A next-generation human lymphatic filariasis vaccine candidate, rBmHAXT, for clinical development

Abstract

This study was conducted to develop a robust, scalable manufacturing process for the candidate vaccine rBmHAXT for human lymphatic filariasis (LF). During scale-up production, rBmHAXT showed significant antigen aggregation leading to a loss of purified vaccine antigens. This project aims to create an improved formulation suitable for industrial-scale production while maintaining robust protection. We generated three variants: (1) ∆Cys, in which all cysteinyl residues were mutated to serinyl. (2) GS, which has a flexible glycine-serine linker inserted between each of the component antigens, and (3) GS/∆Cys, a third variant with a combination of both the cysteine deletion and the addition of linkers. We then evaluated the immunogenicity and efficacy of each variant in a mouse model. We demonstrated that the ΔCys mutant retained immunogenicity and vaccine efficacy of the parent tag-free rBmHAXT protein. We also performed an accelerated stability study. All preparations remained stable at 4°C, and the ΔCys variant remained stable even at 25°C throughout the study (6 weeks). The ∆Cys protein was stable with equivalent potency in mice. Therefore, ∆Cys is an optimal candidate for progression to cGMP (Current Good Manufacturing Practices) manufacturing and human clinical trials as a vaccine for lymphatic filariasis.

Introduction

Lymphatic filariasis (LF) is a mosquito-transmitted tropical parasitic infection affecting over 81 million people in 51 countries worldwide, with an additional 882 million people in 44 countries at risk of acquiring the disease^1–3. There are currently no vaccines available to prevent or control this infection. Preventive chemotherapy consists of three medications currently administered through mass drug administration (MDA), the current approach to controlling this infection in endemic regions. The MDA approach reduced the disease burden in a few countries but has not eliminated the disease, despite 24 years of implementation⁴. Only partial disease elimination has been achieved due to a variety of issues, including patient non-compliance, shorter half-life of the drugs used in MDA, limited effects on adult worms, and the reemergence of the infection in several endemic regions^5–8. An effective vaccine complementing MDA would significantly benefit the eradication of the disease, especially since a distribution infrastructure has been established for MDA⁹.

In previous studies, we screened a phage-displayed cDNA expression library from infective-stage larvae of Wuchereria bancrofti using serum samples from putatively immune (Endemic Normal) individuals to identify several potential vaccine candidates for LF. After screening multiple antigens for vaccine efficacy, we narrowed the panel to the four antigens (HSP12.6, ALT-2, TPX-2, and TSP-LEL) that provided the highest protection. We combined the DNA sequence of the four proteins in tandem to create the BmHAXT gene and subsequently expressed the tetravalent fusion protein, rBmHAXT, in competent E. coli cells. We tested the vaccine potential of rBmHAXT protein in mice, gerbils, and non-human primates before deciding to initiate the product development. Since the LF parasite is a multicellular organism that employs multiple strategies to evade the immune system, we decided to combine several antigens to obtain better protection. Compared to a monovalent, bivalent, and a trivalent vaccine formulation, a tetravalent fusion of the four antigens gave over 85% protection in the mice model and over 57% protection in the non-human primate models. Since these worms do not multiply within the human host, a substantial decrease in worm establishment can significantly reduce infection-associated pathology, as demonstrated in the monkey trials¹⁰. Additionally, lower worm burdens and decreased fecundity of female worms were observed in both the jird model and the monkey model^10,11. This suggested that the tetravalent vaccine also has transmission-blocking potential that can synergize with MDA. Given the promising results from preclinical studies, we initiated scaled-up production of the tag-free recombinant vaccine antigen for industrial manufacturing under cGMP conditions. During this process, we found significant protein aggregation, especially following purification, even at refrigerated conditions (2–8 °C). To improve the recovery of less aggregated recombinant protein, we generated three variants of the antigen, as described in this report, by mutating all 17 cysteine residues to serine and/or inserting a linker (GS) between each component antigen in the tetravalent formulation. The reasoning behind mutating all cysteine residues was: (1) this will reduce the formation of disulfide bonds that could increase aggregation due - in part - to aberrant disulfide bond formation during the refolding process; (2) since rBmHAXT is a fusion of four independent protein sequences, the potential to form disulfide bonds between a cysteinyl residue in one protein to a cysteinyl residue in another protein due to close proximity during folding can be avoided; (3) similarly, aggregation of rBmHAXT may potentially suggest that the individual proteins are not able to achieve their native fold and therefore would reduce potency since antibodies may not recognize the natively folded protein in the parasitic worm; (4) the sulfur atoms in the cysteines can oxidize themselves leading to suboptimal stability; and finally (5) the cysteinyl residues are probably not critical for the generation of a protective immunity since the serinyl residues likely bind to the same MHC molecules in a similar manner and do not interfere with that binding through the formation of interchain disulfides^12,13. We tested each variant for its vaccine potency, stability, immunogenicity, and efficacy compared to the parent vaccine candidate. Our results showed that one variant, in which all cysteinyl residues were mutated to serinyl residues, had reduced aggregation, improved stability, and demonstrated greater potency as a vaccine candidate.

We also evaluated three different adjuvants; AL019 (a combination of Aluminum hydroxide and GLA) (AAHI, Seattle, WA), AlT4™ (Aluminum hydroxide and 3D-(6-acyl) PHAD™) (PAI Life Sciences Inc., Seattle, WA); and PAI-RI (a RIBI-like adjuvant composed of a squalane emulsion containing cell wall skeleton from Mycobacterium phlei, monophosphoryl lipid A (MPLA) from Salmonella minnesota R595, and trehalose 6,6′-dimycolate (TDM) from Mycobacterium bovis) (PAI Life Sciences Inc.), for their abilities to promote a robust immune response to the rBmHAXT (∆Cys) vaccine antigen.

Results

Expression levels of rBmHAXT variants

Following expression and purification, we analyzed the purity of the four proteins and observed that the cysteine-deleted (ΔCys) variants were significantly more pure, particularly with respect to the primary monomer band at 65 kDa (Fig. 1). The rBmHAXT (ΔCys) and rBmHAXT (GS/ΔCys) also seemed to be significantly improved with respect to degradation bands in the 16-50 kDa region. The wild-type protein exhibited the greatest amount of initial aggregation and the greatest number of bands below 65 kDa. Expression levels of the variants containing the glycine-serine linkers were consistently lower than levels observed with rBmHAXT (ΔCys).

Fig. 1. Expression and purity of rBmHAXT mutants.

Approximately 2 µg of each of the rBmHAXT variants, wild-type antigen (Lane 1), rBmHAXT (GS) (Lane 2), rBmHAXT (ΔCys) (Lane 3), and rBmHAXT (GS/ΔCys) (Lane 4) were reduced and separated on a Novex™ Wedge Well™ 4–20% Tris-glycine SDS PAGE gel and stained with Simply Blue™ Safe stain. SeeBlue PLUS2 molecular weight markers were used to determine the molecular mass. A major band at 60 kDa (monomer) was observed in all samples. However, the wild-type protein also showed several bands in the 140–250 kDa range (likely aggregates) and several fragments in the 10–50 kDa range. Comparatively, rBmHAXT (ΔCys) exhibited fewer aggregates and fragments compared to all other mutants.

ΔCys variant of rBmHAXT is more stable

We performed accelerated stability studies at five (5) temperatures for 6 weeks. For brevity and clarity, in Fig. 2, we show the percentage of monomer remaining in the solution after incubating each of the variants at different temperatures (−80 °C, −20 °C, 4 °C, 25 °C, and 42 °C) on days (0, 3, 7, 21, and 42). Complete gel images and the calculations are provided as supplementary data (Supplementary Fig. 1, Supplementary Table 1, and Supplementary Table 2). After one week, the wild-type antigen as well as the rBmHAXT (GS) were nearly completely aggregated at 42 °C, whereas more than half of rBmHAXT (ΔCys) remained as monomers. At 3 weeks, both rBmHAXT and rBmHAXT (ΔCys) were fully aggregated at 42 °C. At 6 weeks, over 90% of the rBmHAXT (ΔCys) protein remained intact at 25 °C. The addition of a GS linker did not appear to have the intended effect of promoting protein folding and decreasing aggregation. Instead, it seemed to have the opposite effect, resulting in greater aggregation, which was evident after 3 days and more pronounced at 25 and 42 °C (Fig. 2). Moreover, the aggregates appeared to be slightly larger (~300–400 kDa) than those observed for rBmHAXT (ΔCys).

Fig. 2. Stability of rBmHAXT mutants at various temperatures.

Single-use aliquots of each of the rBmHAXT mutants, wild-type antigen (w/t), rBmHAXT (GS), rBmHAXT (ΔCys), and rBmHAXT (GS/ΔCys) were stored at −80 °C, −20 °C, 4 °C, 25 °C, and 42 °C for 6 weeks (completion of the experiment). On days 0, 3, 7, 21, and 42, approximately 2 µg of each sample was resolved on a reduced 4–20% Tris-glycine gel and stained with Simply Blue™ Safe stain. SeeBlue PLUS2 molecular weight markers were used to determine the molecular mass. The figure shows the percentage of monomers remaining in each tube at the indicated temperatures and time points. Our results show that the percentage of monomers in the wild-type protein decreased with time and increased temperature. On day 42 at 25 °C, more than 50% of the monomers in wild-type antigen and the rBmHAXT (GS) were lost, whereas nearly all rBmHAXT (ΔCys) and rBmHAXT (GS/ΔCys) monomers were intact at 25 °C.

All the variants of rBmHAXT induced significant antibody titers

The results showed that compared to control mice that were given adjuvant alone, all mice vaccinated with wild-type antigen or rBmHAXT (∆Cys) or rBmHAXT (GS), or rBmHAXT (GS/∆Cys) showed significantly (p < 0.005) higher titers of rBmHAXT-specific IgG antibodies compared to controls in the sera of mice (Fig. 3) confirming that ∆Cys or GS or the combined modification did not significantly alter the immunogenicity of the wild-type vaccine antigen. These studies also suggested that the immune epitopes of the wild-type vaccine antigen were not altered despite the cysteine mutation or the addition of the GS linker sequence.

Fig. 3. Titers of antigen-specific IgG levels after three immunizations with rBmHAXT variants.

Titers of antigen-specific IgG were evaluated in serum samples of mice immunized three times with the respective antigens using a capture ELISA. Samples were analyzed in duplicate wells. Our results show that the titers of antigen-specific IgG antibodies were significantly higher in all vaccinated groups compared to the adjuvant control group (1:20,000). The rBmHAXT (ΔCys) group had a slight but significant increase in IgG titers at 1:100, 1:1000 and at 1:2000 compared to the other vaccinated groups. Statistical significance between groups was determined by two-way ANOVA with Tukey’s multiple comparisons test. Each data point represents mean ± SD value. N = 5 mice per group.

In our ELISAs, we coated wells with the parent wild-type rBmHAXT antigen. Significant binding of the rBmHAXT (ΔCys)-specific IgG to the parent his-tagged rBmHAXT suggests that the rBmHAXT (ΔCys) retains the immunogenic epitopes of the parent antigen. At 1:5,000 dilutions, IgG titers were significantly (p < 0.05) higher in the sera of rBmHAXT (∆Cys) immunized mice than in those of wild-type vaccine antigen immunized mice. IgG antibodies were detectable even at 1:20,000 dilutions of the serum samples in both wild-type vaccine antigen and rBmHAXT (∆Cys) immunized mice. These findings confirm that the redesigned rBmHAXT (∆Cys) retains immunogenicity relative to the wild-type protein.

Antibody isotype analysis also showed that the levels of IgG1, IgG2a, IgG2b, and IgG3 isotypes were elevated in the sera of all immunized mice (Fig. 4) and levels were highly significant (****p < 0.0001). However, IgE antibody levels were at background levels, indicating that none of the variants induced an IgE response in mice. In summary, these studies showed that the variant proteins had comparable immunogenicity to the original construct as measured by total IgG or IgG subclasses.

Fig. 4. Levels of antigen-specific IgG antibody isotypes and IgE antibodies in immunized mice.

Antibody isotype profiling of vaccinated mice revealed that levels of all IgG isotypes were significantly elevated (****p < 0.0001) across all groups compared with the adjuvant control group. However, no significant differences (NS) in antibody isotype levels were observed between the immunized groups. IgE levels were the same as background readings in all immunized mice. Statistical significance was determined by two-way ANOVA with Tukey’s multiple comparison test. Each bar represents the mean ± SD value. N = 5 mice per group.

Antigen-specific splenocytes from vaccinated animals secreted high levels of IL-17A and IFN-γ

Cytokine levels in the culture supernatants of spleen cells were determined using a cytokine bead array. Our results showed that, when comparing wild-type vaccine antigen+AL019 and rBmHAXT + AlT4™, a similar pattern of cytokine responses (IL-17A, IFN-γ, IL-10) was observed (Fig. 5).

Fig. 5. Secreted levels of cytokines in the culture supernatants of rBmHAXT-specific splenocytes.

Spleen cells collected from immunized mice were stimulated with 5 μg of wild-type vaccine antigen for 72 h, and culture supernatants were assayed in a flow cytometer using a cytokine bead array kit from BD BioSciences. Values shown are fold increases over the values from control mice that received only the adjuvant. The secreted levels of IL-17A, IL-2, IL-10, and IFN-γ in wild-type vaccine antigen plus AL019 and wild-type vaccine antigen plus AlT4™ groups showed similar patterns of increases. However, the cells from rBmHAXT (ΔCys) plus AlT4™ immunized animals secreted nearly 150-fold higher levels of IL-17A and about 50-fold higher levels of IFNγ, suggesting that a potent Th17/Th1 response is induced by rBmHAXT (ΔCys) plus AlT4™ immunization. Statistical significance was determined by two-way ANOVA with Tukey’s multiple comparisons test. Each bar represents the mean ± SD value. *p < 0.001. N = 5 mice per group.

Compared to the AL019 adjuvant, the AlT4™ adjuvant appears to further boost the levels of these secreted cytokines, potentially suggesting that AlT4™ may have a slightly better adjuvanting effect for rBmHAXT compared to AL019. When the cytokine response to rBmHAXT plus AlT4™ was compared with rBmHAXT (∆Cys) plus AlT4™, IL-17A and IFN-γ responses were substantially higher in rBmHAXT (∆Cys) plus AlT4™ compared to wild-type vaccine antigen plus AlT4™ suggesting that the ∆Cys mutant promoted better Th17/Th1 responses following vaccination, consistent with the lower IgG1 levels seen with this combination (Figs. 4 and 5). The IL-10 response was slightly lower, but several-fold higher than the controls. Spleen cells from the other mutants secreted very low levels of the cytokines tested. These findings show that AlT4™ is a more efficient adjuvant for rBmHAXT and, in combination with rBmHAXT (∆Cys), AlT4™ promoted a robust cellular response.

T_CM cells were generated in the spleen of mice vaccinated with rBmHAXT or its ΔCys mutant protein

To evaluate if the vaccination generated memory T cells, we cultured 1 × 10⁶ splenocytes from each mouse with 5 µg/ml of rBmHAXT protein for 72 h at 37 °C. Central memory T cells express CD62L and CCR7 receptors. CD62L or L-selectin is a marker that distinguishes central memory (T_CM, CD62L+) from effector memory (T_EM, CD62L⁻) T cells. After incubation, cells were harvested and stained with CD3/CD62L/CCR7 antibodies and analyzed on a Becton Dickinson FACSMelody™ flow cytometer. After gating for T (CD3+) cells, cells that were dual-positive for CD62L and CCR7 were counted as T_CM cells. Our results showed that mice vaccinated with wild-type vaccine antigen or rBmHAXT (∆Cys) generated a significant percentage of T_CM cells irrespective of the adjuvants (AL019 or AlT4™) used (p≤0.0001) (Fig. 6).

Fig. 6. Percentage increase in rBmHAXT-specific T central memory (T_CM) cells in the spleen of immunized animals.

Animals immunized with parent protein or ΔCys protein showed significantly higher increases in the percent of rBmHAXT-specific T_CM cell population in their spleens. However, animals immunized with rBmHAXT(GS), rBmHAXT(GS/ΔCys), or rBmHAXT(ΔCys) plus RI group had only very small increases in the rBmHAXT-specific T_CM cells in their spleens. Each bar represents the mean ± SD value. N = 5 mice per group.

Vaccine-induced protection

We then evaluated the protective potential of the vaccine antigens using a challenge model. Consistent with the immunogenicity studies, challenge studies in mice showed that the protection rate was comparable or even higher with rBmHAXT (ΔCys) than with the parent wild-type vaccine antigen given with AL019 (90.7%) or with AlT4™ (88%) (Fig. 7). Mice immunized with rBmHAXT (ΔCys) plus AlT4™ showed 97% protection compared to adjuvant control (8%). The rBmHAXT (GS) group (73%), the rBmHAXT (GS/ΔCys) group (49%), and the rBmHAXT (ΔCys) plus PAI-RI group (41%) showed low protection rates. These findings suggested that immunization with rBmHAXT (ΔCys) plus AlT4™ adjuvant had greater vaccine efficacy than the parent wild-type vaccine antigen. This study reinforced that the cysteine mutation did not affect the vaccine efficacy of rBmHAXT.

Fig. 7. Immunization with rBmHAXT and its variants conferred significant levels of protection in vaccinated animals.

Vaccine-induced protection was determined by calculating the percentage of larval death as a measure of protection in immunized animals. Following three immunizations, the percentage of protection in rBmHAXT(ΔCys) plus AlT4™ and wild-type vaccine antigen plus AL019 was significantly higher (***p < 0.0001) compared to the adjuvant group. The vaccine-induced protection was also significantly (*p < 0.05) higher in the rBmHAXT(GS) plus AlT4™ group and in the wild-type plus AlT4™ group (**p < 0.005) compared to the control group. However, the level of protection in these groups was significantly lower compared to the rBmHAXT(ΔCys) and rBmHAXT vaccinated animals. Statistical significance was determined by one-way ANOVA. N = 5 mice per group.

Discussion

During scale-up in the product development process, we observed significant aggregation and fragmentation of our original protein antigen following purification. To address this, we generated three mutants of rBmHAXT: rBmHAXT ΔCys, rBmHAXT GS, and rBmHAXT GS/ΔCys. With the ΔCys variant, we tested our hypothesis that rBmHAXT, by virtue of its non-natural fusion of 4 protein domains, has a propensity to aggregate. This is likely because adjacent cysteinyl residues form disulfide bonds between separate protein chains rather than the native disulfide bonds required for proper folding. We therefore replaced all cysteinyl residues with serinyl residues. Our second hypothesis was that the direct fusion of the four proteins in the wild-type protein hampers the freedom and space for each subunit to properly fold, and this misfolded state would promote aggregation. Therefore, GS variant proteins containing a 15-amino-acid flexible glycine-serine linker were also produced to allow the chains to fold independently of one another. We produced all four proteins and analyzed them for expression levels, stability, and resistance to aggregation using a 5-temperature, 6-week accelerated stability program. We tested the potency, immunogenicity, and vaccine efficacy of each variant to select the best rBmHAXT candidate for industrial manufacturing and clinical studies. Our results show that among the three variants tested, rBmHAXT (ΔCys) is superior to the parental (his-tagged or tag-free) wild-type vaccine antigen in terms of immunogenicity and vaccine efficacy in a mouse model.

Recombinant proteins - especially cysteine-rich proteins - tend to form protein aggregates leading to the formation of inclusion bodies (IBs) during translation in E. coli^12,14. Formation of IBs can be both problematic and beneficial for the scale-up and manufacture of a recombinant parasite protein^14–16. One problem is that the IBs must be solubilized in a chaotrope such as urea and then refolded ex vivo to form active protein monomers if an enzymatically active protein is desired. However, IBs are heavier than solutes in lysed bacteria, and a significant benefit of IB formation is that, once the bacteria are lysed, the IB fraction can be purified separately from the soluble fraction, yielding an initially semi-pure recombinant protein. In the context of rBmHAXT, our process leverages and, in fact, promotes the formation of IBs. The goals of our study were to produce a high-purity antigen that primarily forms stable monomers, remains non-aggregated for extended periods, and elicits a robust, protective vaccine response. We are not looking for enzymatic activity or a properly folded, soluble protein, because rBmHAXT is a synthetically engineered, non-native protein composed of 4 diverse domains. Wildtype vaccine antigen tended to aggregate significantly, which we hypothesize was due to the 17 cysteinyl residues forming non-natural disulfide bonds, that is, cysteines in one subdomain forming a disulfide bond with cysteines in a second domain. We also hypothesized that the non-natural nature of the protein made it difficult to fold in vivo in E. coli, especially when large amounts are being translated during IPTG induction. Slight misfolding during this process can lead to aggregation and the formation of IBs. Thus, we hypothesized that a flexible linker would help reduce protein aggregation. However, contrary to our hypothesis, the introduction of flexible glycine-serine linkers did not improve folding and instead led to increased aggregation and lower protein yields. This contrasted with our hypothesis that, by increasing the distance between the component proteins, the subdomains would have the space and freedom to fold independently of each other, leading to decreased aggregation and increased expression in the soluble fraction. Similarly, the combination of the GS linkers and the cysteine mutations was also not as beneficial as the cysteine modification alone. While it was surprising that isolating the domains using a flexible GS linker did not promote better immune responses to the antigens, this may be because hydrophobic areas may have had an easier time pairing across domains, which enhanced misfolding rather than encouraging more native-like folds. We demonstrated that we could purify the recombinant monomer protein to >95% purity. Thus, these findings clearly suggest the advantage of mutating all cysteine residues to prevent protein aggregation and enable more efficient purification.

The vaccine is intended for subjects living in tropical countries where LF infection is highly prevalent. Therefore, the vaccine protein needs to remain stable in the event of interrupted cold storage and preferably at higher temperatures during transportation and storage. Vaccines can become unstable during storage, and this instability can reduce the safety and efficacy of the vaccines being deployed in certain areas^15,16. The Expert Committee for Biological Standardization provides guidance on thermal stability and shelf life of vaccines during manufacturing^17,18. Based on these guidelines and to advance the vaccine to clinical use, we tested the stability and shelf life of our recombinant vaccine antigens in this study. Our results show that the rBmHAXT ΔCys monomer remained stable, even at 25 °C, throughout the 42-day accelerated stability study (Fig. 2), whereas all other formulations began to substantially lose the monomer by day 7 at 25 °C, mostly due to aggregation. These data suggest that the improved vaccine is more stable across a broader temperature range, which is advantageous for transportation and storage.

Following any improvement to the vaccine antigen, the formulation needs to be tested to determine whether the vaccine retains the immunogenicity and vaccine efficacy of the original formulation^19,20. The mouse is a good model to evaluate the immunogenicity and efficacy of vaccine candidates against lymphatic filariasis^21–24. When we tested the immunogenicity of all three mutants in a mouse model, we found that they were highly immunogenic. The rBmHAXT ΔCys elicited a robust IgG response comparable to that of the parent wild-type vaccine antigen, and the IgG isotype responses were also similar to those of the parent wild-type vaccine antigen, confirming that the new mutants are similarly immunogenic to the original formulation. Our studies also demonstrated that IgG antibodies generated against each mutant cross-react with the parent his-tagged wild-type vaccine antigen, as we used the parent antigen to coat the wells in our ELISAs (Fig. 3), suggesting that the key immunogenic epitopes are not altered by the mutations or sequence alterations. This implied that we could use any of the variants as a vaccine candidate to replace the parent formulation.

It was critical to determine if any or all the mutants could confer protective immune responses against LF. The preclinical studies reported here confirmed that all mutants elicited protective immune responses at varying levels. However, the protective immune responses generated following vaccination with rBmHAXT ΔCys were higher than those with the other mutants and even the original unmodified wild-type vaccine antigen. This suggested that the cysteine mutations had little influence on the immunoprotective epitopes of the wild-type vaccine antigen. This also suggested that the cysteinyl residues are not critical for vaccine-induced immunity or that the serinyl residues, due to their chemical similarity, are suitable mimics of the native cysteines. Analysis of splenocytes from immunized mice reveals that all immunized mice generated T central memory cells. Deletion of cysteine residues in rBmHAXT did not interfere with the generation of memory T cells in vaccinated animals (Fig. 6).

Several previous studies show that both antibodies and cellular responses are critical for host immunity against lymphatic filariasis^24–30. Similarly, our previous studies also showed that when infective larvae of B. malayi were incubated with immune sera and PBMC or peritoneal cells from mouse³¹, gerbil¹¹, rhesus macaque^10,32 or humans³³, numerous macrophages and lymphocytes bound to the larval surface leading to the larval death within 24–48 h⁹. If cells alone, immune serum alone, or IgG-depleted immune human serum were used, this killing does not happen³³. This suggests that both IgG and cells (mainly macrophages) are critical for killing the larvae.

There was also a significant increase in the percentage of T central memory (T_CM) cells in the spleen of all rBmHAXT immunized animals that received AL019 or AlT4™ adjuvant. C-C receptor 7 (CCR7) is a chemokine receptor that mediates the homing of central memory T cells to the spleen and lymph nodes. CCR7 is absent in effector memory T cells. The central memory cells circulate through the lymphatic system and can self-renew, providing significant immunity against pathogens. This robust T_CM response, along with the high level of protection achieved, indicates the induction of long-term immunity following rBmHAXT vaccination.

Adjuvants play a critical role in eliciting and skewing immune responses in favor of maximum protection^34–36. In this study, we also tested three adjuvant formulations to determine which provides the best protective response. While the TLR4 agonist-on alum adjuvants Al019 and AlT4™ are relatively similar and activate inflammatory responses via alum and also activate TLR4 with the respective agonists, the other adjuvant, PAI-RI, is a potent combination adjuvant with multiple mechanisms of action. The two TLR4-on-Alum formulations gave significantly better results than the PAI-RI adjuvant. This could be because PAI-RI induces an excessive response, leading to immune downregulation, consistent with the lower responses observed in our assays. The new formulation developed by our group, AlT4™, yielded the best results, suggesting significant potential as an adjuvant for human vaccines against lymphatic filariasis.

In conclusion, our studies demonstrate that mutating cysteine residues in our vaccine protein can significantly improve the quality of the vaccine antigen. The cysteine mutation did not affect protein expression; in fact, it improved the recovery of the purified antigen. Our studies also showed that the ∆Cys mutation significantly improved the stability of the vaccine antigen—including when subjected to higher temperatures. Similarly, vaccine efficacy studies in a mouse model showed that mutation of all cysteine residues in rBmHAXT ∆Cys did not affect the immunogenicity, potency, or vaccine efficacy; if anything, these changes boosted responses. Significantly, we also showed that a next-generation TLR4-on-alum adjuvant, AlT4™, elicited the best responses in our test systems. Overall, our results demonstrate that we now have an optimal version of our LFGuard™ candidate and are poised for cGMP production and human testing of this important vaccine.

Methods

Design and cloning of rBmHAXT (tag-free)

To ensure traceability from clone development throughout process development, all cloning, expression, and purification reagents were carefully sourced and documented to ensure the absence of animal-derived products used in the process. Tag-free rBmHAXT was prepared as described by Melendez et al.³⁷.

Design and cloning of rBmHAXT (ΔCys)

To address the aggregation and degradation issues, we designed rBmHAXT (ΔCys) by replacing all cysteine codons in the BmHAXT sequence with serine codons. There is a total of 17 cysteinyl residues in rBmHAXT interspersed amongst the four component protein sequences. For the mutagenesis, cysteine codons with the nucleotide sequence TGC were mutated to AGC serine codons by changing the thymine base in the first position to an adenine base while cysteine codons with the nucleotide sequence TGT were mutated to TCT serine codons by changing the guanine base in the second position to a cytosine base (Supplementary Table S3).

The redesigned rBmHAXT (ΔCys) gene (Accession # PZ091447) was manufactured as a GeneBlock™ gene fragment from Integrated DNA Technologies (IDT, Coralville, IA). The rBmHAXT (ΔCys) was cloned into the pET29a(+) expression vector (Millipore Sigma, Burlington, MA). Plasmids were transformed into E. coli Turbo cells (NEB, Ipswich, MA) and the transformants were selected on LB plates supplemented with 50 ug/mL of kanamycin sulfate (LB-Kan). The transformants were screened by PCR for correctly sized inserts, and the presumptive positive sequences were confirmed by Sanger sequencing. Plasmids from a single confirmed clone were transformed into the commercial E. coli expression strain HMS174 (DE3) and transformants selected on LB-Kan plates. Resulting colonies were screened for expression of a ~60 kDa band corresponding to the predicted size of rBmHAXT (ΔCys).

Preparation of additional variants of rBmHAXT

Two other variant proteins, termed rBmHAXT (GS) (Accession # PZ091448) and rBmHAXT (GS/ΔCys) (Accession # PZ091449), were designed as possible next-generation candidates (Supplementary Data 1). Both rBmHAXT (GS) and rBmHAXT (GS/ΔCys) include three 12-amino acid flexible glycine-serine (GS) linkers (Gly-Gly-Gly-Ser-Gly-Gly-Gly-Ser-Gly-Gly-Gly-Ser) inserted between each of the four protein sequences in the fusion. For rBmHAXT (GS/ΔCys), we mutated all 17 cysteine residues in rBmHAXT (GS) to serine residues (Supplementary Table S3). The only difference was in the numbering for each cysteine residue, as it differed slightly for rBmHAXT (GS/ΔCys) due to the insertion of the glycine-serine linkers. Like rBmHAXT (ΔCys), both rBmHAXT (GS) and rBmHAXT (GS/ΔCys) were designed in silico, produced as GeneBlocks, and cloned and purified as described below.

Process development

The process for fermentation and purification of rBmHAXT (ΔCys) was the same as described by Melendez et al.³⁷, with two modifications: first, we did not use dithiothreitol in the purification of rBmHAXT (ΔCys) since we had deleted all the cysteine residues in this protein and therefore would not need a reducing agent for disulfides. The second modification was to reduce the washing stringency (NaCl concentration reduced from 180 to 160 mM) during the purification washing steps. This step was taken because some rBmHAXT ΔCys protein was being washed away; reducing the salt concentration improved final yields without decreasing overall purity.

E. coli cell pellets were thawed and resuspended in 5 mL lysis buffer (50 mM Tris and 0.5% Triton X-100, pH 8.0) per gram of wet cell paste and mixed by vortexing and disruption by pipetting until no visible clumps were observed. The suspension was passed through an LM10 microfluidizer (Microfluidics Corp., Westwood, MA) 3 times at 15,000 psi, with intermittent cooling between passes. The pelleted IB fraction was solubilized in 20 mL of solubilization buffer [50 mM tris, 8 M urea pH 8.0] per gram of IB and rolled gently at 4 °C for 16–20 h. The solubilized crude IB solution was clarified by centrifugation at 15,000 × g for 3 h at 4 °C. The supernatant containing the solubilized rBmHAXT (ΔCys) was decanted to a fresh container and stored at −80 °C until purification.

Purification and diafiltration of recombinant proteins

Capto Q ImpRes ion exchange chromatography was used to purify rBmHAXT (ΔCys) as described previously³⁷. Triplicate 1 µg loads of rBmHAXT (ΔCys) and BSA standards were analyzed by reducing SDS-PAGE and quantified by ImageJ densitometry analysis to confirm both concentration and purity. Identity was confirmed by Western blot analysis with monkey anti-rBmHAXT antisera¹⁰ at 1/10,000 and detected with a 1/10,000 dilution of HRP-conjugated Goat anti-Monkey IgG (H + L) secondary antibody (Thermo Fisher Scientific, Rockford, IL.). Residual E. coli host cell proteins were detected by Western analysis using a 1/1000 dilution of Rabbit anti-E. coli Host Cell Protein (HCP) polyclonal antibody (Rockland Immunochemicals, Inc., Limerick, PA) with a 1/2000 dilution of HRP-conjugated Donkey anti-Rabbit IgG (H + L) secondary antibody (Southern Biotech, Birmingham, AL). Residual endotoxin was measured using the Limulus amebocyte lysate (LAL) assay and read using the Endosafe® Nextgen PTS Reader (Charles River Laboratories, Worcester, MA).

Stability of rBmHAXT protein and its variants

A short-term stability study was performed to compare the tag-free rBmHAXT with the three variants: rBmHAXT (ΔCys), rBmHAXT (GS), and rBmHAXT (GS/ΔCys). The stability study was performed by diluting each of the two proteins to 0.5 mg/mL in 20 mM Tris pH 8.0. Approximately 40 aliquots of 0.1 mL of each protein were placed at the five indicated temperatures. Protein stored at −80 °C in a Revco Ultra Low freezer was used as a control, as there should be no change in aggregation or degradation at this temperature. Storage at −20 °C was in a standard freezer. Storage at 4–8 °C was used to simulate a typical refrigerator. The 25 °C temperature simulates typical room-temperature conditions. Finally, 42 °C was chosen to promote forced (accelerated) degradation and to stress-test the proteins at a higher temperature. At each of the seven time points, one aliquot of each of the proteins was removed from each of the five storage conditions, and 1 µg total protein was resolved on a reducing SDS-PAGE gel and stained with SimplyBlue Safe Stain (Thermo Fisher Scientific).

Cell banking

A research cell bank of 162 vials of E. coli HMS174 (DE3) strain containing the plasmid rBmHAXT ΔCys clone 6B in vector pET29a was laid down. The clone was grown in animal product-free LB broth supplemented with 50 μg/mL of kanamycin sulfate (LBK) to an OD600 of ~1.0. After confirming the culture purity by microscopic observation and growth on both selective and non-selective media, the culture was mixed 1:1 with sterile LBK broth containing 20% plant-derived glycerol. The cell bank was stored at −80 °C. Confirmation of expression before and after cell banking was performed, and the expressed protein was confirmed to be localized to the insoluble fraction (IBs) as expected. Using the newly produced research cell bank (RCB), a 200-vial master cell bank was prepared at the University of Nebraska Biological Process Development Facility (BPDF) under conditions suitable for future cGMP manufacturing. We confirmed the expression of rBmHAXT ΔCys protein from the newly created MCB (data not shown).

Adjuvants

In this study, we used three different adjuvants: (i) Aluminum hydroxide combined with GLA (AL019) obtained from the Access to Advanced Health Institute (AAHI, Seattle, WA); (ii) Aluminum hydroxide combined with a TLR4 agonist (AlT4™) was prepared by PAI Life Sciences (Seattle, WA); as was (iii) a potent adjuvant based on the original RIBI adjuvant system³⁸ termed PAI-RI. Both AL019 and AlT4™ are produced by mixing the TLR4 agonist with a detergent and adsorbing the mixture onto aluminum hydroxide (alum) particles. These alum particles consist of nanometer crystals that assemble into aggregates of several micrometers and provide a stable particulate formulation for the sustained release of the adjuvant. In vivo data indicate that TLR4-alum directs the quality of the immune response to a balanced Th1/Th2 response. AlT4™ differs from AL019 in that the TLR4 agonist has no acyl chain on the 3 position, making it more similar to the naturally derived TLR4 agonist 3D-MPL from GSK plc. PAI-RI is a potent adjuvant composed of a mixture of cell wall skeleton, a TLR4 agonist, a squalane emulsion adjuvant, and the mincle agonist trehalose 6,6′-dimycolate (TDM).

The alum-adsorbed TLR4 agonist is produced in a two-step process. First, a micellar suspension of the TLR4 agonist is produced (MiT4^TM), which is also used as a TLR4-only control. This suspension is then adsorbed to the surface of Alhydrogel™ by mixing. 3D-(6-acyl) PHAD (Croda/Avanti Polar Lipids, Yorkshire, UK) is added to a flask, then combined with 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) and dissolved in chloroform. Chloroform is then evaporated, water for injection is added, and the mixture is placed in a sonicating bath at 60 °C. Periodically, samples are taken and analyzed by Dynamic Light Scattering using a Malvern Zetasizer™. Once a suitable particle size averaging approximately 120 nm is obtained, the MiT4™ is sterile filtered and aliquoted. Next, Alhydrogel™ (Croda, Yorkshire, UK) is washed in water and sedimented by centrifugation at 3,500 rpm for 25 min at 4 °C. MiT4™ from the previous step is then added, and the mixture is vortexed at room temperature for 25 min until thoroughly combined. Once combined, the AlT4™ adjuvant is aliquoted into single-use vials and capped for storage at 2–8 °C.

The final adjuvant, PAI-RI, was originally produced as a replacement for RIBI as this adjuvant is no longer commercially available. PAI Life Sciences prepared the custom RIBI-like adjuvant composed of a squalane emulsion containing cell wall skeleton from Mycobacterium phlei, monophosphoryl lipid A (MPLA) from Salmonella minnesota R595, and TDM from Mycobacterium bovis³⁸. PAI-RI has recently shown significant promise in preclinical studies for a tri-antigen syphilis vaccine³⁹.

Animals and parasites

Eight-week-old male BALB/c mice were purchased from Taconic Biosciences (Hudson, NY, USA) and quarantined for 2 weeks before the start of the experiments. The infective larval stage (L3) of B. malayi was purchased from the TRS Laboratories (Athens, GA).

Ethics statement

Use of animals in this study was approved by the institutional animal care and use committee (IACUC) of the University of Illinois, Rockford (Approval # 2323915-2, dated 6/20/2025). Euthanasia of animals was performed with carbon dioxide gas followed by cervical dislocation as per the recommendation of the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Vaccination protocol

Thirty-five (35) mice were divided into seven (7) groups of five mice each. Group 1 mice received 10 µg of the TLR4 on Alum (AlT4™) adjuvant in 100 µl of PBS. Group 2 mice received 25 µg of rBmHAXT plus 10 µg of AL019 adjuvant; Group 3 mice received 25 µg of rBmHAXT plus 10 µg of AlT4™ adjuvant; Group 4 mice received 25 µg of rBmHAXT (∆Cys) plus 10 µg of AlT4™ adjuvant; Group 5 mice received 25 µg of rBmHAXT (GS) plus 10 µg of AlT4™ adjuvant; Group 6 mice received 25 µg of rBmHAXT GS/ΔCys) plus 10 µg of AlT4™ adjuvant; and Group 7 mice received 25 µg of rBmHAXT (∆Cys) plus 10 µg of PAI-RI adjuvant. All injections were given subcutaneously (SC) into the right flank region of each mouse on days 0, 14, and 28.

Collection of blood

Approximately 100 µL of whole blood was collected from the submandibular vein of each mouse on day 0 (pre-immune), day 14 (before first booster), day 28 (before second booster), and on day 48 (before challenge). Mice were anesthetized with a ketamine/xylazine formulation (0–100 mg/kg/5–10 mg/kg) before collecting the blood. Serum samples were prepared and stored at −80 °C for serological analysis.

Titer of antigen-specific IgG

The titers of rBmHAXT-specific IgG in serum samples were evaluated using an indirect ELISA as described previously³¹. Briefly, the wells of a 96-well plate were coated overnight at 4 °C with 1 µg/mL of wild-type rBmHAXT vaccine antigen. After washing the plates with phosphate-buffered saline containing TWEEN-20 (PBST), the wells were blocked with 3% bovine serum albumin. Following this, diluted (1:100, 1:1000, 1:5000, 1:10,000, 1:20,000, and 1:40,000) serum samples were added and incubated for 1 h at room temperature. Following incubation, the plates were washed with PBST, and HRP-conjugated chicken anti-mouse IgG antibody (Thermo Fisher Scientific) at a 1:10,000 dilution was added as the secondary antibody. Following 1 h incubation at room temperature, plates were washed with three rounds of PBST and distilled water, then incubated with 1-step Ultra TMB-ELISA substrate and 0.16 M H₂SO₄ (Thermo Fisher Scientific), and optical density was determined at 450 nm using a BioTek Synergy 2 ELISA reader.

Levels of antigen-specific antibody isotypes

Levels of rBmHAXT-specific antibody isotypes (IgG1, IgG2a, IgG2b, IgG3, IgE, IgM and IgA) were determined in the serum samples using an indirect ELISA as described above³¹. Respective isotype-specific biotinylated goat anti-mouse antibodies (Sigma) at 1: 10,000 dilution and streptavidin-HRP (1:20,000) were used. Color was developed with 1-step Ultra-TMB. The reaction was stopped using 0.16 M H₂SO_4, and absorbance was determined at 450 nm using a BioTek Synergy 2 ELISA reader.

Challenge studies

To determine vaccine-induced protection, we used a micropore chamber challenge method as described by Chauhan et al.^31,37. Briefly, 20 infective larvae of Brugia malayi were placed in a micropore chamber and surgically implanted into the peritoneum of each mouse. Mice were given slow-release Meloxicam and Buprenorphine subcutaneously about four hours before the surgery for pain management, and Ketamine/Xylazine formulation (0–100 mg/kg/5–10 mg/kg) was used for anesthesia before surgery. Seventy-two hours following implantation, the chambers were recovered from the mice, and the total number of larvae recovered was counted. The larvae were then examined under a phase-contrast microscope for cell adherence and larval death. Larvae that were transparent, straight, and with no movement were counted as dead. Larvae that were active, coiled, and translucent were counted as live.

Secreted levels of cytokines from antigen-stimulated splenocytes

After removing the chambers, spleens were collected, and single-cell suspensions were made from each mouse. Approximately, 2 × 10⁶ cells in duplicate wells were stimulated with 5 µg/ml of respective rBmHAXT variant for antigen-specific proliferation. Cells stimulated with 2 µl/ml concanavalin A (ConA) were used as a control for non-specific stimulation. Unstimulated splenocytes were kept as a negative control for the assay. After 72 h of incubation, culture supernatants were collected and levels of IL-2, IL-4, IL-6, IFNγ, TNFα, IL-10, and IL-17A were determined using a cytokine bead array kit (BD BioSciences, San Jose, CA).

T cell subsets in antigen-stimulated splenocytes

Isolated splenocytes were washed and labeled with fluorescently labeled anti-mouse CD3 (APC) and within the CD3-gated population, the CD62L (PE/Cya7) and CCR7 (PE) positive T cells were identified as T-central memory cells. The percent population of each cell type was determined using a flow cytometer. Briefly, cells were incubated with FcγRII blocker in staining buffer (2% FBS + 0.1% sodium azide) for 30 min at 4 °C. Following washing of the cells, all three fluorescent-labeled antibodies were added and incubated for 1 h at 4 °C in the dark. After washing, cells were fixed in 4% paraformaldehyde and analyzed using a BD FACSCalibur™ (BD Biosciences) flow cytometer.

Statistical analyses

All graphs were generated using GraphPad Prism version 7.0 (GraphPad Software, San Diego, CA). Statistical analyses were performed using SPSS version 26.0 (IBM Corporation, Armonk, NY). Data were assessed for normality using the Shapiro–Wilk test; datasets with a p-value > 0.05 were considered normally distributed. Based on the results of the normality test, appropriate parametric or non-parametric statistical tests were applied to evaluate differences between groups. A p-value < 0.05 was considered statistically significant and served as the alpha level for hypothesis testing to determine whether the null hypothesis could be rejected.

Author contributions

D.C., R.K., and S.A.G. planned the experiments. J.D., C.M.P.-C., and S.A.G. prepared the various mutants of rBmHAXT vaccine antigens. C.M.P.-C., D.C., and S.A.G. generated adjuvants for the study. J.K. performed the accelerated stability analysis and calculations of the monomers. N.S., V.K., N.C., and R.K. performed all the mouse studies. D.C., R.K., and S.A.G. drafted the manuscript.

Data availability

All data sets presented in this study are included in the article.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41541-026-01497-7.