A detailed understanding of how activation of innate immunity can be exploited to generate more effective vaccines is critically required. However, little is known about how to target adjuvants to generate safer and better vaccines. In this study, we describe an adjuvant that, through complement activation and binding to follicular dendritic cells (FDC), dramatically enhances germinal center (GC) formation, which results in greatly augmented Ab responses. The nontoxic CTA1-DD adjuvant hosts the ADP-ribosylating CTA1 subunit from cholera toxin and a dimer of the D fragment from Staphylococcus aureus protein A. We found that T cell-dependent, but not -independent, responses were augmented by CTA1-DD. GC reactions and serum Ab titers were both enhanced in a dose-dependent manner. This effect required complement activation, a property of the DD moiety. Deposition of CTA1-DD to the FDC network appeared to occur via the conduit system and was dependent on complement receptors on the FDC. Hence, Cr2−/− mice failed to augment GC reactions and exhibited dramatically reduced Ab responses, whereas Ribi adjuvant demonstrated unperturbed adjuvant function in these mice. Noteworthy, the adjuvant effect on priming of specific CD4 T cells was found to be intact in Cr2−/− mice, demonstrating that the CTA1-DD host both complement-dependent and -independent adjuvant properties. This is the first demonstration, to our knowledge, of an adjuvant that directly activates complement, enabling binding of the adjuvant to the FDC, which subsequently strongly promoted the GC reaction, leading to augmented serum Ab titers and long-term memory development.
Adjuvants are vital components of most nonliving vaccines. Despite this, the modes of action, including the function of clinically used adjuvants, are still incompletely known (1). Even aluminum-containing adjuvants, used in a majority of vaccines, have just recently been explored from a mechanistic point of view and both direct, via the NALP3 inflammasome, and indirect actions, via uric acid, are thought to explain its adjuvant function (2, 3). Thus, recent progress in adjuvant research has given us new insights into possible mechanisms of action, but more studies are required to get a better understanding of how adjuvants impact on the immune response. We know that most adjuvants activate the innate immune system through proinflammatory receptors, which include TLRs and nucleotide-binding oligomerization domain-like receptors (4). These receptors are stimulated following immunization, as dendritic cells (DCs), in particular, take up Ag and undergo maturation and subsequent migration to the draining lymph nodes, where Ag is presented to T cells (2, 5). The DCs also imprint necessary homing, effector, and memory functions on the adaptive immune response needed for protection (6). However, current adjuvants are poorly targeted to the DCs or other cells critical for augmenting an immune response, and a frequently encountered problem when designing new vaccines is the degree of reactogenicity of the vaccine and insufficient immunogenicity, especially in subunit vaccines (7, 8). An attractive solution to this problem could be specific targeting of the adjuvant to improve the safety profile and greatly promote the efficacy of vaccine adjuvants (9, 10). This notion has recently attracted increased attention and has also gained wider acceptance, but the selection of targets for the adjuvant effect are only beginning to be explored (11).
To circumvent the toxicity problem with cholera toxin (CT), which is one of the most potent experimental adjuvants we know of today, we developed the CTA1-DD adjuvant. This adjuvant combines the ADP-ribosylating CTA1 subunit of CT with a d-dimer, derived from Staphylococcus aureus protein A (12). The fusion protein retains the strong adjuvant properties of CT, but contrary to CT, it cannot bind to the GM1-ganglioside receptor used by CT, and, hence, CTA1-DD has no adverse or toxic side effects when tested in mice or nonhuman primates (12–14). Although CT and related holotoxins are powerful adjuvants targeted to distinct ganglioside receptors present on all cells of the innate immune system, we do not yet know precisely which cells and how the CTA1-DD adjuvant augments an immune response (15). It improves immune responses to many infectious disease-relevant Ags and by enhancing a wide range of functions from an effective priming of CD8 and CD4 T cells to greatly augmented Ab responses in a TLR-independent manner (13, 16). Hence, we have hypothesized that the adjuvant must critically affect key events or processes in the priming of an immune response, possibly through mechanisms that could be different from that of most adjuvants operating via TLR stimulation.
A hallmark of T cell-dependent (TD) humoral immune responses is the formation of germinal centers (GC). CTA1-DD, similar to CT, is a strong inducer of GC formations (17). The GC provides a specialized microenvironment within the B cell follicles that facilitate rapid B cell expansion, Ig class-switch recombination, and somatic hypermutation recombination that promote the generation of high-affinity B cell clones (18). Ultimately, the GC reaction generates memory B cells and long-lived plasma cells through selection processes that we recently discovered were upregulated by the CTA1-DD adjuvant (19, 20). The follicular DCs (FDC) play a central role in the GC reaction; they retain Ag in the form of immune complexes (IC) within the GC via complement receptors or Fc receptors FcγIIB (CD32) and FcεRII (CD23) (21, 22). Ags presented as IC on FDC have been shown to be highly stimulatory and provide a depot of Ag, a quality thought to be important for selecting high-affinity B cell clones in the GC, although this concept still remains somewhat controversial (23–26). FDC are also a source of several modulating and survival signals that help maintain the GC, including CXCL13, BAFF, and IL-6, which are secreted by the FDC upon activation (27, 28). Previously, we demonstrated that CTA1-DD exerted unimpaired adjuvant function in FcεR−/− and FcγRII−/− mice, excluding that FcR were involved in the adjuvant function (17). However, the FDC network also carries complement receptors, known to bind complement fragments, and C3d-conjugated proteins were, indeed, found to be at least 1000-fold more immunogenic than unconjugated proteins (29).
In the current study, we have explored the role of the complement system for the adjuvant function of CTA1-DD following intranasal (i.n.), i.p., or i.v. administrations. The core function of the complement system revolves around complement C3 and its cleavage products (30). Mice deficient in C3 or complement receptors type 1 and 2 (CR1 and CR2) have clearly demonstrated a key role for this system in transporting and retaining IC to the FDC network (31, 32). CR1 (CD35) and CR2 (CD21) are alternative splice forms encoded by the Cr2 locus (33); CD35 binds C3b and C4b and is found on a number of cell types including myeloid cells and lymphocytes, whereas CD21 is mainly found on B cells and FDC and binds iC3b, C3d, and C3dg (34). We investigated whether the distribution of CTA1-DD in the spleen following i.v. injections and the interactions with the complement system could help explain the adjuvant function of this molecule. To this end, we made use of mice deficient in critical complement functions and immunized these mice with CTA1-DD plus 4-hydroxy-3-nitrophenyl acetyl-hapten conjugated to chicken γ-globulin (NP-CGG) to evaluate T cell-dependent Ab responses (35).
Materials and Methods
Mice and immunizations
C57BL/6 and BALB/c mice were obtained from Taconic Farms (M&B, Lille Skensved, Denmark) or Charles River Laboratories (Kisslegg, Germany). C3−/− and TNFR1−/− mice on the C57BL/6 background and Cr2−/− and DO11.10 TCR-transgenic mice on the BALB/c background were bred in ventilated cages at the Laboratory for Experimental Biomedicine, University of Gothenburg (Göteborg, Sweden). All mice were maintained under specific pathogen-free conditions. C4−/− mice were kindly provided by Prof. Michael Carroll (Harvard Medical School, Boston, MA) and maintained under specific pathogen-free conditions at The Rockefeller University (New York, NY). Experiments at The Rockefeller University were performed in accordance with National Institutes of Health guidelines. Age- and sex-matched mice were immunized i.v. in the tail vein with 5 or 30 μg NP-CGG and 1 μg NP-dextran or NP-Ficoll (Biosearch Technologies) together with adjuvants as indicated. Ag was administered with a number of different concentrations of CTA1-DD (1, 5, or 25 μg) as described or with 5 μg CTA1R7K-DD or 5 μg CTA1-OVA-DD in a total volume of 200 μl PBS. For Ribi adjuvant (MPL+TDM+CWS Adjuvant System, M6661; Sigma-Aldrich) immunizations, 5 μg NP-CGG/dose was admixed with Ribi according to the manufacturer’s instructions. Primary immunizations were followed by an i.p. booster containing only Ag without adjuvants at day 10 with 25 or 30 μg NP-CGG or 1 μg NP-dextran or NP-Ficoll in 200 μl PBS. Serum samples and spleen cells were taken on day 18. For i.n. immunizations, mice were immunized three times 1 wk apart and sacrificed at day 28, and Ag/adjuvant formulations were given in 20 μl PBS. GC formation was analyzed 12 d after injection with 10 μg CTA1-DD together with 10 μg NP-CGG i.p.
Preparation of fusion proteins
CTA1-DD, CTA1R7K-DD, or CTA1-OVA-DD (carrying one copy if the immunodominant OVA323–339 peptide) were produced.
Briefly, the fusion proteins were expressed in Escherichia coli DH5 cells, transformed with the expression vectors for the different fusion proteins, and grown in 500 ml cultures overnight in SYPPG medium with 100 μg/ml carbenicillin at 37°C. The cells were harvested by centrifugation, and the fusion proteins, produced as inclusion bodies, were washed before extraction by treatment with 8 M urea. After refolding the proteins by slowly diluting them 35–40 times in Tris-HCl (pH 7.4) at 4°C, the fusion proteins were purified in two steps, by ion exchange and size exclusion chromatography. After concentration and sterile filtration, the purified fusion proteins were stored in PBS at −80°C until use. Fusion proteins were routinely tested for the presence of endotoxin by end-point chromogenic limulus amebocyte lysate methods (LAL Endochrome; Charles River Endosafe, Charleston, SC). Endotoxin levels were <100 endotoxin units/mg.
To visualize CTA1-DD in tissue sections, mice were administered 10 μg CTA1-DD i.v., and spleens were collected at 30 min, 2, 6, or 24 h after administration. Spleens were snap-frozen in OTC-embedding medium (Sakura) using isopentane cooled by liquid nitrogen. Tissue sections (7 μm) were prepared on microslides using a Zeiss cryostat (Carl Zeiss). Sections were air dried, fixed in acetone for 10 min, and rehydrated in PBS. Sections were blocked using horse serum diluted 1/20 for 15 min and then incubated with the Abs of interest: biotinylated anti-B220 (BD Pharmingen), biotinylated anti-CD11c (BD Pharmingen), purified anti-macrophage receptor with a collagenous structure (MARCO) (Serotec), purified anti–MOMA-1 (Serotec), FITC labeled anti-CR1/CR2 (BD Pharmingen), purified anti-Mfge8 (MBL International), purified anti-collagen type I (Millipore), and purified anti-laminin (Sigma-Aldrich). Biotinylated Abs were detected using Alexa Fluor 405, Alexa Fluor 488 (Invitrogen), or streptavidin-TXRD (DakoCytomation). Secondary Abs were donkey anti-rat–Cy3 or donkey anti-rabbit–Cy5 (Jackson ImmunoResearch Laboratories). CTA1-DD was detected by using a chicken anti–CTA1-DD Ab (Agrisera) in combination with anti-chicken–DyLight488 (Jackson ImmunoResearch Laboratories). Cell nuclei were labeled using TO-PRO (Invitrogen). Confocal microscopy was performed at the Centre for Cellular Imaging using the Zeiss LSM 510 META system and LSM software (Carl Zeiss). All images within an experiment were equivalently manipulated using Photoshop (Adobe Systems) to adjust brightness and contrast. Frequency of germinal center induction was evaluated using the ImageJ software (National Institutes of Health) available online (36).
Detection of anti-NP Abs in serum
Serum Abs were measured using ELISA. Briefly, microtiter plates (MaxiSorp; Nunc) were coated with 10 μg NP3- or NP7-BSA/ml (Biosearch Technologies) in PBS at 4°C overnight. After washing with PBS and blocking with 0.1% BSA/PBS for 30 min at 37°C, serum samples were diluted 1/50 in 0.1% BSA/PBS and loaded to the plates. A series of 3-fold dilutions of the different serum samples in corresponding subwells were performed. Plates were then incubated for 2 h at room temperature (r.t.). After washing with PBS-Tween, alkaline phosphatase-conjugated goat anti-mouse IgM, IgG, IgG1, IgG2b, IgG2c, or IgG3 Abs (Southern Biotechnology Associates) were added to the wells, diluted 1/500 in 0.1% BSA/PBS. Plates were incubated for 2 h at r.t. Ag-specific Abs were visualized using phosphatase substrate tablets, nitrophenyl disodium salt hexahydrate (Sigma-Aldrich) dissolved at 1 mg/ml in ethanolamine buffer (pH 9.8) added to the wells. Absorbance was measured at 405 nm using a Multiscan MS spectrophotometer (Olympus). The linear part of the curve was used for calculating titers at a cutoff value of 0.4.
Detection of anti-NP Ab-forming cells in the spleen
The frequency of NP-specific Ab-forming cells (AFCs) from the spleen was assessed by using ELISPOT. Ninety-six–well nitrocellulose ELISPOT plates (Millipore) were coated with 10 μg/ml NP7-BSA in PBS at 4°C overnight. Plates were then blocked with 0.1% BSA in PBS for 30 min at 37°C. Spleens were forced thru a nylon mesh using a syringe plunger. Erythrocytes were lysed using ammonium chloride, and the cells were washed in HBSS (Life Technologies) three times. Splenocytes (2,000,000 cells/well) were added to the wells in Iscove’s total medium, and 3-fold serial dilutions in the corresponding subwells were performed. Plates were incubated at 37°C, 5% CO2. After a 3-h incubation, nitrocellulose filters were washed two times with PBS and PBS-Tween. Alkaline phosphatase-conjugated goat anti-mouse IgG or IgM Abs (Southern Biotechnology Associates) were added to the wells, diluted 1/500 in 0.1% BSA/PBS, and plates were incubated at 4°C overnight. After washing, spots were visualized using SIGMAFAST BCIP/NBT (Sigma-Aldrich), and plates were incubated until spots could be seen clearly. The number of spots was automatically enumerated using an ImmunoSpot Series I Analyzer (Cellular Technology).
Complement activation assay
Microtiter plates (MaxiSorp; Nunc) were coated with 1% BSA, aggregated mouse IgG1 for the classical pathway (2.5 μg/ml; BDPharmingen, aggregated by incubation at 63°C for 30 min), or zymosan for the alternative pathway (200 μg/ml; Sigma-Aldrich) diluted in PBS at 4°C overnight. Wells were washed with PBS and blocked with 1% BSA/PBS for 2 h at r.t. Dilutions of fresh mouse sera in GVB++ (2.5 mM veronal buffer [pH 7.3], 150 mM NaCl, 0.1% gelatin, 1 mM MgCl2, and 0.15 mM CaCl2) for the classical pathway or Mg2+ EGTA (2.5 mM veronal buffer [pH 7.3], 70 mM NaCl, 140 mM glucose, 0.1% gelatin, 7 mM MgCl2 and 10 mM EGTA) for the alternative pathway were incubated for 30 min at 37°C. Complement deposition was detected using biotinylated Fab fragments from an anti-C3d Ab (DakoCytomation). Fab fragments were produced using a Fab preparation kit (Pierce). Biotinylation of the Fab fragments was performed after dialysis in sodium bicarbonate buffer. The Fab fragments were then mixed with d-biotinoyl-ε-aminocaproicacid-N-hydroxysuccinimide ester according to the manufacturer’s instructions (Roche) for 2 h at r.t. following dialysis against PBS. After 1 h incubation at r.t., the plates were washed in PBS-Tween and incubated with extravidin-HRP (Sigma-Aldrich) for 30 min at r.t. followed by o-phenylenediamine substrate (1 mg/ml; Sigma-Aldrich) in citrate buffer containing 0.04% H2O2. Absorbance was measured at 450 nm using a Multiscan MS spectrophotometer. The linear part of the curve was used for calculating titers at a cutoff value of 0.4.
In vivo treatments
Marginal zone macrophages (MZMØ) were depleted using clodronate liposomes, and mice were treated with 200 μl clodronate liposomes in PBS i.v. 24 h before being immunized. Clodronate was a gift from Roche Diagnostics (Mannheim, Germany). Preparation of liposomes was performed as described (37). For FTY720 treatment, mice were injected i.p. with 0.1 μg FTY720 (Cayman Chemical) in distilled H2O 18 h before being immunized.
Adoptive transfer and bone marrow chimeras
T cells specific for OVA were purified by negative selection using MACS. Spleens from DO11.10 mice were forced thru a nylon mesh using a syringe plunger, and cell suspensions were pooled. Erythrocytes were lysed by incubation with ammonium chloride for 5 min, and cells were then washed in 2% FCS/PBS. Cells were incubated with CD4+ T cell biotin Ab mixture (Miltenyi Biotec) for 10 min at 4°C, and biotin beads were then added for 15 min. The cell suspension was then transferred to a CS MACS column according to the manufacturer’s instructions (Miltenyi Biotec). Cells were then labeled with CFSE (Invitrogen) by resuspending them at 3 × 107/ml in 5 μM CFSE in PBS for 7 min at 37°C. The reaction was quenched with an equal volume of cold FCS, and cells were washed and incubated in HBSS (Life Technologies) for 15 min to allow excess CFSE to leak out. Purity of eluted T cells was controlled using FACS (typically ∼80%). A total of 3 × 106 cells was injected i.v. into recipient mice in 200 μl PBS.
For generating chimeras, Cr2−/− mice were irradiated with 1000 rad. Wild-type (WT) bone marrow cells were flushed from femurs and tibias and injected i.v. into recipient mice. The mice were reconstituted with 2–5 × 106 cells. Mice were rested for 7 wk before the start of the experiment.
DO.11.10 transgenic T cell proliferation in response to CTA1–OVA-DD was investigated using FACS analysis. Spleens were forced thru a nylon mesh using a syringe plunger. Erythrocytes were lysed using ammonium chloride, and the cells were washed in HBSS (Life Technologies) three times. Cells were resuspended in 0.1% BSA/PBS and incubated with the FcR blocking Ab (24G2) for 5 min, followed by staining with directly conjugated Abs for 30 min at 4°C, Alexa Fluor 700-conjugated anti-CD4 (BD Pharmingen), and PE-conjugated anti-KJ1.26 (BD Pharmingen). 7-Aminoactinomycin D (Sigma-Aldrich) was included to eliminate nonviable cells. After labeling, cells were washed twice and analyzed using an LSR II (BD Biosciences). Data analysis was performed using FlowJo software (Tree Star). Live gates were set on lymphocytes by forward and side scatter.
In vitro culture and cytokine assay
Four days after immunization, spleens were removed, and single-cell suspensions were prepared. A total of 2 × 106 cells/ml was cultured in 96-well plates (Nunc) in Iscove’s medium (Biochrom) supplemented with 10% heat-inactivated FCS (Biochrom), 50 μM 2-ME (Sigma-Aldrich), 1 mM l-glutamine (Biochrom), and 50 μg/ml gentamicin (Sigma-Aldrich) for 96 h at 37°C in 5% CO2 either with medium alone or with OVA (Sigma-Aldrich) 2 mg/ml. Supernatants were analyzed for cytokines after 96 h by flow cytometry using a mouse Th1/Th2/Th17 cytometric bead array kit (BD Pharmingen) according to the manufacturer’s instructions.
All statistical analyses were performed using PASW Statistics version 18 for Windows (SPSS). The p values <0.05 were considered to indicate statistical significance: *p < 0.05, **p < 0.01, and ***p < 0.001.
CTA1-DD augments TD Ab responses by influencing GC formations
Whereas the immunoenhancing effects of CTA1-DD have been extensively studied, we still lack an understanding of the mechanisms involved or the target populations responsible for the adjuvant effects (14, 17, 38, 39). Therefore, we set out to investigate the relationship between adjuvanticity and function of CTA1-DD using the NP hapten coupled to TD, CGG (NP-CGG), T cell-independent (TID) type 2 Ags, Ficoll (NP-Ficoll), or dextran (NP-dextran). A simple protocol was used with a single priming i.p. with Ag plus CTA1-DD followed by a boost with Ag alone after 10 d. We found that CTA1-DD greatly augmented the anti-NP serum titers in the NP-CGG–immunized mice, whereas no increase in titers was observed in mice immunized with NP coupled to the TID Ags, Ficoll, or dextran (Fig. 1A, 1B). Moreover, the distribution of IgG-subclasses in the serum anti-NP response was dominated by IgG1, followed by IgG2b, IgG3, and IgG2c, whereas an augmenting effect on IgM anti-NP titers was minimal (Fig. 1C). Confirming previous observations, we found that CTA1-DD adjuvant had significantly increased GC formations as compared with immunized mice given NP-CGG alone [not shown (19)]. Moreover, in agreement with previous studies, the augmenting effect of CTA1-DD was completely dependent on the enzyme activity because the inactive mutant, CTA1R7K-DD, failed to augment Ab titers or GC reactions [not shown (14, 40)]. Noteworthy, as many as 75% of all B cell follicles in the spleen hosted GC following NP-CGG plus CTA1-DD immunizations [not shown (19)]. Finally, there was no difference in magnitude of the adjuvant effect of CTA1-DD when given i.n. as compared with i.v. immunizations (Fig. 1D). Hence, CTA1-DD comparably increased TD-specific immune responses, irrespective of the route of administration, and the dramatic augmentation effect was enzyme dependent and correlated with a strong impact on the GC reaction.
CTA1-DD sequentially localizes to the MZMØ and the FDC network in the B cell follicles
Next, we determined the localization of CTA1-DD following immunization. To reliably control for optimal delivery of adjuvant to the spleen, we selected to use the i.v. route and removed the spleens at different time points. As early as 30 min after injection, we could detect CTA1-DD located to the marginal zone (MZ), as shown by a distinct band outside the marginal sinus, visualized by anti-laminin labeling (Fig. 2A, 2B). We did not observe any significant colocalization to B220+ B cells in the MZ (Fig. 2A). Complementing cell-specific labeling revealed that CTA1-DD was trapped and internalized by MZMØ as assessed by colocalization with the MZMØ marker, MARCO (Fig. 2B–D). No colocalization was found when labeling with the CD11c or MOMA-1–specific mAbs detecting DC and metallophilic macrophages, respectively (Fig. 2E). Nor did CD4-, Gr1-, F4/80-, or CD11b-specific mAbs colocalize with CTA1-DD at 30 min after injection (data not shown). Surprisingly, at 2 h, the adjuvant was not only present in the MZ, but also appeared within the B cell follicle (Fig. 2F). This was more manifest at 6 h following injection when CTA1-DD was no longer detectable in the MZ and could only be seen in the B cell follicle, distinctly colocalized with the FDC markers Mfge8 and CR1/CR2, and forming a network-like appearance (Fig. 2G, 2H and data not shown) (41). Because TNFR1−/− mice have B cell follicles but lack FDC, we analyzed whether the accumulation of CTA1-DD to the B cell follicle required a FDC network (42). Indeed, we found that no CTA1-DD was observed in the B cell follicles in TNFRI−/− mice (Fig. 2I). The distribution of CTA1-DD in vivo was independent of the enzymatic activity because the CTA1R7K-DD mutant exhibited an identical distribution, suggesting that the DD element was the critical moiety for the localization to the FDC network (Fig. 2J). Moreover, to reach the FDC in the B cell follicle, CTA1-DD would have to be transported from the MZ to the FDC or directly be distributed through the conduit system to the B cell follicle. Of note, the CTA1-DD remained detectable in the FDC network for at least 24 h, but had disappeared from the MZ already at 6 h, suggesting that the distribution to the MZ may not have been required for the localization to the FDC (data not shown). Indeed, using an anti-collagen I Ab, we found conduits that carried CTA1-DD (Fig. 2K) in agreement with previous studies, demonstrating that small substances can be transported into the B cell follicles via the conduit system (43).
The functional impact of the CTA1-DD adjuvant is dependent on the FDC but independent of the MZ localization
To functionally separate the early appearance of CTA1-DD in the MZ from the late presence of the adjuvant to the FDC network, we undertook experiments to eliminate the MZMØs. To this end, we used clodronate (dichloromethylene-bisphosphonate) liposomes to selectively deplete this cell subset (37, 44). We used a dose of clodronate that selectively depleted the MZMØ and MZB cells, as the latter depend on the MZMØ, leaving red pulp macrophages and DCs unaffected (Fig. 3A–D) (45). The early distribution of CTA1-DD to the MZ was now dramatically altered, with small amounts of CTA1-DD randomly distributed mostly to the red pulp (Fig. 3E, 3F). Despite this, we observed unchanged accumulation of CTA1-DD to the FDC network even in clodronate-treated mice (Fig. 3G, 3H). The lack of MZMØ and the absence of CTA1-DD in the MZ confirmed that this localization had no impact on the adjuvant effect of CTA1-DD following i.v. immunizations with NP-CGG (Fig. 3I). Anti-NP IgG1 serum titers were not significantly different in clodronate or untreated mice, ruling out that MZMØ and the MZ localization was important for the adjuvant effect. In addition, these results suggested that MZ B cells were not involved in transporting CTA1-DD to the FDC network, because clodronate is also known to eliminate the MZ B cells (45). However, to specifically address whether MZ B cells were involved in transporting CTA1-DD, we used FTY720, a sphingosine 1-phosphate receptor 1 agonist that downregulates the receptor, which results in the complete loss of B cells from the MZ (46, 47). Convincingly, this treatment had no impact on the adjuvant function of CTA1-DD, although the B cells were absent from the MZ after treatment with FTY720 (Fig. 4A–C). Hence, we concluded that MZ B cells were not involved in transporting CTA1-DD to the FDC network. Of note, the localization of CTA1-DD to the MZ after 30 min or to the FDC network after 6 h was not affected by FTY720 (Fig. 4D, 4E).
Localization of CTA-DD adjuvant to the FDC is dependent on complement and complement receptors
Having demonstrated that CTA1-DD localizes to the FDC network, we asked through which mechanism this association was established. The FDC is known to bind IC through FcR and complement receptors (21). Because we have previously documented that CTA1-DD does not form significant IC, and its adjuvant effect was intact in FcεR and FcγRII-deficient mice, we turned our attention to the complement system (17). First, we asked whether CTA1-DD could activate complement. We found that CTA1-DD activated complement both through the classical and alternative pathways and that it was the DD moiety that was responsible for this activity, because a fusion protein without the DD element failed to activate complement (Fig. 5A, 5B). As a negative control, we used C3-deficient serum, which showed no activity with CTA1-DD in the assay. When C3-deficient mice were analyzed for an immunoenhancing activity, we found a very poor effect of CTA1-DD, suggesting that complement activation was indeed involved in the adjuvant function of CTA1-DD (Fig. 5C). Furthermore, C3-deficient mice failed to localize CTA1-DD to the FDC network (Fig. 5D, 5E). As C3 is the central component of the complement system and CTA1-DD could activate complement through both the classical and alternative pathways, we asked if CTA1-DD would be deposited on the FDC network in C4−/− mice, as C4 discriminates between the classical and the alternative pathway (48). Strikingly, the C4−/− mice exhibited an identical accumulation of CTA1-DD to the FDC network as that seen in WT mice, indicating that the alternative rather than the classical pathway could be responsible for the complement-dependent retention of CTA1-DD to the B cell follicle in vivo (Fig. 5F, 5G).
To critically address the link between complement activation and FDC-localization, we investigated to what extent complement receptor binding was a prerequisite for the FDC localization and an immunoenhancing effect of CTA1-DD. We analyzed whether mice lacking complement receptors (Cr2−/− mice) were defective in FDC localization and whether CR2 deficiency abrogated the adjuvant function. Firstly, we found that there was no accumulation of CTA1-DD to the FDC network at any time point, although the early appearance of CTA1-DD in the MZ was still prominent in Cr2−/− mice (Fig. 6A–D). Following i.n. immunizations, Cr2−/− mice failed to respond with increased serum anti-NP titers, and a dose range of CTA1-DD given i.v. demonstrated greatly reduced adjuvant function, as only a low level of Ab response was achieved (Fig. 6E, 6F). However, in striking contrast, the adjuvant effect of Ribi, an unrelated TLR4 ligand, monophosphoryl lipid A-dominated adjuvant, gave anti-NP IgG1 responses that were comparable to those seen in WT mice (Fig. 6G) (49). Thus, Cr2−/− mice demonstrated a selective defect in the adjuvant function of CTA1-DD, whereas the enhancing effect of Ribi was intact in these mice, indicating that CTA1-DD, as opposed to Ribi, was dependent on complement and complement receptor expression. In fact, it appeared that the FDC localization, dependent on complement activation and CR2 expression, was a dominant feature of the adjuvant effect of CTA1-DD. The influence of CTA1-DD on GC formations was also lost in Cr2−/− mice, as seen in spleen sections taken from mice 12 d after an i.p. priming with NP-CGG and CTA1-DD (Fig. 6H). Finally, the lack of GC formations resulted in poor development of long-term memory B cells specific for NP, as revealed by the failure to elicit a rapid increase in splenic anti-NP IgG1 AFC and serum anti-NP IgG1 Ab responses after 100 d following a challenge immunization with NP-CGG and Ribi in Ag-primed Cr2−/− as compared with WT mice (Fig. 6I).
Unaltered adjuvant effects of CTA1-DD on CD4 T cell priming in Cr2−/− mice
CTA1-DD had lost most of its adjuvant effect on Ab production in Cr2−/− mice. However, because C3 is known to influence the immune response in several ways, also involving priming of T cells, we analyzed if the augmenting effect on CD4 T cell priming also was reduced in Cr2−/− mice, which partly could have explained the poor TD-Ab response observed (50, 51). For these experiments, we used a fusion protein containing the p323 peptide from OVA, CTA1-OVA-DD, to prime Cr2−/− and WT mice after adoptive transfer of CFSE-labeled OVA transgenic DO.11.10 T cells (52). However, we found no reduction in the adjuvant effect of CTA1-DD in Cr2−/− mice, which exhibited, if anything, enhanced CD4 T cell expansion compared with that seen in WT mice (Fig. 7A). This notion was further confirmed by the similar pattern of cell divisions determined by CFSE dilutions and cytokine production in isolated DO.11.10 splenic T cells following priming with CTA1-DD adjuvant in Cr2−/− and WT mice (Fig. 7B–D). Thus, CTA1-DD adjuvant function on CD4 T cell priming did not appear to require the complement system.
CR2 expressed on the FDC is required for a full adjuvant effect of CTA1-DD
Given that CD4 T cell priming was unaffected in the Cr2−/− mice, we asked if the augmenting effect of CTA1-DD was dependent on CR2-expression on the B cells, because this receptor also functions as a coreceptor to the BCR, lowering the threshold of activation (53, 54). To this end, we made bone marrow chimeras with CR2 expressed on the B cells, but not on the FDC, and analyzed the effect of i.p. immunizations with CTA1-DD adjuvant on NP-CGG responses in the chimeric mice. We found that chimeric mice lacking CR2 exclusively on the FDCs, but not on the B cells, were still poor responders to the adjuvant effect, similar to that seen in Cr2−/− mice (Fig. 8A). Immunofluorescence labeling of spleen sections confirmed that the chimeric mice lacked CR2 on the FDC and that they were unable to trap CTA1-DD, similar to what we found with Cr2−/− mice (Fig. 8B, 8C). Thus, CR2 expression was needed on the FDC to allow for full adjuvant function of CTA1-DD.
The present study is the first, to our knowledge, to unravel the critical involvement of the complement system for the augmenting effect of an adjuvant. This observation has several implications for our understanding of the complexity of the adjuvant function and highlights the superior effect of adjuvant targeting to achieve strong enhancement with little or no adverse reactions. We found that CTA1-DD accumulated and bound to the FDC network in the splenic B cell follicles only hours after injection. The association with FDCs was dependent on the CR2 expression, as demonstrated in Cr2−/− mice, which failed to bind CTA1-DD to the FDC network. Furthermore, CTA1-DD activated complement through the DD moiety, an effect that preferentially engaged the alternative pathway as supported by results in C4−/− mice and stimulated enhanced GC reactions subsequent to binding to the CR2 receptors on the FDC. The GC-augmenting effect was not only dependent on binding, but required an intact ADP-ribosylating function, because an enzymatically inactive mutant, CTA1R7K-DD, failed to promote GC while still binding to the FDC. Hence, complement activation, CR2 binding, and ADP ribosylation were all required for the strong enhancing effect of CTA1-DD on the GC reaction and for augmenting specific serum Ab responses. Interestingly, Cr2−/− mice did not completely fail to respond to the Ab-enhancing effect of CTA1-DD, as significantly higher titers were recorded in these mice compared with those that were immunized without adjuvant, albeit the Cr2−/− mice had few detectable GC. Moreover, and also independent of CR2 binding, we still observed a fully retained adjuvant effect of CTA1-DD on CD4 T cell priming in Cr2−/− mice, suggesting that the mechanism of adjuvanticity involved at least three critical events responsible for a strong TD-Ab response; namely, the FDC-dependent GC reaction, the support of an extrafollicular Ab response, and the priming of CD4 T cells. The GC effect required complement and complement receptors, whereas the latter two were independent of complement, clearly demonstrating the complexity of the adjuvant function.
Following injection, the accumulation of CTA1-DD was restricted to two different compartments of the spleen, the MZ and the FDC network. The localization to these sites was kinetically distinct and separated in time; MZ macrophages hosted most of the adjuvant at 30 min, whereas the FDC network accumulated CTA1-DD after 2 h, which also remained detectable for at least 24 h in this location, whereas it was completely gone from the MZ already at 6 h following injection. The MZ localization appeared not to influence the adjuvant function, as clodronate liposome depletion of the MZMØ had no effect on the accumulation of CTA1-DD to the FDC or the Ab-augmenting effect (37). Indeed, although MZ macrophages are known to trap a variety of Ags and can clear blood-borne pathogens, they do not seem to be important for priming of T cells (55–59). Although we had proven that the deposition of the CTA1-DD to the MZ was not important for the adjuvant effect, we asked to what extent this localization was still critical for transporting the adjuvant from the MZ to the B cell follicle and the FDC. A recent study elegantly showed that Ag could be shuttled from the MZ to the FDC by MZ B cells in a CXCR5- and sphingosine 1-phosphate receptor-dependent manner (60). However, depletion of B cells from the MZ, through treatment with FTY720, had no effect on the FDC localization or the adjuvant effect of CTA1-DD. Besides, we were unable to detect colabeling of CTA1-DD with MZ B cells, and our chimeric mice expressing CR2 on the MZ B cells were unable to accumulate CTA1-DD in the B cell follicle. Rather, it seemed that CTA1-DD was transported to the B cell follicle via the conduit system because we detected CTA1-DD in the conduits in sections of spleen from immunized mice. This agrees well with previous studies that have documented the ability of the conduit system to deliver small Ags (turkey egg lysozyme, 14 kDa) to the B cell follicle, resulting in accumulation on the FDC after 6 h, whereas negligible amounts of large Ags (keyhole limpet hemocyanin, 450 kDa) could be detected on the FDC at this time (43). CTA1-DD is 37 kDa in size and exhibited a pattern of distribution similar to the small Ag used by Roozendaal et al. (43).
The lack of an effect of CTA1-DD to enhance Ab responses to T cell-independent (TID) Ags NP-Ficoll and NP-dextran supports a TLR-independent function, because such responses probably require TLR stimulation (61, 62). In preliminary experiments, we have also found that the adjuvant function of CTA1-DD was independent of TLR signaling, because GC reactions and augmented serum Ab levels were unimpaired in MyD88- and TLR4-deficient mice (J. Mattsson, unpublished observation). These observations confirm previous studies showing that CTA1-DD acted independently of TLR signaling and also had no polyclonal B cell-activating effect in vivo (13, 19). In the current study, we consistently failed to demonstrate an effect of CTA1-DD on GC reactions or Ab responses to NP-Ficoll or NP-dextran, responses known to depend on extrafollicular Ab production primarily by B1b cells and possibly MZ B cells (63). It should be noted, though, that this result is at variance with an earlier report, in which we showed that both GC reactions and serum Abs were upregulated to DNP-dextran by CTA1-DD (17). A simple explanation of this conflicting finding could be that our DNP-dextran stimulated GC reactions in WT mice and in this sense was not strictly TID, as has been reported by others (64–66). Also, haptenation of Ags is known to alter and increase immunogenicity in a TLR-independent fashion, hence such Ags may acquire new properties compared with nonhaptenated Ags (67). Thus, the CTA1-DD adjuvant could have promoted anti-DNP responses by acting on the GC (17, 68). Nevertheless, we can conclude that CTA1-DD effectively promoted TD Ab responses, but it failed to act on TID responses generated by B1 or MZ B cells. At present, we do not know how to account for the augmenting effect of CTA1-DD on extrafollicular Ab production.
CTA1-DD does not form significant IC, and mice deficient in FcεR and FcγRII do not demonstrate a reduced adjuvant function of CTA1-DD (17). Thus, the binding of CTA1-DD to FDC was independent of IC formation and did not require FcR expression. By contrast, we found that this binding required CR1/2 expression and can only occur secondary to complement activation. We demonstrated that both the classical and alternative pathways could be activated by CTA1-DD in vitro. However, the fact that C4−/− mice exhibited unaltered accumulation of CTA1-DD to the FDC argued for that the alternative pathway was sufficient for this deposition in vivo. The TNFR1−/− mice have a defective splenic microarchitecture and lack FDC, which results in that they cannot form GC (42). Consistent with this, we did not observe CTA1-DD in the B cell follicles of TNFR1−/− mice. Moreover, despite an intact FDC network, C3 or Cr2-deficient mice could not bind CTA1-DD to the FDC and subsequent development of GC failed. Thus, the consequence in all three models was a poor augmenting effect of CTA1-DD on specific Ab responses, clearly giving evidence of the critical function of the FDC and the GC for the full adjuvant effect of CTA1-DD. Of note, human complement activation was also observed with CTA1-DD, suggesting that the adjuvant may well function in humans through a complement-dependent pathway (J. Mattsson, unpublished observation).
In agreement with other studies, we found no impairment of T cell response in Cr2−/− mice following administration of the adjuvant, ruling out a defect in T cell-priming ability of CTA1-DD in these mice (50). Furthermore, because it is well known that CR2 signaling in B cells can be critical for an optimal activation, amply documented in a number of studies, we addressed this by constructing chimeric mice (53, 54, 69). However, we found that chimeric mice that lack CR1/2 on the FDCs, but not on the B cells, were still equally impaired in their Ab response, confirming that the defect associated with poor adjuvant function was restricted to the FDC. The augmenting effect of CTA1-DD on the GC reaction was, hence, strictly dependent on CR2 expression on the FDC network. By contrast, Cr2−/− mice showed unimpaired function of Ribi, demonstrating that an unrelated adjuvant, which does not rely on complement receptor expression, was effective at augmenting Ab responses. This also agrees well with previous studies reporting that other adjuvants, aluminum hydroxide and CFA, were fully active in Cr2−/− mice (70). Thus, contrasting with these TLR-dependent adjuvant systems, CTA1-DD appears to be unique in that it exploits the complement system as part of the adjuvant function. In addition, because soluble Ag (i.e., NP-CGG) did not localize to the FDC prior to immunization (J. Mattsson, unpublished observations), whereas the CTA1-DD readily distributed to the FDC within hours in WT mice, it is clear that the poor adjuvant function in CR2−/− mice is not because of a lack of Ag deposition to the FDC but rather the effect of disrupted binding of CTA1-DD to the FDC. Noteworthy, in comparison with Ribi and aluminum hydroxide, we recently showed that CTA1-DD gave stronger GC reactions and, better retained serum Ab titers and stronger expansion of long-lived plasma and memory B cells, which, taken together, explain the exceptional potency of the CTA1-DD adjuvant system (19).
The role of the FDC network is critical for the GC reaction (71). These cells have recently been isolated and studied in vitro (72, 73). The implications of adjuvant binding to the FDC have previously not been analyzed (25, 26). Of note, CTA1-DD not only binds to the FDC, but also the adjuvant effect is dependent on the ADP-ribosylating enzyme, which are both needed for the GC-promoting function. FDCs are known to express TLR4 and can be activated upon challenge with LPS, thus promoting accessory activity and enhancing IgG responses (74). Upon activation, FDC increase expression of adhesion molecules (VCAM-1, ICAM-1, and mucosal addressin cell adhesion molecule-1), CXCL13, BAFF, and cytokines (IL-1β, IL-5, IL-10 and IL-15) and upregulate TLR4 expression (26, 72, 73). Whether CTA1-DD activated FDC in a TLR-independent fashion and which promoting effects it had leading to upregulated GC reactions will be studied in detail in future experiments. However, CTA1-DD probably promoted the GC reaction by establishing a favorable microenvironment after binding to the FDC. This could involve many different effects and the release of cytokines, chemokines, growth factors, or adhesion molecules. In fact, several factors could be influencing the GC-promoting effect. There is established and emerging data to suggest that CD40, ICOS, IL-21, programmed cell death-1, CD95, IFN regulatory factor 4, and Bcl-6 are among the factors that critically affect the GC reaction (75–77). Also, enhanced production of BAFF or a proliferation-inducing ligand could upregulate the GC reaction (78). Moreover, CXCL13 would promote B cell migration to the FDC network, and CXCL2 may influence movement of activated B cells within the GC (26, 71). Whether CTA1-DD modulates expression of any of these factors has not yet been investigated. However, in a previous report, we documented that one of the direct effects of the adjuvant on BCR-activated B cells was to augment Bcl-2 expression, arguing in favor of an antiapoptotic effect (17). This could well be involved in the larger GC sizes and frequencies observed with CTA1-DD. Future studies will address these possibilities.
This study described a novel mechanism for adjuvanticity based on direct complement activation by the DD moiety and described in detail the mechanisms behind the augmenting effect. We believe these findings could be further explored to design future targeted adjuvant vectors that would help improve the next generation of vaccines.
We thank Prof. Anna Blom, University of Lund, Lund, Sweden, for expert advice on complement activation and Jörgen Elgqvist and Elin Cederkrantz at the Department of Radiation Physics, Sahlgrenska University Hospital, Göteborg, Sweden, for excellent assistance in the generation of chimeric mice. We also thank the Centre for Cellular Imaging and the Laboratory for Experimental Biomedicine facility at the University of Gothenburg for excellent help and expert advice from the staff.
This work was supported by the Swedish Research Council, the Swedish Cancer Foundation, the Swedish International Development Cooperation Agency, Global Health and Vaccination Research, the Norwegian Research Council, LUA/ALF, a Swedish Foundation for Strategic Research Mucosal Immunobiology and Vaccine Center grant, the Juvenile Diabetes Research Foundation, and European Union Grant QLK2-CT-2001-01702.
Abbreviations used in this article:
follicular dendritic cells
macrophage receptor with a collagenous structure
marginal zone macrophages
4-hydroxy-3-nitrophenyl acetyl-hapten conjugated to chicken γ-globulin
The authors have no financial conflicts of interest.