Prevention of Peripheral Tolerance by a Dendritic Cell Growth Factor: Flt3 Ligand as an Adjuvant

DO11.10 TCR mice (23) were bred in a pathogen-free facility according to the National Institutes of Health guidelines. BALB/c and C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and kept in a specific pathogen-free facility. For adoptive transfers, age- and sex-matched BALB/c recipients were given 2.5 × 10⁶ DO11.10 TCR transgenic T cells intravenously as described previously (8).

Mice (2–10/group) were injected once daily (subcutaneously at the nape of the neck) with 1 μg mouse serum albumin (MSA) plus 10 μg of recombinant human FL (human Chinese hamster ovary cell–derived) for 9 consecutive d. Chicken OVA (Sigma Chemical Co., St. Louis, MO) was freshly prepared in PBS, filtered through a 0.45-μm pore size filter, and deaggregated by centrifugation at 15,000 g for 10 min. Footpad injections were given in a volume of 25 μl. In the tolerance experiments, intraperitoneal injections were given in a volume of 200 μl. OVA was emulsified in CFA (Sigma Chemical Co.), as described previously (8). The M158 antibody against murine CD40L was prepared at Immunex (Seattle, WA), and found to be endotoxin-free. This was injected in a volume of 25 μl in the footpad. A rat IgG antibody (M132) made at Immunex was used as an isotype control.

Cell suspensions were prepared from the popliteal draining lymph nodes and incubated on ice with PE- labeled anti-CD4 (PharMingen, San Diego, CA) and FITC-labeled KJ1-26 mAb (24), as previously described (8). A FACScan^® flow cytometer (Becton Dickinson, San Jose, CA) was used to collect and analyze 20,000 events that had the light scatter properties of lymphocytes.

Various times after priming with OVA in the hind footpads, 5 × 10⁵ popliteal lymph node cells were plated in triplicate in 96-well flat-bottomed plates (Costar, Cambridge, MA) in 200 μl of DMEM complete supplemented with 5% FBS, together with different concentrations of OVA 323–339 peptide. Proliferative responses were assessed after 72 h of culture in a humidified atmosphere of 5% CO₂ in air. Cultures with pulsed with 0.5 μCi [³H]thymidine for 5 h and the cells were harvested onto glass fiber sheets for counting on a gas-phase β counter. For cytokine assays, aliquots of culture supernatants were removed after 72 h, pooled, and assayed for the presence of IFN-γ, IL-2, IL-4, and IL-10 by ELISA.

96-well ELISA plates (Maxisorp; Nunc, Naperville, IL) were coated overnight with 1 μg/well OVA in PBS at 4°C, blocked with PBS/5% FBS, and washed with PBS/0.1% Tween 20. Serum samples were diluted in PBS/5%FBS (starting at 1:100), and threefold dilutions were made. Plates were incubated for 2 h at room temperature, washed, and incubated with alkaline phosphatase-conjugated anti-IgG1 (1:2,000; PharMingen), anti-IgG2a, anti-IgG2b or anti-IgM (1:1,000; PharMingen) detecting antibodies for an additional 2 h at room temperature. Plates were washed again, and enzyme activity was detected with p-nitrophenyl phosphate disodium (Sigma Chemical Co.). The amount of reaction product was assessed on an ELISA plate reader at an OD of 405 nm using the Deltasoft program (DeltaPoint, Monterey, CA). Multiple point analysis was performed on each set of isotype titrations using the BIOASSAY program (Immunex, Seattle, WA), selecting a maximum value for each isotype and determining for each sample the dilution giving half-maximal OD value, thus generating arbitrary unit per milliliter values as previously described (24a).

IFN-γ, IL-2, IL-10, and IL-4 were quantified by ELISAs adapted from PharMingen protocols. In brief, Nunc ELISA plates (Maxisorp; Nunc) were coated with 100 μl of the following antibodies diluted in 0.1 M NaHCO₃ for 6 h: R46A2 (mIFN-γ at 1 μg/ml); 11B11 (mIL-4 at 1 μg/ml); and purified anti-mIL-2 and anti-mIL-10 (PharMingen) at 2 μg/ml. The plates were then washed in PBS/0.1% Tween 20 and blocked with PBS/5% FBS for 6 h. Serial dilutions (twofold) of the samples were made in PBS/5% FBS and added to the plate overnight at 4°C. Twofold, serial dilutions of the standards (PharMingen) were made at the following starting concentrations: mIFN-γ (18 ng/ml); mIL-4 (500 pg/ml); mIL-10 (10 ng/ ml); mIL2 (10-ng/ml). After overnight incubation, the plates were washed, and biotinylated anti-cytokine antibodies (PharMingen) were added at the following dilutions: mIFN-γ at 0.5 μg/ml; mIL-4 at 0.25 μg/ml; mIL-10 at 10 ng/ml; mIL-2 at 10 ng/ml. 2 h later the plates were washed and incubated with streptavidin-horseradish peroxidase (Zymed, San Francisco, CA) diluted in PBS/Tween 20 at 1:2,000. 2 h later, the plates were washed and the bound peroxidase was detected with TMB Microwell substrate (Kirkegaard & Perry Labs., Gaithersburg, MD). The amount of reaction product was assessed on an ELISA plate reader at an OD of 450 nm. The detection limits were: 100 pg/ml for IFN-γ, 100 pg/ml for IL-2, and 20 pg/ml for IL-10.

We examined whether FL-treated mice had enhanced potential to mount antigen-specific immune responses against soluble antigens. C57BL/6 or BALB/C mice were treated with either FL or MSA (control) for 9 d. This FL treatment regimen results in up to a 30-fold expansion of DCs in spleen and various other tissues (21, 22). These animals were then injected subcutaneously with various doses of soluble OVA or OVA in the adjuvant RIBI (OVA/RIBI) on the final day of FL treatment. 3 wk later, the mice were boosted with a second injection of OVA or OVA/RIBI and the serum anti-OVA antibody titers were measured 7 d later. FL treatment dramatically enhanced IgG2a anti-OVA antibody titers in both C57BL/6 and BALB/C mice, compared with MSA-treated controls (Fig. 1, A and B). The titers of IgG2a OVA-specific antibody in FL-treated mice were, in some cases, increased up to 10,000-fold, and were as high as those in mice immunized with OVA/RIBI. In the BALB/C strain, in two experiments, a minor subset of the MSA-treated mice showed a significant IgG2a response when injected with 300 μg of soluble OVA (Fig. 1 B). A lower (but detectable) level can also be seen in a single C57BL/6 mouse, in response to 100 μg of OVA. In both strains, the minor subsets of responders gave titers above the threshold of detectability in this assay (<5 U/ml). The reason(s) for this aberrant response is at present unclear, but may simply reflect variability between mice.

IgG1 anti-OVA antibody titers in both strains of FL-treated mice were elevated 3–10-fold over control mice at the 1,000-μg dose of antigen (P < 0.01). However, little to no enhancement of IgG1 was observed at lower antigen doses, with the exception of a subset of responders at the lowest antigen dose in FL-treated C57BL/6 mice. The IgM titers also showed increases at the higher doses. There were also marked increases in the IgG2b anti-OVA titers in FL-treated BALB/C mice, but more modest increases in the C57BL/6 strain. The IgE and IgG3 titers were unaffected by FL treatment (data not shown). Titers of total Ig (IgM, IgG1, IgG2a, IgG2b, IgG3, and IgE) were unaffected by FL treatment, suggesting that the effect of FL occurred in an antigen-specific way (data not shown), with a preferential skewing towards an IgG2a response.

To investigate the effect of FL treatment on antigen-specific T cell responses, we used TCR transgenic mice that contain rearranged TCR-α and TCR-β genes encoding a TCR specific for OVA 323–339 bound to I-A^d class II MHC molecules (DO11.10 mice; reference 23). TCR transgenic T cells were adoptively transferred into syngeneic BALB/C recipients, such that they constituted a small but detectable proportion of all T cells (8, 9). In this system, the fate of the OVA-specific, transgenic T cells could be followed with the KJ126 clonotypic antibody (24). The reconstituted mice were injected with 10 mmol OVA in the footpad, in either the presence or absence of FL treatment (Fig. 2,A). The CD4⁺, OVA-specific T cell response in the draining lymph node was monitored by flow cytometry. Injection of OVA elicited a significant clonal expansion of the KJ126⁺CD4⁺ cells in the draining lymph nodes, consistent with previous reports (9). FL treatment significantly enhanced the percentage of KJ126⁺CD4⁺ cells in vivo, 7 d after challenge (2.1% soluble OVA versus 5.9% soluble OVA + FL; Fig. 2, B and C). There was a corresponding increase in the absolute numbers of KJ126⁺CD4⁺ cells at day 7 (Fig. 2 D). Thus, FL treatment enhanced the clonal expansion of antigen-specific T cells induced by administration of a soluble protein antigen. This two- to threefold increase in the frequency and the absolute numbers of such cells is similar to the enhancement observed with a natural adjuvant such as LPS in this adoptive transfer system (9, 25). Interestingly, this increase is apparent only at day 7, when the response is on the decline. Therefore, it is conceivable that FL is somehow retarding the elimination of cells rather than, or in addition to, enhancing their expansion.

FL treatment alone, in the absence of OVA injection, failed to elicit any response above the basal levels. The absolute number and activation state (assessed by CD69, CD44, CD45Rb, and CD62L expression), of the non- antigen-specific (KJ126⁻CD4⁺) T cells were unaffected by FL treatment (data not shown), suggesting that FL treatment enhances T cell responses in an antigen-specific manner.

Previous work in this system has shown that productive T cell immunity is elicited only when the antigen is injected with an adjuvant, and that injections of soluble antigens only result in a transient and abortive clonal expansion, in which the antigen-specific T cells cannot be efficiently restimulated in vitro with peptide (8, 9, 25). Therefore, we examined the in vitro proliferative capacity of the OVA-specific T cells from the mice injected with OVA (with or without FL) by culturing single-cell suspensions of the draining lymph nodes from the various cohorts of mice, in the presence of varying concentrations of OVA 323–339 peptide. As shown in Fig. 3,A, mice that received an injection of OVA without FL treatment had greatly diminished responses compared with those that received FL treatment, or those challenged with OVA/CFA, at each of the time points examined. As observed previously, injection of soluble OVA renders the mice unresponsive to an in vitro challenge with OVA 323–339 (8). However, FL treatment prevented this unresponsiveness (Fig. 3,A). Furthermore, FL treatment alone (in the absence of OVA) did not increase the proliferative capacity of T cells above the background controls. This suggests that the observed adjuvant effects of FL are indeed antigen-specific, and not caused by some generic effect of FL on all T cells. It should be noted that the concentration of OVA 323–339 peptide required to elicit a half maximal response diminished with time. Thus, at day 2, half maximal response was achieved with 0.01 μM of peptide, whereas at day 6 only 0.0001 μM peptide was required (Fig. 3 A). This may reflect a state of transient refractoriness in the T cells after antigen-mediated stimulation in vivo.

Consistent with their increased proliferative capacity, OVA-specific T cells from mice injected with OVA plus FL made significantly higher levels of IL-2, IFN-γ, and IL-10, when re-stimulated in vitro (Fig. 3,B). IL-4 was not detectable. The enhanced IFN-γ production is consistent with the dramatic increase in OVA-specific IgG2a titers observed (Fig. 1), and is likely to be mediated by IL-12 produced by the DCs in vivo (22, 26, 27).

There is strong evidence that CD40–CD40L interaction is crucial for humoral and cellular immune responses (28, 29). We examined the potential role CD40–CD40L interaction could play in mediating the enhanced immune responses, since DCs from FL-treated mice express significant levels of CD40 (21, 22). FL- or MSA-treated mice were injected in the footpad with OVA (Fig. 2,A), and on days −1, 0, +1, and +2 200 μg of the M158 antibody (antagonistic against murine CD40L) was injected intraperitoneally. Antigen-specific T cell responses were measured on day 5. As shown in Fig. 4,A, treatment with anti-CD40L mAb, but not the control antibody, significantly blunted the clonal expansion of KJ126⁺CD4⁺ cells in vivo. The dramatic effect that FL treatment has on the in vitro proliferation of OVA-specific T cells is also greatly reduced as a result of blocking the CD40–CD40L pathway in vivo (Fig. 4 B). This is consistent with reports ascribing a pivotal role to CD40–CD40L interaction in the onset of T cell–dependent immune responses (30–33).

Injection of soluble antigens into mice is known to induce antigen-specific T and B cell tolerance rather than immunity (2–6, 9). It has been proposed that this outcome may be due to antigen presentation to T cells by poorly stimulatory, resting B cells (11, 12, 14). Since the profound expansion in DCs caused by FL treatment dramatically enhances antigen-specific T and B cell responsiveness to soluble antigens, we investigated whether FL treatment could prevent the tolerance induction caused by systemic injection of soluble antigen. BALB/C mice, which had been reconstituted previously with transgenic T cells, were injected intraperitoneally with 1 mg of OVA, in either the presence or absence of FL treatment (Fig. 5,A). 7 d later, mice were challenged with OVA in CFA (OVA/CFA) in the footpads. The OVA-specific T cell response was monitored by flow cytometry as described above. OVA/CFA elicits an antigen-specific T cell response, and, as shown previously (8), injection of OVA before challenge with OVA/CFA induces a state of impaired responsiveness, evidenced by a reduction in the frequency and the absolute numbers of KJ126⁺CD4⁺ cells, compared with the OVA/CFA controls (Fig. 5, B and C). In contrast, mice that received OVA in the presence of FL treatment responded even better than the mice that had not been pretreated with OVA before challenge immunization with OVA/CFA. Therefore FL treatment appears to prevent the nonresponsiveness caused by soluble antigen.

Thus the in vivo expansion in DCs caused by FL treatment is able to enhance immune responses dramatically against a soluble antigen via a mechanism involving CD40– CD40L interaction, and bypass the induction of peripheral tolerance caused by systemically administered soluble antigen. These results are consistent with the recently described ability of FL to augment both antitumor immunity (34) and rejection of allogeneic transplants (35). They also support the recent observations of Gong et al. (36) that hybridoma DCs can abrogate T cell tolerance to tumor antigens in transgenic mice. However, our results differ from a recent report that describes the effect of FL in enhancing oral tolerance to fed soluble antigens at certain low doses of antigen (37). The contrasting effects of FL treatment on T cell tolerance to orally versus systemically administered antigens is reminiscent of the effects that LPS has on tolerance induced by these different routes of antigen administration. Although LPS is known to abrogate peripheral tolerance (8, 17), several reports describe its ability to enhance oral tolerance (e.g., 38). The reason(s) for the differences between our observations and those of Viney et al. (37) are unclear at present, but may well reside in the functional differences in DC subsets, or in differences in the microenvironments between the gut and peripheral lymphoid organs.

The exact mechanism by which FL treatment enhances immune responses is unclear. It is useful to consider this question in the context of the recent observation that the establishment of T cell anergy by soluble antigens in vivo may be dependent on the interaction between B7 molecules on APCs and CTLA-4 on T cells (39). This mechanism of T cell tolerance is likely to occur when the APC expresses very low levels of B7 molecules, resulting in a preferential interaction with the high affinity CTLA-4 receptor rather than with CD28. In FL-treated mice, there is an enormous expansion of CD86⁺ DCs in the T cell zones and marginal zones of spleens (22). It is possible that this increased surface area and density of CD86⁺CD40⁺class II⁺ APCs in the T cell zones alters the balance between tolerance and immunity and results in soluble antigen being presented to T cells in an immunogenic way.

In summary, these data provide evidence for DCs playing a pivotal role in maintaining the equilibrium between immunity and self-tolerance in vivo, and highlight the potential for FL as a vaccine adjuvant.

We thank Bill Fanslow (Immunex) for providing us with M158 and isotype control; Jeff Fitzner and colleagues at Immunex for synthesis of the OVA 323–339 peptide; Alan Alpert, Daniel Hirschstein, and Steve Brady for help with flow cytometry; Gary Carlton for help with graphics; and Dana Schack and our colleagues for help with FL injections and animal husbandry. We also thank Drs. Gus Nossal and our colleagues at Immunex for their comments on the manuscript; and Drs. Polly Matzinger and Jacques Banchereau for discussion and critical comments.

 
• CD40L

  CD40 ligand

 
• DC

  dendritic cell

 
• FL

  Flt3 ligand

 
• MSA

  mouse serum albumin