Characterization of a lysozyme-major histocompatibility complex class II molecule-loading compartment as a specialized recycling endosome in murine B lymphocytes.

We have previously identified an intracellular compartment involved in the association between processed lysozyme and IAk major histocompatibility complex class II molecules (called the lysozyme-loading compartment (LLC)). Here, we show that the LLC polypeptide composition analyzed by two-dimensional gel electrophoresis shares similarities with that of early endosomes, but not with that of late endosomes. The transferrin receptor, a well known marker for both early and recycling endosomes, colocalizes with IAk molecules in LLC. Moreover, both transferrin and fluid-phase markers have access to LLC after 15 min of internalization. In the presence of concanamycin B, SDS-stable dimer formation and transport of class II molecules out of LLC are impaired. In contrast, nocodazole treatment has no effect. These results suggest that LLC is a specialized compartment of the recycling pathway involved in lysozyme loading and in the targeting of lysozyme-major histocompatibility class II complexes toward the cell surface.

Recently, specialized sites for peptide loading have been described in both human and murine B cells. In human B cells, these specialized sites have been termed MIIC or MIIC-related compartments (11,12). MIIC appear as dense multimembrane vesicles (0.2-0.5-m diameter) containing lysosomal markers such as Lamp1, CD63, and ␤-hexosaminidase, suggesting that they correspond to late endocytic compartments. In murine B cells, several different compartments for peptide loading have been described. Some of them are similar to MIIC-related compartments containing late endosomal-lysosomal markers (13,14). Other have been identified as low density membrane vesicles (CIIV) positive for the transferrin receptor, a well known marker for both early and recycling endosomes, and negative for lysosomal markers (15). Additionally, it has been reported that antigen processing and loading on MHC class II molecules might involve several compartments along the endocytic pathway characterized by their low to dense densities (16).
We previously showed that, in the presence of hen egg lysozyme (HEL), newly synthesized SDS-stable class II molecules are first detected in a dense compartment called the lysozyme-loading compartment (LLC) (17). LLC does not have the characteristics of early endosomes (EE), late endosomes (LE), and lysosomes. Indeed, it is negative for the Rab4, Rab5, and Rab7 GTPases and does not contain detectable amounts of lysosomal markers.
In this paper, we show that LLC is connected to the recycling pathway. Bidimensional gel mapping reveals that its protein composition is similar to that of EE, but different from that of LE. At the ultrastructural level, LLC membrane fraction corresponds to a homogeneous population of 0.5-m diameter multimembrane vesicles containing both IA k molecules and the transferrin receptor. LLC is accessible to transferrin and the horseradish peroxidase fluid-phase marker after 15 min of internalization. In the presence of concanamycin B, an inhibitor of the vacuolar H ϩ -ATPase, the formation of SDS-stable dimers in LLC and the transport of class II molecules out of LLC are impaired. In addition, the transport of class II molecules from LLC is independent of the microtubule network. Altogether, these data suggest that LLC is a specialized compartment of the recycling pathway involved both in the loading of lysozyme and in the routing of the lysozyme-MHC class II complexes to the cell surface.
Antibodies-The mouse anti-IA k monoclonal antibody 10.2.16, recognizing both ␣␤-dimers and ␣␤⅐Ii complexes (16), and the rat antitransferrin receptor monoclonal antibody H129-121-6.8 were provided by Drs. S. Meresse  Subcellular Fractionation-Subcellular fractionation of 2A4 B lymphoma cells has been previously detailed (17). Briefly, cells that had been precultured for 2 h with 2 mg/ml HEL were gently homogenized, and post-nuclear supernatants were prepared. Post-nuclear supernatants were adjusted to 40.6% sucrose; loaded at the bottom of an SW 60 centrifugation tube; and then sequentially overlaid with 1.5 ml of 35% sucrose, 1 ml of 30% sucrose, and finally 0.5 ml of homogenization buffer (1 mM MgCl 2 , 250 mM sucrose, and 3 mM imidazole, pH 7.4). Gradients were centrifuged for 60 min at 35,000 rpm at 4°C. Early and late endosome-enriched fractions were collected at the 35/30% sucrose interface and at the uppermost region of the 30% sucrose cushion, respectively. LLC was separated from early endosomes by loading the early endosome-enriched fractions recovered from the flotation gradient onto a 27% Percoll gradient (17). LLC fractions correspond to Percoll gradient fractions 9 -11, whereas Rab5-positive EE and Rab7-positive LE correspond to fractions 3-7.
Two-dimensional Gel Electrophoresis-2A4 cells were incubated in prewarmed cysteine/methionine-deficient RPMI 1640 medium containing 5% dialyzed fetal calf serum. After 1 h of incubation at 37°C, cells were pelleted at room temperature and resuspended in 3 ml of the cysteine/methionine-deficient medium containing 0.5 mCi of an 80% [ 35 S]methionine and 20% [ 35 S]cysteine mixture and 0.5 mCi of [ 35 S]cysteine in the presence of HEL for 18 h at 37°C. EE, LE, and LLC were prepared as described above, adjusted to 40.6% sucrose, and submitted to a second flotation step gradient to remove the excess of Percoll. After centrifugation, LLC and EE were collected at the 35/30% sucrose interface, whereas LE were harvested from the 30% sucrose/homogenization buffer interface. Membrane fractions were incubated with 1% Nonidet P-40 in phosphate-buffered saline for 30 min at 4°C and centrifuged in a TLA-100.3 rotor at 70,000 rpm for 30 min in a TLA centrifuge (Beckman Instruments) at 4°C. Supernatant proteins were submitted to trichloroacetic acid precipitation (10% final concentration) for 1 h at 4°C. After centrifugation, pellets were washed three times with acetone and resuspended in 9.8 M urea, 4% (w/v) Nonidet P-40, 2% (v/v) Ampholine, pH 7-9, and 100 mM dithiothreitol. A combination of isoelectric focusing and SDS-polyacrylamide gel electrophoresis was used to resolve proteins in two dimensions. Tube gels used for the isoelectric focusing gels were 25 cm long and 2.5 mm in internal diameter. Isoelectric focusing gels were run at 1200 V for 17 h. The pH gradient was linear between pH 4.5 and 7.4. The second dimension resolving gels were 10% (w/v) acrylamide and were prepared for autoradiography using Intensify (DuPont NEN).
Electron Microscopy-LLC were prepared after flotation and Percoll step gradients. LLC fractions were pooled and submitted to a second flotation gradient to eliminate the Percoll. After centrifugation, LLC membrane pellets were treated for electron microscopic observation. For class II molecule and transferrin receptor labeling, LLC fixed in 8% paraformaldehyde were resuspended in 10% gelatin before infusion with a mixture of 1.8 M sucrose and 20% polyvinylpyrrolidone (18) and frozen in liquid nitrogen. For single labeling experiments, thawed ultrathin cryosections were labeled using either the mouse monoclonal anti-class II molecule antibody followed by rabbit anti-mouse IgG (Sigma) and protein A-coated colloidal gold or the rat monoclonal antitransferrin receptor antibody followed by rabbit anti-rat IgG (Cappel) and protein A-coated colloidal gold (19). Double labeling was performed using the anti-transferrin receptor antibody followed by rabbit anti-rat IgG and 15-nm gold-conjugated goat anti-rabbit IgGs (GAR15) (Bio-Cell). Then, sections were incubated with the anti-class II molecule antibody followed by 20-nm gold-conjugated goat anti-mouse IgGs (GAM20) (BioCell). For Epon-embedded sections, LLC were fixed using 2% glutaraldehyde in 0.2 M cacodylate buffer. After centrifugation for 30 min at 100,000 ϫ g at 4°C in an Airfuge (Beckman Instruments), post-fixation of membrane pellets was performed using 2% OsO 4 and 1.5% potassium ferricyanide in H 2 O for 30 min, followed by an overnight incubation with 1.5% uranyl acetate in water (20). The LLC membrane pellets were dehydrated in an ethanol series before embedding and polymerization. Epon sections were stained using lead citrate.
Distribution of Horseradish Peroxidase in LE, EE, and LLC-2A4 cells were incubated for 5 min at 37°C with 1.5 mg/ml horseradish peroxidase. Then, horseradish peroxidase was chased for 10, 25, and 40 min at 37°C (17). LE, EE, and LLC were prepared, and external media were collected and assayed for horseradish peroxidase enzymatic activity (21). To study the effect of concanamycin B on horseradish peroxidase internalization and transport, cells were pretreated for 2 h with 1 M concanamycin B, followed by horseradish peroxidase internalization, and were chased in the presence of 1 M concanamycin B.
Distribution of Transferrin-conjugated Peroxidase in LE, EE, and LLC-After endogenous transferrin depletion, cells were incubated for 1 h at 4°C with 12 ml of a 20 g/ml solution of mouse transferrinconjugated peroxidase (Pierce). After extensive washings, cells were separated into six different samples corresponding to 0, 5, 10, 15, 30, and 45 min of internalization performed by incubating the cells at 37°C in culture medium. After each time point, external media were collected, and cells were washed with ice-cold phosphate-buffered saline and incubated with 25 mM acetic acid, 0.1% bovine serum albumin, pH 4.5, in phosphate-buffered saline for 30 min at 4°C to remove surfacebound transferrin. After homogenization, LE, EE, LLC, and external media were assayed for peroxidase enzymatic activity associated with the mouse transferrin (21).
Effects of Nocodazole and Concanamycin B on MHC Class II Molecule Transport-For nocodazole experiments, 10 8 cells were metabolically labeled for 20 min at 37°C, and radioactivity was chased for 30, 60, 120, and 240 min. In each sample, nocodazole was added at a 10 M final concentration only after a 30-min chase to allow the transport of class II molecules along the secretory pathway. After subcellular fractionation, class II molecules were immunoprecipitated from fractions 9 -11 of the Percoll gradient. In the presence of concanamycin B, cells were preincubated for 2 h at 37°C with both 1 M concanamycin B and 2 mg/ml HEL; metabolically labeled for 20 min; and chased at 37°C for 30, 60, and 240 min in the presence of concanamycin B and HEL. After subcellular fractionation, 0.8-ml Percoll gradient fractions were collected, pooled two by two, and lysed with 1% Nonidet P-40 in the presence of protease inhibitors, and class II molecules were immunoprecipitated. Samples were treated under nonboiling conditions for the detection of SDS-resistant class II molecules (17).
Western Blotting-Solubilized EE, LE, and LLC proteins were loaded onto 12% SDS-polyacrylamide gel and transferred onto Immobilon-P membranes before incubation with the anti-transferrin receptor antibody H129-121-6.8, detected by an indirect chemiluminescence system as described previously (22).

LLC Protein Composition Is Closely Related to
That of EE-To characterize the relationship among LLC, EE, and LE, we analyzed their respective protein compositions. EE, LE, and LLC were prepared from cells metabolically labeled to equilibrium and analyzed using high resolution two-dimensional gels (23,24) and autoradiography. The polypeptide patterns are shown in Fig. 1. We can detect ϳ150 polypeptides in LLC. About 100 are found in the three compartments (EE, LE, and LLC). Among these, actin, heat shock proteins, and tubulins can be identified by their typical mobilities (24). The remaining 50 LLC polypeptides were analyzed in more detail. We do not find polypeptides specifically associated with LLC. Strikingly, all of them can be detected in EE, but not in LE, and can be separated in three groups depending on both their electric mobility and molecular mass. The first group corresponds to basic and low molecular mass proteins (10 -15 kDa). Among them, the SNARE-like 14-kDa protein cellubrevin, shown to be enriched in the recycling endosomes (25,26), and the p11 protein, known to be associated with annexin II in early endosomes (27), can be identified by their typical electrophoretic mobility. The second group of proteins has an acidic mobility with a molecular mass of 10 -30 kDa, whereas the third group corresponds to glycoproteins of high molecular mass .
LLC Contains Both the Transferrin Receptor and IA k Molecules-LLC membrane fractions were loaded onto a second flotation gradient to eliminate the Percoll (see "Experimental Procedures"), allowing electron microscopic observations. Fig.  2a shows that LLC membrane vesicles are ϳ0.5 m in diameter, with internal membrane structures containing IA k molecules. As LLC protein composition was closely related to that of EE, we also investigated the presence of the transferrin receptor, a well known marker for EE and recycling endosomes (28 -30). Immunolabeling using the anti-transferrin receptor antibody shows that the transferrin receptor is also present in LLC vesicles (data not shown) and colocalizes with IA k molecules (Fig. 2b). The presence of the transferrin receptor in LLC was confirmed by Western blotting on isolated membrane fractions (Fig. 3).
LLC Is Accessible to the Horseradish Peroxidase Fluid-phase Marker and Transferrin-To determine the kinetics of horseradish peroxidase transport in EE, LE, and LLC, horseradish peroxidase was internalized for 5 min at 37°C and chased for various periods of time (Fig. 4A). After 5 min of internalization at 37°C, horseradish peroxidase enzymatic activity is first detected in low density EE. No significant horseradish peroxidase activity is detected in either LLC or LE. After a 15-min chase, horseradish peroxidase is clearly shifted from EE toward LLC, showing that LLC is accessible to horseradish peroxidase only after 15 min of internalization. After a 1-h chase, horseradish peroxidase is detected in LE, whereas almost none remains in EE and LLC.
The presence of the transferrin receptor in LLC prompted us to investigate the transferrin accessibility to LLC, EE, and LE. Fig. 4 (C and D) shows clearly that ϳ20% of internalized transferrin transits through LLC between 10 and 15 min. At 30 min, transferrin is no longer detectable in LLC, whereas small amounts start to appear in LE. These results favor the hypothesis defining LLC as a recycling compartment.

Inhibition of Vacuolar ATPase Prevents SDS-stable Dimer Formation and Induces Accumulation of Class II Molecules in LLC-
The proton pump vacuolar ATPase, known to generate the acidic luminal environment of endosomes, lysosomes (31)(32)(33)(34), and the Golgi apparatus (35), is directly inhibited by low concentrations of concanamycin B or bafilomycin A 1 (36 -39). Upon concanamycin B treatment, it has been shown that the formation of endosomal carrier vesicles ensuring the transport between early and late endosomes in baby hamster kidney cells is impaired (40). In contrast, in the HepG2 human hepatoma cell line, bafilomycin A 1 did not prevent the transport of horseradish peroxidase to LE, but almost totally inhibited the transport from LE to lysosomes (39). We first investigated the effects of concanamycin B on horseradish peroxidase endocytosis (Fig.  4B). Concanamycin B causes a 25% decrease in the amount of horseradish peroxidase internalized in EE after 5 min at 37°C (data not shown), probably by reducing the rate of horseradish peroxidase internalization as previously observed (39). After horseradish peroxidase internalization in EE, horseradish peroxidase recycling from EE to the external medium and horseradish peroxidase transport from EE to LE are partially inhibited (Fig. 4B). In contrast, LLC is accessible to horseradish peroxidase after 5 min of internalization followed by a 15-min chase as in control cells, suggesting that concanamycin B does not alter the transport of horseradish peroxidase to LLC (Fig.  4B). However, horseradish peroxidase transport out of LLC is slowed down, consistent with the reduced amounts of recycled horseradish peroxidase in the external medium.
We then investigated the effect of concanamycin B on the FIG. 2. LLC vesicles contain IA k molecules and transferrin receptor. After flotation and Percoll gradients, LLC were submitted to another flotation gradient step to remove the Percoll and centrifuged before sectioning. LLC ultrathin cryosections were labeled in a by using the anti-class II molecule antibody followed by rabbit anti-mouse IgG and 15-nm protein A-gold. In b, double immunolabeling shows a colocalization between the transferrin receptor (15-nm gold) and class II molecules (20-nm gold). Bars ϭ 200 nm (magnification ϫ 50,000 (a) and 37,500 (b)).

FIG. 3. Transferrin receptor present in LLC, EE, and LE membrane fractions.
10 g of LLC, EE, and LE proteins were loaded onto 12% SDS-polyacrylamide gel and transferred onto Immobilon-P membranes before incubation with the anti-transferrin receptor antibody, detected by an indirect chemiluminescence system (ECL, Amersham Corp.).

FIG. 1. Protein composition of LLC, EE, and LE.
Cells were metabolically labeled to equilibrium. After homogenization, EE, LE, and LLC were prepared. LLC was recovered from the high density membrane fractions 9 -11, and EE from the low density membrane fractions 3-7 when Percoll gradients were loaded with early endosomal fractions. LE corresponded to the low density membrane fractions 3-7 when Percoll gradients were loaded with late endosomal fractions. LLC, EE, and LE proteins were then analyzed by isoelectric focusing in the first dimension (direction of electrophoresis was from left ((Ϫ)-end) to right ((ϩ)-end)) followed by SDS-polyacrylamide gel electrophoresis in the second dimension and by autoradiography. Diamonds indicate the actin position in the three fractions. Arrows indicate the three groups of polypeptides that are strictly common to LLC and EE. The molecular mass markers are indicated by squares (97, 69, 46, 30, and 14 kDa from top to bottom).
formation of SDS-stable dimers in LLC and the transport of class II molecules out of LLC (Fig. 5). 2A4 cells were metabolically pulse-labeled for 20 min and chased for 30, 60, or 240 min in the presence or absence of 1 M concanamycin B. After a 30-min chase, class II molecules are detected only in membrane fractions 4 -8 of the Percoll gradient. No difference is observed in the presence or absence of concanamycin B. After a 1-h chase, class II molecules appear in the dense fractions 10 -12 of the Percoll gradient, indicating that the transport of class II molecules to LLC occurred. Even though this transport is not impaired in the presence of concanamycin B, concanamycin B inhibited SDS-stable dimer formation (compare fractions 10 -12 in the presence and absence of concanamycin B). Interestingly, after a 4-h chase, unstable class II molecules (␣and ␤-chains) accumulate in LLC in the presence of concanamycin B, whereas without concanamycin B, SDS-stable dimers are no longer detected in LLC, but are found in a low density membrane compartment as described previously (17). These data show that the formation of SDS-stable dimers normally occurs in an acidic environment and that the transport of class II molecules out of LLC is inhibited by concanamycin B.
In addition, we analyzed the effects of concanamycin B on the morphology of LLC at the electron microscopic level (Fig. 6). Without concanamycin B, LLC is composed of multivesicular and multilaminar vesicles containing membrane bud profiles with internal vesicles (Fig. 6a). Interestingly, in the presence of concanamycin B, we observe a central core containing intramembrane structures with extended tubules at the LLC periphery (Fig. 6b). One explanation might be that the budding of transport vesicles generated from LLC is impaired in the presence of concanamycin B, leading to the formation of tubular structures, as has already been observed for the formation of endosomal carrier vesicles (40).
Nocodazole Does Not Prevent SDS-stable Dimer Transport out of LLC-It is now well established that the transport of endocytosed materials from early to late endosomes depends on an intact microtubule network, whereas the recycling pathway from early endosomes to the plasma membrane does not (41-  B (Conc B). 0.8-ml Percoll gradient fractions were collected and pooled two by two; proteins were solubilized; and class II molecules were immunoprecipitated. Samples were treated under nonboiling conditions to preserve SDS-stable dimers before SDS-polyacrylamide gel electrophoresis and autoradiography. ␣ and ␤ correspond to the ␣and ␤-subunits of IA k class II molecules, respectively, and C (compact forms) corresponds to SDS-stable dimers of IA k molecules.
FIG. 6. Morphology of LLC vesicles in presence or absence of concanamycin B. Cells were incubated with or without 1 M concanamycin B for 6 h at 37°C and then homogenized. After flotation and Percoll gradients, fractions 9 -11 were pooled and submitted to another flotation gradient step to remove the Percoll. After pelleting and fixation with 2% glutaraldehyde, the LLC membrane pellets were embedded in Epon and sectioned for electron microscopy. Bar ϭ 500 nm (magnification ϫ 42,000). Arrowheads indicate the budding and tubular structures in a and b, respectively. 46). We tested the effect of nocodazole on (i) the transport of class II molecules in LLC, (ii) the formation of SDS-stable dimers, and (iii) their transport out of LLC (Fig. 7). Cells were pulse-labeled for 20 min, and radioactivity was chased for 1, 2, and 4 h. For each experiment, after a 30-min chase, nocodazole was added to allow the exit of class II molecules from the Golgi apparatus since it is believed that depolymerization of microtubules affects the transport of macromolecules in this organelle (47,48). Fractions 9 -11 of the Percoll gradient were pooled, and class II molecules were immunoprecipitated. As shown in Fig. 7, nocodazole treatment affects neither the transport of class II molecules to the LLC compartment nor the formation of SDS-stable dimers. Class II molecules (unstable and compact forms) appear in LLC after a 1-h chase, as already shown in Fig. 5. After a 4-h chase, SDS-stable dimers are no longer detected in LLC. These results suggest that the transport of class II molecules through LLC is independent of the microtubule network. DISCUSSION We have previously shown that LLC is involved in the formation of SDS-stable dimers between HEL and IA k molecules. LLC is a dense compartment different from EE (Rab4/Rab5positive), LE (Rab7-positive), and lysosomes (17). In the present paper, we have further characterized LLC, and we propose that LLC is a specialized compartment of the recycling pathway in murine B cells.
LLC corresponds to 0.5-m diameter vesicles containing multivesicular and multilaminar structures (Fig. 2). These structures were previously observed in B lymphocytes (12,49) and in Langerhans cells (50,51), and it has been postulated that multivesicular and multilaminar vesicles might correspond to various stages of MIIC maturation (12,49,50) even if the relationship between these different stages still remains unclear. One of the major LLC characteristics is that this compartment is morphologically different from tubulovesicular EE and LE (28,43,(52)(53)(54). It has also been shown that transport of endocytic material from EE to LE involves the multivesicular endosomal carrier vesicles (43). Their formation can be prevented by concanamycin B treatment (40), an inhibitor of the proton pump vacuolar ATPase. In our study, the biogenesis of LLC was not affected by concanamycin B (Fig. 6), indicating that LLC is probably different from endosomal carrier vesicles.
Strikingly, we present morphological and biochemical evidence that LLC contains the transferrin receptor (Figs. 2 and  3), a well known marker for both early and recycling endosomes, suggesting that LLC is associated with the early endosomal network, but is different from EE. More indications come from biochemical studies. Comparative analyses of the protein compositions of LLC, EE, and LE indicate that the LLC polypeptide pattern analyzed by two-dimensional gel electrophoresis is closely related to that of EE, but is different from that of LE (Fig. 1), suggesting that LLC may be generated from EE. This has been also suggested for the peptide-loading compartment CIIV (15,55) in murine B cells, whose characteristics are very similar to those of LLC except for their physical properties (high density for LLC versus low density for CIIV). Comparative studies of their respective protein compositions would be helpful to address the similarity between LLC and CIIV. More information is given by fluid-phase internalization of horseradish peroxidase and transferrin receptor-mediated endocytosis. Indeed, the kinetics of internalization of both the horseradish peroxidase fluid-phase marker and transferrin demonstrate that LLC is rapidly accessible to horseradish peroxidase in a 10 -15-min range, but after EE (Fig. 3).
We then studied the role of LLC in the HEL-IA k class II molecule association. When 2A4 B cells are treated with concanamycin B, the formation of LLC is not affected, and class II molecules are detected in LLC after a 1-h chase (Figs. 4 and 5). In contrast, the formation of SDS-stable dimers is impaired, as described previously in human B cells (38). This shows that the inhibition of SDS-stable dimer formation depends on the direct inactivation of the vacuolar proton pump, confirming that low pH is required for antigen processing and association with class II molecules in vitro (56,57). After a 4-h chase, concanamycin B induces an accumulation of unstable class II molecules in LLC (Fig. 5). This strongly suggests that the transport of class II molecules out of LLC occurs only after SDS-stable dimers have been generated, leading to the conclusion that the transport mechanism of class II complexes from LLC to the plasma membrane can be triggered by the tight association of HEL with class II molecules.
We then used nocodazole, a microtubule-depolymerizing agent known to affect the transport of endocytosed material from EE to later stages of the endocytic pathway, but not the recycling of receptors and solutes from EE to the plasma membrane (41)(42)(43)(44)(45)(46). In our study, nocodazole did not have any effect on class II molecule transport in and out LLC (Fig. 7), suggesting that those transport mechanisms occur in a microtubuleindependent manner. The facts that LLC contains the transferrin receptor and possesses a protein composition similar to that of EE and that nocodazole does not affect the transport of class II molecules in and out of LLC strongly suggest that LLC might be a compartment related to the recycling pathway.
Without underestimating the possibility that part of the class II molecules transit directly from the trans-Golgi network to LLC, we favor a model of the transport of class II molecules from the trans-Golgi network to LLC through EE. This model hypothesis is also supported by several lines of evidence. Newly synthesized ␣␤⅐Ii complexes can be digested by transferrincoupled neuraminidase before reaching the cell surface (58,59). In mouse B cells, at least 50% of the ␣␤⅐Ii complexes transit through early endosomes (16,60). We propose that, when leaving EE, class II molecules could be transported to the recycling LLC compartment, where they associate with HEL peptides. The formation of HEL-class II SDS-stable complexes might induce their targeting to the plasma membrane. This process could involve a direct fusion of LLC with the plasma membrane or vesicle-mediated membrane traffic between LLC and the plasma membrane. In human B cells, multivesicular MIIC have the propensity to fuse directly with the plasma membrane and to deliver internal vesicles called exosomes in the extracellular medium able to stimulate Th lymphocytes (49). This direct fusion has not been described in murine B cells, yet. However, LLC membrane fractions prepared for electron mi- Cells were metabolically labeled for 20 min and chased for 60, 120, or 240 min in the presence or absence of nocodazole (Noc). After a 30-min chase, 10 M nocodazole was added to the chase medium. LLC fractions were collected and pooled, and class II molecules were immunoprecipitated and analyzed as described in the legend to Fig. 5. ␣ and ␤ correspond to the ␣and ␤-subunits of IA k class II molecules, respectively, and C (compact forms) corresponds to SDS-stable dimers of IA k molecules.
croscopy show the presence of budding areas with internal membranes (Fig. 6). The significance of these buds is not clear, but they could generate vesicular compartments containing internal vesicles en route to the cell surface in a microtubuleindependent manner. Upon fusion with the plasma membrane, these vesicular compartments will release their content of vesicles in the external medium. Therefore, they could constitute a potential source of microvesicles with IA k molecules exposed at their surface interacting with T lymphocytes. Another possibility would be that these vesicles could fuse with EE or the tubular recycling compartment before reaching the plasma membrane since, after a 4-h chase, SDS-stable dimers cosediment with Rab4-and Rab5-positive early endosomal fractions (Fig. 5) (17). Further studies are required to investigate the mechanisms of the biogenesis of these vesicles. In particular, in vitro assays reconstituting LLC biogenesis and vesicle transport out of LLC will certainly be an important issue for the coming years.