Modular complement assemblies for mitigating inflammatory conditions

Edited by Matthew V. Tirrell, University of Chicago, Chicago, IL, and approved February 22, 2021 (received for review September 3, 2020)
April 5, 2021
118 (15) e2018627118


Current treatments for chronic inflammatory conditions rely on biologic drugs, commonly monoclonal antibodies that interfere with inflammatory signaling pathways. These drugs have made enormous contributions to the treatment of inflammatory diseases but still possess considerable drawbacks, including high cost that limits access in low-resource settings, the requirement for regularly repeated injections, and uneven efficacy. As an alternative to such biologics, active immunotherapies, in which an individual is induced to generate their own therapeutic antibodies, offer considerable potential advantages. Here, we report chemically defined nanomaterials inducing therapeutic responses in two models of inflammation in mice. The materials were produced by coassembling defined T cell epitopes, B cell epitopes, and an engineered fragment of complement protein C3dg into defined nanomaterials.


Complement protein C3dg, a key linkage between innate and adaptive immunity, is capable of stimulating both humoral and cell-mediated immune responses, leading to considerable interest in its use as a molecular adjuvant. However, the potential of C3dg as an adjuvant is limited without ways of controllably assembling multiple copies of it into vaccine platforms. Here, we report a strategy to assemble C3dg into supramolecular nanofibers with excellent compositional control, using β-tail fusion tags. These assemblies were investigated as therapeutic active immunotherapies, which may offer advantages over existing biologics, particularly toward chronic inflammatory diseases. Supramolecular assemblies based on the Q11 peptide system containing β-tail–tagged C3dg, B cell epitopes from TNF, and the universal T cell epitope PADRE raised strong antibody responses against both TNF and C3dg, and prophylactic immunization with these materials significantly improved protection in a lethal TNF-mediated inflammation model. Additionally, in a murine model of psoriasis induced by imiquimod, the C3dg-adjuvanted nanofiber vaccine performed as well as anti-TNF monoclonal antibodies. Nanofibers containing only β-tail–C3dg and lacking the TNF B cell epitope also showed improvements in both models, suggesting that supramolecular C3dg, by itself, played an important therapeutic role. We observed that immunization with β-tail–C3dg caused the expansion of an autoreactive C3dg-specific T cell population, which may act to dampen the immune response, preventing excessive inflammation. These findings indicate that molecular assemblies displaying C3dg warrant further development as active immunotherapies.
The protein C3dg, a late product of the complement cascade, functions as a key interface between innate and adaptive immunity. It has received considerable interest as a molecular adjuvant, but its utility in immunotherapies has yet to be fully realized. When C3dg is attached to protein antigens, it enhances immunogenicity by engaging both humoral and cell-mediated immunity. It has long been understood that the binding of the C3dg fragment to an antigen promotes B lymphocyte activation through coengagement of the B cell receptor (BCR) and complement receptor 2 (CD21). The simultaneous engagement of these signals lowers the threshold for activation of B cells and induces isotype switching, somatic hypermutation, and B cell memory (1, 2). More recently, C3dg has also been found to be rich in T cell epitopes that, when presented by major histocompatibility complex class II molecules to autoreactive T helper cells, lead to an enhancement of cell-mediated immunity (3). These effects of C3dg are further enhanced when an antigen is linked with multiple copies of C3dg or C3d (a fragment of C3dg which exhibits similar behavior) (410). Some C3d-adjuvanted vaccine platforms rely on cross-linking to create clusters of C3d-decorated antigen, producing complexes as great as 20-mers, but the random nature of assembly is difficult to control and may occlude the CD21-binding domain (11). An alternate method involves genetic assembly, expression, and purification of individual constructs linking C3dg and an antigen, but this is both time-consuming and restricted by the size of the expression product, often limiting the maximum C3d copy number to three or fewer (2, 8, 9). Thus, the potential of C3dg as an adjuvant is limited without ways of controllably assembling multiple copies of C3dg molecules into vaccine platforms.
The capacity to enhance both humoral and cell-mediated immunogenicity makes C3dg attractive in a broad range of vaccine applications. In the present study, we explored its use toward therapeutic active immunotherapy, which promises to offer important advantages over existing biologics such as monoclonal antibodies (mAb), particularly toward chronic inflammatory diseases. Despite the significant impact mAb-based therapies have made on the treatment of diseases such as rheumatoid arthritis, Crohn’s disease, and psoriasis, these drugs have considerable shortcomings, including the need for repeated injections and resultant patient compliance issues, tolerability, cost, primary nonresponse, and secondary loss of response due to the generation of anti-drug antibodies (ADAs). ADAs that develop in response to frequently administered mAbs not only lead to a reduction in drug efficacy but also contribute to serious adverse immune reactions such as hypersensitivity (12). Active immunotherapies address these drawbacks in principle by stimulating the immune system to produce its own therapeutic antibodies against problematic self-molecules. This mode of treatment, if successful, would require fewer doses, likely leading to improved patient compliance and tolerance, in addition to polyclonal responses that may improve therapeutic efficacy. Despite its advantages, a common concern with active immunotherapy is the risk of undesirable antibody persistence. It should be noted, however, that in the absence of boosting, antibody titers generally steadily decline over time in previously investigated active immunotherapies (13). Active immunotherapy platforms under investigation have previously comprised conjugates of antigen to carrier proteins such as keyhole limpet hemocyanin (14) or virus-like particles (15, 16) that provide exogenous T cell epitopes, facilitating the production of T-dependent immune responses (17). Many such platforms additionally require adjuvants to achieve a therapeutic response (18), which commonly induce some degree of inflammation and may not necessarily induce the desired T helper phenotype or antibody subclass/isotype (19).
Previously, our group reported an anti-TNF active immunotherapy based on a supramolecular peptide system. In this work, exogenous T cell epitopes and TNF B cell epitopes were coassembled into nanofibers using the self-assembling peptide Q11 (Ac-QQKFQFQFEQQ-NH2) (19). The TNF B cell epitope was chosen based on its previously reported ability to raise antibodies that react specifically against the soluble form of TNF but only weakly bind to membrane-bound TNF (15, 20). This choice was made to maximize the production of neutralizing antibodies against the circulating cytokine, while minimizing the potential for autoimmunity or compromising the ability to fight bacterial infections. In this previous work, optimized immunizations afforded partial protection in mice in a model of lipopolysaccharide (LPS)-induced inflammation. Surprisingly, the addition of a CpG adjuvant to this system, intended to enhance antibody titers, instead led to a diminished therapeutic effect, which we attributed to an overly inflammatory immune phenotype induced by CpG. This finding highlighted one of the key advantages of peptide assemblies, which can generate antibody responses with minimal or no associated inflammation in the absence of adjuvants (2123). In the present work, we sought to enhance these antibody responses using C3dg, exploiting its known abilities to engage humoral and cellular immunity and hypothesizing that if induced C3dg-specific T helper cells were sufficiently noninflammatory, the cytokines they produce may additionally help quell aberrant cycles of inflammation found in chronic inflammatory diseases. In this way, we sought to utilize C3dg through three potential mechanisms: augmenting antibody titers against the target molecule via direct interactions with B cells, utilizing built-in T cell epitopes to further help B cell activation, and maintaining anti-inflammatory phenotypes in responding T cells at sites of complement activation.
To create defined molecular assemblies, we tagged proteins with β-tails, which facilitate the coassembly of multiple different expressed proteins into supramolecular nanofibers with tight compositional control (24). We designed supramolecular nanofibers containing defined quantities of peptide B cell epitopes, peptide T cell epitopes, and β-tail–tagged C3dg (Fig. 1 AC). Nanofibers containing all three components significantly improved protection in the lethal TNF-mediated inflammation model. Additionally, in a murine model of psoriasis induced by imiquimod, the C3dg-adjuvanted nanofiber vaccine performed as well as the current standard of care, anti-TNF mAB. Nanofibers containing only β-tail–C3dg and lacking the TNF B cell epitope also showed improvements in both models, suggesting that supramolecular C3dg, by itself, played an important therapeutic role.
Fig. 1.
Engineered β-tail–C3dg integrates into self-assembled peptide nanofibers. (A) Schematic representation of the β-tail’s transitional behavior into a β-sheet conformation (24). (B) Schematic of nanofiber components. (C) Schematic of coassembled nanofibers. (D) SDS-PAGE of expressed, purified β-tail–C3dg protein. (E) Negative-stained TEM images of nanofibers containing 20 μM β-tail–C3dg (prepared at 2 mM total Q11 peptide concentration and diluted to 0.2 mM for imaging). (F) β-tail–C3dg was integrated into Q11 nanofibers over the range of 12.5 to 50 μM in a β-tail–dependent manner, as measured by loss of protein from the supernatant. (G) Quantification of β-tail–C3dg and C3dg assembly into nanofibers, expressed as the percentage of total protein incorporated. n = 3, ± SD for F and G. (Scale bar, 200 nm.) Statistical significance (G) was tested by multiple t tests with Holm–Sidak correction, **P < 0.01, ***P < 0.001.


Supramolecular β-tail–C3dg Assembles into Peptide Nanofibers with Gradated Control.

In Escherichia coli, β-tail–C3dg was cloned and expressed (sequence in SI Appendix). Previous work established the utility of the β-tail in producing β-sheet nanofibers capable of displaying strictly controlled combinations of multiple proteins while maintaining their native folding and biological activity (24). Because of the β-tail peptide’s slow transition from an α-helix to a β-sheet (24) (Fig. 1A), β-tail–C3dg was expressed with minimal aggregation, or misfolding, and was recovered from E. coli in yields of 8 to 10 mg/L bacterial culture. The identity of purified β-tail–C3dg was confirmed via SDS-polyacrylamide gel electrophoresis (PAGE) (Fig. 1D) and dot blot, using a primary antibody against the native C3dg structure (SI Appendix, Fig. S1A). Q11 peptide nanofibers are effectively sedimented via centrifugation, and protein incorporation can be measured by monitoring the loss of β-tail-C3dg in the supernatant of pelleted nanofibers (24). Using this technique, the integration efficiency of β-tail–C3dg was found to vary depending on the protein concentration, from 96 to 84 to 70% as the concentration of β-tail–C3dg was increased from 12.5 to 25 to 50 μM, respectively (Fig. 1 F and G). The upper threshold of assembly was found to be around 1 β-tail–C3dg per 60 Q11 peptides, considering around 30% of the protein remained in the supernatant when 50 μM β-tail–C3dg was assembled with 2 mM Q11 peptide. With this in mind, we chose 20 μM β-tail–C3dg for initial immunization formulations owing to the >90% incorporation efficiency at this concentration. C3dg protein expressed without the β-tail was pelleted with the Q11 nanofibers to a far lesser but still measurable degree, indicating some nonspecific adsorption for nontagged proteins, likely reflecting the extensive surface area of the nanofiber network. However, the β-tail was required for the reliable integration of high concentrations of protein into the nanofibers in a controlled, gradated manner. Assembling Q11 and β-tail–C3dg together produced well-formed nanofibers by transmission electron microscopy (TEM) (Fig. 1E), and their morphology did not appear to be influenced by varying amounts of β-tail–C3dg (SI Appendix, Fig. S1). They also did not appear to be perturbed by the Q11-conjugated B and T cell epitopes employed later in the study (SI Appendix, Fig. S1 BH). This was consistent with previous observations of Q11-based peptides bearing a wide range of N-terminal epitopes, which have been shown repeatedly to assemble stoichiometrically into nanofibers (19, 22, 23, 25, 26). Previous work also established the colocalization of β-tail fusion proteins within Q11 nanofibers via immunogold labeling and TEM (24).

β-tail–C3dg Enhances B Cell Responses to Peptide Nanofibers In Vitro and In Vivo.

The content and multivalency of C3dg in the nanofibers had a direct effect on B cell activation in vitro, as measured by BCR-induced Ca2+ influx in splenic B cells using the calcium-sensitive dye Fluo-4 and flow cytometry (Fig. 2). The adjuvanting effect of C3dg has been known to be enhanced with an increasing copy number (2, 46), caused by the simultaneous binding of multiple CD21/CD19 surface receptors and BCR by polyvalent antigens (27, 28). Whereas the addition of unmodified Q11 did not affect BCR-induced Ca2+ responses in splenic B cells, nanofibers bearing increasing concentrations of β-tail–C3dg elicited progressively stronger and faster Ca2+ influx responses (Fig. 2). To confirm that the observed effect was due to CD21 interactions rather than nonspecific protein interactions with the BCR, nanofibers containing β-tail–GFP (green fluorescent protein) were assessed as an irrelevant recombinant protein control. β-tail–GFP nanofibers did not induce significant Ca2+ responses, suggesting that it was the specific β-tail–C3dg–CD21/CD19 engagement that drove the BCR-induced Ca2+ responses in a dose-dependent manner (Fig. 2 A and B). Additionally, assembled C3dg (β-tail–C3dg) was able to activate B cells more rapidly and to a greater extent than soluble, unassembled C3dg, which elicited a delayed response of reduced magnitude (Fig. 2 C and D). These results collectively indicated the enhanced capacity of C3dg to activate B cells when assembled into nanofibers, and it reinforced the appropriateness of 20 μM β-tail–C3dg in the nanofibers for subsequent experiments, as this concentration exhibited both reliable assembly and strong B cell activating capacity.
Fig. 2.
β-tail-C3dg activates B cells in vitro in a dose-dependent manner and enhances humoral responses in vivo. (A) Nanofibers containing increasing amounts of β-tail–C3dg enhanced B cell activation in a dose-dependent manner. Signal was normalized by subtracting the initial mean florescent intensity measurement from all subsequent data points. Arrow indicates time of nanofiber application. (B) Total calcium signaling integrated over 90 s poststimulation for the same groups as in A. (C) β-tail–C3dg assembled into Q11 nanofibers elicited faster and stronger calcium signaling than soluble C3dg. (D) Total calcium signaling integrated over 90 s poststimulation for the same groups as in C. OVA peptide-specific (E) and C3dg protein-specific (F) IgG responses in the sera of mice immunized with formulations containing OVAQ, soluble C3dg, and supramolecular β-tail–C3dg. TNF peptide-specific (G) and C3dg protein-specific (H) IgG responses in the sera of mice immunized with various formulations containing TNFQ, PADREQ, soluble C3dg, and β-tail–C3dg. (AD) Combination of two independent experiments in which sample sizes were n = 2 mice. In E, mice were immunized at week 0 and boosted at week 3 and 6. In F, mice were immunized at week 0 and boosted at week 4. n = 5 for each group in EH. Mean ± SD is shown. Statistical significance was tested by one-way ANOVA with Tukey’s multiple comparison test (B), Student’s two-tailed, unpaired t test (D), or two-way ANOVA with Tukey’s multiple comparison test (EH). ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.
We next investigated the immunogenicity-enhancing effects of β-tail–C3dg when coassembled into nanofibers with other peptide epitopes. The peptide OVAQ contains the OVA323–339 mixed B cell/T cell epitope conjugated to the N terminus of the self-assembling Q11 peptide, and it raises strong antibody responses in nanofiber form (22, 23). When β-tail–C3dg was coassembled into nanofibers along with OVAQ and delivered to C57BL/6 mice in subcutaneous immunizations, these mixed nanofibers induced significantly greater OVA-specific antibody responses compared to formulations without β-tail–C3dg or with unassembled (soluble) C3dg (Fig. 2E). Prior to boosting, only formulations containing either β-tail–C3dg or soluble C3dg raised measurable responses, and after two boosts, nanofibers containing both OVA and β-tail–C3dg raised the strongest responses. The IgG subclass of the anti-OVA peptide antibody response was altered by β-tail–C3dg immunization content as well, leading to increases in titers of both IgG2b and IgG2c (SI Appendix, Fig. S2A). Formulations containing C3dg also raised antibody responses against C3dg itself (Fig. 2 F and H), indicating that the C3dg protein contains at least one B cell epitope. It should be noted that the plates were coated with native C3dg (not containing the β-tail or linker), meaning that these antibodies bind to native C3dg protein itself rather than the added motifs. Importantly, these antibodies did not bind to the progenitor proteins of C3dg, complement C3, or C3b (SI Appendix, Fig. S3 C and D), indicating that the dominant B cell epitope(s) in C3dg were not available in these proteins upstream of it in the complement cascade. We additionally assayed for serum levels of C3 in immunized mice and found no significant differences between unimmunized and immunized mice (SI Appendix, Fig. S3A). We also measured the total hemolytic activity of pooled serum from immunized mice using sensitized sheep erythrocytes. Hemolytic activity can be diminished if complement components in the classical pathway (C1, C4, C2, and C3) or terminal pathway (C5 through C9) are blocked or cleared. None of the groups tested exhibited a compromised complement cascade (SI Appendix, Fig. S3B). Thus, we conclude that the anti-C3dg antibodies observed did not detrimentally alter the complement system.
The addition of C3dg also led to a more rapid, robust humoral response in a nanofiber formulation containing TNFQ, comprising Q11 conjugated to an N-terminal B cell epitope from mouse TNF4–23, found on the surface of soluble TNF. Some formulations also included the peptide PADREQ, consisting of Q11 conjugated to the synthetic pan–DR epitope, PADRE. The group immunized with soluble C3dg and TNFQ formulation raised the lowest anti-TNF antibody titers, averaging a titer of ∼2.5 by week 9. The TNFQ/PADREQ formulation raised slightly higher titers, indicating that the assembled T cell epitope helped facilitate antibody production. Interestingly, the formulation containing β-tail–C3dg and TNFQ raised higher titers than the TNFQ/PADREQ group. This phenomenon could be attributed to assembled C3dg providing T cell help similarly to PADRE. Alternatively, it could also reflect the direct costimulatory effect of CD19/CD21 engagement by C3dg on the B cell response. Like in the OVA study, the formulation containing β-tail–C3dg, PADREQ, and TNFQ not only raised the greatest anti-TNF IgG by week 9 but also raised significant titers by week 2, prior to boosting (Fig. 2G). Isotyping revealed that the formulations containing all three of the aforementioned components also enhanced the magnitude of IgG2b and IgG3 responses compared to the other formulations (SI Appendix, Fig. S2B).
Our group previously showed that immunization with nanofibers bearing the TNF4–23 B cell epitope did not diminish the ability of mice to clear Listeria infections (19). Here, we found that the addition of β-tail–C3dg to the nanofiber formulation did not alter this result, despite the enhanced anti-TNF IgG response observed with these formulations (SI Appendix, Fig. S7). This was not unexpected as we deliberately chose an epitope that was found to raise antibodies against soluble TNF rather than transmembrane TNF (20), which is expressed on several activated lymphocytes. We did note in the first group of five mice evaluated that one mouse receiving the full immunization formulation exhibited increased Listeria burden. We reevaluated this finding with 10 more mice, without observing increased susceptibility.

Immunization Protects from TNF-Mediated Inflammation in Mice.

In a model of TNF-mediated inflammation, mice receiving prophylactic immunization with formulations containing β-tail–C3dg exhibited significant improvements in survival and for some formulations the prevention of shock-like symptoms (Fig. 3A). In this model, LPS is delivered intraperitoneally and induces symptoms including weight loss and hypothermia (2932). Without immunization, this model was lethal for 90% of mice in control groups, with mice removed from the study upon reaching predetermined cutoffs (20% body weight loss or 32 °C body temperature). In this model, the body temperature of challenged mice steadily decreases for about 12 h, and if mice survive, the temperature recovers over the course of the next 3 d. Strikingly, we found that immunization with the β-tail–C3dg/TNFQ/PADREQ formulation provided complete protection, with these mice exhibiting minimal symptoms and appearing nearly unaffected by the challenge (Fig. 3H). The body temperature loss at the 12 h timepoint for these mice was not statistically different from that of the negative control mice who were not challenged with LPS (Fig. 3B). All formulations containing β-tail–C3dg or soluble C3dg also had significantly smaller temperature differentials compared to the unimmunized positive control group, indicating that C3dg, by itself, may have played an important role in enhancing therapeutic efficacy.
Fig. 3.
Immunization with peptide nanofibers protected in a model of acute TNF-mediated inflammation, with the optimized formulation preventing the development of shock-like symptoms. (A) The difference between the highest and lowest measured temperature per mouse. (BI) Body temperature and overall survival of individual mice in each group. Letters at the lower right-hand corner of each graph indicate other groups that are statistically different (P < 0.05). (I and J) The lowest recorded temperature of each mouse graphed against its anti-TNF (J) and anti-C3dg (K) titers from serum collected 1 wk before the LPS Challenge. Combination of two independent experiments in which n = 5 each (soluble C3dg + TNFQ group was only included in one experiment). Statistical comparisons of temperature changes (A) and temperature/titer correlations (I and J) were made using one-way ANOVA with Tukey’s multiple comparison test. Statistical comparisons of survival between all groups in CH were made using log-rank test; ns = not significant, *P < 0.05, ****P < 0.0001.
Mice immunized with nanofibers containing only Q11 and β-tail–C3dg or β-tail–C3dg and PADREQ (no additional TNF antigen) also exhibited reduced temperature loss, with an overall survival rate of 90% (Fig. 3 E and G). These mice exhibited no anti-TNF antibodies but did have high anti-C3dg antibody titers (Fig. 2H), suggesting that the responses raised against C3dg itself may have had therapeutic value. We also observed a correlation between temperature and anti-TNF (Fig. 3J) or anti-C3dg titers (Fig. 3K). As the titers against each antigen decreased, the mice were more likely to experience body temperature loss. While a causal relationship was not possible to draw from this experiment, both of the correlations between lowest recorded temperature and α-TNF or α-C3dg titer were statistically significant. These results indicated that these antibodies were correlated with a reduction in shock-like symptoms. The correlation between α-C3dg titer and temperature was weaker than that of the α-TNF titer, indicating that these α-C3dg antibodies may play a role, but there are likely other factors that more strongly contribute to protection. To once again ensure the anti-C3dg antibodies were not affecting the complement cascade, we assayed for C5a levels in the serum and intraperitoneal space at three time points during a repeated LPS challenge. The protein C5a, an anaphylatoxin, is a downstream product of the complement cascade and is implicated in the mediation of endotoxic fever (33). We found no significant difference in the C5a levels between unimmunized mice and mice immunized with β-tail–C3dg at any of the time points (SI Appendix, Fig. S5), once again reinforcing the conclusion that nanofiber immunizations were not inhibiting the complement cascade broadly. Nevertheless, immunization with β-tail–C3dg but no TNF B cell epitope significantly reduced or prevented inflammation in the LPS challenge model, leading us to investigate further the therapeutic value of nanofibers containing solely C3dg.

Immunization with Nanofibers Containing β-tail–C3dg Causes the Production of C3dg-Specific Autoreactive Helper T Cells.

Next, we examined whether C3dg would promote a productive T helper response. Although it is not presently well understood what parameters specify the range of T cell phenotypes for fibrillized peptide nanofiber vaccines, we have previously found that T helper phenotype influences therapeutic efficacy. Previous investigations of unadjuvanted Q11-based materials using the OVA epitope have led to unpolarized (23) or TH2-slanted (26) T cell responses; however, the addition of an adjuvant such as CFA or CpG causes a shift to a Th1-slanted response (19, 21). As mentioned earlier, we have previously found that antibody titers are not the only requirement for therapeutic efficacy in active immunotherapies against autologous targets, and that other factors influencing the phenotype of the immune response appear to be important considerations (19).
Here, the addition of β-tail–C3dg or C3dg into immunization formulations containing OVAQ did not strongly alter the TH2-slanted response when lymphocytes harvested from spleens were restimulated with peptide OVA (Fig. 4 AC). These cells remained polarized toward TH2 with a slight increase in the number of IFN-γ secreting cells. Additionally, when restimulated with C3dg, lymphocytes from mice immunized with formulations containing C3dg largely secreted IL-4, indicating the presence of TH2-biased, C3dg-specific helper T cells.
Fig. 4.
β-tail–C3dg in nanofibers induces the production of TH2-biased, C3dg-specific helper T cells. Cytokine-secreting cells from mice immunized with OVAQ formulations (AC) and TNFQ formulations (DF) were quantified ex vivo using ELISpot. Lymphocytes from spleens were suspended and restimulated with peptide epitope or C3dg. n = 5 for all groups; data points represent individual mice. In AC, statistical significance was tested by multiple two-tailed t tests with Holm–Sidak correction. In DF, statistical significance was tested by one-way ANOVA with Tukey’s multiple comparison test. ns: not significant, *P < 0.05, **P < 0.01, ***P < 0.001.
Interestingly, the addition of β-tail–C3dg into the TNFQ/PADREQ nanofiber formulation caused a shift in the T cell response to PADRE (Fig. 4F). The formulation without β-tail–C3dg raised a strong IL-4–dominant TH2 phenotype, whereas nanofibers containing β-tail–C3dg enhanced IFN-γ secretion upon PADRE restimulation, leading to a relatively unpolarized response. The alteration of the PADRE-specific response was particularly interesting. We previously reported that nanofibers with varying T cell epitope content elicit altered T cell responses with specific formulations altogether abolishing the response (19); therefore, it is not surprising that the addition of the C3dg would enhance or alter the response. Once again, when restimulated with C3dg, we observed strong IL-4 secretion and a TH2-polarized phenotype. These results are consistent with previous findings that C3dg contains T cell epitopes capable of stimulating autoreactive C3dg-specific T helper cells that have escaped thymic deletion (3). These C3dg-specific autoreactive T cells would be circulating in the periphery, poised to respond to C3dg bound to foreign antigens.
We hypothesized that nanofiber immunizations increased the population of TH2-biased, C3dg-specific T helper cells, and these cells became activated at sites of inflammation due to the increased presence of C3dg (a result of the activation of the complement cascade). The activated T cells may then be secreting anti-inflammatory cytokines, thus reducing local or systemic inflammation. This hypothesis would explain why there was improvement in mice that were immunized with nanofibers containing β-tail–C3dg but no TNF epitope. We tested this hypothesis in the LPS model by measuring the levels of IL-10 and TNF in the serum or intraperitoneal lavage of unimmunized mice or mice immunized with nanofibers containing β-tail–C3dg 2 h postchallenge. We saw that the β-tail–C3dg-immunized mice had significantly greater levels of IL-10 in the lavage and significantly reduced levels of TNF in the serum and lavage compared to the unimmunized mice (SI Appendix, Fig. S6). This finding suggested that the secretion of anti-inflammatory cytokines may have contributed to the overall therapeutic effect. It is also possible that mice immunized with nanofibers containing β-tail–C3dg, TNF B cell epitope, and PADRE T cell epitope in the same nanofiber experienced the synergistic effect of having both an anti-TNF antibody response as well as an anti-complement TH2-biased T cell response. Alternatively, the anti-C3dg antibodies that were raised upon immunization may have contributed to the therapeutic effect seen. At present, we do not know how these antibodies would affect the pathogenesis of inflammation, but other studies have shown that complement C3 is involved in the development psoriasis-like skin inflammation induced by short-term treatment with imiquimod (34). In prior studies, mice lacking C3 displayed less skin inflammation compared to wild type (WT), indicating that C3 plays a proinflammatory role in the disease model. While anti-C3dg antibodies produced in response to the nanofiber immunizations do not bind and clear C3, it is possible that anti-C3dg antibodies have an alternative effect on the proinflammatory role played by C3.
We have not observed that the T cell responses described herein are detrimental to the function of the complement cascade in innate immunity. We were initially concerned with raising an autoimmune response against an abundant plasma protein; however, we tested several parameters of complement function, and immunized mice showed no impairment in complement function compared to unimmunized mice (SI Appendix, Fig. S3). We also found that immunized mice did not become susceptible to bacterial infection in a Listeria monocytogenes (Lm) challenge (SI Appendix, Fig. S7). All of these results indicated that the C3dg-specific T cells may play a beneficial role in the therapeutic treatment of chronic inflammation without compromising antimicrobial immune responses.

Immunization Significantly Reduces Symptoms of Local Inflammation in a Model of Imiquimod-Induced Psoriasis.

Finally, we investigated the ability of peptide nanofiber immunizations to reduce inflammation in a model of imiquimod-induced psoriasis. In this model, the repeated application of imiquimod induces clinical and histological changes such as erythema, scaling, and induration, approximating the human phenotype of psoriasis (35, 36). The skin reaction seen in mice is shown to be dependent on IL‐23 and IL‐17RA signaling, which are both known to play a pivotal role in psoriasis (3638). Additionally, TNF plays a significant role in imiquimod-induced inflammation (35), suggesting that immunization against TNF could have a significant effect on this model’s pathogenesis.
Mice immunized with the β-tail–C3dg/TNFQ/PADREQ formulation that performed the best in the LPS challenge exhibited the least amount of epidermal thickening among the immunized mice groups (Fig. 5 A and B). This group exhibited significantly reduced epidermal thickening compared to the unimmunized control mice and was not significantly different from the group treated with a monoclonal antibody against TNF (Fig. 5B), a current clinical treatment for psoriasis. Anti-TNF antibodies were also observed in all mice receiving vaccinations with the TNF B cell epitope peptide (SI Appendix, Fig. S8). The dispensability of the TNF epitope was confirmed using TNF knockout (KO) mice, which were immunized with a formulation containing β-tail–C3dg/TNFQ/PADREQ. These mice also exhibited reduced epidermal thickening (Fig. 5 C and D). These mice do not express TNF; so, both the pathology of imiquimod-induced psoriasis and the therapeutic efficacy of the C3dg component appear to be independent of TNF.
Fig. 5.
Immunization with peptide nanofibers reduces the amount of epidermal thickening in a model of imiquimod-induced psoriasis. Representative histology images and epidermal thickening of skin collected from C57BL/6 (AB) and TNF-KO (CD) mice are shown. (B) Mice treated with anti-TNF antibody 3 and 0 h prior to the first application of imiquimod exhibited similar levels of reduction in epidermal thickening when compared to mice immunized with β-tail–C3dg/TNF/PADRE nanofibers. (D) TNF-KO mice immunized with the β-tail–C3dg/TNF/PADRE formulation also showed reduced epidermal thickening. Data points represent individual mice. (B) Combination of two independent experiments in which n = 5 for each group (not all groups were included in the repeated experiment). (D) Single experiment with n = 5 for each group. Statistical significance was tested using one-way ANOVA with Tukey’s multiple comparison test. ns: not significant, *P < 0.05, **P < 0.01, ****P < 0.0001. (Scale bar, 200 μm.)

C3dg-Specific CD4+ T Cells, rather than α-C3dg Antibodies, Play a Crucial Role in Therapeutic Efficacy.

In order to gain more mechanistic insight into the unexpected therapeutic benefit of immunization with β-tail–C3dg, we investigated the individual contribution of α-C3dg antibodies to protection in an LPS challenge. Serum collected from unimmunized mice and serum collected from β-tail–C3dg-immunized mice were transferred to naïve recipients (Fig. 6 A and B). One day after the passive transfer, serum was collected from the recipients to measure their individual levels of α-C3dg antibodies (Fig. 6C). The mice were then challenged with LPS as described earlier. The recipient mice were unimmunized; therefore, any effects observed can be attributed to the presence or absence of α-C3dg antibodies. Interestingly, neither group exhibited protection in the LPS challenge, with all of the mice falling below the humane endpoint body temperature (32 °C) by hour 12 of the challenge. This result indicates that α-C3dg antibodies did not play a predominant role in the prevention of shock-like symptoms in this model.
Fig. 6.
CD4+ T cells play a critical role in the prevention of shock-like symptoms, whereas α-C3dg antibodies do not provide protection in an LPS challenge. Serum from (A) unimmunized mice or (B) β-tail–C3dg immunized mice was administered via tail vein injection on day −1 before the LPS challenge. (C) C3dg protein-specific IgG responses in the sera of mice (D and E) collected immediately prior to the start of the LPS challenge. (DF) Mice immunized with β-tail–C3dg assembled into nanofibers or unimmunized mice were treated either with a CD4-depleting antibody or isotype control at days −3 and −1 before the start of the LPS challenge. (D and E) Mice were immunized at week 0 and boosted at week 4. n = 5 for each group. (C) Statistical significance tested by Student’s two-tailed, unpaired t test. Statistical comparisons of survival between groups DF were made using log-rank test. Letters at the lower right-hand corner of each graph indicate other groups that are statistically different (P < 0.05). ****P < 0.0001.
Next, we explored the role of CD4+ T cells and specifically C3dg-specific autoreactive CD4+ T cells. Mice were immunized with nanofibers containing β-tail–C3dg at weeks 0 and 4, and a week later, immediately prior to an LPS challenge, CD4+ T cells were depleted using a monoclonal antibody treatment (Fig. 6D and SI Appendix, Fig. S9). None of these mice exhibited protection. Comparatively, β-tail–C3dg-immunized mice that received the isotype control antibody demonstrated mild shock-like symptoms, with an overall survival rate of 100% (Fig. 6E). Unimmunized mice that received the same isotype control, however, did not exhibit protection, with an overall survival rate of 0%. This experiment demonstrates that C3dg-specific CD4 T cells played a crucial role in preventing endotoxic shock, and it appears that this cell population, generated by immunization with nanofibers containing β-tail–C3dg, is the main contributor to the therapeutic benefit that we observed in the previous LPS challenge (Fig. 3), as well as the imiquimod model (Fig. 5) when the TNF B cell epitope was absent from the immunization.
This result, though surprising, is supported by evidence of other natural negative feedback mechanisms of the immune system. Autoreactive T cells that have escaped thymic deletion can be converted to function as regulatory T cells and may moderate effector immune responses as a means of controlling inflammation (39). For example, the presence of regulatory T cell epitopes in the Fc region of IgG can have a tolerizing effect in some immunoglobulin therapies (40). This demonstrates that other autologous molecules have been shown to have therapeutic effects by T cell-dependent mechanisms. We hypothesize that the T cell epitopes present in C3dg played a similar role: By immunizing with β-tail–C3dg, the population of C3dg autoreactive T cells was expanded and exerted influence over the immune responses observed herein.


Supramolecular β-tail–C3dg enables the display of complement C3dg in β-sheet nanofibers alongside B and T cell epitopes. With this strategy, humoral and adaptive immune responses can be enhanced and tuned. In this work, β-tail–tagged C3dg, B cell epitopes from TNF, and the universal T cell epitope PADRE not only raised strong antibody responses against both TNF and C3dg but also expanded a population of C3dg autoreactive T cells. These nanofibers were strongly protective in a model of acute TNF-mediated inflammation, preventing the development of shock-like symptoms. Additionally, in a murine model of imiquimod-induced psoriasis, immunization with optimized nanofiber formulations was therapeutically equivalent to treatment with a monoclonal antibody against TNF, a current standard immunotherapy. We found that immunization with nanofibers containing βtail–C3dg did not affect the function of the complement cascade, and mice did not become significantly susceptible to bacterial infection with Lm. The modular nature of the peptide nanofiber platform allowed an investigation into the role played by each vaccine component and enabled the discovery of a potentially beneficial role played by C3dg autoreactive T cells. For these reasons, fibrillar peptide assemblies displaying C3dg warrant further development as active immunotherapies for chronic inflammatory conditions.


Peptide Synthesis.

Peptide sequences are shown in SI Appendix. All peptides were synthesized using standard Fmoc solid-phase synthesis protocols. Peptides were purified via reverse-phase high-performance liquid chromatography, and peptide molecular weight was verified by matrix-assisted laser desorption/ionization (MALDI) mass spectrometry with α-cyano-4-hydroxycinnamic acid as the matrix.

Cloning, Expression, and Purification of Proteins.

See SI Appendix for detailed cloning, expression, and purification methods. Plasmids containing a fusion protein were constructed using a pET-24a+ vector modified for the seamless fusion of genes (41). The gBlocks containing the β-tail–C3dg gene along with 5′ and 3′ overhang sequences (80 bp) homologous to the vector insert site were purchased from Integrated DNA Technologies. The gBlock was inserted into the BseRI-digested pET-24a+ vector via Gibson Assembly. The sequence of each recombinant fusion was confirmed by sequencing performed by Genewiz. Fusion proteins were expressed in BL21 E. coli (New England Biolabs) and purified using metal-affinity chromatography on HisPur cobalt resin (Thermo Scientific). Endotoxin content was reduced using Triton X-114 cloud-point precipitation, according to previously reported methods (42). All immunizing formulations contained less than 1 endotoxin units per mL, as confirmed using the Limulus Amebocyte Lysate kit (Lonza).

Nanofiber Preparation.

Nanofiber formation generally followed previously published protocols (19, 24). All immunizations contained a total peptide concentration of 2 mM in phosphate-buffered saline (PBS). For immunizations containing OVA epitopes, 1 mM OVAQ was mixed with 1 mM Q11. For TNF immunizations, nanofibers contained 1 mM TNFQ, 0.05 mM PADREQ, and the balance Q11 to equal 2 mM total peptide. To form nanofibers, dry lyophilized peptides were first intermixed by vortexing for 30 min and then dissolved in sterile water to a total concentration of 10 mM peptide. For immunizations containing β-tail fusion proteins, aqueous peptide solutions were next diluted fourfold with 1× PBS containing the β-tail–C3dg at a total protein concentration of either 2.5 or 25 μM. Further, 10× PBS was added, bringing the final peptide concentration to 2 mM (a 0.2-fold dilution). The solution was then incubated 4 °C overnight (24) to ensure equilibrated self-assembly. To prepare formulations without β-tail fusion proteins, the intermixed lyophilized peptides were dissolved in sterile water at a total concentration of 2.5 mM and incubated overnight at 4 °C. Peptide solutions were then diluted to a final concentration of 2 mM with 10× PBS and incubated for at least 3 h at room temperature (43).

Characterizing Protein Integration into Nanofibers.

Q11 nanofibers that were formed overnight in the presence of β-tail fusion proteins were sedimented by centrifugation at 12,000 × g for 5 min. The supernatant was removed and analyzed for β-tail–C3dg or C3dg content via SDS-PAGE with Coomassie blue staining. Band density was converted to protein concentration using a standard curve. Protein concentration in the nanofibers was reported as the difference between the protein concentration in solution during nanofiber assembly and the protein concentration in the supernatant after centrifugation.


TEM was used to visualize the morphology of peptide nanofibers with and without the integration of β-tail–C3d, similar to previously reported methods (24). Briefly, formed nanofibers were diluted to 0.2 mM total peptide in PBS to ensure low enough density on the TEM grid for visualization. A total of 5 μL of nanofiber suspensions were deposited onto Formvar/carbon-coated 400 mesh copper grids (Electron Microscopy Sciences), washed, stained with 1% wild type/volume uranyl acetate in water, and dried. Samples were imaged on an FEI Tecnai Spirit TEM.

In Vitro B Cell Activation.

Naive, unimmunized mice were euthanized, and the spleens were harvested. Single-cell suspensions were prepared, and cells were incubated with 10 μM cell permeant Fluo-4 AM (Thermo Fisher) for 30 min at 37 °C. Cells were washed and stained for B220 (BD Bioscience, catalog number 553092). Flow cytometry was performed using a BD FACSCanto II, and B220+ B cells were gated and analyzed. Baseline readings were acquired for 30 s, after which the splenocytes were stimulated with unmodified Q11 nanofibers (2 mM), Q11 nanofibers (2 mM) containing β-tail–GFP (25 μM), Q11 nanofibers (2 mM) containing increasing amounts of β-tail–C3dg (12.5 to 50 μM), or ionomycin (2 μg/mL, a positive control to elicit maximal responses). The initial baseline geometric mean fluorescent intensity was subtracted from all subsequent data points to normalize all individual samples.

Mice and Immunizations.

All animal experiments were performed at Duke University under a protocol approved by the Duke Institutional Animal Care and Use Committee (protocol number A264-18-11). Female WT C57BL/6 mice were purchased from Harlan–Envigo laboratories, and female TNF-KO (JAX stock number 003008) (44) were purchased from Jackson Laboratories. Procedures were in compliance with the National Research Council's Guide for the Care and Use of Laboratory Animals, Eighth Edition. Mice 8 to 12 wk old were used for experiments. Mice were anesthetized and given two 50 μL subcutaneous injections near the shoulder blades for each primary and booster immunization. The OVA groups were boosted on weeks 3 and 6. The imiquimod-induced psoriasis and Listeria challenge groups received a single boost at week 4. The LPS challenge groups were boosted on weeks 4 and 10.


Plates were coated with 1 μg/mL streptavidin in PBS overnight at 4 °C, washed, and then coated with 20 μg/mL biotinylated peptide epitopes (OVA or TNF) in PBS. Alternatively, plates were coated with 20 μg/mL C3dg, murine C3, or C3b (Complement Technologies catalog number M113 and M114, respectively) overnight at 4 °C. Background wells were coated with the same solution lacking the target peptide or protein. Plates were washed with PBS containing 0.05% Tween 20 (PBST) and then blocked with PBST containing 1% bovine serum albumin (PBST-BSA) for 1 h at room temperature. Serum was serially diluted PBST-BSA in 10-fold steps from 1:102 to 1:106, applied to coated wells, and incubated for 2 h at room temperature. To detect total IgG, HRP-conjugated, Fc-γ fragment-specific goat anti-mouse IgG (Jackson Immunoresearch, catalog number 115-035-071) was used as the detection antibody. For antibody isotyping, HRP-conjugated IgG subtype-specific (i.e., IgG1, IgG2b, IgG2c, and IgG3) antibodies were utilized (Southern Biotech catalog number IgG1: 1071-05, IgG2b: 1091-05, IgG2c: 1078-05, IgG3: 1101-05). Plates were developed with TMB substrate and absorbance was measured at 450 nm with a SpectraMax M5 plate reader (Molecular Devices). The background absorbance was subtracted, and the titer was defined as the highest dilution that produced an absorbance signal higher than 0.2 (∼the mean + 3 × SD of naïve mouse serum). A titer of 1 indicates no measurable antibody titer.

C3 and C5a ELISAs.

The C3 enzyme-linked immunosorbent assays (ELISA) kit (ab157711) was purchased from Abcam, and the C5a ELISA kit was purchased from Invitrogen. The concentration of these proteins in the serum and intraperitoneal lavage were measured according to the manufacturers’ protocols.

IL-10 and TNF ELISAs.

An IL-10 ELISA kit (BMS614INST) and a TNF ELISA kit (BMS607) were purchased from Invitrogen. Cytokine concentration in the serum and intraperitoneal lavage were measured according to the manufacturer’s protocols.

CH50 Assay.

Serum from five mice was pooled for each immunization condition and serially diluted with gelatin veronal buffered saline containing 0.15 mM CaCl2 and 0.5 mM MgCl2 (GVB++) (Complement Technologies catalog number B102). Sensitized sheep erythrocytes (Complement Technologies catalog number B200) were suspended at 5 × 108 cells/mL in GVB++ and kept on ice. A total of 200 μL diluted serum was added to a series of 12 × 75 glass tubes. A total of 200 μL suspended sheep erythrocytes were added to each tube, and tubes were transferred to a 37 °C water bath for 60 min (cells resuspended every 10 min). Included were a tube with 200 μL GVB++ and cells as a 0% lysis negative control and 200 μL water and cells as a 100% lysis positive control. After 60 min, the tubes were removed from the bath and placed on ice. After 5 min on ice, the tubes were centrifuged for 3 min at 700 × g. Supernatant was transferred to a 96-well plate, and absorbance was measured at 451 nm with a SpectraMax M5 plate reader (Molecular Devices). The proportion of lysed cells was calculated by subtracting the negative control value from all other readings (including the 100% lysis value), then plotted data = the optical density of each assay tube/optical density of the 100% lysis tube.

T Cell ELISpot.

One week after final boost, the mice were euthanized, and the spleens were collected. Single-cell suspensions were prepared and plated at 0.5 × 106 cells per well (200 μL) in a precoated 96-well plate (Millipore). Cells were stimulated with TNF peptide (5 μM), PADRE peptide (1 μM), or β-tail–C3dg (1 μM). Cells were stimulated in a CO2 incubator at 37 °C for 48 h. To detect IL-4 and IFN-γ secreting cell spots, biotinylated anti-mouse IL-4 (BD, catalog number 551878) and IFN-γ (BD, catalog number 551881) detection antibody pairs, streptavidin-alkaline phosphatase (Mabtech, 3310-10) and Sigmafast BCIP/NBT substrate (Sigma, B5655) were used sequentially following the manufacturer’s general guidelines and concentrations (BD Biosciences). PBS was used as a negative control, and ConA (Sigma, C5275) was used as a positive control. Plates were imaged, and spots were counted by ZellNet Consulting using a Zeiss KS enzyme-linked immunospot (ELISpot) system.

Passive Transfer.

Serum was collected from unimmunized or β-tail–C3dg-immunized mice and pooled. A total of 100 μL serum was injected into a naïve, age-matched recipient mouse via tail vein injection. The next day, a small amount of serum was collected and analyzed via ELISA to ensure the mice exhibited expected levels of α-C3dg antibodies.

CD4 T Cell Depletion.

On days −3 and −1 prior to LPS Challenge, mice received 500 μg GK1.5 anti-CD4 monoclonal antibody (BioXcell BE0003-1) via intraperitoneal (i.p.) injection. Depletion was verified via flow cytometry (SI Appendix, Fig. S9).

Intraperitoneal LPS Challenge.

LPS challenge was conducted as reported previously (19). Briefly, mice were challenged with 1 mg/mL LPS (serotype 055:B5) from E. coli (Sigma) in sterile PBS intraperitoneally at a dose of 10 mg/kg 1 wk after the final booster immunization (week 10). The mice were monitored for 72 h, and temperature and body weight were recorded. Humane endpoints were set at the following: 1) >20% weight loss; 2) significant hypothermia (rectal temperatures below 32 °C); or 3) inability to ambulate, eat, and drink. Immediately before euthanasia, blood was collected for analyzing the serum anti-TNF peptide and anti-C3dg IgG concentrations.

Imiquimod Treatment and Assessment.

C57BL/6 mice receiving a therapeutic TNF-neutralizing antibody (BioXCell BP0058) were injected i.p. with 250 μg mAb on day −3 and day 0 of the imiquimod challenge. Mice at 13 to 17 wk of age received a daily topical dose of 0.125 g 5% imiquimod cream (6.25 mg imiquimod [Perrigo]) on the plucked back skin for five consecutive days. Mice were euthanized on day 6, and biopsies were taken for histological examination. Then, 10 μm cryosections were collected (10 sections from each mouse) and stained with Toluidine blue. Epidermal thickness was evaluated using ImageJ software, and the reported thicknesses are the average of eight measurements taken from two tissue sections. The spleens of the euthanized mice were removed and weighed. The experiments with the WT mice were performed twice by two different researchers with n = 5 mice. The researcher measuring the epidermal thickness was blinded to the groups.

Lm Challenge.

Mice receiving a therapeutic TNF-neutralizing antibody (BioXCell BP0058) prior to Lm challenge were employed as a positive control, and mice immunized with sterile PBS were included as a negative control. Mice receiving the TNF-neutralizing antibody were injected with a 500 μg dose intraperitoneally 3 h prior to challenge. Mice were challenged 1 wk after the last nanofiber immunization with 105 colony-forming units (CFU) of Lm (eGFP strain ATCC 35152) (200 μL) by i.p. injection. Spleens were harvested 48 h later and placed in sterile 0.05% Tween 20 in PBS. Tissues were homogenized with a Tissue Tearor handheld homogenizer (Biospec Products). The homogenized sample was serially diluted at 1:10 dilutions and 25 μL were plated on Brain Heart Infusion Agar (Sigma-Aldrich). The plates were incubated for up to 48 h in a 37 °C incubator. The Lm colonies were counted, and the total CFU/mg spleen tissue were calculated using the following formula:

Statistical Analysis.

Sample sizes and replicates are stated in the figure legends. Error bars represent means ± SD unless indicated otherwise, and data points represent individual mice. ANOVA with Tukey’s post hoc comparisons were employed as indicated in the figure legends, with the exception of the survival experiments, which were analyzed using log-rank tests. Blinding was performed on the imiquimod-induced psoriasis experiment (the researcher measuring the epidermal thicknesses was blinded to the group treatments). Analyses were performed using Graphpad Prism software.

Data Availability

All study data are included in the article and/or SI Appendix.


This research was supported by the NIH (NIBIB 5R01EB009701) and Duke University. K.M.H. was supported by NIH Training Grant T32GM008555. L.S.S., S.H.K., and C.N.F. were supported by the NSF Graduate Research Fellowship Program under Grant No. DGE-1644868. MALDI was performed on an instrument supported by the North Carolina Biotechnology Center, Grant 2017-IDG-1018. We thank Garrett Kelly for his help with protein expression and purification methods.

Supporting Information

Appendix (PDF)


D. T. Fearon, The complement system and adaptive immunity. Semin. Immunol. 10, 355–361 (1998).
P. W. Dempsey, M. E. Allison, S. Akkaraju, C. C. Goodnow, D. T. Fearon, C3d of complement as a molecular adjuvant: Bridging innate and acquired immunity. Science 271, 348–350 (1996).
A. S. De Groot et al., C3d adjuvant effects are mediated through the activation of C3d-specific autoreactive T cells. Immunol. Cell Biol. 93, 189–197 (2015).
T. D. Green, D. C. Montefiori, T. M. Ross, Enhancement of antibodies to the human immunodeficiency virus type 1 envelope by using the molecular adjuvant C3d. J. Virol. 77, 2046–2055 (2003).
T. M. Ross, Y. Xu, T. D. Green, D. C. Montefiori, H. L. Robinson, Enhanced avidity maturation of antibody to human immunodeficiency virus envelope: DNA vaccination with gp120-C3d fusion proteins. AIDS Res. Hum. Retroviruses 17, 829–835 (2001).
T. M. Ross, Y. Xu, R. A. Bright, H. L. Robinson, C3d enhancement of antibodies to hemagglutinin accelerates protection against influenza virus challenge. Nat. Immunol. 1, 127–131 (2000).
S. T. Test, J. Mitsuyoshi, C. C. Connolly, A. H. Lucas, Increased immunogenicity and induction of class switching by conjugation of complement C3d to pneumococcal serotype 14 capsular polysaccharide. Infect. Immun. 69, 3031–3040 (2001).
I. Watanabe et al., Protection against influenza virus infection by intranasal administration of C3d-fused hemagglutinin. Vaccine 21, 4532–4538 (2003).
X.-L. Wang, D.-J. Li, M.-M. Yuan, M. Yu, X.-Y. Yao, Enhancement of humoral immunity to the hCG β protein antigen by fusing a molecular adjuvant C3d3. J. Reprod. Immunol. 63, 97–110 (2004).
J. F. Bohnsack, N. R. Cooper, CR2 ligands modulate human B cell activation. J. Immunol. 141, 2569–2576 (1988).
R. H. Carter, D. T. Fearon, Polymeric C3dg primes human B lymphocytes for proliferation induced by anti-IgM. J. Immunol. 143, 1755–1760 (1989).
T. T. Hansel, H. Kropshofer, T. Singer, J. A. Mitchell, A. J. T. George, The safety and side effects of monoclonal antibodies. Nat. Rev. Drug Discov. 9, 325–338 (2010).
M. F. Bachmann, P. Whitehead, Active immunotherapy for chronic diseases. Vaccine 31, 1777–1784 (2013).
E. Assier et al., Modulation of anti-tumor necrosis factor alpha (TNF-α) antibody secretion in mice immunized with TNF-α kinoid. Clin. Vaccine Immunol. 19, 699–703 (2012).
B. Chackerian, D. R. Lowy, J. T. Schiller, Conjugation of a self-antigen to papillomavirus-like particles allows for efficient induction of protective autoantibodies. J. Clin. Invest. 108, 415–423 (2001).
K. M. Frietze, D. S. Peabody, B. Chackerian, Engineering virus-like particles as vaccine platforms. Curr. Opin. Virol. 18, 44–49 (2016).
A. Link, M. F. Bachmann, Immunodrugs: Breaking B- but not T-cell tolerance with therapeutic anticytokine vaccines. Immunotherapy 2, 561–574 (2010).
A. Bavoso et al., Aldehyde modification and alum coadjuvancy enhance anti-TNF-α autovaccination and mitigate arthritis in rat. J. Pept. Sci. 21, 400–407 (2015).
C. Mora-Solano et al., Active immunotherapy for TNF-mediated inflammation using self-assembled peptide nanofibers. Biomaterials 149, 1–11 (2017).
G. Spohn et al., A virus-like particle-based vaccine selectively targeting soluble TNF-alpha protects from arthritis without inducing reactivation of latent tuberculosis. J. Immunol. 178, 7450–7457 (2007).
J. S. Rudra, Y. F. Tian, J. P. Jung, J. H. Collier, A self-assembling peptide acting as an immune adjuvant. Proc. Natl. Acad. Sci. U.S.A. 107, 622–627 (2010).
J. S. Rudra et al., Modulating adaptive immune responses to peptide self-assemblies. ACS Nano 6, 1557–1564 (2012).
J. Chen et al., The use of self-adjuvanting nanofiber vaccines to elicit high-affinity B cell responses to peptide antigens without inflammation. Biomaterials 34, 8776–8785 (2013).
G. A. Hudalla et al., Gradated assembly of multiple proteins into supramolecular nanomaterials. Nat. Mater. 13, 829–836 (2014).
J. S. Rudra et al., Self-assembled peptide nanofibers raising durable antibody responses against a malaria epitope. Biomaterials 33, 6476–6484 (2012).
R. R. Pompano et al., Titrating T-cell epitopes within self-assembled vaccines optimizes CD4+ helper T cell and antibody outputs. Adv. Healthc. Mater. 3, 1898–1908 (2014).
D. T. Fearon, M. C. Carroll, Regulation of B lymphocyte responses to foreign and self-antigens by the CD19/CD21 complex. Annu. Rev. Immunol. 18, 393–422 (2000).
P. K. Mongini, M. A. Vilensky, P. F. Highet, J. K. Inman, The affinity threshold for human B cell activation via the antigen receptor complex is reduced upon co-ligation of the antigen receptor with CD21 (CR2). J Immunol. 159, 3782–3791 (1997).
V. Lehmann, M. A. Freudenberg, C. Galanos, Lethal toxicity of lipopolysaccharide and tumor necrosis factor in normal and D-galactosamine-treated mice. J. Exp. Med. 165, 657–663 (1987).
A. Bessede et al., Aryl hydrocarbon receptor control of a disease tolerance defence pathway. Nature 511, 184–190 (2014).
K. Tateda, T. Matsumoto, S. Miyazaki, K. Yamaguchi, Lipopolysaccharide-induced lethality and cytokine production in aged mice. Infect. Immun. 64, 769–774 (1996).
J. E. Juskewitch et al., LPS-induced murine systemic inflammation is driven by parenchymal cell activation and exclusively predicted by early MCP-1 plasma levels. Am. J. Pathol. 180, 32–40 (2012).
S. Li, V. M. Holers, S. A. Boackle, C. M. Blatteis, Modulation of mouse endotoxic fever by complement. Infect. Immun. 70, 2519–2525 (2002).
C. Giacomassi et al., Complement C3 exacerbates imiquimod-induced skin inflammation and psoriasiform dermatitis. J. Invest. Dermatol. 137, 760–763 (2017).
H. Vinter et al., Tumour necrosis factor-α plays a significant role in the Aldara-induced skin inflammation in mice. Br. J. Dermatol. 174, 1011–1021 (2016).
L. van der Fits et al., Imiquimod-induced psoriasis-like skin inflammation in mice is mediated via the IL-23/IL-17 axis. J. Immunol. 182, 5836–5845 (2009).
M. A. Lowes et al., Psoriasis vulgaris lesions contain discrete populations of Th1 and Th17 T cells. J. Invest. Dermatol. 128, 1207–1211 (2008).
A. Di Cesare, P. Di Meglio, F. O. Nestle, The IL-23/Th17 axis in the immunopathogenesis of psoriasis. J. Invest. Dermatol. 129, 1339–1350 (2009).
J. A. Bluestone, A. K. Abbas, Natural versus adaptive regulatory T cells. Nat. Rev. Immunol. 3, 253–257 (2003).
A. S. De Groot et al., Activation of natural regulatory T cells by IgG Fc-derived peptide “Tregitopes”. Blood 112, 3303–3311 (2008).
J. R. McDaniel, J. A. Mackay, F. G. Quiroz, A. Chilkoti, Recursive directional ligation by plasmid reconstruction allows rapid and seamless cloning of oligomeric genes. Biomacromolecules 11, 944–952 (2010).
G. A. Hudalla et al., A self-adjuvanting supramolecular vaccine carrying a folded protein antigen. Adv. Healthc. Mater. 2, 1114–1119 (2013).
J. Z. Gasiorowski, J. H. Collier, Directed intermixing in multicomponent self-assembling biomaterials. Biomacromolecules 12, 3549–3558 (2011).
M. Pasparakis, L. Alexopoulou, V. Episkopou, G. Kollias, Immune and inflammatory responses in TNF alpha-deficient mice: A critical requirement for TNF alpha in the formation of primary B cell follicles, follicular dendritic cell networks and germinal centers, and in the maturation of the humoral immune response. J. Exp. Med. 184, 1397–1411 (1996).

Information & Authors


Published in

Go to Proceedings of the National Academy of Sciences
Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 118 | No. 15
April 13, 2021
PubMed: 33876753


Data Availability

All study data are included in the article and/or SI Appendix.

Submission history

Published online: April 5, 2021
Published in issue: April 13, 2021


  1. vaccine
  2. self-assembly
  3. immunoengineering
  4. immune engineering
  5. active immunotherapy


This research was supported by the NIH (NIBIB 5R01EB009701) and Duke University. K.M.H. was supported by NIH Training Grant T32GM008555. L.S.S., S.H.K., and C.N.F. were supported by the NSF Graduate Research Fellowship Program under Grant No. DGE-1644868. MALDI was performed on an instrument supported by the North Carolina Biotechnology Center, Grant 2017-IDG-1018. We thank Garrett Kelly for his help with protein expression and purification methods.


This article is a PNAS Direct Submission.
See online for related content such as Commentaries.



Biomedical Engineering Department, Duke University, Durham, NC 27708
Lucas S. Shores
Biomedical Engineering Department, Duke University, Durham, NC 27708
Biomedical Engineering Department, Duke University, Durham, NC 27708
Zachary J. Bernstein
Biomedical Engineering Department, Duke University, Durham, NC 27708
Biomedical Engineering Department, Duke University, Durham, NC 27708
Chelsea N. Fries
Biomedical Engineering Department, Duke University, Durham, NC 27708
Marisha S. Madhira
Biomedical Engineering Department, Duke University, Durham, NC 27708
Caslin A. Gilroy
Biomedical Engineering Department, Duke University, Durham, NC 27708
Ashutosh Chilkoti
Biomedical Engineering Department, Duke University, Durham, NC 27708
Joel H. Collier1 [email protected]
Biomedical Engineering Department, Duke University, Durham, NC 27708


To whom correspondence may be addressed. Email: [email protected].
Author contributions: K.M.H., C.A.G., and J.H.C. designed research; K.M.H., L.S.S., N.L.V., Z.J.B., S.H.K., and M.S.M. performed research; A.C. contributed new reagents/analytic tools; K.M.H., L.S.S., and J.H.C. analyzed data; and K.M.H., C.N.F., and J.H.C. wrote the paper.

Competing Interests

Competing interest statement: J.H.C. is an inventor on a US patent describing the β-tail technology.

Metrics & Citations


Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.

Citation statements



If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Cited by


    View Options

    View options

    PDF format

    Download this article as a PDF file


    Get Access

    Login options

    Check if you have access through your login credentials or your institution to get full access on this article.

    Personal login Institutional Login

    Recommend to a librarian

    Recommend PNAS to a Librarian

    Purchase options

    Purchase this article to get full access to it.

    Single Article Purchase

    Modular complement assemblies for mitigating inflammatory conditions
    Proceedings of the National Academy of Sciences
    • Vol. 118
    • No. 15







    Share article link

    Share on social media