable, possibly owing to removal of low affinity/highly cross-reactive antibodies by the affinity purification protocol), affinity-purified antibody against the 50-kD protein recognized both Igα as well as the 50-kD protein. Because this antiserum was originally generated against recombinant whole murine Igα, these results demonstrate that at least some anti-Igα antibodies specifically recognize the 50-kD putative CIIV-marker protein, demonstrating that these proteins are immunologically related (i.e., that Igα and the 50-kD protein minimally share one cross-reactive epitope). Additionally, an antiserum raised against the cytoplasmic tail of Igα (20) also demonstrated cross-reactivity to the 50-kD protein, suggesting an immunological relationship between the cytoplasmic tail of Igα and some region of the 50-kD protein (data not shown).
Because the 50-kD putative CIIV-marker protein does not exhibit a decrease in apparent molecular weight upon treatment with either endoglycosidase H or F (Drake, J.R., unpublished results), we have foregone the more traditional gp50 designation in favor of p50Igα (the Igα superscript indicates the immunological relationship of the protein to the Igα subunit of the BCR). Although the structure and function of p50Igα remains unknown, it is unlikely to represent a highly modified form of Igα, since it was also detected in Igα-negative J774 macrophage-like cell line (Drake, J.R., unpublished results). Moreover, p50Igα is unlikely to be an artifact of proteolytic activity because it can be detected in detergent extracts of whole cells prepared in the presence of a cocktail of protease inhibitors (Drake, J.R., unpublished results). Additionally, preliminary analysis of J774 cells and the murine B cell hybridoma 2C3E1 (21) suggests a restricted distribution of p50Igα to CIIV in these cells (Drake, J.R., and P. Webster, unpublished results).
Considering the presence of p50Igα in the BCR-negative J774 macrophage-like cell line, p50Igα, unlike Igα, may not be a component of the BCR protein complex. Correspondingly, preliminary analysis has failed to reveal any physical association between p50Igα and the BCR of A20μWT cells (Drake, J.R., unpublished results). Therefore, a more thorough understanding of the possible function of p50Igα will have to await its eventual purification and sequencing or cDNA cloning. Interestingly, previous Southern blot analysis of the murine MB-1 gene (which codes for Igα) indicated the presence of an additional, Igα-related gene (22), possibly that coding for p50Igα. Most importantly, the identification of p50Igα as a putative marker for CIIV graphically demonstrates the unique biochemical nature of these novel class II-containing vesicles and provides us with a tool to study their origin, fate, and relationship to other intracellular compartments.
Previously, we demonstrated that in murine B cells, class II molecules are restricted to relatively early endocytic compartments (i.e., endosomes and CIIV), with little or no class II found in high density lysosomes (14). Because peptide–class II complexes have been demonstrated to form only in low density (i.e., nonlysosomal) compartments in these cells (13, 14), the presence of BCR-internalized antigen in both class II–positive endosomes as well as CIIV suggests that antigen processing and class II peptide loading may occur at either or both of these sites. Given the predominant role of newly synthesized class II molecules in BCR-mediated antigen processing and presentation (6–8) and the fact that CIIV are an intermediate in the pathway of transport of newly synthesized class II molecules to the cell surface (14, 23), our results strongly suggest a role for CIIV in BCR-mediated antigen processing and class II peptide loading.
Why might B cells possess a novel endocytic compartment for the processing of BCR-internalized antigens? In contrast with endosomes, lysosomes, and MIIC, which readily accumulate nonselectively internalized fluid phase endocytic tracers (10, 24), CIIV do not readily accumulate proteins internalized by fluid phase endocytosis (14). Within endosomes, lysosomes, and MIIC, the vast excess of nonantigenic peptides (i.e., peptides derived from the proteolytic degradation of fluid phase plasma proteins) may effectively compete with antigenic peptides (i.e., peptides derived from BCR-internalized antigen) for binding to class II molecules, preventing the efficient formation of antigenic peptide–class II complexes within these compartments. The absence of these nonantigenic peptides from CIIV may allow for more efficient formation of antigenic peptide–class II complexes in these vesicles. Additionally, because peptide loading onto class II is a relatively slow process (25), the relatively slow transport of newly synthesized class II molecules through CIIV, with class II molecules residing in CIIV for up to 2 h (14), may provide the necessary time for the loading of these class II molecules with antigen-derived peptides.
The authors would like to thank L. Schaefer for technical assistance, L. Ryan for photographic darkroom assistance, and L. Chicoine for assistance with sectioning and labeling samples for immunoEM analysis. Additionally, the authors would like to thank R. Mitchell (Harvard University, Cambridge, MA) and M. Nussenzweig (Rockefeller University, New York) who generously provided A20μWT cells and careful guidance as to their use.
The research described in this paper was supported by grants from the Public Health Service.
Abbreviations used in this paper: BCR, B cell receptor; CIIV, class II vesicles; ECL, enhanced chemiluminescence; FFE, free flow electrophoresis; huBCR, phosphorylcholine-specific human mIgM BCR; immunoEM; immuno-electron microscopy; LDM, low density membranes; MIIC, MHC class II–enriched compartment; muBCR, murine IgG2a BCR; NHS–LC–biotin, sulfosuccinimidyl-6-(biotinamido) hexanoate; PC, phosphorylcholine; PC–RGG–125I, PC-modified Fab fragments of rabbit γ globulin labeled with 125I; PC–OVA, PC-modified ovalbumin; PM, plasma membrane.
Address correspondence to James R. Drake at The Trudeau Institute, 100 Algonquin Avenue, P.O. Box 59, Saranac Lake, New York 12983. Phone: 518-891-3080; FAX: 518-891-5126; E-mail: email@example.com