Although it is accepted that particulate Ags are more immunogenic than soluble Ags in vivo, it is unclear whether this effect can be explained solely through enhanced uptake by APCs. In this study we demonstrate that vesicle size modulated the efficiency of Ag presentation by murine macrophages and that this effect was accompanied by a profound change in trafficking of Ag. Ag prepared in large particles (560 nm) was delivered into early endosome-like, immature phagosomes, whereas smaller vesicles (155 nm) and soluble Ags localized rapidly to late endosomes/lysosomes. However, peptide/class II complexes could be detected in both compartments. Phagosomes formed on uptake of large vesicles recruit Ag-processing apparatus while retaining the characteristics of early endosomes. In contrast, smaller vesicles bypassed this compartment, appeared to go more rapidly to lysosomal compartments, and exhibited reduced Ag-presenting efficiency. We conclude that the ability of phagocytosed, particulate Ag to target early phagosomes results in more efficient Ag presentation.

Although it is accepted generally that particulate Ags are inherently more immunogenic than soluble Ags (1, 2, 3, 4), it is unclear whether this enhanced immunogenicity is due solely to the increased efficiency of uptake by APCs. This feature of particles makes them attractive adjuvants or delivery systems for vaccine Ags, and consequently, a number of formulation technologies, such as microparticles, liposomes, emulsions, virus-like proteins, and chemically polymerized Ags, have been developed to exploit this phenomenon (4). Formulation of Ag in particles has also been demonstrated to qualitatively affect the immune response to Ags, with particulate Ags favoring Th1-type responses and soluble Ags promoting Th2 (2, 5, 6). Despite the significance and widespread application of this observation in vaccine design, it remains unclear exactly how rendering an Ag particulate can affect its immunological profile.

In vivo studies have demonstrated that particles are trafficked from the site of injection to local lymph nodes by mononuclear phagocytes (7, 8, 9, 10). This suggests that particulate Ag/adjuvant formulations may act by targeting Ag to phagocytic APCs in vivo, resulting in more efficient presentation to T cells (10, 11). In vitro studies have indicated that the targeting effect mediated by particulate Ag may be due to the higher Ag concentration compared with soluble Ag (3). However, particulate Ags have also been shown to influence phagocytic APCs directly via effects on intracellular trafficking of Ag to Ag-processing compartments (12). Furthermore, the ability of particles to target Ag into phagocytic, rather than pinocytic, uptake mechanisms has also been shown to affect APC function (6, 13, 14).

In the current studies we have characterized the consequences of variation in Ag uptake and subsequent trafficking, processing, and presentation of soluble Ag or Ag prepared in large (>200 nm) and small (<200 nm) lipid vesicles. We confirm that large and small lipid vesicles select different mechanisms of endocytic uptake by macrophages and demonstrate that subsequent processing of Ag prepared in large, but not small, lipid vesicles occurs extensively in prelysosomal compartments. These distinct differences in endocytic trafficking are linked to more efficient presentation of Ag prepared in large vs small lipid vesicles.

Ultra-high purity OVA was purchased from Worthington Biochemical (Freehold, NJ), and Alexa Fluor 488-labeled OVA was obtained from Molecular Probes (Eugene, OR). HRP was purchased from Sigma-Aldrich (Poole, U.K.). Ags were labeled with hydrazide derivatives of fluorochromes (Molecular Probes) by periodate activation of carbohydrate moieties (15). Paramagnetic iron-dextran was prepared by dextran (40 kDa; Sigma-Aldrich) precipitation of magnetite from a mixture of FeCl2/FeCl3 (both from Sigma-Aldrich) as described previously (16). Rat anti-lysosome-associated membrane glycoprotein (anti-LAMP-1) mAb was prepared from 1D4B hybridoma cells (Developmental Hybridoma Bank, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD.). Anti-procathepsin D was generated by immunization of rabbits with a synthetic peptide (ILKGPITKYSMQSS) representing the murine pro-sequence of cathepsin D (15). Rabbit anti-mouse H2-DMa antiserum was a gift from J. Monaco (University of Cincinnati, Cincinnati, OH) (17), and rabbit anti-IFN-γ-induced lysosomal thiol-reductase (anti-GILT) antiserum was a gift from P. Cresswell (Section of Immunobiology, Yale University School of Medicine, New Haven, CT). Mouse anti-FITC (clone 4-4-20) was purchased from Molecular Probes, and rat anti-transferrin receptor (CD71, clone C2) and anti-CD4 (clone L3T4) were purchased from BD Biosciences (Oxford, U.K.). Secondary Abs for Western blotting were purchased from Sigma-Aldrich (anti-rat), or New England Biolabs (Beverly, MA; anti-rabbit and anti-biotin), and gold-labeled secondary Abs for immunoelectron microscopy were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).

Macrophages were derived from bone marrow cells isolated from BALB/c or CBA/ca mice (Harlan Olac, Bicester, U.K.) and were cultured for 7–10 days in DMEM supplemented with 10% FCS, 5% horse serum, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate (Invitrogen Life Technologies, Paisley, U.K.), and 20% L-929 conditioned medium (BMMo medium). Where appropriate, macrophages were activated by incubation for 24 h with 50 U/ml IFN-γ (R&D Systems, Oxon, U.K.).

Lipid vesicles were prepared from 1-monopalmitoyl glycerol, cholesterol, and dicetyl phosphate (Sigma-Aldrich) in the molar ratio 5:4:1, as described previously (6). In some experiments the lipid envelopes of vesicles were radiolabeled by incorporation of [3H]cholesterol (Amersham Biosciences, Aylesbury, U.K.) or were opsonized by incorporation of FITC-labeled dihexadecanoyl-glycerophosphoethanolamine (Molecular Probes), followed by treatment with mouse anti-FITC (Molecular Probes). Ags were entrapped in lipid vesicles by repeated cycles of freezing and thawing, and defined sizes of vesicles were prepared by extrusion through decreasing pore size polycarbonate filters at 60°C using a thermobarrel extruder as described previously (6). Nonentrapped Ag was removed by centrifuging at 100,000 × g for 45 min, and the protein concentrations of the vesicle suspensions were determined using a modified ninhydrin assay (6). The particle size of the resulting lipid vesicles were demonstrated to be routinely 560 ± 60 nm after extrusion through 800-nm pore size filters and 155 ± 10 nm after extrusion through 100-nm pore size filters. Lipid vesicles were postformation labeled with molecular gold linked to palmitic acid (Palmitoyl Nanogold; Nanoprobes, Yaphank, NY) to label only the outer envelope.

Macrophages were cultured in BMMo medium on sterile no. 1.5, 13-mm diameter coverslips (BDH, Poole, U.K.) and placed in 24-well tissue culture dishes (Costar, Cambridge, MA). Where appropriate, cells were pulsed for 1 h with 1 mg/ml FITC-dextran (Sigma-Aldrich). After a minimum 6-h chase in BMMo medium, cells were incubated for 4 min with lipid vesicles with or without FITC-dextran (1 mg/ml). After chasing with BMMo medium, coverslips were removed, washed in PBS (Invitrogen Life Technologies), and placed on slides, and the edges were sealed with nail varnish. Cells were viewed under a ×63 objective, and images were captured by a CCD camera. Where appropriate, series of z-sections through the cells were digitally deconvolved using Openlab software (Improvision, Coventry, U.K.).

Macrophages were processed for immunoelectron microscopy as previously described (18). Briefly, cells were incubated with soluble FITC-HRP and/or HRP-biotin prepared in lipid vesicles for 10 min, chased for 0 or 20 min at 37°C, then fixed for 1 h on ice with 1% glutaraldehyde in 200 mM PIPES/0.5 mM MgCl2 (pH 7.1). Macrophages were encapsulated in gelatin and infiltrated with 2.3 M sucrose/20% polyvinyl pyrrolidone in PIPES/MgCl2. The samples were trimmed and frozen, and sections were prepared on an RMC (Tucson, AZ) MT7/CR21 cryoultra-microtome. Sections were probed with rabbit anti-FITC Ab (Molecular Probes), rat anti-LAMP-1 mAb, or mouse anti-biotin Ab (New England Biolabs) and detected with appropriate gold-labeled secondary Abs (Jackson ImmunoResearch Laboratories). To reveal 1.4 nm palmitoyl nanogold labeling of the lipid vesicle membrane, macrophages were grown on Thermanox coverslips (Nunc, Naperville, IL) incubated with golden lipid vesicles for 10 min, and fixed immediately in 2.5% glutaraldehyde for 30 min at 37°C. After washing in 50 mM glycine/PBS, cells were brought gradually into distilled water and gold enhanced (GoldEnhance; Nanoprobes) for 5 min at 4°C. Cells were then routinely processed for transmission electron microscopy through 1% aqueous OsO4 and 1% aqueous uranyl acetate before dehydration in an ethanol series and embedding in Spurrs epoxy resin. Ultrathin sections (80 nm thick) were uranyl and lead stained before electron imaging.

Macrophages (∼2 × 107) were incubated with soluble HRP or HRP prepared in lipid vesicles prepared with [3H]cholesterol for 4 min at 37°C. After chase periods of 0–40 min, cells were harvested in ice-cold homogenization buffer (250 mM sucrose, 0.5 mM EDTA, 0.5 mM EGTA, 20 mM HEPES, and 0.05% gelatin). Microsomes were prepared by disrupting cells by repeated passage through a 25-gauge needle and low speed centrifugation at 100 × g to remove nuclei. Density gradient electrophoresis (DGE)3 apparatus was used to separate microsome extracts based on charge and buoyant density as described previously (15, 19). Microsomes prepared in DGE buffer (250 mM sucrose, 1 mM EDTA, 0.5 mM EGTA, and 10 mM triethanolamine) and 10% Ficoll were layered onto a solution of 12% Ficoll/DGE buffer, and a continuous gradient of 0–8% Ficoll/DGE buffer was layered on top. Microsomes were then separated into the gradient by application of 12 mA constant current for 2 h. Three separate fractions could be isolated, which were then analyzed for the presence of HRP enzyme activity by dot blot, [3H]cholesterol by scintillation counting and the endosome markers described above by Western blot.

Macrophages were pulsed with soluble OVA or OVA prepared in lipid vesicles, chased for various periods of time, fixed with 1% paraformaldehyde/PBS (Sigma-Aldrich), and quenched with 0.06% glycine-glycine/PBS (Sigma-Aldrich). The presentation of OVA by macrophages was quantified using the DO11.GFP T cell hybridoma supplied by Dr. D. Underhill (Institute for Systems Biology, Seattle, WA). DO11.GFP is derived from the DO11.10 T hybridoma that recognizes residues 323–339 of OVA in the context of MHC class II (I-Ad). However, DO11.GFP has been modified to express GFP under the control of a NFAT-responsive promoter (20). Previous studies have demonstrated that levels of Ag presentation are directly proportional to the intensity of T cell-associated GFP fluorescence detected by FACS (20). The presence of MHC/OVA323–339 peptide complexes in DGE-isolated cellular organelles could also be determined after fixation of organelles as previously described (21).

Macrophages were incubated with lipid vesicles loaded with iron-dextran and chased, and microsome extracts were prepared as described above. Compartments containing iron-dextran-loaded lipid vesicles were isolated by magnetic separation (MACS; Miltenyi Biotec, Bisley, U.K.) and then characterized by Western blot analysis.

Our previous studies indicated that different sizes of lipid vesicles have similar capacities to be internalized by peritoneal macrophages (6). Fig. 1,a confirms this observation, demonstrating that Ag prepared in both sizes of particles associates with bone marrow macrophages to a greater extent than a similar amount of soluble Ag. Furthermore, this was true for both surface binding as well as internalization of lipid vesicles (Fig. 1,b). The ability of OVA-pulsed macrophages to present Ag was determined after various chase periods (Fig. 1,c). In each case, peak presentation occurred with macrophages fixed after 6-h chase. In line with Fig. 1 a, both large and small vesicles resulted in enhanced Ag presentation compared with soluble Ag; however, despite similar levels of uptake, large vesicles resulted in significantly higher levels of presentation, indicating that Ag in large vesicles was processed more efficiently. The differential effect that vesicle size has on Ag presentation indicates that the increased efficiency of presentation is due to two, functionally separable phenomena: the increased dosage achieved by particulate delivery and the enhanced presentation dependent on vesicle size.

FIGURE 1.

Presentation of Ag prepared in large (560 nm) lipid vesicles is more efficient than Ag prepared in small (155 nm) lipid vesicles. a, Uptake of Alexa Fluor 488-labeled OVA in a solution prepared in large or small lipid vesicles was determined. Macrophages were incubated with these preparations for the indicated times, and internalization was determined as the mean fluorescence intensity by FACS analysis. The vesicle-associated Ag was internalized similarly, regardless of vesicle size, and this level of uptake was significantly higher than soluble Ag at all time points tested (p < 0.01). b, Both internalization and surface binding of large (560 nm) and small (155 nm) lipid vesicles containing Alexa Fluor-488 labeled OVA were similar. Bone marrow-derived macrophages were incubated with these preparations for the indicated times, and cell-associated signal was determined as described above. Surface binding was determined by incubating the preparations with the macrophages at 4°C to block internalization. Levels of surface-bound and total cell-associated signal (incubation at 37°C) were similar regardless of vesicle size. c, Levels of OVA-derived 323–339 peptide/class II complexes was determined by fixing cells after various chase periods and incubating them with the DO.11.GFP hybridoma. Both vesicle preparations induced greater Ag presentation than soluble Ag (p < 0.05), although despite similar levels of uptake, large lipid vesicles could induce higher levels of Ag presentation at 6 h than small vesicles (p < 0.05). This effect was dependent on Ag being prepared inside the lipid vesicles (data not shown).

FIGURE 1.

Presentation of Ag prepared in large (560 nm) lipid vesicles is more efficient than Ag prepared in small (155 nm) lipid vesicles. a, Uptake of Alexa Fluor 488-labeled OVA in a solution prepared in large or small lipid vesicles was determined. Macrophages were incubated with these preparations for the indicated times, and internalization was determined as the mean fluorescence intensity by FACS analysis. The vesicle-associated Ag was internalized similarly, regardless of vesicle size, and this level of uptake was significantly higher than soluble Ag at all time points tested (p < 0.01). b, Both internalization and surface binding of large (560 nm) and small (155 nm) lipid vesicles containing Alexa Fluor-488 labeled OVA were similar. Bone marrow-derived macrophages were incubated with these preparations for the indicated times, and cell-associated signal was determined as described above. Surface binding was determined by incubating the preparations with the macrophages at 4°C to block internalization. Levels of surface-bound and total cell-associated signal (incubation at 37°C) were similar regardless of vesicle size. c, Levels of OVA-derived 323–339 peptide/class II complexes was determined by fixing cells after various chase periods and incubating them with the DO.11.GFP hybridoma. Both vesicle preparations induced greater Ag presentation than soluble Ag (p < 0.05), although despite similar levels of uptake, large lipid vesicles could induce higher levels of Ag presentation at 6 h than small vesicles (p < 0.05). This effect was dependent on Ag being prepared inside the lipid vesicles (data not shown).

Close modal

Localization of different sizes of lipid vesicles in macrophages by fluorescence microscopy was pursued with cells pulse-chased with FITC-dextran to label lysosomes (12) and then pulsed with lipid vesicles loaded with Texas Red-labeled HRP. Small lipid vesicles were clearly identified inside macrophages, and some degree of colocalization with lysosomes could be detected after a 4-min pulse (Fig. 2, a and b). After 20-min chase, virtually all the signal from small lipid vesicles was detected in FITC-dextran-positive lysosomes (Fig. 2, c and d). In contrast, although some of the signal associated with large lipid vesicles could be detected in lysosomes at 0 and 20 min chase, colocalization was incomplete, particularly in lipid vesicle-containing compartments around the periphery of the cells (Fig. 2, e–h).

FIGURE 2.

Small (155 nm) lipid vesicles rapidly colocalize with FITC-dextran-labeled lysosomes, whereas large lipid vesicles (560 nm) do not. The lysosomes of macrophages were labeled by 1-h pulse and overnight chase with FITC-dextran. Cells were then incubated with lipid vesicles containing Texas Red-HRP. a and b, Small lipid vesicles rapidly colocalized with FITC-dextran in perinuclear lysosomes, and colocalization was complete in 20 min (c and d). In contrast, large lipid vesicles occupy peripheral compartments, which do not contain FITC-dextran (c and d) even after 20-min chase (e and f).

FIGURE 2.

Small (155 nm) lipid vesicles rapidly colocalize with FITC-dextran-labeled lysosomes, whereas large lipid vesicles (560 nm) do not. The lysosomes of macrophages were labeled by 1-h pulse and overnight chase with FITC-dextran. Cells were then incubated with lipid vesicles containing Texas Red-HRP. a and b, Small lipid vesicles rapidly colocalized with FITC-dextran in perinuclear lysosomes, and colocalization was complete in 20 min (c and d). In contrast, large lipid vesicles occupy peripheral compartments, which do not contain FITC-dextran (c and d) even after 20-min chase (e and f).

Close modal

When macrophages were pulsed with a mixture of Texas Red-labeled HRP prepared in lipid vesicles and FITC-dextran in solution, a similar pattern of distribution was observed. Almost complete colocalization of small lipid vesicles with FITC-dextran was observed after pulse (Fig. 3,a), and this colocalization was complete after 20-min chase (Fig. 3,b). These markers colocalized in the perinuclear areas of cells, consistent with these compartments being lysosomes. In contrast, large vesicles did not colocalize with the fluid phase marker after pulse (Fig. 3,c) or 20-min chase (Fig. 3d), and as shown in Fig. 2, compartments containing solely lipid vesicle-associated label were seen at the periphery of these cells. To characterize the mechanism of uptake of the different substrates, cells were incubated with cytochalasin D for 20 min before and during pulse with lipid vesicles and fluid phase markers. This treatment clearly did not reduce internalization of FITC-dextran or small lipid vesicles (Fig. 3,e), but did block internalization of large lipid vesicles (Fig. 3 f).

FIGURE 3.

Small (155 nm), but not large (560 nm), lipid vesicles colocalize with markers of fluid phase endocytosis and can be internalized independently of phagocytosis. Macrophages were pulsed with a mixture of FITC-dextran (70 kDa) as a marker of fluid phase pinocytosis and the various sizes of vesicle containing Texas Red-HRP. a and b, Although small lipid vesicles colocalize with FITC-dextran in perinuclear compartments (probably lysosomes), large lipid vesicles occupy peripheral compartments, which do not contain FITC-dextran. c and d, This was particularly clear after 20-min chase. e and f, Uptake of small lipid vesicles and fluid phase markers was not inhibited by treatment of macrophages with cytochalasin D. In contrast, uptake of large lipid vesicles was inhibited, indicating that these particles were ingested by a mechanism involving actin skeleton rearrangement, probably phagocytosis.

FIGURE 3.

Small (155 nm), but not large (560 nm), lipid vesicles colocalize with markers of fluid phase endocytosis and can be internalized independently of phagocytosis. Macrophages were pulsed with a mixture of FITC-dextran (70 kDa) as a marker of fluid phase pinocytosis and the various sizes of vesicle containing Texas Red-HRP. a and b, Although small lipid vesicles colocalize with FITC-dextran in perinuclear compartments (probably lysosomes), large lipid vesicles occupy peripheral compartments, which do not contain FITC-dextran. c and d, This was particularly clear after 20-min chase. e and f, Uptake of small lipid vesicles and fluid phase markers was not inhibited by treatment of macrophages with cytochalasin D. In contrast, uptake of large lipid vesicles was inhibited, indicating that these particles were ingested by a mechanism involving actin skeleton rearrangement, probably phagocytosis.

Close modal

Immunoelectron microscopy was performed to examine the localization of large and small lipid vesicles in bone marrow macrophages in more detail. These studies were performed comparably to those detailed in Fig. 3; however, FITC-HRP was used to label the internal compartments of lipid vesicles, and biotin-HRP was used as a marker of fluid phase uptake; these markers were detected on cryosections with anti-FITC/18-nm gold and avidin/12-nm gold conjugates, respectively. Immediately after a pulse with a mixture of soluble biotin-HRP and small lipid vesicles containing FITC-HRP, both markers colocalized in compartments that had a characteristic multivesicular appearance (Fig. 4,a). After 20-min chase, both markers continued to colocalize, and the multivesicular appearance of these compartments was more pronounced (Fig. 4,b). In contrast, large lipid vesicles did not colocalize with the accumulated soluble marker at either time point (Fig. 4, c and d), and the compartments occupied by these lipid vesicles had a transparent appearance and were morphologically quite distinct from the multivesicular compartments occupied by small vesicles. The multivesicular compartments occupied by small vesicles had a similar appearance to prelysosomal compartments that have been associated with Ag loading onto class II (MHC class II) (22). These compartments also contained the lysosomal marker LAMP-1, confirming its identity as a prelysosomal/lysosomal compartment (22), whereas LAMP-1 was considerably less abundant in the compartments occupied by the large lipid vesicles. Large and small lipid vesicles were shown to remain intact in macrophage endosomes through labeling of the peripheral membrane with palmitoyl nanogold (Fig. 4, g and h). Although macrophages pulsed with small lipid vesicles contained multiple lipid vesicles per endosome, single lipid vesicles occupied the endosomes of macrophages pulsed with large lipid vesicles. Furthermore, the endosome membrane was tightly apposed to the large lipid vesicles, unlike small lipid vesicles, which occupied spacious compartments.

FIGURE 4.

Small (155 nm), but not large (560 nm), lipid vesicles traffic into multivesicular, LAMP-1-positive compartments containing fluid phase markers. Macrophages were pulsed with a mixture of biotin-labeled HRP, as a marker of fluid phase pinocytosis, and the various sizes of vesicle containing FITC-HRP. a and b, Although small lipid vesicles colocalize containing FITC-HRP (18 nm gold) with biotin-HRP (12 nm gold) in electron-dense compartments, large lipid vesicles occupy electron transparent compartments, which do not contain biotin-HRP. c and d, This was particularly evident after 20-min chase. e and f, After 20-min chase, the small lipid vesicles containing FITC-HRP (18 nm gold) occupied multivesicular compartments, which were LAMP-1 positive (6 nm gold), both characteristics of classical peptide-loading compartments (MHC class II). In contrast, large lipid vesicles were found in LAMP-1-negative electron-transparent compartments. g and h, Endosomes formed by internalization of small and large golden lipid vesicles by bone marrow macrophages were also morphologically distinct. g, Multiple small lipid vesicles could be identified in macrophage endosomes that appear spacious (h). In contrast, large lipid vesicles were taken up into single endosomes, and the membrane (as identified by gold staining) appeared closely apposed to the endosome membrane. Representative images from two independent experiments are shown. Magnifications: a, ×53,000; b, ×54,000; c, ×34,000; d, ×48,000; e, ×55,000; f, ×37,000; g, ×48,125; h, ×48,000.

FIGURE 4.

Small (155 nm), but not large (560 nm), lipid vesicles traffic into multivesicular, LAMP-1-positive compartments containing fluid phase markers. Macrophages were pulsed with a mixture of biotin-labeled HRP, as a marker of fluid phase pinocytosis, and the various sizes of vesicle containing FITC-HRP. a and b, Although small lipid vesicles colocalize containing FITC-HRP (18 nm gold) with biotin-HRP (12 nm gold) in electron-dense compartments, large lipid vesicles occupy electron transparent compartments, which do not contain biotin-HRP. c and d, This was particularly evident after 20-min chase. e and f, After 20-min chase, the small lipid vesicles containing FITC-HRP (18 nm gold) occupied multivesicular compartments, which were LAMP-1 positive (6 nm gold), both characteristics of classical peptide-loading compartments (MHC class II). In contrast, large lipid vesicles were found in LAMP-1-negative electron-transparent compartments. g and h, Endosomes formed by internalization of small and large golden lipid vesicles by bone marrow macrophages were also morphologically distinct. g, Multiple small lipid vesicles could be identified in macrophage endosomes that appear spacious (h). In contrast, large lipid vesicles were taken up into single endosomes, and the membrane (as identified by gold staining) appeared closely apposed to the endosome membrane. Representative images from two independent experiments are shown. Magnifications: a, ×53,000; b, ×54,000; c, ×34,000; d, ×48,000; e, ×55,000; f, ×37,000; g, ×48,125; h, ×48,000.

Close modal

To define the compartments of macrophages occupied by different sizes of lipid vesicles, we applied density gradient electrophoresis to resolve distinct endosomal compartments as described previously (15, 19). After separation by DGE for ∼2 h, the original microsome extract of macrophages could be resolved into three distinct fractions (Fig. 5,a; I, II, and III), where I is the least and III is the most anodally deflected fraction. The endosomal compartments present in each fraction were defined functionally by pulsing macrophages with HRP and chasing with fresh medium for various times before preparation of microsomes and DGE fractionation. Immediately after pulse, HRP could be detected at low levels in fraction I, and the HRP signal was primarily detected in fraction II (Fig. 5,a). After 10-min chase, the HRP signal was confined to fraction III. Western blot analysis of these fractions demonstrated that the pro-form of cathepsin D could be detected in fractions I and II, but not in fraction III (Fig. 5 b). Conversely, unlike fractions I and II, fraction III was clearly LAMP-1 positive. Together with previous studies (15, 19), these analyses indicate that fraction I contained early endosomes/plasma membrane, fraction II contained middle endosomes, and fraction III contained late endosomes/lysosomes.

FIGURE 5.

Fractionation of the endosomal network by DGE demonstrates that large lipid vesicles and their cargo traffic into early endosome-like phagosomes, whereas small lipid vesicles rapidly localize in late endosomes/lysosomes. Macrophages were pulsed with either soluble HRP or HRP prepared in the various sizes of vesicle, with incorporated [3H]cholesterol. After various chase times, microsome extracts were prepared and separated into three distinct fractions using DGE (I, II, and III, where I is the least and III is the most anodally deflected fraction). a, Immediately after pulse, soluble HRP was detected at low levels in fraction I and mainly in fraction II. After 10-min chase or longer, the HRP signal was confined to fraction III. b, Western blot analysis of the separated fractions indicates high expression of LAMP-1 in fraction III, with little or no expression found in fractions I or II. In contrast, probing the blot with an Ab against the pro-sequence of cathepsin D revealed an inverse distribution relative to LAMP-1. Together with previous studies (1519 ), these analyses indicate that fraction I contained early endosomes, fraction II contained middle endosomes, and fraction III contained late endosomes/lysosomes. c and d, HRP prepared in small vesicles traffics rapidly into late endosomes/lysosomes with similar kinetics as soluble Ag, whereas HRP prepared in large vesicles resides in early endosome compartments for the duration of the experiment, although some signal can be detected in late endosomes/lysosomes after 20 min. e and f, [3H]cholesterol incorporated into the lipid vesicle bilayer follows similar kinetics and trafficking as HRP, suggesting that both the Ag and the lipid vesicle are in the same compartments.

FIGURE 5.

Fractionation of the endosomal network by DGE demonstrates that large lipid vesicles and their cargo traffic into early endosome-like phagosomes, whereas small lipid vesicles rapidly localize in late endosomes/lysosomes. Macrophages were pulsed with either soluble HRP or HRP prepared in the various sizes of vesicle, with incorporated [3H]cholesterol. After various chase times, microsome extracts were prepared and separated into three distinct fractions using DGE (I, II, and III, where I is the least and III is the most anodally deflected fraction). a, Immediately after pulse, soluble HRP was detected at low levels in fraction I and mainly in fraction II. After 10-min chase or longer, the HRP signal was confined to fraction III. b, Western blot analysis of the separated fractions indicates high expression of LAMP-1 in fraction III, with little or no expression found in fractions I or II. In contrast, probing the blot with an Ab against the pro-sequence of cathepsin D revealed an inverse distribution relative to LAMP-1. Together with previous studies (1519 ), these analyses indicate that fraction I contained early endosomes, fraction II contained middle endosomes, and fraction III contained late endosomes/lysosomes. c and d, HRP prepared in small vesicles traffics rapidly into late endosomes/lysosomes with similar kinetics as soluble Ag, whereas HRP prepared in large vesicles resides in early endosome compartments for the duration of the experiment, although some signal can be detected in late endosomes/lysosomes after 20 min. e and f, [3H]cholesterol incorporated into the lipid vesicle bilayer follows similar kinetics and trafficking as HRP, suggesting that both the Ag and the lipid vesicle are in the same compartments.

Close modal

Analysis of the movement of HRP prepared in lipid vesicles through these compartments indicated that small vesicles had a similar temporal distribution as soluble HRP (Fig. 5,c). In contrast, large lipid vesicles caused the arrest of HRP in the early endosome fraction for the duration of the experiment, and HRP signal could only be detected in the lysosome/late endosomes after 20 min of chase (Fig. 5,d). When [3H]cholesterol was used as a marker of the lipid envelope, temporal distribution was observed similar to that seen for HRP prepared in lipid vesicles (Fig. 5, e and f). Although it was possible that these differences may have reflected different efficiencies in separation, total 3H levels detected at each time point or for each size of lipid vesicle did not significantly vary, making this explanation unlikely (data not shown). Furthermore, the DGE data presented also confirm the rapid lysosomal distribution of small lipid vesicles already identified by fluorescence and electron microscopy.

We investigated the location of the compartment where MHC class II molecules become loaded with Ag for each of the vesicle preparations. This was performed by DGE separation of macrophages pulsed for 10 min with OVA prepared in lipid vesicles, followed by detection of intact class II OVA323–339 complexes using the Do.11.GFP hybridoma. After pulse with Ag in large vesicles and 20-min chase, loading of OVA peptide could be detected equally in early endosome and late endosome/lysosome compartments (Fig. 6 a). In contrast, whereas equal levels of class II/peptide complexes could be detected in late endosome/lysosome compartments after pulse with OVA prepared in small vesicles, the levels of complex in early endosome fractions were greatly reduced compared with those in the large vesicle preparation.

FIGURE 6.

Peptide/class II complexes are found in early phagosomes that contain Ag-processing apparatus when Ag is prepared in large, but not small, lipid vesicles. Macrophages were pulsed for 10 min with a constant amount of OVA prepared in either solution (OVA) or lipid vesicles with a mean diameter of 560 nm (560 nm/OVA) or155 nm (155 nm/OVA). a, After 40-min chase, microsomes were prepared from macrophages and subjected to fractionation by DGE. Fractions were then fixed and quenched before the addition of DO.11.GFP T hybridoma cells to detect class II loaded with OVA peptide (aa 323–339). Peptide-loaded class II could be detected in late, but not early, endosome-like fractions when Ag was prepared in small vesicles (155 nm). In contrast, class II with loaded peptide could be detected in both early endosome and late endosome/lysosome compartments when Ag was prepared in large (560 nm) lipid vesicles. Unstimulated levels of DO.11.GFP fluorescence are shown (Control), and presentation could not be detected after treatment with empty lipid vesicles prepared in the absence of OVA (data not shown). b, After various periods of chase with iron-dextran prepared in large lipid vesicles, the magnetically isolated microsome fraction was analyzed by Western blot. Confirming previous studies, the isolated fraction contained early endosome markers throughout the experiment, with LAMP-1 only being acquired after 20-min chase. Molecules involved Ag processing and presentation (class II, H-2M, and GILT) could be detected throughout the experiment.

FIGURE 6.

Peptide/class II complexes are found in early phagosomes that contain Ag-processing apparatus when Ag is prepared in large, but not small, lipid vesicles. Macrophages were pulsed for 10 min with a constant amount of OVA prepared in either solution (OVA) or lipid vesicles with a mean diameter of 560 nm (560 nm/OVA) or155 nm (155 nm/OVA). a, After 40-min chase, microsomes were prepared from macrophages and subjected to fractionation by DGE. Fractions were then fixed and quenched before the addition of DO.11.GFP T hybridoma cells to detect class II loaded with OVA peptide (aa 323–339). Peptide-loaded class II could be detected in late, but not early, endosome-like fractions when Ag was prepared in small vesicles (155 nm). In contrast, class II with loaded peptide could be detected in both early endosome and late endosome/lysosome compartments when Ag was prepared in large (560 nm) lipid vesicles. Unstimulated levels of DO.11.GFP fluorescence are shown (Control), and presentation could not be detected after treatment with empty lipid vesicles prepared in the absence of OVA (data not shown). b, After various periods of chase with iron-dextran prepared in large lipid vesicles, the magnetically isolated microsome fraction was analyzed by Western blot. Confirming previous studies, the isolated fraction contained early endosome markers throughout the experiment, with LAMP-1 only being acquired after 20-min chase. Molecules involved Ag processing and presentation (class II, H-2M, and GILT) could be detected throughout the experiment.

Close modal

To confirm that the Ag processing and presentation apparatus was present in the large lipid vesicle phagosome and not in other early endosome compartments that could comigrate on DGE, we loaded the vesicles with iron-dextran and isolated the lipid vesicle-containing compartment magnetically. Western blotting confirmed the previous DGE results, indicating that large lipid vesicles resided in compartments that only became LAMP-1 positive after 20-min chase and also that these compartments were accessible to transferrin receptor, as a marker of recycling compartments, throughout the time course (23). The isolated phagosomes also contained MHC class II; H2-M, a chaperone of class II peptide loading (17); and GILT, a thiol-reductase necessary for Ag processing before peptide loading (24). Collectively, these data argue strongly that delivery of Ag in large vesicles enhances its retention in a prelysosomal compartment that is capable of efficient processing and presentation of OVA.

Previous studies have demonstrated that uptake of opsonized particles via macrophage FcRs results in increased formation of phagolysosomes (25). Therefore, we prepared different sizes of lipid vesicles containing OVA, opsonized with Ab, and determined their distribution in macrophages. In contrast to the data shown in Fig. 2, after 20-min chase, the lipid vesicles colocalized with lysosomes regardless of their size (Fig. 7,a). When we then determined the degree of Ag presentation induced by each formulation, opsonization of small lipid vesicles had no effect on presentation, suggesting that the opsonized vesicles were not simply activating macrophages or quantitiatively affecting Ag uptake. In contrast, opsonization of large lipid vesicles significantly lowered their efficiency of Ag presentation compared with that of nonopsonized control formulations (Fig. 7 b).

FIGURE 7.

Directing large lipid vesicles into lysosomes by opsonization reduces Ag-presenting efficiency to similar levels as in small lipid vesicles. Large and small lipid vesicles containing OVA were prepared with FITC-labeled lipid. Lipid vesicles were then opsonized by the addition of mouse anti-FITC and incubated with macrophages that had been pulse/chased with Texas Red-HRP to label lysosomes. a, In contrast to unopsonized lipid vesicles (Fig. 2), opsonized lipid vesicles rapidly localized in lysosomes regardless of size. b, Opsonization of large lipid vesicles also significantly reduced the level of OVA presentation (p < 0.05) to a similar extent as observed with small lipid vesicles. Opsonization of small lipid vesicles did not alter localization or Ag presentation efficiency.

FIGURE 7.

Directing large lipid vesicles into lysosomes by opsonization reduces Ag-presenting efficiency to similar levels as in small lipid vesicles. Large and small lipid vesicles containing OVA were prepared with FITC-labeled lipid. Lipid vesicles were then opsonized by the addition of mouse anti-FITC and incubated with macrophages that had been pulse/chased with Texas Red-HRP to label lysosomes. a, In contrast to unopsonized lipid vesicles (Fig. 2), opsonized lipid vesicles rapidly localized in lysosomes regardless of size. b, Opsonization of large lipid vesicles also significantly reduced the level of OVA presentation (p < 0.05) to a similar extent as observed with small lipid vesicles. Opsonization of small lipid vesicles did not alter localization or Ag presentation efficiency.

Close modal

As demonstrated previously, the delivery of Ag inside lipid vesicles enhances the uptake of Ag by APCs compared with soluble Ag (3, 10, 11). Not surprisingly, the delivery of Ags in particulate form to professional phagocytes increases the local Ag concentration (3). However, in this current study we show that the efficiency of presentation of peptides derived from internalized Ags to T cells was also affected by different sizes of lipid vesicle. Ag prepared in large vesicles produced enhanced T cell activation compared with Ag prepared in small vesicles (<200 nm in diameter). This phenomenon correlated directly with vesicle size and was not due to effects on macrophage class II or costimulatory molecule expression (data not shown). Furthermore, the DO.11.GFP T cell hybridoma used in these experiments is known to be costimulator independent (20). These data suggest that Ag prepared in large lipid vesicles is handled by the intracellular processing machinery of macrophages in a more efficient fashion than Ag prepared in small lipid vesicles.

Significantly, the current studies demonstrate that both the uptake processes and the intracellular trafficking of Ag prepared in large vs small particles were also distinct. Internalization of large vesicles was mediated by actin-dependent phagocytosis, whereas internalization of small vesicles could occur independently of actin rearrangement. Similarly, Ag prepared in small lipid vesicles colocalized with soluble tracers that accumulated in lysosomes after 20-min chase or less, whereas Ag prepared in large lipid vesicles appeared not to traffic into lysosomes, but remained in peripheral compartments. Furthermore, the endosomes formed by large and small vesicles were distinct, with single large lipid vesicles being internalized individually into endosomes with tightly apposed membranes, whereas multiple small lipid vesicles could be identified in single endosomes that were spacious. Interestingly, previous studies have demonstrated that small and large latex beads form similarly distinct compartments after endocytosis, and the close apposition of endosome membrane to large beads has been proposed as a possible mechanisms for their lack of fusion with lysosomes (26). Biochemical analysis of Ag trafficking demonstrated that the compartments occupied by large lipid vesicles were characteristic of early endosomes and confirmed that small lipid vesicles rapidly trafficked into late endosomes/lysosomes with similar kinetics as soluble Ags. The biochemical and morphological appearance of this compartment was similar to that of the class II containing compartments (MIIC) previously defined in B cells, macrophages, and dendritic cells thought to be important in processing and loading Ag onto class II (27, 28, 29). Previous studies indicate that Ags prepared in liposomes are also processed in lysosomes and subsequently recycled for presentation (12, 30). This would suggest that large lipid vesicles, through reducing transit of Ag to lysosomes, would confer lower Ag presentation efficiency on entrapped Ags. In the current study, however, we clearly show that Ag prepared in large lipid vesicles can be processed and loaded onto class II in phagosomes that exhibit a paucity of markers for phagolysosomes (e.g., LAMP-1) and remain accessible to early endosome/phagosome markers, such as transferrin receptor. Peptide/MHC II complexes could be detected in these compartments as early as 20 min after Ag pulse, and at this time point, little Ag could be detected in late endosomes/lysosomes, suggesting that processing and loading onto class II in phagosomes did not involve these later compartments. Intriguingly, because peptide/class II complexes can be detected in early endosomes or lysosomes this early after pulse with Ag large or small vesicles, respectively, why it takes an additional 5 h for peptide/class II complexes to exocytose to the cell surface is an interesting question that we are currently investigating.

To confirm our interpretation that these large vesicles remained retarded in early endosomes, we loaded the vesicles with iron-dextran to facilitate the magnetic isolation of large lipid vesicle phagosomes. These preparations contained both the apparatus essential for processing and presentation as well as the endocytic markers consistent with early endosome-like compartments. Significantly, previous studies by Harding et al. (12) used liposomes prepared by extrusion through 0.2-μm pore size filters, that may be taken up by a phagocytosis-independent pathway and trafficked, processed, and presented in a similar manner to small, but not large, lipid vesicles. Those studies are in contrast to more recent studies using larger particles, which demonstrate that processing can occur in phagosomes formed by 1-μm latex beads, without the involvement of MHC class II compartments with late endosomal characteristics (21). In contrast to the phagosome occupied by latex beads, in the current study we demonstrate that Ag processing and loading onto class II can occur in phagosomes that are LAMP-1 negative and still accessible to recycling transferrin receptor. The requirement for trafficking of liposomal Ag to lysosomes made previously was based on the requirement for reduction of disulphide bonds in Ags for processing to begin (30). In this context, it is significant that one of the components found in the large lipid vesicle phagosome was the recently described GILT (24), which has been demonstrated to be essential to perform this function. To our knowledge, this is the first time this enzyme has been identified in phagosomes.

To determine the significance of this endocytic pathway in the increased presentation observed with large vesicles, we attempted to alter the trafficking of large vesicles. Internalization of Ags via macrophage FcRs has previously been shown to induce rapid trafficking to lysosomes (25), and Ab-opsonized large vesicles followed a similar trafficking pathway. Opsonization also reduced the levels of Ag presentation induced by Ag prepared in large lipid vesicles to levels equivalent to those produced by Ag in small lipid vesicles. Although FcR ligation has a number of downstream effects on macrophages (25), our observation that presentation of Ag prepared in small vesicles was unaffected by opsonization suggests that these are not directly involved in Ag processing. Similarly, the lack of effect of opsonization on presentation of Ag prepared in small vesicles suggests that quantitative differences in uptake are not able to explain this observation. Therefore, it seems more likely that the effect of opsonization on presentation of Ag in large vesicles was a result of altered trafficking in macrophages. These data suggest that the ability of nonopsonized particles to increase presentation would be more relevant in the initial T cell inductive phase of the immune response. Although DCs clearly occupy the central role in activating naive T cells, it is well known that macrophages are the major immune system component that interacts with particles in tissues (9, 10). To reconcile these observations, studies have demonstrated that monocytes, especially those phagocytosing particles (31), can be induced to differentiate into DCs after transendothelial migration, mimicking movement into afferent lymph vessels (9).

In this study we have presented evidence that a significant component of the adjuvant effect of particles is linked to the size of the delivery vesicles that not only enhances Ag delivery, but also modulates its processing and presentation to lymphocytes. Given the increased interest in particulate delivery systems to enhance immunogenicity of vaccines, it is extremely important that the factors that modulate uptake, processing, and presentation of Ag by professional APCs be considered in the rational design of appropriate particulate delivery systems.

We thank Peter Cresswell (Yale University School of Medicine, New Haven, CT) and John Monaco (University of Cincinnati, Cincinnati, OH) for their generous gifts of Ab reagents.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

J.M.B. received a Wellcome Trust Career Development Fellowship. D.G.R. is supported by grants from the U.S. Public Health Services. Image analysis equipment used in these studies was provided through a grant from the Royal Society.

3

Abbreviations used in this paper: DGE, density gradient electrophoresis; GILT, IFN-γ-induced lysosomal thiol-reductase; LAMP, lysosome-associated membrane glycoprotein.

1
Raychaudhuri, S., K. L. Rock.
1998
. Fully mobilizing host defense: building better vaccines.
Nat. Biotechnol.
16
:
1025
.
2
Yang, X., R. S. Gieni, T. R. Mosmann, K. T. HayGlass.
1993
. Chemically modified antigen preferentially elicits induction of Th1-like cytokine synthesis patterns in vivo.
J. Exp. Med.
178
:
349
.
3
Dal Monte, P. R., F. C. Szoka, Jr.
1989
. Effect of liposome encapsulation on antigen presentation in vitro: comparison of presentation by peritoneal macrophages and B cell tumors.
J. Immunol.
142
:
1437
.
4
Singh, M., D. O’Hagan.
1999
. Advances in vaccine adjuvants.
Nat. Biotechnol.
17
:
1075
.
5
Newman, K. D., J. Samuel, G. Kwon.
1998
. Ovalbumin peptide encapsulated in poly(d,l lactic-co-glycolic acid) microspheres is capable of inducing a T helper type 1 immune response.
J. Control Release
54
:
49
.
6
Brewer, J. M., L. Tetley, J. Richmond, F. Y. Liew, J. Alexander.
1998
. Lipid vesicle size determines the Th1 or Th2 response to entrapped antigen.
J. Immunol.
161
:
4000
.
7
Smith, J. B., G. H. McIntosh, B. Morris.
1970
. The traffic of cells through tissues: a study of peripheral lymph in sheep.
J Anat.
107
:
87
.
8
Harmsen, A. G., B. A. Muggenburg, M. B. Snipes, D. E. Bice.
1985
. The role of macrophages in particle translocation from lungs to lymph nodes.
Science
230
:
1277
.
9
Randolph, G. J., K. Inaba, D. F. Robbiani, R. M. Steinman, W. A. Muller.
1999
. Differentiation of phagocytic monocytes into lymph node dendritic cells in vivo.
Immunity
11
:
753
.
10
Dupuis, M., K. Denis-Mize, A. LaBarbara, W. Peters, I. F. Charo, D. M. McDonald, G. Ott.
2001
. Immunization with the adjuvant MF59 induces macrophage trafficking and apoptosis.
Eur. J. Immunol.
31
:
2910
.
11
Guery, J., F. Ria, L. Adorini.
1996
. Dendritic cells but not B cells present antigenic complexes to class II-restricted T cells after administration of protein in adjuvant.
J. Exp. Med.
183
:
751
.
12
Harding, C. V., D. S. Collins, J. W. Slot, H. J. Geuze, E. R. Unanue.
1991
. Liposome encapsulated antigens are processed in lysosomes, recycled and presented to T cells.
Cell
64
:
393
.
13
Shibata, Y., W. J. Metzger, Q. N. Myrvik.
1997
. Chitin particle-induced cell-mediated immunity is inhibited by soluble mannan: mannose receptor-mediated phagocytosis initiates IL-12 production.
J. Immunol.
159
:
2462
.
14
Scheicher, C., M. Mehlig, H. P. Dienes, K. Reske.
1995
. Uptake of microparticle-adsorbed protein antigen by bone marrow-derived dendritic cells results in up-regulation of interleukin-1α and interleukin-12 p40/p35 and triggers prolonged, efficient antigen presentation.
Eur. J. Immunol.
25
:
1566
.
15
Beatty, W. L., E. R. Rhoades, H. J. Ullrich, D. Chatterjee, J. E. Heuser, D. G. Russell.
2000
. Trafficking and release of mycobacterial lipids from infected macrophages.
Traffic
1
:
235
.
16
Kronick, P., R. W. Gilpin.
1986
. Use of superparamagnetic particles for isolation of cells.
J. Biochem. Biophys. Methods
12
:
73
.
17
Russell, H. I., I. A. York, K. L. Rock, J. J. Monaco.
1999
. Class II antigen processing defects in two H2d mouse cell lines are caused by point mutations in the H2-DMa gene.
Eur. J. Immunol.
29
:
905
.
18
Russell, D. G..
1994
. Immunoelectron microscopy of endosomal trafficking in macrophages infected with microbial pathogens.
Methods Cell Biol.
45
:
277
.
19
Tulp, A., D. Verwoerd, M. Fernandez-Borja, J. Neefjes, A. A. Hart.
1996
. High resolution density gradient electrophoresis of cellular organelles.
Electrophoresis.
17
:
173
.
20
Underhill, D. M., M. Bassetti, A. Rudensky, A. Aderem.
1999
. Dynamic interactions of macrophages with T cells during antigen presentation.
J. Exp. Med.
190
:
1909
.
21
Ramachandra, L., R. Song, C. V. Harding.
1999
. Phagosomes are fully competent antigen-processing organelles that mediate the formation of peptide:class II MHC complexes.
J. Immunol.
162
:
3263
.
22
Calafat, J., M. Nijenhuis, H. Janssen, A. Tulp, S. Dusseljee, R. Wubbolts, J. Neefjes.
1994
. Major histocompatibility complex class II molecules induce the formation of endocytic MIIC-like structures.
J. Cell Biol.
126
:
967
.
23
SturgillKoszycki, S., U. E. Schaible, D. G. Russell.
1996
. Mycobacterium-containing phagosomes are accessible to early endosomes and reflect a transitional state in normal phagosome biogenesis.
EMBO J.
15
:
6960
.
24
Maric, M., B. Arunachalam, U. T. Phan, C. Dong, W. S. Garrett, K. S. Cannon, C. Alfonso, L. Karlsson, R. A. Flavell, P. Cresswell.
2001
. Defective antigen processing in GILT-free mice.
Science
294
:
1361
.
25
Aderem, A., D. M. Underhill.
1999
. Mechanisms of phagocytosis in macrophages.
Annu. Rev. Immunol.
17
:
593
.
26
De Chastellier, C., L. Thilo.
1997
. Phagosome maturation and fusion with lysosomes in relation to surface property and size of the phagocytic particle.
Eur. J. Cell Biol.
74
:
49
.
27
Peters, P. J., J. J. Neefjes, V. Oorschot, H. L. Ploegh, H. J. Geuze.
1991
. Segregation of MHC class II molecules from MHC class I molecules in the Golgi complex for transport to lysosomal compartments.
Nature
349
:
669
.
28
Harding, C. V., H. J. Geuze.
1993
. Immunogenic peptides bind to class II MHC molecules in an early lysosomal compartment.
J. Immunol.
151
:
3988
.
29
Neefjes, J..
1999
. CIIV, MIIC and other compartments for MHC class II loading.
Eur. J. Immunol.
29
:
1421
.
30
Harding, C. V., D. S. Collins, O. Kanagawa, E. R. Unanue.
1991
. Liposome encapsulated antigens engender lysosomal processing for class II presentation and cytosolic processing for class I presentation.
J. Immunol.
147
:
2860
.
31
Randolph, G. J., S. Beaulieu, S. Lebecque, R. M. Steinman, W. A. Muller.
1998
. Differentiation of monocytes into dendritic cells in a model of transendothelial trafficking.
Science
282
:
480
.