The in vivo mechanisms of action of most vaccine adjuvants are poorly understood. In this study, we present data in mice that reveal a series of critical interactions between the cholera toxin (CT) adjuvant and the dendritic cells (DC) of the splenic marginal zone (MZ) that lead to effective priming of an immune response. For the first time, we have followed adjuvant targeting of MZ DC in vivo. We used CT-conjugated OVA and found that the Ag selectively accumulated in MZ DC following i.v. injections. The uptake of Ag into DC was GM1 ganglioside receptor dependent and mediated by the B subunit of CT (CTB). The targeted MZ DC were quite unique in their phenotype: CD11c+, CD8α−, CD11b−, B220−, and expressing intermediate or low levels of MHC class II and DEC205. Whereas CTB only delivered the Ag to MZ DC, the ADP-ribosyltransferase activity of CT was required for the maturation and migration of DC to the T cell zone, where these cells distinctly up-regulated CD86, but not CD80. This interaction appeared to instruct Ag-specific CD4+ T cells to move into the B cell follicle and strongly support germinal center formations. These events may explain why CT-conjugated Ag is substantially more immunogenic than Ag admixed with soluble CT and why CTB-conjugated Ag can tolerize immune responses when given orally or at other mucosal sites.
Activation of innate immune responses is a prerequisite for an adjuvant function and a much needed component in any vaccine (1). Although we have seen exceptional expansion of our knowledge about receptor-mediated activation of innate responses over the last 5 years, we still lack a detailed understanding of how most adjuvants work in vivo (1, 2). A majority of adjuvants are microbial products that activate innate responses through pattern recognition receptors, which lead to the release of proinflammatory cytokines and up-regulate costimulatory molecules on the APC (3, 4). Although B cells and macrophages are known to act as APC, dendritic cells (DC)3 are considered the key APC for priming of naive T cells (5, 6). The difficulty in targeting DC in vivo has limited our knowledge about the priming events that determine whether Ag stimulation will result in a tolerogenic or immunogenic outcome (7, 8). Immature DC that reside in tissues are known to take up Ag and, if maturation occurs, migrate to regional lymph nodes or the spleen (5). In the secondary lymphoid tissues, the DC immigrants, expressing strong costimulation, may be inherently stimulatory, but whether resident or poorly activated immigrants are tolerogenic is currently a much debated issue (7, 8, 9). In particular, we lack in vivo information about DC at specific anatomical sites, such as the marginal zone (MZ) of the spleen, the lamina propria of the mucosal membranes, or the conduit system in the peripheral lymph nodes (10, 11, 12).
Recent experiments using DC targeting in vivo have indicated that DC can be modulated to direct tolerogenic or immunogenic priming of naive T cells depending on the degree of inflammatory signals released at the site of Ag exposure or as a result of Ag dose (7, 8, 13). A proinflammatory environment would license the naive T cells to develop into effective helper cells for B cell immunity, whereas an anti-inflammatory environment would support the development of regulatory T cells. Relatively few studies have investigated in vivo the functions of APC exposed to adjuvants, and little is known about the maturation and migration of specific APC following exposure to immunomodulators (13, 14). Using an adoptive transfer model of TCR transgenic T cells, it was shown recently that oral feeding of protein leads to clonal expansion of naive T cells, but in the absence of adjuvant, feeding fails to generate B cell help, while in the presence of adjuvant strong B cell help develops (15). Thus, both the tolerogenic and the productive T cell response to fed Ag involved clonal expansion, but the quality of the primed T cells was different from that of T cells in adjuvant-exposed mice. Following the priming, the tolerized T cells failed to enter B cell follicles, but upon challenge with Ag they did enter follicles, but still failed to provide adequate B cell help (15).
Cholera toxin (CT) is one of the best studied and most effective experimental adjuvants known today (16). The mechanism for its adjuvant effect is thought to involve the modulation of APC, but it is poorly understood which APC are functionally targeted in vivo (14, 16, 17, 18). All nucleated cells, including all professional APC, can bind the toxin via the GM1 ganglioside receptor present in the cell membrane (16, 19). Previous reports have documented both a proinflammatory and an anti-inflammatory effect of CT (13, 20). From several studies, including our own work, CT exposure of APC has an augmenting effect on IL-1 and IL-6 production, whereas in other studies a down-regulating effect on IL-12 and promoting effect on IL-10 production have been reported and would have an anti-inflammatory effect (13, 17, 20, 21, 22). In fact, investigators have used the CT adjuvant to generate Th1 cells, but most reports have shown a bias for Th2 cells, and recently also regulatory Tr-1 cells (16, 20, 23). Although data reported by different groups on cytokine production by CT-targeted cells, especially IL-1 and IL-12, are not consistent, it is clear that CT is a strong adjuvant and may induce both a pro- and anti-inflammatory effect in vivo (13, 17, 22). How this dual pro- and anti-inflammatory ability of CT is regulated is still debated, but a different cytokine release and milieu might explain the observed dual behavior. Interestingly, the B subunit of CT (CTB) is a well-documented carrier for the induction of mucosal Ag-specific tolerance (24), and thus, CT and its derivatives have the capacity to be used both as vehicles for tolerance as well as for stimulation of productive immunity.
The CT holotoxin consists of an enzymatically active component (CTA1) linked to a pentamer of CTB that binds to the GM1-ganglioside receptor (16). The ADP-ribosyltransferase activity of the CTA1 moiety hosts strong adjuvant ability and is important for the adjuvant effect, although enzymatically inactive mutants of CT, and the related Escherichia coli heat-labile enterotoxin, have also been found to exert some enhancing function in vivo (16, 25). Because CTB binds to essentially all nucleated cells carrying the GM1 receptor, it has been difficult to explain how the holotoxin exerts adjuvant function in vivo (16). Immature DC exposed to CT ex vivo are known to undergo maturation and expression of costimulatory molecules, CD80 and CD86, together with CCR7 and CXCR4 (13, 17, 20), that would dramatically influence the migration and function of DC, affecting the outcome of Ag priming from a tolerogenic to a productive immune response.
The present study was undertaken to investigate in vivo the targeting and immunomodulating ability of CT. We have compared the effects of CT or CTB on priming of immunity in normal mice and focused on the adjuvant properties, as reflected in Ag delivery and modulation of DC maturation and migration. An adoptive transfer model using TCR transgenic T cells enabled us to follow the interaction of DC with naive T cells and the subsequent migration of activated T cells into the B cell follicle, leading to the development of germinal center (GC) reactions. We constructed probes to follow the targeting of APC in vivo by chemically conjugating soluble OVA to CT (CT-OVA) or CTB (CTB-OVA). Both conjugates were exclusively dependent on the GM1 ganglioside receptor binding for the modulating effect (19); thus, the conjugates were ideal tools to compare the impact of targeting and ADP-ribosyltransferase activity on APC function in vivo.
Materials and Methods
Preparation of conjugates
CT-OVA and CTB-OVA conjugates were prepared with modifications, as described in detail (21). Briefly, 2 mg of CT or rCTB (provided by Dr. J. Holmgren, Department of Medical Microbiology and Immunology, Göteborg, Sweden) was solubilized in dH2O, followed by addition of 8 mg/ml N-succinimidyl-(3-(2-pyridyl)-dithio)propionate (SPDP) (Pharmacia-Upjohn). Four times higher molar amounts of OVA were solubilized in 0.1 M phosphate/0.1 M NaCl buffer and coupled to SPDP, as described above. The CT/CTB-SPDP and OVA-SPDP were mixed at room temperature for 24 h, dialyzed against PBS, and concentrated on a centrifuge column. The final concentration of the conjugates was assessed by GM1 ganglioside ELISA against a standard preparation of CT or CTB (30). The relative OVA content in the conjugates was assessed by rabbit anti-OVA Abs and anti-rabbit Ig-HRP Abs using the GM1 ganglioside ELISA (30). The molar ratio of OVA to CT or CTB was always 4:1. In some experiments, we used biotinylated CTA1-DD adjuvant (35). FITC-labeled rCTB (List Biological Laboratories) or OVA was used at 5 and 40 μg/dose, respectively, to detect freshly isolated MZ DC.
Eight- to 12-wk-old age-matched female BALB/c mice were purchased from M&B, and DO11.10 TCR transgenic and Rag2−/− mice were bred and maintained at the animal facility Experimentell Biomedicin at Göteborg University. Mice were kept in the pathogen-free animal facility and routinely monitored by health screening according to the Federation of European Laboratory Animal Science Association recommendations.
Mice were immunized with different concentrations of CT-OVA, CTB-OVA, or OVA (Sigma-Aldrich) i.v. alone or when indicated, also followed by an i.p. challenge with 200 μg of OVA. Doses given were based on the CT or CTB content, according to concentrations, determined by GM1-ELISA in relation to a standard preparation, and the dose of OVA in the conjugates was 4 times that of CT or CTB, on a molar level, as indicated, with or without 2 μg of CT. Footpad injections s.c. with 5 μg of conjugates were undertaken to analyze CD11c+ cells after 24 h in the draining popliteal lymph node. Ex vivo pulsing of cultured DC for 2 h with 40 μg/ml OVA, or 5 μg/ml CT-OVA, CTB-OVA, or CTB-OVA + 1 μg/ml free CT adjuvant was performed before i.v. injection into naive BALB/c mice. Seven days later, the mice were challenged i.p. with 200 μg of OVA before sacrifice on day 14. Five to six mice were included in each group. Adoptively transferred age-matched BALB/c mice received splenocytes i.v. from transgenic DO11.10 mice at 107 cells per mouse. After 24 h, mice were immunized with a single dose of conjugates or OVA at doses given above. When indicated, CT or CTB conjugates were preincubated for 2 h at room temperature with saturating amounts (10 nmol/ml) of blocking soluble GM1.
The mice were bled before sacrifice, and sera were assayed by ELISA. Briefly, microtiter plates (Nunc) were coated with 3 ng/ml GM1 ganglioside or 200 μg/ml OVA in PBS, as described (26). Sera were diluted in PBS. Anti-OVA or CTB Abs were followed by HRP-labeled rabbit anti-mouse Ig Abs (DakoCytomation) and visualized using 0-phenylenediamine substrate (1 mg/ml)/0.04% H2O2 in citrate buffer (pH 4.5). The reactions were read in a spectrophotometer (Flow Laboratories) at 450 nm. The anti-OVA or CT/CTB serum titers were defined as the interpolated OD readings on the linear part of the curve with an absorbance of 0.4 above background and given as log10 titer means ± SE of each group.
ADP-ribosyltransferase enzymatic activity
The activity was determined using the NAD-agmatine assay, as described earlier (27). Samples of 10 μg of CT, OVA, CTB-OVA, or CT-OVA were diluted 2-fold, and the enzymatic activity was assessed. The relative activity was expressed in mean cpm of three experiments with SD <5%.
Careful phenotypic analyses on isolated spleenocytes, MACS-enriched CD11c+ DC, or the murine D1 cell line were undertaken (28). D1 cells were cultured in IMDM (Sigma-Aldrich) supplemented with 10% FCS and 30% R1 medium (GM-CSF-transfected NIH-3T3 fibroblast-conditioned medium). Cells were treated with OVA (40 μg), CT-OVA, or CTB-OVA (5 μg/ml) for different times (2 or 24 h), and the cell surface expression of the costimulatory molecules, CD80, CD86, and CD40 (BD Pharmingen), or binding and uptake of OVA into the targeted cells was assessed by FACS. For determination of uptake of OVA after exposure to CT-OVA of splenocytes, we treated cells with saponin for intracellular labeling with anti-OVA. The FACS analysis of OVA uptake was performed with gated CD11chighCD11blow or CD11bhighCD11c− DC and macrophages, respectively. Freshly isolated CD11c+ cells were enriched to 85–90% purity by MACS, according to the manufacturer’s instructions (Miltenyi Biotec). For FACS analysis of CT-targeted CD11c+ cells, we injected mice with a single dose of FITC-labeled CTB or OVA, at 5 or 40 μg, respectively.
BALB/c or DO11.10 mice were given an i.v. injection of CT-OVA and CTB-OVA (2.5 μg), CTA1DD biotin (10 μg), CT biotin (5 μg), heat-labile enterotoxin β subunit (LTB) (5 μg; provided by Dr. M. Lebens, Department of Medical Microbiology and Immunology, Göteborg, Sweden), or mutated LTB (EtxB (G33D) (5 μg; provided by Dr. T. Hirst, Department of Pathology and Microbiology, Bristol, U.K.), and spleens were removed after 2, 24, 48, and 96 h. Spleens were embedded in Tissue Tek (Compound Miles), and frozen microslides were prepared using a cryostat (Zeiss). Sections were fixed in acetone, air dried, and then double labeled with: Texas Red-, PE-, FITC-, or biotin-conjugated anti-IgM (Southern Biotechnology Associates), B220, CD11c, CD11b, CD19, CD8α, CD4, MHC II I-Ad, CD40, CD80, CD86, CD68 mAbs, peanut agglutinin, DEC205 mAb (Serotec), and anti-laminin (Sigma-Aldrich) at optimal dilutions and different combinations, as indicated. Rabbit anti-OVA Abs were followed by FITC-labeled anti-rabbit Ig. Biotinylated CT or CTA1DD were visualized by Texas Red-streptavidin conjugates (Vector Laboratories). Total macrophages, MZ macrophages, metallophilic macrophaghes or DC were detected using F4/80 (BD Pharmingen) and MOMA-2 (Serotec), ER-TR9, MOMA-1 cysteine-rich mannose receptor (CR-Fc; detected by a human fusion protein (gift from Dr. C. Mueller, Laboratoire d’Immunologie, Clinque et Cellulaire, INSERM, Paris, France)), respectively. MOMA-2 (Serotec), F4/80 (BD Pharmingen) were used. For detection of all or transgenic T cells, anti-CD3 and the clonotypic KJ1-26 mAb were used, respectively. Cells in division were detected by anti-human Ki-67 (BD Pharmingen) and a control Ab. Staining of single-cell suspensions; DC were first incubated with CT- or CTB-OVA and thereafter plated on poly(L) lysine-treated coverslips and detected by rabbit anti-OVA Abs. Photography and evaluation of tissue stainings were performed using a Leica DM LB microscope.
Ag presentation assay
Different densities of freshly isolated and MACS-enriched splenic CD11c+ DC from CT-OVA, CTB-OVA, OVA, or unimmunized mice were cultured together with 105 DO11.10 SCID transgenic CD4+ T cells with or without 1 μM OVAp323 peptide in 96-well cultures for 96 h. Cells were pulsed for 6 h with [3H]thymidine to assess proliferation.
We used Student’s t test for analysis of significance. ∗, Denotes p < 0.05.
Ag conjugation to CT greatly augments immunogenicity
Following chemical conjugation of OVA to CT or CTB, we investigated the immunogenicity of adjuvant-coupled Ag with that of Ag given alone. Serum anti-OVA Ab responses in mice immunized i.v. with CT-OVA were strikingly augmented compared with those seen in CTB-OVA, OVA alone, or OVA admixed with CT adjuvant (Fig. 1). Doses (based on the content of CT or CTB) ranging from 0.5 to 5 μg CT-OVA were highly effective, while CTB-OVA required at least 10-fold higher doses to give comparable anti-OVA titers (Fig. 1,A). Mice injected with OVA admixed with CT adjuvant required >100-fold higher dose of OVA than CT-conjugated OVA to give similar serum titers, and a 200-fold higher dose of OVA alone failed to stimulate anti-OVA Ab responses (Fig. 1,B). Thus, CT-OVA was significantly more effective at stimulating Ab production than CTB-OVA, OVA admixed with CT, or OVA alone. Importantly, the enzymatic activity of the CT-OVA conjugate was unaltered compared with unconjugated CT, whereas CTB-OVA and OVA were completely devoid of ADP-ribosyltransferase activity (Fig. 1,C). Both CT-OVA and CTB-OVA conjugates carried comparable molar concentrations of OVA, as assessed by ELISA and total protein content (Fig. 1 D). Gel electrophoresis and Limulus amebocyte lysate test confirmed that the conjugates were essentially pure and that all protein preparations contained <1 pg/mg contaminating endotoxin.
Both CT and CTB target DC in vitro
To identify possible mechanisms that could explain the differential adjuvant effects of CT and CTB, we assessed whether CT and CTB conjugates delivered Ag to target cells with comparable efficiency. Therefore, DC of the D1 cell line were exposed to OVA, CT-OVA, or CTB-OVA and then analyzed for OVA content using an anti-OVA polyclonal antiserum and labeled secondary Ab (28). CT-OVA or CTB-OVA conjugates were similarly delivered to the target cells in vitro (Fig. 2). In contrast, soluble OVA given at 100-fold higher doses was ineffective (Fig. 2,C). The microscopic analysis of OVA was complemented by FACS and the mean fluorescent intensity (MFI), which demonstrated equal capacity of CT and CTB conjugates to deliver Ag to the target cells (Fig. 2 D). Also, freshly isolated bone marrow DC exposed to CT- or CTB-OVA conjugates confirmed this result (data not shown). Thus, CT and CTB conjugates were effective delivery vehicles for Ag loading of DC. The difference between OVA alone and the conjugated OVA was the presence of the GM1 receptor-binding element, CTB, in both conjugates.
Injected CT- and CTB-conjugated Ag colocalizes to the MZ of the spleen
Next, we investigated the distribution of Ag after i.v. injection of the conjugates in mice and focused on differences between CT and CTB as delivery vehicles for the deposition of OVA. Frozen sections of spleens were analyzed for the presence of OVA at different time points following injection, using the polyclonal anti-OVA detection system described above. Already 15 min after an i.v. injection of CT-OVA (2.5 μg/dose) we found significant numbers of OVA-containing cells in a striking band formation in the MZ area of the spleen (Fig. 3,A). At 2 h, the deposition was most marked, but OVA was clearly detectable at 24 h and at reduced levels even at 48 h postinjection (data not shown). The deposition was similar following injection of CTB-OVA, as no major difference in distribution or labeling intensity was detectable compared with CT-OVA (Fig. 3,B). The splenic distribution of OVA+ cells was dependent on the conjugate because no OVA could be detected following injection of even very high doses (500 μg/dose) of OVA alone (Fig. 3,C). Moreover, the band-formed distribution of OVA was identical with the distribution of labeled CT (Fig. 3,E) or CTB (data not shown) when given i.v. alone, suggesting that the CTB-mediated GM1 ganglioside receptor binding was responsible for the accumulation of OVA to the MZ. This was also confirmed, as nearly no OVA could be detected if conjugates were first preincubated with saturating amounts of soluble GM1 ganglioside before injection, blocking the receptor interaction with the cells of the MZ (Fig. 3,D). The binding to ganglioside GM1, as a prerequisite for accumulation in MZ, was also verified using conjugates made with LTB or rLTB/G33D, a mutant of rLTB (the B subunit of E. coli heat-labile toxin), which does not bind to GM1 (data not shown). Importantly, the binding ability of CT or CTB to GM1 ganglioside receptors was not restricted to the MZ cells, but all nucleated cells avidly bound the conjugates when applied directly to the sections (Fig. 3,E, inset). In contrast, injected i.v. labeled CT/CTB were localized exclusively to the MZ (Fig. 3,E). Thus, the specific accumulation of OVA in the MZ of the spleen appeared to be dependent on the route of entry into the spleen rather than on a unique receptor-mediated cellular binding via CTB in the MZ. In contrast, biotin-labeled CTA1-DD, which binds Ig and does not bind GM1 ganglioside, was found to accumulate in the B cell follicle area and was not found to colocalize with the OVA-containing cells in the MZ after i.v. injections (Fig. 3 F). Taken together, the impaired localization of OVA to the MZ in the GM1 ganglioside-treated CT-OVA preparation, the failure to accumulate rLTB/G33-OVA conjugates, and the lack of accumulation of labeled, nonbinding CTA1-DD to the MZ strongly support a GM1 ganglioside-mediated mechanism for the accumulation of CT-conjugated Ag to the MZ.
GM1 receptor-targeted Ag accumulates in DC of the MZ
The cells that stained brightly for presence of OVA were phenotypically characterized using markers for macrophages (anti-F4/80, MOMA-1, MOMA-2, CD11b, ER-TR9 mAbs), B cells (anti-B220 or anti-CD19 mAbs), or DCs (anti-CD11c, anti-CD8α, antiCD11b, anti-CD4 mAbs). We found that OVA-containing cells labeled strongly with CD11c-specific Abs, but were negative for markers unique to B cells or macrophages. Confocal microscopy revealed that CD11c-positive cells were also colabeled with anti-OVA (Fig. 4,A), indicating that our target population was indeed MZ DC. The OVA-containing CD11c+ cells were first observed in the MZ (Fig. 4,B), but over time (at 24 h), more OVA-containing cells were found to be colocalizing with the CD11c+ population in the T cell zone of the spleen, as illustrated in Fig. 4, E and F. At 72 h or more after injection, no OVA was detectable in the spleens (data not shown). Noteworthily, the target population appeared to be different from the DC subtype expressing CR-Fc+, a population that has been associated with the transport of Ag to the B cell follicle (Fig. 4,C) (29). Moreover, the OVA-enriched cells were negative for MOMA-1 (Fig. 4,C) and ER-TR9 (data not shown), known to label metallophilic macrophages and MZ macrophages, respectively (30, 31). CD11c+, OVA-containing cells were found to express intermediate levels of MHC class II; some cells also had low levels of DEC205, but none of the other DC markers were expressed by the target population, including CD11b, CD8α, and B220 (Table I). Thus, GM1 ganglioside receptor-mediated Ag delivery and uptake favored exceptional accumulation of Ag in MZ DC, whereas other cells in the MZ appeared not to enrich CT-conjugated Ag, despite carrying the GM1 ganglioside receptor. These results indicate that CT-conjugated Ag is targeted primarily to DC in vivo, which was further supported by the finding that OVA was selectively carried by CD11c+ DC in the popliteal lymph node following s.c. injections of CT-OVA conjugates (Fig. 4 G).
|Marker (mAb) .||DC Subtype .|
|Marker (mAb) .||DC Subtype .|
Mice received i.v. injections with 5 μg of CT-OVA conjugate and sacrificed 2 h later. Spleens were removed and frozen sections were processed for anti-OVA staining (green) and labeling with subset-specific Abs (red) prior to microscopic analysis, as described in Materials and Methods. The compiled data from several experiments are shown, and the labeling pattern was also confirmed by FACS analysis of freshly isolated CD11c+ cells at 2 h after CTB-FITC injection (see Fig. 5).
An extended analysis of the spleen of CT-OVA-injected mice revealed that a substantial number of MZ DC was targeted by CT-OVA (Fig. 5). In fact, in a whole spleen, we calculated that ∼25% of all CD11c+ cells were carriers of OVA. These DC were all located to the MZ of the spleen at 2 h following injection of CT-OVA. To more objectively document the distribution of OVA to CD11c+ cells, we used FACS analysis and CTB-FITC, which replaced CT-OVA, as we had confirmed that CT or CTB targeted the same MZ DC as CT-OVA in our previous experiments (Fig. 3, A and E). At 2 h after i.v. injections, the MACS-enriched CD11c+ cells (>85% pure) carried CTB-FITC in 20%, as depicted in the dot plot in Fig. 5. The CD11c− cells had not bound or taken up CTB-FITC.
Thus, CT appeared to host an exceptional targeting ability for CD11c+ DC in vivo, which was clearly GM1 ganglioside receptor mediated. To reconcile the fact that OVA accumulation occurred in DC and not in macrophages in vivo, despite similar ability to bind CT ex vivo (Fig. 3,E), we analyzed whether these cell types differed in their ability to take up and accumulate OVA when presented with the CT-OVA conjugate. Previous studies had indicated that macrophages were impaired in their Ag-processing ability after CT exposure, and we speculated that this might be relevant to explain the differential accumulation of OVA to DC in vivo (35). Therefore, we incubated splenocytes from Rag2−/− mice with CT-OVA and analyzed by FACS the membrane and intracellular accumulation of OVA at various times. We gated on CD11chighCD11b− or CD11bhighCD11clow and found only a slight increase in MFI in CT-OVA-exposed saponin-treated DC, as compared with that seen with similarly treated macrophages (geometric mean 3.2 vs 2.9) (Fig. 5 B). This difference clearly did not reflect the selective accumulation of OVA to MZ DC that we observed in vivo.
DC exposed to CT-, but not CTB-Ag conjugates undergo maturation and effectively stimulate immune responses following injection
Because the CT-conjugated OVA was significantly more immunogenic than CTB-conjugated OVA, irrespective of a similar ability to deliver Ag to the target cells, we analyzed whether the conjugates differed with regard to their immunomodulating effects on DC. To this end, we used the D1 cells again and asked whether CT- or CTB-exposed DC exhibited differences in the expression of maturational markers. The D1 cell line represents immature growth factor-dependent mouse splenic DC, which fully mature in response to bacteria or inflammatory cytokines, reflecting that D1 cells mimick the maturational process of DC in vivo (28). Indeed, we found striking differences, with CT conjugates strongly promoting DC maturation, as shown by up-regulation of CD80, CD86, CD40, and MHC class II (Fig. 6). CTB had some minor effect on MHC class II expression, but not on the other markers of DC maturation (Fig. 6). Moreover, as the CT-OVA-exposed cells were injected into mice, they effectively stimulated anti-OVA immunity, while CTB-OVA-exposed DC were inefficient (Table II). CTB-OVA-pulsed DC were no better than OVA alone-treated DC, and only the addition of intact CT to CTB-OVA-treated DC restored the OVA-priming ability of the conjugate, indicating that the ADP-ribosylating property of CT was required for augmenting DC maturation and function (Table II). Importantly, however, the injected D1 cells did not accumulate in the MZ, and neither did freshly isolated splenic DC loaded ex vivo with CT-OVA conjugate (data not shown).
|No. .||Anti-OVA Ig .|
|OVA||2.1 ± 0.05|
|CTB-OVA||2.5 ± 0.3|
|CT-OVA||3.8 ± 0.3b|
|CTB-OVA + CT||4.1 ± 0.5|
|No. .||Anti-OVA Ig .|
|OVA||2.1 ± 0.05|
|CTB-OVA||2.5 ± 0.3|
|CT-OVA||3.8 ± 0.3b|
|CTB-OVA + CT||4.1 ± 0.5|
DC were treated ex vivo for 2 h with 5 μg/ml CT-OVA, CTB-OVA, CTB-OVA + CT (1 μg/ml), OVA (50 μg) alone, or cell culture medium, and thoroughly washed in medium prior to transfer to naive mice. The animals were subsequently challenged i.p. with OVA alone and sacrificed 7 days later. Serum was collected from individual mice, and serum anti-OVA Abs were determined. The relative concentration of total anti-OVA Abs was expressed as mean log10 titers ± SE of four to five mice in each group. The anti-OVA Ab response was significantly higher in CT-OVA compared with CTB-OVA-treated DC-injected mice (
, p < 0.05). One representative experiment of three is shown.
CT-Ag conjugates dramatically promote the maturation of DC in vivo
The strong promoting effect of CT on DC maturation was seen also in vivo following injection of the conjugates. We found that 24 h after injection of the CT-OVA conjugates, the T cell zone was loaded with CD11c+ cells that brightly stained with anti-CD86 mAb (Fig. 7, A and D). This was in contrast to mice injected with CTB-OVA conjugates (Fig. 7,B) or OVA alone (data not shown), which demonstrated few CD86-expressing cells in the T cell zone. Most CD86+ cells were located to the MZ of the spleen in CTB-OVA- or OVA-injected mice. Thus, CT-OVA immunization resulted in redistribution of CD11c+ cells, from the MZ to the T cell area, but these cells were negative for CD11b+, which instead was found to be increased on cells outside of the MZ (Fig. 7,E). In addition, FACS analysis of isolated spleen DC 24 h after injection revealed that CT-OVA dramatically augmented the level of CD86 expression, whereas CTB-OVA did not alter the expression level compared with that seen after injection of OVA alone, confirming our microscopic findings (Fig. 7,C). Interestingly, CD80 was not up-regulated on the CD11c+ cells in the T cell zone following injection of CT conjugates; rather, the CD80+ cells were found outside of the T cell and MZ (Fig. 7,F). To investigate whether the CD86+ DC from CT-OVA-immunized mice also mediated an augmented ability to present OVA peptides to T cells, we injected mice with OVA alone or with the different conjugates and isolated DC by MACS to >80% purity at 20 h following the injections. OVA peptide (p323)-specific DO11.10 CD4+ T cells were cultured in vitro with the differently in vivo treated DC. We found that only DC from CT-OVA-injected mice stimulated significant T cell proliferation, whereas DC from CTB-OVA- or OVA-injected mice were poor stimulators of T cell proliferation not significantly different from DC from untreated mice (Fig. 7 G). Saturating amounts of peptide (1 μM) gave similar T cell proliferation in all cultures, demonstrating the presence of equal numbers of DO11.10 T cells in the cultures. Thus, CT-OVA effectively delivers Ag to DC and strongly promotes DC maturation, greatly potentiating an effective Ag presentation to T cells in vivo.
CT-Ag conjugates stimulate expansion and migration of specific T cells into the B cell follicle, resulting in augmented GC reactions
We exploited the D011.10 TCR transgenic mouse model further and performed a more detailed analysis of the consequences of CT-directed immunomodulation (15). We found that the splenic T cell zones were filled with CD86+CD11c+ cells 24 h after injection of CT-OVA conjugates in DO11.10 mice (Fig. 7, H and I). This was not seen with CTB-OVA-injected mice (data not shown). Furthermore, after adoptive transfer of DO11.10 T cells into BALB/c mice, we measured the expansion and migration of specific T cells in the spleen. The difference between CT- and CTB-OVA conjugates on DC maturation and function, observed earlier, was clearly seen also in the adoptive transfer model. The KJ1-26+ T cells were accumulating in the T cell zone of the spleen of the CT-OVA-injected mice, whereas the CTB conjugate had minimal effects on KJ1-26+ T cell numbers (Fig. 8, A and B) as compared with OVA-immunized or untreated mice. The KJ1-26+ T cells labeled with anti-Ki67 mAb indicated that they were undergoing expansion in situ in the T cell zone (Fig. 8,C, upper panel). The increase in peptide-specific T cells was most pronounced at 96 h, whereafter the KJ1-26+ T cell numbers were reduced in the T cell zone, but increased in the B cell follicles. After 6 days, significant numbers of KJ1-26+ T cells were seen in the B cell follicles in CT-OVA-immunized mice (Fig. 8,D). By contrast, CTB-OVA-immunized mice exhibited only few KJ1-26+ T cells in the B cell follicle (Fig. 8,E). Concomitant with this, GC were formed in the CT-OVA-immunized mice, and after 12 days large GC were observed in these mice, while CTB-OVA or nonimmunized mice had no or few and small GC (Fig. 8, G and H). These results suggest that the mechanism for CT adjuvanticity involves targeting of MZ DC, leading to activation, maturation, and migration of DC to the T cell zone. This is followed by expansion of specific T cells and movement of T cells into the B cell follicle, finally resulting in greatly augmented GC reactions. Thus, the key adjuvant events appear to be the targeted delivery of Ag and the ADP-ribosyltransferase activity of the CTA1 acting on the MZ DC.
Although many studies have addressed the mechanism for the adjuvant function of the bacterial ADP-ribosylating enterotoxins CT and heat-labile enterotoxin, this is the first report to demonstrate that CT acts as a targeting and immunomodulating vector for GM1 ganglioside-expressing DC in the MZ of the spleen. Previous studies have suggested that DC may be one of the major cellular targets for the CT adjuvant, but none have monitored the targeted DC closely and followed the in vivo priming process in detail, as was done in the present study (13, 17, 18, 20). Our approach was to follow the accumulation of Ag in targeted cells and to detect alterations in APC function induced by the ADP-ribosyltransferase activity of the CT adjuvant. This strategy was possible after we chemically coupled OVA to CT or CTB, which share the ability to bind to the GM1 ganglioside receptor. Notably, CT differs from CTB only in that the CTA1 enzyme, known to be critical for the adjuvant function, is missing in the latter (16). With this approach, we demonstrated that CT as well as CTB deliver large amounts of Ag to the MZ DC following i.v. injections, and that the accumulation of Ag was dependent on the binding to the GM1-ganglioside receptor. The targeted DC appeared to undergo maturation and migration to the T cell zone in response to the CTA1 enzyme. Using the DO11.10 transgenic T cells, we found that targeting of CT-OVA, but not CTB-OVA, to MZ DC resulted in effective priming and expansion of naive T cells concomitant with high expression of CD86 on CD11c+ cells in the T cell zone. The encounter with DC exposed to CT adjuvant seemed to instruct T cells to migrate to the B cell follicle and to promote the development of large GC reactions. Thus, for the first time, a whole sequence of events could be followed after administration of CT adjuvant by monitoring the in vivo deposition of Ag and the accumulation of Ag to the MZ DC.
The present study provides an explanation as to how it is possible that CT can act as a powerful adjuvant despite that it can bind to all GM1 ganglioside-carrying, i.e., all nucleated, cells in the body. In fact, the OVA accumulated exclusively in MZ DC when CT-OVA was injected, arguing for a selective effect on DC, sparing B cells and macrophages in the MZ. However, assessing the density of GM1 receptors on MZ cells, through direct application of labeled CTB onto spleen tissue sections, we failed to reveal any difference in receptor density between DC, macrophages, or B cells. Given that blood flowing into the sinuses of the spleen filters out through the MZ and the red pulp before emptying into the venous sinuses, most Ags are likely to be trapped in the MZ (10, 31). Depending on the type of Ag, soluble or particulate, MZ cells may be differently engaged. Macrophages have been shown to play an important role in trapping of Ags especially for capturing of polysaccharide Ags (32). Also, MZ B cells have been implicated in uptake of blood-borne Ags (33). In the light of the present findings, it is likely that soluble Ags that are taken up by DC would benefit the strongest from the use of the CT adjuvant, whereas other adjuvant mechanisms may be more effective at enhancing responses toward other types of Ags. Our work and that of others support this notion (34). We have demonstrated that whereas the adjuvant effect of CT is dependent on CD40 signaling, i.e., more active at enhancing T-dependent type of responses, the CTA1-DD adjuvant, which is targeted to B cells, also greatly augments T-independent type of responses (27). The differences in deposition of the two adjuvants in relation to the MZ documented in the present study would agree with this finding.
Because CT interacts via the GM1 ganglioside receptor, large quantities of Ag accumulated in the MZ DC. This allowed for relatively small amounts of Ag (<1 μg) to stimulate significant immune responses. At least a 100-fold higher dose of OVA, when simply admixed with CT, was required to give a comparable response to that of the CT-OVA conjugate. By comparison, the CTA1-DD molecule, which binds to B cells via Ig, did not accumulate in the MZ DC after i.v. injection, suggesting that GM1 receptor-mediated uptake is essential for effective loading of soluble Ag into the MZ DC. Nolte et al. (11) have shown that the deposition of Ag in the splenic white pulp and MZ is dependent on a conduit system and restricted by molecular size. However, we do not believe that the CT-mediated targeting of OVA to the MZ DC was merely an effect of molecular size, although the CTA1-DD is 37 kDa and the CT-OVA conjugate is much larger. Why macrophages and B cells in the MZ did not accumulate CT-OVA cannot be explained by the results from the present study. Clearly, both subsets carry GM1 ganglioside receptors and appear to bind CTB equally well ex vivo compared with DC (21). One may speculate that CT, in fact, triggers enhanced accumulation of Ag in DC over time because of facilitated uptake and reduced processing of Ag compared with that occurring in macrophages. Matousek et al. (35) showed that macrophages exposed to CT, but not CTB, reduced their processing activity. However, at variance with our in vivo finding, we failed to show a difference between macrophages and DC with regard to intracellular accumulation of OVA in vitro following CT-OVA exposure. Therefore, in future experiments, we will use various deletional models to eliminate DC or macrophages in vivo, hoping to better explain the selective accumulation of Ag to MZ DC. It should be emphasized that ∼20–25% of all DC in the spleen were targeted, clearly demonstrating the powerful targeting ability of GM1-binding CT-linked Ag for the delivery to the MZ DC. However, in addition, s.c. injections of CT-OVA in the footpad targeted DC that migrated to the popliteal lymph node.
Immature DC in the blood can efficiently capture and transport Ag to the spleen (36). However, we believe that CT was specifically taken up by MZ DC because we detected OVA already at 15 min after CT-OVA injections. At 24 h, DC migrated from the MZ into the T cell zone, which was dependent on the CTA1-enzymatic activity, as CTB conjugates failed to cause this migration of CD11c+CD86-expressing DC into the T cell zones. The targeted MHCIIintDEC205intCD11b−CD8α− MZ DC appeared to be quite unique in that they were CD11b− and CD8α−. A recent publication reported on splenic MZ CD11b+CD8α−DC, which internalized circulating apoptotic cells and acquired CD8α during their later mobilization to T cell areas (10). Although we did not detect CD11b or CD8α on the cell surface of our OVA-high DC in the MZ, it is possible that these cells may express these markers at later time points. Interestingly, CD11b+ cells accumulated outside of the T cell zone and the MZ subsequent to CT-OVA injections, which may be mechanistically important as it argues against the involvement of macrophages in the adjuvant effect of CT.
Previous reports on CT and its ability to stimulate cytokine, chemokine, and chemokine receptor production by DC have shown that these cells express CXCR4 and CCR7, i.e., CT may promote colocalization of DC with naive T cells (17). The effect on the chemokines and chemokine receptor expression may help explain the difference in migrating ability between CT-OVA- and CTB-OVA-targeted MZ DC. In this context, a lack of an effect of CT-OVA on chemokine and chemokine receptor expression in B cells or macrophages in the MZ also would answer the question as to why only DC appear to carry CT-conjugated Ag to the T cell zone. Recent findings in CD38-deficient mice have indicated that this ADP-ribosylating ectoenzyme may be critical for the regulation of adjuvant responsiveness in DC, as lack of CD38 negatively affected T cell-priming efficiency and humoral immunity (37). Thus, CT could provide selectivity by replacing CD38 activity in DC, and thereby, affect chemokine receptor signaling through CCR7 or CXCR4 and the ability to migrate to, e.g., the T cell zone in peripheral lymph nodes or spleen. This notion agrees well with the fact that CTB-OVA, devoid of enzymatic activity, delivered Ag to the MZ DC, but failed to affect DC maturation and migration.
A notable finding was that delivery of CTB-OVA did not affect the differentiation stage of the targeted DC, i.e., no expression of CD86 or migration of DC to the T cell zone and eventually only few T cells were licensed to migrate into the B cell follicle. The latter observation is in agreement with T cells tolerized by oral Ag, which fail to enter into the B cell follicle and provide B cell help (15). In fact, CTB-OVA given orally has been proven one of the most potent ways of induction of T cell tolerance (24). Taken together, these findings argue for the importance of CTA1-dependent maturational signals provided by the enzymatically intact holotoxin and acting on the targeted MZ DC. A hallmark of the effect of CT adjuvant was the development of large GC, expansion sites for the specific B cell response, and important for the generation of memory, Ig class-switching, and somatic hypermutations (38). A similar drive on GC formation is found with the CTA1 enzyme alone, as can be seen when the CTA1-DD adjuvant is used to replace the holotoxin (27). Thus, the enlarged GC appear to be a common denominator for the CTA1-dependent adjuvant mechanism (27, 38). Previously, it has been documented that CD40 is important for the development of GC (39). Recent investigations by Gray and coworkers (40) showing that DC control the migration of T cells to B cell follicles, a mechanism dependent on CD40/OX40L interactions, appear to apply particularly well to the effect of CT-OVA conjugates. CD40 or OX40L expression on MZ DC may be critical for the adjuvant effect of CT. Ongoing studies are, therefore, addressing the effects of CT on OX40L and CD40 expression in MZ DC following immunizations.
We acknowledge the Centre for Cellular Imagining at University of Göteborg for excellent facilities and equipment for the confocal studies. We also thank Dr. Eric Lycke for critical reading of the manuscript.
The authors have no financial conflict of interest.
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.
This study was supported by the Swedish Research Council, the Swedish Cancer Foundation, European Union Grants QLK2-CT-1999-00228 and QLK2-CT-2001-01702, the Foundation Clas Groshinskys Memorial fund, Sven & Dagmar Saléns Foundation, the Swedish Strategic Foundation, Swedish Foundation for International Cooperation in Research and Higher Education Stiftelsen For Internationalisering Av Hogre Utbildning Och Forskning, Marie Curie Individual Fellowship QLK-CT-2000-51129, and Ministero dell’Istruzione, dell’Universita e della Ricerca grants (Programma di Ricerca di Rilevante Interesse Nazionale and Fondo per gli Investimenti della Ricerca di Base).
Abbreviations used in this paper: DC, dendritic cell; CT, cholera toxin; CTA, A subunit of CT; CTB, B subunit of CT; GC, germinal center; LTB, heat-labile enterotoxin B subunit; CR-Fc, cysteine-rich mannose receptor; MFI, mean fluorescent intensity; MZ, marginal zone; SPDP, N-succinimidyl-(3-(2-pyridyl)-dithio)propionate; int, intermediate.