Visualization of immune pathways that enhance the neutralizing antibody response to vaccines after primary immunization

ABSTRACT

We examined the relationship between the association of a vaccine antigen with immune cells in secondary lymphoid organs shortly after immunization and the resulting neutralizing antibody response induced by that antigen using three antigenic forms of anthrax protective antigen (PA) that induce qualitatively different antibody responses. The three PA forms used were wild-type PA, which binds to anthrax toxin receptors and elicits a robust antibody response that includes both neutralizing and non-neutralizing antibodies; a receptor-binding-deficient (RBD) mutant form of PA, which does not bind cellular receptors and elicits only barely detectable antibody responses; and an engineered chimeric form of PA, which binds cholera toxin receptors and elicits a robust total antibody response but a poor neutralizing antibody response. We found that both wild-type PA and the PA chimera associated with immune cells in secondary lymphoid organs after immunization, but the RBD mutant PA exhibited minimal association, revealing a relationship between antigen binding to toxin receptors on immune cells after immunization and subsequent antibody responses. A portion of wild-type PA that bound to immune cells was cell surface-associated and maintained its native conformation. Much lower amounts of conformationally intact PA chimera were associated with immune cells after immunization, correlating with the lower neutralizing antibody response elicited by the PA chimera. Thus, binding of an antigen to receptors on immune cells in secondary lymphoid organs after immunization and maintenance of conformational integrity of the cell-associated antigen help dictate the magnitude of the resulting neutralizing antibody response, but not necessarily the total antibody response.

IMPORTANCE

Many vaccines protect by the induction of antibodies that neutralize the action of the pathogen. Here, we followed the fate of three antigenic forms of a vaccine antigen in secondary lymphoid organs after immunization to investigate events leading to a robust neutralizing antibody response. We found that the magnitude of the neutralizing antibody response, but not the total antibody response, correlates with the levels of conformationally intact antigen associated with immune cells in secondary lymphoid organs after primary immunization. We believe that these results provide important insights into the genesis of neutralizing antibody responses induced by vaccine antigens and may have implications for vaccine design.

INTRODUCTION

Vaccines provide protection against many diseases by inducing a protective neutralizing antibody response. However, relatively little is known about the immunological events that favor the induction of neutralizing, rather than non-neutralizing, antibodies. For the optimal production of neutralizing antibodies, B cells need to come into contact with a conformationally intact form of the antigen to activate the subset of B cells capable of producing antibodies that recognize and, therefore, have the potential to neutralize the native form of the antigen. The events that occur after vaccination that promote the interaction of the intact or native form of antigen to cognate B cells in secondary lymphoid organs are only beginning to be understood (1–4).

Bacterial toxins have proved to be powerful tools for studying important cellular processes and pathways such as eukaryotic signal transduction, intracellular trafficking, exocytosis, and actin cytoskeleton regulation (5–7). In addition, inactivated forms of bacterial toxins are often the major antigens contained in vaccines against toxin-mediated diseases for which the primary mode of protection is the induction of a toxin-neutralizing antibody response (8). Here, we used the receptor-binding component of anthrax toxin to investigate and visualize immunological pathways that lead to enhanced neutralizing antibody responses to vaccines.

Anthrax toxin is comprised of a cellular receptor-binding component known as protective antigen (PA) as well as two catalytically active components, lethal factor (LF) and edema factor (EF) (9). Anthrax toxin acts by binding to specific protein receptors on the cell surface (10, 11). The full-length form of PA (83 kDa), referred to here as PA83, is then proteolytically activated by a cellular protease (12) resulting in cleavage to a 63 kDa form, referred to here as PA63. PA63 heptamerizes and then binds LF and/or EF and is endocytosed (13). Upon exposure to the acidic environment of the endosome, the heptamer forms a membrane-spanning pore, which facilitates the transport of LF and/or EF into the cell (14). Once inside the cell, LF and EF express their toxic catalytic activities (13, 15). In the absence of LF and/or EF, PA is non-toxic but can induce a protective neutralizing antibody response (16). For this reason, PA is a major component of anthrax vaccines (17).

Previously (18), we compared the antibody responses induced by immunization with three different forms of PA (depicted in Fig. 1A), each having different receptor-binding characteristics. The first PA form, wild-type PA, binds to anthrax toxin receptors. Two protein receptors for the anthrax toxin have been identified. They are tumor endothelium marker-8 (TEM8) (10) and capillary morphogenesis protein-2 (CMG2) (11). PA receptors are found on the surface of immune cells as well as other cell types (19). Wild-type PA induces a robust total antibody response as measured by ELISA (Enzyme-Linked Immuno Sorbent Assay), which includes both neutralizing and non-neutralizing antibodies, as well as a robust neutralizing antibody response as measured by the ability of the antibodies to neutralize the action of anthrax toxin on cells using a toxin-neutralizing antibody (TNA) assay (20, 21). The second PA form used in this study is a receptor-binding-deficient (RBD) mutant of PA containing specific mutations (N682A and D683A) in its receptor-binding region that render it incapable of binding to cellular PA receptors (22, 23). RBD mutant PA induces only barely detectable total and neutralizing antibody responses (18). While the RBD mutant PA induces a barely detectable antibody response, we have previously demonstrated that the RBD mutant PA retains native PA conformational structure as well as the immunogenic epitopes of PA (18). The third PA form utilized in this study was engineered to efficiently bind to and enter cells but through a different receptor-mediated pathway than that taken by wild-type PA. This engineered protein, which has been described previously (18), is composed of the RBD mutant PA fused to the A2 peptide of cholera toxin. The co-expression of this fusion protein with the B subunit of cholera toxin (CTB) results in a chimeric protein consisting of the RBD mutant PA non-covalently tethered to the CTB pentamer (24) through the A2 peptide (RBD PA-A2-CTB). Rather than binding to anthrax toxin receptors, this PA chimera binds to cholera toxin receptors on the cell surface (18), which have been identified as ganglioside GM1 (25, 26). The PA chimera induces a robust total antibody response, but a poor neutralizing antibody response (18). Those previous results demonstrated that the antibody response to PA can be influenced by its ability to bind to specific cellular receptors. In this study, we followed the fate of these three different forms of PA antigen after primary immunization of mice. Specifically, we used both flow cytometry and biochemical techniques to visualize the association of these antigens with immune cells in both the spleen and draining lymph nodes, important sites where antigen is presented to cognate B cells (1), during the first 24 hours after primary immunization in order to gain insights into immune pathways that are associated with the induction of a robust neutralizing antibody response after vaccination.

Fig 1.

PA antigens used in this study and the antibody titers that they induce. (A) The PA antigens used in this study are depicted. Shown are wild-type PA, RBD mutant PA, and the PA chimera (RBD PA-A2-CTB). The amino acid mutations (N682A and D683A) that render RBD mutant PA incapable of binding to cellular receptors are shown in green. The PA chimera (RBD PA-A2-CTB) is composed of RBD mutant PA fused to the A2 peptide of cholera toxin (shown in blue), which is non-covalently tethered to the B subunit of cholera toxin (shown in magenta) (18). (B) Comparison of total anti-PA antibody titers (ELISA titers) and TNA titers (ED₅₀) induced by the different PA antigens used in this study. Groups of 20 mice were immunized once intraperitoneally with either wild-type PA, RBD mutant PA, or RBD PA-A2-CTB chimera protein as described in Materials and Methods. Twenty-eight days after immunization, mice were bled, and test serum samples were analyzed by ELISA or the toxin-neutralizing antibody assay as described in Materials and Methods. Each value represents the geometric mean of the individual titers of 20 mice. Ninety-five percent confidence intervals are shown in parenthesis.

RESULTS

In this study, we compared the fates of three different forms of PA (depicted in Fig. 1A) in the spleen and draining lymph nodes after immunization of mice with these antigens. As we reported previously (18), each PA form induces a distinct antibody response. For ease of comparison of these antibody responses here, we conducted an additional immunization study to repeat our previously published results (18) and have presented those results in Fig. 1B.

As seen in Fig. 1B and as previously reported (18), wild-type PA, which binds to anthrax toxin receptors, induces a robust total antibody response and a robust neutralizing antibody response. The receptor-binding-deficient mutant form of PA (RBD mutant PA) that lacks the ability to bind cellular receptors produces antibody responses (both total and neutralizing) that are essentially at the limits of quantitation of both assays. Thus, as we previously demonstrated (18), the loss of the ability of PA to bind to its cellular receptor(s) results in a severely diminished antibody response. Our previously reported finding that CMG2-null mice (CMG2^−/−) that lack one of the cellular receptors for PA exhibit a significantly lower antibody response to PA as compared to CMG2-sufficient mice further supports a role for binding of PA to its receptor in the antibody response to PA (27).

The PA chimera induces a robust total antibody response that is the same as or slightly greater than that induced by wild-type PA (Fig. 1B). However, unlike wild-type PA, the PA chimera induces a poor neutralizing antibody response. Taken together, this study and our previous published work (18) demonstrate that the neutralizing antibody response induced by the PA chimera is 4–10 times lower than that induced by wild-type PA while the total antibody response induced by the PA chimera is similar to the response induced by wild-type PA. Of note, the PA chimera can be specifically cleaved to the proteolytically active form in a manner similar to wild-type PA and maintains the ability to form heptamers (18) indicating preservation of the native structure of the PA portion of the chimera.

Because encounters of antigens with immune cells in secondary lymphoid organs are important for the initiation of a humoral immune response [reviewed in reference (2)], we investigated whether the differences in antibody titers induced by the different PA antigens might be related to differences in the association of the PA forms with immune cells in secondary lymphoid organs after immunization.

Association of PA proteins with immune cells after immunization as assessed by flow cytometry

As a first step, we performed flow cytometry analyses to evaluate the association kinetics of wild-type PA, RBD mutant PA, and the PA chimera with immune cells from the spleen and draining lymph nodes derived from immunized mice. Groups of mice were immunized with fluorescently labeled (DyLight 488) PA proteins. After immunization, spleens and draining lymph nodes (mediastinal, tracheobronchial, and mesenteric lymph nodes) were collected. Single cell suspensions of splenocytes and lymph node cells were stained with an antibody panel to identify different cell types as shown in Fig. 2A. We analyzed B, T, and myeloid cell populations for association with DyLight-labeled PA antigens using flow cytometry (Fig. 2B and C).

Fig 2.

Kinetics of association of PA antigens with immune cells in the spleen and draining lymph nodes analyzed by flow cytometry. Groups of mice were immunized as described in Materials and Methods with either wild-type PA, RBD mutant PA, or the PA chimera labeled with DyLight 488. The spleen and the draining lymph nodes (mediastinal, tracheobronchial, and mesenteric) were collected at 1, 2, 4, and 24 hours after immunization. At the indicated timepoints, single cell suspensions were prepared from spleens and lymph nodes and stained with an antibody panel to identify B, T, and myeloid cells and analyzed by flow cytometry. After the exclusion of aggregates, fragments, and dead cells, B cells were identified as CD45⁺ CD19⁺ B220⁺, T cells were identified as CD45⁺ CD19⁻ B220⁻ TCRβ⁺, and myeloid cells were identified as CD45⁺ CD19⁻ B220^+/− TCRβ⁻ NK1.1^+/− CD49b⁻ (A). PA-positive B, T, and myeloid cells from the spleen (B) or draining lymph nodes (C) were quantified at the indicated timepoints; unimmunized mice (PA⁻) served as the negative control. The total number of viable splenocytes was similar at each timepoint and for each antigen with the average being 2.2 × 10⁸ viable cells of which an average of 41% were B cells, 47% were T cells, 4% were myeloid cells, and the remainder were NK cells. The viable lymph node cells averaged 4.9 × 10⁶ of which an average of 23% were B cells, 74% were T cells, 2% were myeloid cells, and the remainder were NK cells. Data are from one complete time course experiment. Results are representative of similar experiments.

We first examined the association of PA antigen with B cells, T cells, and myeloid cells of splenocytes from mice immunized with wild-type PA (Fig. 2B, left panels). In the experiment shown in this figure, approximately 48% of the total myeloid cell population was positive for a fluorescent PA signal within 1 hour after immunization. The association of fluorescent wild-type PA with myeloid cells peaked 4 hours after immunization to 54% of total myeloid cells and then declined to 14% at 24 hours after immunization. The percentages of B cell and T cell subsets of splenocytes that were positive for wild-type PA signals were lower than those observed for the myeloid subset, peaking 2 hours after vaccination at about 20% for B cells and 6% for T cells and declining afterward. In contrast to the association of wild-type PA with the immune cells, we found much less or no association of fluorescent RBD mutant PA with either myeloid cells, B cells, or T cells after immunization (Fig. 2B, middle panels). The kinetics of the association of fluorescent PA chimera with splenic myeloid cells were similar to those observed for wild-type PA (Fig. 2B, right panels). However, the percentages of total myeloid cells that were fluorescently positive were generally lower than that observed for wild-type PA at corresponding timepoints. Moreover, the percentages of B and T cells associated with the fluorescent PA chimera were generally low throughout the time course examined.

We next analyzed the association of PA antigens with B cells, T cells, and myeloid cells of draining lymph nodes (mediastinal, tracheobronchial, and mesenteric lymph nodes) after immunization (Fig. 2C). Wild-type PA association with B cells, T cells, and myeloid cells from draining lymph nodes followed the same kinetics as observed with the respective subsets of splenocytes (compare Fig. 2B and C). The percent positivity rates of PA were similar or slightly higher for B cells from lymph nodes compared to those from spleens, but overall lower for T cells and myeloid cells. Essentially, no association of the RBD mutant PA was detected with lymph node cells (Fig. 2C, middle panel). The PA chimera was found to exhibit some association with B cell, T cell, and myeloid cells from lymph nodes, but not to a large extent.

These results demonstrate that after immunization, wild-type PA as well as the PA chimera protein appear rapidly in the spleen and draining lymph nodes. This association is dependent on the ability of the antigen to bind to cellular receptors.

Association of PA proteins with immune cells assessed using biochemical analysis

While the flow cytometry studies described above demonstrate the association of wild-type PA and the PA chimera with B cells, T cells, and myeloid cells in secondary lymphoid organs shortly after immunization, the fluorescent signal from flow cytometry does not reveal whether these proteins are present in their intact forms or as degraded forms such as proteolytic fragments or processed peptides. To further explore forms of the antigen associated with immune cells, we developed and employed a strategy, presented in Fig. 3A, to determine whether at least a portion of the PA proteins associated with immune cells in the spleen and lymph nodes after immunization remains intact.

Fig 3.

Kinetics of association of PA antigens with immune cells in the spleen and draining lymph nodes as analyzed by immunoblotting. (A) Strategy for the visualization of PA antigens associated with splenocytes and immune cells of the draining lymph nodes cells isolated from groups of mice immunized with biotinylated wild-type PA, RBD mutant PA, or the PA chimera (RBD PA-A2-CTB). Groups of five mice each were immunized. Spleens and draining lymph nodes (mediastinal, tracheobronchial, and mesenteric) were collected at 1, 2, 4, and 24 hours after immunization as indicated. Splenocytes and immune cells from the lymph nodes were isolated, and biotinylated PA was captured from the clarified cell lysates as described in detail in Materials and Methods. Captured biotinylated PA from each lysate was subjected to SDS-PAGE and visualized by immunoblotting using a monoclonal anti-PA antibody. (B) Immunoblots of PA proteins captured from splenocyte lysates. (C) Immunoblots of PA proteins captured from lymph node cell lysates. The positions of PA83 and PA63 are indicated by arrows. Results are representative of triplicate experiments.

Groups of mice were immunized with biotin-labeled PA proteins. At specified timepoints, spleens and draining lymph nodes were harvested, and cell lysates were prepared. Biotin-labeled PA proteins present in the cell lysates were captured, and the PA proteins were visualized by immunoblot analysis. As shown in Fig. 3B and C (left panels), the full-length PA83 and the proteolytically activated, cleaved PA63 forms of PA were observed in both splenocyte and lymph node cell lysates collected from mice immunized with wild-type PA up to 4 hours after immunization. At 24 hours, only a faint band corresponding to the PA63 form of PA was seen in splenocyte lysates (Fig. 3B, left panel). In contrast, the RBD mutant PA protein was not detected at any timepoint after immunization (Fig. 3B and C, middle panels). Bands corresponding to the full-length and cleaved forms of the PA subunit of the chimera were observed in splenocyte lysates 1 hour after immunization (Fig. 3B, right panel), although these bands were much less intense than those observed for wild-type PA. Neither the full-length nor the cleaved PA63 form of the PA subunit of the chimera was detected in lymph node cell lysates (Fig. 3C, right panel). These results suggest that after immunization, intact forms of both wild-type PA and the PA subunit of the chimera are associated with immune cells in the spleen and lymph nodes; however, the observed association of the PA chimera was significantly less than that of wild-type PA. These interactions must occur in a toxin receptor-binding-dependent manner since no association was detected between the RBD mutant PA and the immune cells. The findings that the association of these intact forms of the antigens with the immune cells occurred in a receptor-binding-dependent manner, that both wild-type PA and the PA portion of the chimera were properly cleaved to PA63 forms, and that these antigen forms demonstrated the prolonged stability required for visualization by the biochemical technique used here strongly suggest that these antigen forms have retained a significant conformational structure.

Association of intact forms of PA proteins with specific immune cell subsets

We next identified the specific subsets of immune cells (B cells, T cells, or myeloid cells) with which intact forms of wild-type PA and the PA chimera associate in the spleen and lymph nodes using the methodology depicted in Fig. 4A. We immunized groups of mice with either biotin-labeled wild-type PA or biotin-labeled PA chimera, harvested the spleen after 1 hour, and prepared the splenocyte suspensions. Specific cell subsets (B cells, T cells, and myeloid cells) were isolated, cell lysates were prepared, and biotin-labeled proteins were captured and visualized as described in Materials and Methods.

Fig 4.

Analysis of the association of wild-type PA and the PA chimera (RBD PA-A2-CTB) with specific immune cell types and localization of wild-type PA to the cell surface of myeloid cells. (A) Strategy for the analysis of the association of the wild-type PA and RBD PA-A2-CTB with specific immune cell types. (B) Immunoblot analysis of the association of the wild-type PA and the RBD PA-A2-CTB chimera with specific immune cell types. B cells, T cells, and myeloid cells from spleens of mice immunized with biotinylated PA antigen were isolated as described in Materials and Methods. After purification, each immune cell type was lysed and centrifuged, and biotinylated PA antigen was captured using streptavidin magnetic beads as described in detail in Materials and Methods. Samples containing biotinylated PA were subjected to SDS-PAGE, and PA proteins were visualized by immunoblot analysis using an anti-PA antibody. The positions of PA83 and PA63 are indicated. Note that the full-length and proteolytically activated (PA63) forms of the PA chimera migrate slightly slower than the respective forms of wild-type PA due to the fusion of the PA portion of the chimera with the A2 peptide of cholera toxin (see Fig. 1A). (C) Strategy for the analysis of cell surface-localized wild-type PA. (D) Immunoblot analysis of wild-type PA localized to the cell surface of myeloid cells. Myeloid cells were purified from total splenocytes of mice immunized with wild-type PA as described in Materials and Methods. Myeloid cell surface proteins were biotinylated using the cell impermeable Sulfo-NHS-SS-Biotin and were isolated as described in Materials and Methods. Cell surface-localized biotinylated PA was visualized by immunoblot analysis performed using an anti-PA antibody. Lane 1, myeloid cell surface-localized PA; lane 2, total splenocyte cell-associated PA. The positions of PA83 and PA63 are indicated. Results are representative of triplicate experiments.

As shown in Fig. 4B, both the full-length and proteolytically activated forms of wild-type PA and the PA subunit of the chimera were seen associated with myeloid cells with little, if any, intact protein associated with either B cells or T cells. The signal intensities for both the full-length PA83 and proteolytically activated PA63 forms of wild-type PA captured from the myeloid lysates were significantly stronger than those observed for the corresponding forms of the PA chimera. Note that the full-length and proteolytically activated forms of the PA chimera migrate slightly slower than the respective forms of wild-type PA due to the fusion of the PA portion of the chimera with the A2 peptide of cholera toxin (see Fig. 1A). Similar experiments to examine the association of intact PA proteins with cell subsets from lymph nodes could not be performed due to the sensitivity limit of the methodology used. These results provide direct visualization of intact PA protein antigen associated with splenic myeloid cells after immunization.

Cellular localization of intact forms of PA after vaccination

As described above, intact forms of wild-type PA and, to a lesser extent, chimeric PA subunit were observed associated with both spleen and lymph node myeloid cells after immunization. Therefore, we next addressed whether these intact forms of antigen are cell surface-associated, where they might be accessible for presentation to other cells involved in the antibody production pathway, or if they are located only inside the cell, where their accessibility might be limited. The strategy that we used to visualize myeloid cell surface-associated forms of wild-type PA is described in Fig. 4C. We performed this experiment with wild-type PA only and not with the PA chimera due to the sensitivity limit of the assay. We immunized groups of mice with unlabeled wild-type PA and purified the myeloid cell fraction from splenocytes harvested 4 hours after immunization. Cell surface proteins were labeled with biotin in vitro using Sulfo-NHS-SS-Biotin, an amine-reactive biotinylating reagent that cannot permeate the cell membrane. The use of this reagent ensures that only PA present on the surface of myeloid cells will be labeled. After labeling, myeloid cells were lysed, and the biotinylated PA, representing the surface-localized PA, was captured and analyzed by immunoblot analysis. As shown in Fig. 4D, lane 1, we detected myeloid cell surface-localized full-length PA (83 kDa) as well as the cleaved 63 kDa form. As a positive control and comparator (Fig. 4D, lane 2), we also captured total cell-associated (surface-localized and intracellular) wild-type PA from the splenocyte lysate prepared from a separate group of mice immunized with biotinylated wild-type PA using the strategy described for Fig. 3A. Strikingly, for cell surface-localized PA (Fig. 4D, lane 1), the full-length PA83 form was more abundant than the cleaved PA63 form in contrast to that observed for total splenocyte cell-associated PA (Fig. 4D, lane 2). This result is not entirely unexpected since heptamerized PA63 can undergo endocytosis whereas PA83 does not (13). These results demonstrate that full-length PA can be detected on the cell surface of splenic myeloid cells 4 hours after immunization.

Cell surface-associated PA retains its native structure

Next, we assessed whether myeloid cell surface-localized PA retains its native structure. To do this, we exploited the fact that the combination of cell surface-bound PA with LF results in the formation of a biologically active toxin, which acts on cells to cleave a short peptide from mitogen-activated protein kinase (MEK) at its extreme N-terminus (28). The native structure of PA is required for this activity to ensure the completion of all steps in the intoxication process (i.e., PA heptamerization, PA binding of LF, PA pore formation, and PA-facilitated LF transport across the cell membrane). The strategy used for this approach is depicted in Fig. 5A. We isolated the splenic myeloid cell subset from a group of mice immunized with wild-type PA, which, as was shown in Fig. 4D, contains surface-associated PA. If that surface-associated PA retains its native structure, we would expect it to be able to combine with LF to form an active anthrax toxin. As a control, we also isolated the splenic myeloid cell subset from a group of unimmunized mice. We then added LF to the cells in vitro. The cleavage of myeloid cell MEK was assessed by immunoblot analysis with an N-terminal MEK-specific antibody. Since the antibody recognizes a region at the extreme N-terminal of MEK that would be cleaved off by the action of internalized LF, cleavage of the protein would result in the diminishment of signal intensity of the MEK band as assessed by immunoblot analysis. The results are shown in Fig. 5B. Duplicate immunoblots of myeloid cell lysates were probed with the anti-N-terminal MEK antibody that recognizes only uncleaved MEK (left panel) and an anti-total MEK antibody that recognizes both uncleaved and LF-cleaved MEK (right panel). For the blot probed with the anti-N-terminal MEK antibody (left panel), a substantial decrease in MEK signal intensity was seen for splenic myeloid cells from immunized mice to which LF was added (PA⁺LF⁺, lane 4) as compared to the cells from immunized mice to which no LF was added (PA⁺LF⁻, lane 2) demonstrating functional anthrax toxin activity. In contrast, little decrease in MEK signal intensity was observed for splenic myeloid cells from unimmunized mice to which LF was added (PA⁻LF⁺, lane 3) as compared to the cells from unimmunized mice to which no LF was added (PA⁻LF⁻, lane 1). We believe that the small decrease in signal intensity observed when comparing lane 3 to lane 1 could be due to a small amount of non-specific internalization of LF occurring in the absence of PA as has been reported earlier (13). In contrast, no differences were seen in total MEK signal intensities in the blot shown in Fig. 5B, right panel, which was probed with an anti-total MEK antibody confirming that the total cellular MEK levels were the same for all treatments. These results demonstrate that at least some of the myeloid cell surface-associated PA observed after immunization retains its native conformation.

Fig 5.

Detection of functional PA on the surface of myeloid cells after immunization. (A) Strategy for the detection of functional PA on the cell surface of myeloid cells after immunization. (B) Myeloid cells were isolated from unimmunized mice (PA⁻) or mice immunized with wild-type PA (PA⁺) and either treated with LF (LF⁺) or left untreated (LF⁻). Cells were then washed, lysed, and subjected to immunoblot analysis using either an antibody that recognizes the extreme N-terminus of MEK and, therefore, recognizes only the uncleaved form of MEK (N-terminal-specific MEK antibody, left panel) or an antibody that recognizes both uncleaved and LF-cleaved MEK (total MEK antibody, right panel). Since only a very short peptide at the N-terminus of MEK is removed by LF cleavage (29), both uncleaved and LF-cleaved MEK migrate at the same position on the gel. Lane 1, myeloid cell lysate from unimmunized mice, cells not treated with LF; lane 2, myeloid cell lysate from wild-type PA immunized mice, cells not treated with LF; lane 3, myeloid cell lysate from unimmunized mice, cells treated with LF; lane 4, myeloid cell lysate from the wild-type PA immunized mice, cells treated with LF. The positions of uncleaved MEK and total (both uncleaved and LF-cleaved) MEK are indicated. Results are representative of triplicate experiments.

DISCUSSION

In this study, we used both flow cytometry and biochemical techniques to examine the association of PA antigens with immune cell subsets in secondary lymphoid organs after primary immunization. The results presented in this study indicate that the association of the PA antigen in vivo with immune cells is receptor-binding-dependent since the RBD mutant PA showed minimal or no interaction with the immune cells whereas both wild-type PA and the PA chimera, which bind to cellular receptors, exhibited measurable levels of association with immune cells. Wild-type PA and the PA chimera associated with immune cells after immunization, and both elicited robust total antibody responses; however, the antibody response elicited by RBD mutant PA, which exhibited minimal or no association with immune cells, was essentially undetectable. These data suggest that the antibody response induced by an antigen after primary immunization can be enhanced by binding of the antigen to cell surface receptors on immune cells.

We found that wild-type PA and the PA chimera rapidly associated with immune cells in both the spleen and draining lymph nodes after immunization. The extent of association of wild-type PA or the PA chimera with immune cells that was observed depended on the technique used to assess the association. Such a dependence is not necessarily surprising since flow cytometry and the biochemical techniques used in this study to visualize association might be expected to detect different forms of the antigen, with flow cytometry being expected to detect both intact and degraded forms such as proteolytic fragments or processed peptide forms and the biochemical techniques detecting only intact forms or large fragments.

Using biochemical techniques, we observed the association of full-length PA83 and proteolytically activated PA63 with the immune cells of the spleen and draining lymph nodes after primary immunization. These antigen forms are likely to have maintained much of their native conformational structure given the ability of these forms to associate with immune cells in a receptor-binding-dependent manner, the specificity of cleavage of the full-length antigen to its PA63 proteolytically activated form, and the prolonged stability of these large forms of the antigen required for detection by this technique. Therefore, the view of antigen fate visualized by these biochemical techniques might be the most pertinent to understanding events that lead to a neutralizing antibody response, since only a conformationally intact form of the antigen would be expected to efficiently induce neutralizing antibodies.

Using the biochemical techniques employed in this study, we found that full-length PA83 and the proteolytically activated PA63 form of wild-type PA associated with both splenocytes and immune cells of the draining lymph nodes after immunization to a much greater extent than did the respective forms of the PA chimera. The association of wild-type PA with immune cells of the spleen and draining lymph nodes was clearly observed up to 4 hours after immunization. In contrast, we observed a weak association of the PA chimera with splenocytes only at the 1-hour timepoint. No association of the PA chimera with immune cells of the draining lymph nodes was observed at any timepoint. These data demonstrate a striking correlation between the association of PA83 and PA63 forms of the antigen with immune cells in secondary lymphoid organs and the subsequent neutralizing antibody response induced by the antigen (Fig. 1B and 3B and C). For the reasons described above, these large intact forms of the antigen likely retain a significant conformational structure. In fact, we determined that at least a portion of the surface-localized PA associated with splenic myeloid cells after immunization fully retained its native structure as demonstrated by the manifestation of anthrax toxin activity when combined with LF (Fig. 5). These results extend previous findings in that we have demonstrated a direct relationship between the amount of conformationally intact antigen associated with immune cells in secondary lymphoid organs after primary immunization and the magnitude of the neutralizing antibody response induced by that antigen. Our data also demonstrate that the total antibody response, unlike the neutralizing antibody response, is not necessarily related to the amount of conformationally intact antigen associated with immune cells since both wild-type PA and the PA chimera induce similar robust total antibody responses [Fig. 1B and reference (18)].

We found that full-length PA83 and the cleaved PA63 form of the PA antigen were associated with myeloid cells rather than B or T cells (Fig. 4) suggesting that PA binding to the surface of an APC might be a critical step in the generation of the neutralizing antibody response seen upon primary immunization with wild-type PA. These results suggest possible pathways for delivery of the PA antigen to cognate B cells that produce neutralizing antibodies. The APC-bound antigen may be presented in its native form to cognate B cells by direct contact between the two cell types. Alternatively, PA-associated immune cells may act as antigen transport cells, delivering the antigen to another cell type further along the pathway toward the presentation of antigen to cognate B cells. The concept of intermediary transport cells was demonstrated previously for immune complexes by Phan et al. (30) who showed that non-cognate B cells could transport immune complexes from macrophages in the subcapsular sinus to follicular dendritic cells (DCs) for further presentation of the immune complex-bound antigen to cognate B cells.

Our results also indicate that the specific receptor to which an antigen binds likely determines the magnitude of the neutralizing antibody response induced by the antigen, since wild-type PA induces a more robust neutralizing antibody response than the PA chimera. The receptors for wild-type PA, TEM8 (10) and CMG2 (11), are both proteins. After PA binds to its receptor, it is eventually endocytosed and ultimately trafficked to the late endosomes (31). In contrast, the PA chimera binds to ganglioside GM1, the receptor for cholera toxin, which follows a retrograde transport pathway (25, 26). The specific receptor may dictate the cell type(s) to which the antigen binds, the amount of antigen bound, the extent of time that the intact antigen remains on the cell surface, and the internalization pathway, all of which could influence the antibody response elicited by the antigen.

The results presented here pertain to primary immunization. For booster immunization, the formation of immune complexes and binding of those complexes to complement and/or Fc receptors would be expected to play an important role, if not a major role, in this process. Nonetheless, enhancing the primary response to vaccination and especially the neutralizing antibody response would likely result in an augmented protective booster response. Our preliminary results of booster studies suggest that this appears to be the case (Verma and Burns, unpublished observations).

The results presented in this study suggest that the receptor-mediated association of antigen with immune cells—especially myeloid cells—of the spleen and draining lymph nodes, surface association of the antigen, and maintenance of the conformational integrity of the antigen may increase the presentation of the native form of PA to cognate B cells, thus enhancing the neutralizing antibody response. Many vaccines contain antigens derived from microbial components that have cell-binding capabilities; thus, our findings may extend to other vaccine antigens. We believe that the results presented here provide insights into the genesis of neutralizing antibodies and yield information that could be applied to vaccine design.

MATERIALS AND METHODS

Construction and expression of genes encoding wild-type PA, RBD mutant PA, and RBD PA-A2-CTB chimera

Genes encoding wild-type PA and RBD mutant PA (N682A and D683A) were generated as described previously (18). The different PA constructs and PA chimera were expressed and purified essentially as previously described (18, 32).

Labeling of PA proteins

Wild-type PA, RBD mutant PA, and PA chimera proteins were conjugated with DyLight 488 following the manufacturer’s instructions (DyLight 488 NHS-Ester kit; Thermo Scientific, Rockford, IL). DyLight-labeled wild-type PA, RBD mutant PA, and PA chimera proteins with a conjugation ratio of ~1–1.2 (one fluorescent dye molecule/PA molecule) were used. To ensure that the labeling of PA proteins did not compromise the cellular association and internalization competence of these proteins, we compared the internalization of the labeled wild-type PA and the PA chimera protein with the respective unlabeled PA proteins in vitro using the DC2.4 cell murine dendritic cell line as described previously (18). No differences were observed in the internalization efficiency for labeled and unlabeled PA proteins (data not shown).

For biotin labeling, different PA proteins were conjugated to the biotin molecule using the EZ-Link NHS-SS-Biotin kit following the manufacturer’s instructions (Thermo Scientific, Rockford, IL). Biotin-labeled wild-type PA, RBD mutant PA, and PA chimera proteins with a conjugation ratio of ~1 were used.

Immunization studies and analysis of serum samples by TNA assay and ELISA

Groups of 20 mice (6-week-old female CD-1 mice) were immunized once intraperitoneally with equimolar amounts of PA proteins, i.e., wild-type PA (15 µg of antigen/mouse), RBD mutant PA (15 µg of antigen/mouse), or PA chimera protein (26 µg of antigen/mouse). The doses of different PA proteins used for immunization were adjusted so that equimolar amounts of PA proteins were administered to the mice. A control group of mice was immunized with PBS (Phosphate-Buffered Saline). Immunizations and serum collections were carried out by Cocalico Biologicals, Inc. (Reamstown, PA) in compliance with the guidelines of their Institutional Animal Care and Use Committee. Sera were collected 28 days after immunization and analyzed by the TNA assay, which measures the ability of sera to neutralize the action of anthrax lethal toxin (LF+PA), using J774A.1 cells essentially as described previously (32, 33). Total anti-PA antibodies were measured by ELISA essentially as described previously (32). For TNA and ELISA, a four-parameter logistic regression model was used to fit the data points generated when the absorbance was plotted against the reciprocal of the serum dilution. For TNA, the ED50, which is defined as the reciprocal of the serum dilution at 50% inhibition, was determined as previously described (20). For ELISA, the inflection point of the curve, which indicates a 50% response, was reported as the antibody titer.

Analysis of association of PA proteins with immune cells by flow cytometry

To visualize the association of PA proteins with immune cells using flow cytometry, groups of three mice were immunized via the intraperitoneal route with equimolar amounts of PA proteins, i.e., wild-type PA (25 µg of antigen/mouse), RBD mutant PA (25 µg of antigen/mouse), or PA chimera protein (44 µg of antigen/mouse) labeled with DyLight 488. An unimmunized group was also included in all experiments as a negative control. After immunization, the spleen and draining lymph nodes (mediastinal, tracheobronchial, and mesenteric) were collected at 1, 2, 4, and 24 hours. The spleen and lymph nodes were manually disrupted through cell strainers, and red cell lysis was performed using ACK lysis buffer (GIBCO BRL). Cells were washed using PBS containing 2% FBS (Fetal Bovine Serum), and single cell suspensions were used for flow cytometry. Briefly, cells were incubated with anti-CD16/CD32 (Fc block, BD Pharmingen, San Diego, CA) and stained with the Live/Dead staining kit (Invitrogen, Carlsbad, CA) according to the manufacturer’s instruction. Cells were then washed in a flow cytometry buffer (PBS with 2% FBS) and stained for cell surface markers. Antibody concentrations were optimized for use in multi-color staining protocols as required. The following antibodies were used: anti-B220 (clone RA3-6B2), anti-CD19 (clone 6D5), anti-TCRβ (clone H57-597), anti-NK1.1 (clone PK136), anti-CD49b (clone DX5), and anti-CD45 (clone 30-F11). The antibodies were purchased either from BD Pharmingen or BioLegend (San Diego, CA). After staining, cells were washed and resuspended in 0.5% paraformaldehyde in a flow cytometry buffer. A minimum of 30,000 total events were counted using an analytical Cytek Aurora flow cytometer (Cytek Biosciences, Fremont, CA). Data analyses were performed using FlowJo (Tree Star, Inc. BD) software.

Analysis of the association of PA proteins with immune cells by immunoblot analysis

To visualize the association of PA proteins with immune cells using immunoblot analysis, groups of five mice were immunized with equimolar amounts of biotin-labeled PA proteins as described above for flow cytometry. An unimmunized group was also included in all experiments as a negative control. Spleen and lymph node cell suspensions from immunized and unimmunized mice were prepared as described above. Approximately, 3–4 × 10⁸ splenocytes and 4–5 × 10⁶ lymph node cells from each group were lysed using ice-cold MPER lysis buffer (Invitrogen) containing 1× Halt Protease inhibitor cocktail (Pierce) and universal nuclease (Pierce). After lysis, the clear lysates collected after centrifugation were incubated with Dynabeads M-280 streptavidin-coated beads (Invitrogen) at 4°C to capture the biotin-labeled PA proteins. After capture, the Dynabeads M-280 streptavidin-coated beads were washed extensively with PBS and resuspended in 1× SDS-PAGE lysis buffer, boiled, and analyzed by immunoblotting using an anti-PA antibody (anti-PA monoclonal antibody 18720; QED, Biosciences) to visualize the captured biotin-labeled PA present in the cell lysates.

Purification of specific immune cell subsets (B cells, T cells, and myeloid cells)

Total splenocytes isolated from groups of five mice immunized with wild-type PA and PA chimera were used for the purification of each cell type, i.e., B cells, T cells, and myeloid cells. A total of approximately 3.6 × 10⁸ splenocytes were used for the purification of each cell type. B cells were purified using the Dynabeads mouse Pan B (B220)-positive selection strategy. Briefly, total splenocytes were incubated with prewashed Dynabeads mouse Pan B (B220). After binding, beads were washed with PBS containing 0.1% BSA (Bovine Serum Albumin) and 2 mM EDTA. For T cell purification, the Dynabeads Untouched Mouse T cells (Invitrogen) kit was used. Briefly, total splenocytes were incubated with a rat monoclonal IgG antibody mixture to bind to B cells, DCs, NK cells, monocytes/macrophages, and granulocytes. After incubation, cells were washed and incubated for 15 min at room temperature with Dynabeads coated with a polyclonal sheep anti-rat IgG antibody to capture all unwanted non-T cells. Supernatants containing the untouched T cells were collected. For the purification of myeloid cells, total splenocytes were incubated with Fc receptor blocking antibody and followed by incubation with MACS Miltenyi Pan DC, CD11b, and anti-Ly6G microbeads, as per the manufacturer’s instructions, to capture mainly DCs, monocytes/macrophages, and neutrophils. After incubation, the splenocyte cell suspension was purified using the MACS Miltenyi LS column. The enriched myeloid cell suspension was further purified by incubation with Dynabeads mouse Pan B (B220) and Dynabeads mouse Pan T (Thy1.2) with rotation for 30 min at 4°C. After incubation, the Dynabeads were separated, and the purified myeloid cell suspension was collected. The enrichment and purification of myeloid cell suspension were confirmed by flow cytometry. Approximately 3.4–3.6 × 10⁸ splenocytes, collected from groups of five mice each, were used to purify B cells, T cells, and myeloid cell fractions. The purified fractions of each of the cell subtypes were lysed, and biotin-labeled PA proteins associated with these cell subtypes were captured using Dynabeads M-280 streptavidin-coated beads and loaded onto the gel in their entirety for immunoblotting.

Detection of surface-associated PA with myeloid cells

Mice (five mice/group) were immunized with unlabeled wild-type PA (25 µg of PA protein/mice; IP route). Spleens were collected after 4 hours of immunization, and splenocytes were prepared and purified for the enrichment of myeloid cells, as described above. For the labeling of wild-type PA present on the surface of myeloid cells, cells were incubated with 1.0 mg/mL of membrane-impermeable Sulfo-NHS-SS-Biotin (Thermo Fisher Scientific) in chilled ice water for 1 hour. After labeling, cells were washed extensively with ice-cold PBS and lysed using ice-cold MPER lysis buffer containing 1× Halt protease inhibitor cocktail and universal nuclease. After lysis, the clear lysates, collected after centrifugation, were incubated with Dynabeads M-280 streptavidin-coated beads to capture the surface-associated biotin-labeled wild-type PA. After capture/binding, the Dynabeads M-280 streptavidin-coated beads were washed with PBS and resuspended in 1× SDS-PAGE lysis buffer and used for immunoblot analysis by using anti-PA antibodies to visualize the captured biotin-labeled PA present in the lysates. As a positive control, myeloid cells purified from a group of mice immunized with biotin-labeled wild-type PA were used.

Detection of functional PA associated with myeloid cells

Mice (five mice/group) were immunized with unlabeled wild-type PA (25 µg of PA protein/mouse; IP route). Spleens were collected after 4 hours of immunization from the immunized and unimmunized mice and processed to prepare total splenocytes and purified/enriched populations of myeloid cells, as described above. Myeloid cells were treated with 1 µg/mL anthrax LF at 37°C. After treatment, cells were washed extensively and lysed with MPER lysis buffer. Equal amounts of protein lysates from the immunized and unimmunized groups (treated or not treated with LF) were loaded onto the SDS-PAGE gels. Immunoblot analysis was performed using antibodies specific for N-terminal MEK and total MEK to detect LF-mediated N-terminal MEK cleavage mediated by internalized LF.

Statistical analyses

Statistical analyses were performed using GraphPad Prism software (version 6; GraphPad Prism Software, Inc., La Jolla, CA). For the TNA assay, all non-responder mice were assigned an ED50 value of 18 (1/2 the limit of quantitation of the TNA assay). For the ELISA, all non-responder mice were assigned an antibody titer value of 5 (1/2 the limit of quantitation of the ELISA).