The Adjuvant Combination of dmLT and Monophosphoryl Lipid A Activates the Canonical, Nonpyroptotic NLRP3 Inflammasome in Dendritic Cells and Significantly Interacts to Expand Antigen-Specific CD4 T Cells Free

Infectious diseases are the leading cause of death in infants and children and the second leading cause of death worldwide (1–3). Vaccines are the most efficient tool in protecting people from infectious diseases, ideally conferring pathogen-specific sterilizing immunity that is long-lasting (4). Most vaccines consist of inactivated pathogens or recombinant subunits derived from nonliving pathogen Ags, many of which are poorly immunogenic (5). To increase the immunogenicity of these Ags, adjuvants are added to vaccines where they can elicit potent immune responses, yet most adjuvanted vaccines continue to use aluminum salts (aluminum hydroxide [alum]), the adjuvant of choice for >100 y. While alum excels at eliciting a potent Ab response, it is poor at inducing cell-mediated immunity and thus is ineffective against a variety of intracellular pathogens, especially at the mucosal surface where a tissue-specific Th1 and/or Th17 T cell response would often be more desirable (6–8). Because only a few vaccine adjuvants are used in approved vaccine formulations in humans, there exists a limited repertoire of tools to tailor the most effective immune response during vaccination (9). We can address the need for novel adjuvants by combining two or more well-characterized adjuvants with a vaccine Ag to generate a tunable, highly effective immune response that is best suited to control a particular pathogen. For instance, certain combinations could potentially allow us to exploit the properties of each individual adjuvant to elicit broad T and B cell immunity that can target cellular and humoral immunity to the mucosa, where most pathogens gain entry. Such an immune response would be particularly useful for the control of respiratory pathogens such as SARS-CoV-2 and influenza or intestinal pathogens such as Salmonella.

Various newer adjuvants have recently been assessed for use in the next generation of vaccines. For example, monophosphoryl lipid A (MPL-A), the biologically active lipid A portion of the Gram-negative bacterial cell wall component LPS or endotoxin, has garnered recent attention. Although the immunomodulatory effects of LPS and MPL-A are similar, the latter exhibits an ∼100-fold reduction in cellular toxicity, making MPL-A an attractive vaccine adjuvant capable of inducing a robust immune response to coadministered vaccine Ags. MPL-A has been used in combination with alum for more than a decade as the adjuvant AS04, which is used in the approved vaccine formulations for human papillomavirus and hepatitis B virus (10, 11). It is also used in combination with QS-21 in the shingles vaccine, Shingrix (12). Unlike LPS, which acts on TLR4 and predominantly signals in a MyD88-dependent fashion, MPL-A signaling is primarily MyD88-independent and utilizes Toll/IL-1R domain–containing adaptor inducing IFN-β (TRIF) to induce the production of type I IFNs and upregulate IFN-inducible genes (13). It is speculated that the reduced toxicity of MPL-A can be attributed to its MyD88 independence, which results in a less severe inflammatory response downstream of TLR4 signaling (14). Subsequently, TLR4 ligation activates NF-κB and protein kinase cascades that result in the production of proinflammatory cytokines such as IL-12, TNF, IL-6, and pro–IL-1β and the subsequent polarization of, predominantly, Th1 CD4 T cells (12–14). Although Th1 cells are efficient at combatting some infections, namely viral infections, other infections are resistant to Th1 immunity, and so adjuvant choice can greatly affect vaccine efficacy. For example, we and others have shown that the ADP-ribosylating adjuvant double mutant heat-labile toxin (dmLT) can promote both systemic and mucosal Th17 immune responses that are more effective at combatting infections caused by extracellular bacteria and fungi (15–18). Parenteral (i.m., intradermal, transcutaneous) delivery of dmLT avoids the reactogenicity that plagues the holotoxins, such as heat-labile toxin and cholera toxin (17). dmLT is a classical AB₅ toxin with an enzymatic A subunit that is noncovalently associated with a pentameric B subunit. Whereas the B-subunit is responsible for cellular entry via GM1 ganglioside binding, the A subunit ADP-ribosylates Gsα, leading to the irreversible activation of adenylate cyclase, accumulation of cAMP, and the activation of protein kinase A (PKA), all of which have been shown to be required for adjuvanticity (17–19). In addition to the cAMP/PKA axis, ADP-ribosylating adjuvant activity has been shown to potentially require caspase-1, inflammasome-dependent IL-1 signaling (16, 18).

The mechanism of action for many of these newer adjuvants remains unclear, particularly when combined. Some adjuvant combinations such as AS04 activate inflammasomes, which are multiprotein complexes that assemble in the cytosol of epithelial and immune cells and involve the detection of danger-associated molecular patterns or pathogen-associated molecular patterns (20–22). Although there are several types of inflammasome complexes that sense distinct perturbations, their triggering results in the activation of caspases, particularly caspase-1, which subsequently cleaves and releases the active form of the inflammatory cytokines IL-1β and IL-18. Activated caspase-1 also cleaves and activates gasdermin D (GSDMD), a pore-forming protein that permeabilizes the plasma membrane, causing pyroptosis and allowing for the release of activated IL-1β as well as additional danger-associated molecular patterns (23–25). Because this event is inflammatory and can be detrimental to the host, this process is highly regulated and the initiation requires two steps, the first of which is priming (signal 1). In the context of the canonical inflammasome in innate immune cells, priming occurs during the sensing of microbial components, such as LPS, leading to the activation of NF-κB and the subsequent upregulation of NOD-like receptor family pyrin domain containing 3 (NLRP3) and pro–IL-1β (20). Once primed, the inflammasome is triggered by signal 2, which includes an array of stimuli including particulate matter such as salt crystals, extracellular ATP, pore-forming toxins, ionic flux, and mitochondrial or lysosomal damage. Once triggered, the inflammasome sensors NLRP3 and ASC (apoptosis-associated speck-like protein containing a CARD) interact and nucleate monomers of pro–caspase-1, which, when in close proximity to one another, initiate self-cleavage into active caspase-1, which can then cleave pro–IL-1β into biologically active IL-1β (20, 26). It is by this mechanism that the vaccine adjuvant AS04 likely works, where MPL-A acts as the priming event through TLR4 binding and alum is the trigger to initiate inflammasome assembly, resulting in a more potent and effective cellular and humoral immune response than if either adjuvant were used alone.

Despite the ability of MPL-A and dmLT to serve alone as potent vaccine adjuvants, there are no studies exploring the effect of the two as a combination adjuvant on APCs such as dendritic cells (DCs) or on the resulting Ag-specific CD4 T cell response. In the current study, we investigated the potency of this adjuvant combination and assessed the mechanisms that might regulate the adjuvant effect of combining these adjuvants. We show that the combination of dmLT and MPL-A induces more robust activation of DCs compared with either adjuvant alone, leading to the activation of the NLRP3 inflammasome complex and increased secretion of IL-1β, as well as other hallmark proinflammatory Th1/2/17-polarizing cytokines. Unexpectedly, this inflammasome activation occurred independent of classical GSDMD-mediated pyroptosis. Furthermore, this adjuvant combination is most potent at expanding and activating multifaceted Th1, Th2, and Th17 Ag-specific CD4 T cells. These findings suggest that this combination adjuvant can successfully be used, to our knowledge, as a novel platform to elicit protective immunity against a co-delivered vaccine Ag and this immunity is at least partially dependent on nonpyroptotic NLRP3 inflammasome activation.

This study was carried out in accordance with recommendations from the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Tulane University is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. All experimental procedures involving animals were approved and performed in compliance with the guidelines established by Tulane University School of Medicine’s Institutional Animal Care and Use Committee.

C57BL/6J, NLRP3 knockout (KO) (B6.129S6-Nlrp3^tm1Bhk/J), and caspase-1 KO (B6.Cg-Casp1^em1Vnce/J) mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and were maintained under specific pathogen-free conditions in the vivarium at Tulane University School of Medicine. Mouse femurs from ASC KO and GSDMD KO mice were provided by Dr. Igor Brodsky at the University of Pennsylvania, where they were shipped overnight at −20°C in complete RPMI 1640 (10%) and processed immediately upon arrival at Tulane University.

dmLT (R192G/L211A) was produced from an Escherichia coli clone expressing recombinant protein as previously described (27). Briefly, E. coli cultures were lysed in a microfluidizer, dialyzed, and fractionated by d-galactose affinity chromatography. After elution, purified protein was passed through high-capacity endotoxin removal resin (Thermo Fisher Scientific) and stored lyophilized at 4°C and freshly resuspended in ultrapure H₂O at a concentration of 1 mg/ml prior to use. The endotoxin content of dmLT was measured using a chromogenic endotoxin quant kit (Pierce) and was determined to be 0.02 pg/µg protein. Commercially available MPL-A (InvivoGen) extracted from LPS produced by Salmonella minnesota was stored lyophilized at −20°C, resuspended in DMSO at a concentration of 1 mg/ml, sonicated for 5 min, and vortexed for 1 min immediately prior to use. LPS extracted from S. minnesota (InvivoGen) was stored lyophilized at −20°C, resuspended in ultrapure H₂O at a concentration of 2.5 mg/ml, sonicated for 5 min, and vortexed for 1 min immediately prior to use. ATP (Sigma-Aldrich) was resuspended in ultrapure H₂O at a concentration of 100 mM and stored at −20°C until ready to use. 2W1S peptide (sequence EAWGALANWAVDSA) was purchased as sterile and endotoxin free from GenScript Biotech and resuspended at 4 mg/ml in sterile Dulbecco’s PBS (Life Technologies) immediately prior to use.

DCs were cultured and expanded from the bone marrow of mouse femurs, where the bone marrow was flushed, homogenized, and cultured for 7 d in the presence of 20 ng/ml rGM-CSF (PeproTech) in complete RPMI 1640 (10%) at 37°C, 5% CO₂. For cytokine assays, suspension cells were harvested, washed, and plated in 96-well plates at 1.5 × 10⁵ cells per well prior to treatment. For all other assays, cells were plated in 12-well plates at 1 × 10⁶ cells per well prior to treatment. Cells were treated with escalating doses (0.01–100 µg) of dmLT, MPL-A, or a combination during the course of 1–48 h. Positive controls consisted of treatment with 50 ng of LPS for 24 h for activation or for 3 h with LPS plus an additional 45 min with 5 mM ATP for canonical NLRP3 inflammasome activation. For lactate dehydrogenase (LDH) controls, 10× lysis buffer for maximum LDH activity or dH₂O for spontaneous LDH activity was added 45 min prior to the treatment endpoint. After treatments, DCs were centrifuged at 1600 rpm at 4°C and supernatant for ELISA, IFN-β, cAMP, PGE₂, and LDH assays was collected and stored at −20°C until ready to assay. For flow cytometry, DCs were washed with chilled PBS and single-cell suspensions were subsequently processed for FACS analysis. For Western blots, cells were washed twice with chilled PBS and lysed for 5 min using CelLytic M (Sigma-Aldrich) supplemented with 1× protease/phosphatase inhibitor mixture (Cell Signaling Technology), and lysates were stored at −20°C until ready to assay. For RNA-based experiments, cells were washed twice with chilled PBS and further processed with an RNeasy Plus mini kit (Qiagen), and RNA was eluted in 50 µl of ultrapure water. RNA purity and concentration were determined on a NanoDrop 2000 (Fisher Scientific) and stored at −80°C until ready to assay.

Cell supernatants were assayed for IL-1β–, IL-6–, IL-12–, and TNF-secreted cytokines using 96-well flat-bottom plates (Greiner Bio-One), each coated with the respective mouse anti-cytokine (Invitrogen: IL-1β, clone B122; IL-6, clone MP5-20F3; IL-12, clone C15.6; TNF, clone F3F3D4) and incubated for 1 h at 25°C. Next, wells were blocked with 2% BSA for 1 h at 25°C and then cytokine standards or 1:10 diluted supernatants in 2% BSA were added and incubated overnight at 4°C. The next day, the respective mouse biotin anti-cytokine secondary Ab (Invitrogen: IL-6, clone MP5-32C11; IL-12, clone C17.8; TNF, clone MP6-XT3; MP6-CT22, polyclonal IL-1β) was added for 1 h at 25°C followed by the addition of 1:2000 avidin-HRP in 2% BSA for 30 min at 25°C. ELISAs were developed using a KPL tetramethylbenzidine substrate system (SeraCare), and the absorbance was read at 450 nm. Data were analyzed using a sigmoidal dose response with a least squares fit.

Cell supernatants were assayed for LDH cytotoxicity (Pierce) following the manufacturer’s instructions. Briefly, 50 µl of supernatant was added in triplicate to a 96-well flat-bottom plate (Greiner Bio-One), followed by 50 µl of reaction mixture. The plate was incubated at 25°C in the dark for 30 min before stopping the reaction with 50 µl of stop solution. The absorbance was read at 490 and 680 nm, and the percent cytotoxicity was calculated as follows: % Cytotoxicity = (adjuvant-treated LDH activity − spontaneous LDH activity)/(maximum LDH activity − spontaneous LDH activity).

Single-cell suspensions of bone marrow–derived DCs (BMDCs) were prepared in flow sorter buffer (1× PBS, 2% newborn calf serum, and 0.1% NaN₃) and incubated in Fc Block with 2% mouse and rat serum for 10 min on ice before surface staining with Abs at 1:100 dilutions for 30 min at 4°C. Abs used for flow cytometry were as follows: eFluor 450-CD11c, BV605-CD80, BV510-CD86, FITC-MHC class II (MHC-II), allophycocyanin-CD40, PE-Cy7-CD11b, allophycocyanin-eFluor 780 fixable viability dye. DCs were classified as live cells that were highly positive for the lineage marker CD11c. To identify the population of 2W1S-specific CD4⁺ T cells, enriched cell suspensions were stained with viability allophycocyanin-eFluor 780, lineage-negative (eFluor 450⁻, CD11b⁻, CD11c⁻, F4/80⁻, CD19⁻), PE-Cy7-CD25, PerCP-Cy5.5-CD3, FITC-CD8α, and PE-CD44 Abs. Cells were collected on a LSRFortessa (Becton Dickinson). Data were analyzed using FlowJo software (Tree Star). Tetramer-positive cells were gated as lineage-negative (CD11b⁻, CD11c⁻, F4/80⁻, CD19⁻), CD3ε⁺, CD4⁺, CD8α⁻, CD44^hi, and I-A^b:2W1S-allophycocyanin⁺. Ag-specific cells were enumerated using AccuCheck counting beads (Invitrogen) at a known concentration. The number of Ag-specific T cells was calculated by multiplying the total cell number with the percentage of tetramer-positive cells. All FACS data were analyzed using FlowJo v10.7.1 software (BD Life Sciences).

Cell lysates were quantified for total protein concentration using a Coomassie Plus assay kit (Pierce), and 15 µg of protein was loaded into each well of a 4–12% Bis-Tris 12-well gel (NuPAGE). Gels were run at 200 V for 45 min and then transferred to a nitrocellulose membrane using iBlot transfer stacks (Thermo Fisher Scientific). Successful transfer was verified by Ponceau S stain and then membranes were blocked with 5% BSA in PBS with Tween 20 (PBST) for 1 h at 25°C. Secondary Abs from Cell Signaling Technology were added at 1:1000 in 5% BSA in PBST and incubated overnight at 4°C and are as follows: rabbit anti-mouse IL-1β (31202), cleaved IL-1β (63124), caspase-1 (24232), cleaved caspase-1 (89332), and NLRP3 (15101). Rabbit anti-mouse β-actin (BioLegend 622102) was added at 1:1000 as a normalization control. Anti-rabbit IgG-HRP detection Ab (7074) was added at 1:2000 in 5% BSA in PBST and incubated for 1 h at 25°C. Blots were developed using SignalFire ECL reagent (Cell Signaling Technology), imaged on an Amersham Imager 600 Imager (GE Healthcare Life Sciences), and bands were quantified using ImageJ v1.8.0.

Total RNA was used for deep sequencing using the Illumina TruSeq RNA sample guide (v2). Briefly, mRNA was purified with oligo(dT) beads, fragmented with magnesium and heat-catalyzed hydrolysis, and used as a template for first- and second-strand cDNA synthesis with random primers. The cDNA 3′ ends were adenylated, followed by adaptor ligation and a 15-cycle PCR to enrich DNA fragments. Quantification of cDNA libraries was performed by using a Kapa Biosystems primer premix kit with Illumina-compatible DNA primers. The cDNA libraries were pooled at a final concentration of 1.8 pM. Single-read sequencing was performed on an Illumina Genome Analyzer IIx and NextSeq 500. Sequencing reads were annotated and aligned to the UCSC mouse reference genome (GRCm38/mm10) using the RNA sequencing (RNA-seq) aligner STAR (28). Functional analysis was performed on differentially expressed genes at each time point and were uploaded and analyzed separately using the software package Ingenuity Pathway Analysis (IPA; Ingenuity Systems, Redwood City, CA). The resulting biological functions, canonical pathways, and upstream regulators were filtered by setting a threshold of p < 0.05 using Fisher’s exact test.

Total RNA was used to generate cDNA by adding 100 ng of pure RNA to the iScript cDNA synthesis kit (Bio-Rad), and all primers were purchased from Integrated DNA Technologies. Blank and no reverse transcriptase controls were included to ensure primer specificity. A reaction mixture containing 200 µM each primer and 1 µl of cDNA was added to the iQ SYBR Green supermix (Bio-Rad) and reverse transcriptase–quantitative PCR (RT-qPCR) was performed using the CFX Connect real-time PCR machine (Bio-Rad). Data were analyzed using the ΔΔCt method, normalizing all samples to GAPDH and comparing relative expression levels to those of untreated cells.

BMDC supernatants were assessed for the secretion of the IFN-β after treatment with single and combination adjuvant using the IFN-α/β mouse ProcartaPlex panel. Briefly, treated BMDC supernatant was collected at 1, 2, and 6 h posttreatment, and 50 μl of supernatant was applied per well following the manufacturer’s recommendations. Data were collected using the Bio-Rad Bio-Plex 200 system and analyzed using a five-parameter curve fit.

BMDCs were cultured from the bone marrow of C57BL/6 mice and differentiated using rGM-CSF. Cells were seeded in six-well plates at a concentration of 1 × 10⁶ cells/ml in complete RPMI 1640 (10%) and stimulated with 10 μg/ml dmLT, MPL-A, or in combination for 24 h at 37°C, 5% CO₂. After 21 h, the phosphodiesterase (PDE) inhibitors rolipram and cilostazol were added at 50 μM to prevent conversion of the secondary messenger cAMP. Heat-labile toxin from E. coli was included as a positive control, whereas unstimulated cells were only treated with PDE inhibitors. Total responses were determined by analyzing both the supernatants and cell lysates using the cAMP parameter assay kit (R&D Systems) as per the manufacturer’s recommendations.

BMDC supernatants were assessed for the secretion of PGE₂ after treatment with single and combination adjuvant for 24 h using a PGE₂ parameter assay kit (R&D Systems) as per the manufacturer’s recommendations.

C57BL/6 mice were immunized intradermally in the ear with a single dose of 10 μg of the model Ag 2W1S with and without 1 μg of dmLT, 1 of μg MPL-A, or a combination of the two. Ten days later the cervical lymph nodes (CLNs) were harvested, and single-cell suspensions were obtained by homogenization through a 100-mm nylon mesh filter in cold sorter buffer (1× PBS, 2% newborn calf serum, 1 mM EDTA, and 0.1% sodium azide) and resuspended in Fc Block (clone 2.4G2 SFM supernatant + 2% mouse serum, 2% rat serum, 0.1% NaN₃) prior to incubation with 10 nM I-A^b:2W1S-allophycocyanin MHC-II tetramer in the dark for 1 h at room temperature. 2W1S-specific T cells were enriched by magnetic separation with anti-allophycocyanin magnetic beads (Miltenyi Biotec) as described previously. Briefly, cell suspension was incubated with anti-allophycocyanin microbeads for 30 min on ice in the dark. Cells were then washed, resuspended in cold sorter buffer, and applied to Miltenyi LS columns placed on a QuadroMACS magnet over a nylon mesh. The columns were rinsed two more times with cold sorter buffer. 2W1S tetramer-bound cells were released from the columns by the addition of cold sorter buffer to the columns off the magnet and subsequently processed for analysis by flow cytometry.

Primary statistical analyses were performed using GraphPad Prism v9 by ordinary one-way ANOVA with a Tukey’s multiple comparison test or two-way ANOVA with a Dunnett’s or Sidak’s multiple comparison test to determine statistical significance and p values, as appropriate. Significant differences are noted as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 (as displayed in figures). For selected cell count and secreted cytokine results, secondary analyses were performed to test for an interaction between dmLT and MPL-A treatments. Linear regressions, including terms that indicate whether an animal had been treated with dmLT or MPL-A, alone or in combination, and an interaction between treatments were performed using R v4.2.0 and are shown in Table I.

All data are available in the main text or Supplemental Material. The RNA-seq dataset generated in this publication has been deposited in the National Center for Biotechnology Information GEO database under accession number GSE227294. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE227294

To determine whether the combination of dmLT and MPL-A (hereafter combination adjuvant) would be a viable candidate for use in a vaccine formulation, we immunized wild-type (WT) C57BL/6 mice intradermally with the model CD4 T cell Ag 2W1S with and without dmLT, MPL-A, or a combination of the two. Ten days later we used an I-A^b:2W1S MHC-II tetramer to assess endogenous 2W1S-specific CD4 T cell expansion and activation as well as the functional phenotype of these cells as previously described (Supplemental Fig. 1) (29). To ensure that the 2W1S-specific CD4 T cells were specifically recognized by the tetramer, we initially immunized mice with the combination adjuvant alone in the absence of 2W1S Ag and enriched for tetramer binding cells in the injection site draining CLNs. As shown in Supplemental Fig. 1B, we found that only naive (CD44^lo) 2W1S-specific T cells were detected by the tetramer at very low numbers (average of 127 ± 17.25 naive cells from five mice), as has been reported by ourselves and others (30–32). This gave us confidence that the tetramer was truly detecting all of the 2W1S-specific CD4 T cells in subsequent immunization experiments. Interestingly, we found that MPL-A did not appreciably increase Ag-specific T cell expansion in the draining CLNs when compared with 2W1S alone; however, immunization with dmLT resulted in an increase in expansion in the CLNs, which resulted in a multiplicative effect (p = 0.002) by the addition of MPL-A that was not explained by adding the two results together (Fig. 1A, 1B). The results of this effect are shown in Table I where the significant interactions between MPL-A and dmLT are delineated. To determine whether these Ag-specific CD4 T cells were activated, we assessed cells for the expression the early activation marker CD25. Again, we found that immunization with the combination adjuvant resulted in the greatest number of CD25-expressing Ag-specific CD4 T cells (Fig. 1C, 1D). We have previously shown that dmLT can induce mucosal gut CD4 T cell immunity following parenteral (intradermal) administration that induces upregulation of the gut-homing integrin α₄β₇ on CD4 T cells in the draining lymph nodes (15), so we next investigated how the combination adjuvant effected the expansion of these cells. Surprisingly, Ag-specific α₄β₇⁺ CD4 T cell expansion was exclusively dmLT mediated whereas the addition of MPL-A had no impact on this response (Supplemental Fig. 1C, 1D). These data suggested that the combination adjuvant is most efficient at expanding and activating vaccine-specific CD4 T cells in the draining CLNs and that the inclusion of dmLT in our vaccine formulation continues to allow for the engagement and expansion of mucosal gut-homing phenotype α₄β₇⁺ CD4 T cells. Vaccine responses are often assessed in terms of the humoral immune response. Although not the focus of this study, we nonetheless assessed whether a single priming immunization with each adjuvant or the combination had any effect on Ag-specific Ab responses 10 d after immunization by analyzing 2W1S-specific serum IgG. As expected, because the 2W1S Ag we used is a peptide, no 2W1S-specific Abs were observed in serum (Supplemental Fig. 1E). We did find dmLT-specific IgG in the serum of vaccinated mice where once again the combination displayed the greatest amount of Ab. It has been reported that these anti-dmLT Abs do not affect subsequent adjuvanticity (17). dmLT is an attractive adjuvant because it elicits a multifaceted Th1/2/17 CD4 T cell immune response following immunization. MPL-A has been described to induce predominantly Th1 responses, so we next sought to determine how the combination adjuvant would affect T cell polarization. To do this, we assessed cytokine secretion from tetramer-enriched 2W1S-specific CD4 T cells from the CLNs stimulated with 2W1S Ag for 72 h. We found that 2W1S-specific CD4 T cells from CLNs of mice immunized with 2W1S with or without MPL-A did not exhibit an appreciable recall response to 2W1S (Fig. 2); however, animals immunized with dmLT and the combination adjuvant displayed a mixed Th response and secreted cytokines associated with Th1 (IL-2, IFN-γ), Th2 (IL-5, IL-13), and Th17 (IL-17A, IL-22) polarization, with the combination adjuvant exhibiting the highest level of secretion for most cytokines assessed. Additional testing of the interaction between dmLT and MPL-A indicated a significant interaction between dmLT and MPL-A for these cytokines that was not simply explained as an additive effect. The statistical results of this interaction are shown in Table I. Taken together, these data demonstrated that the combination adjuvant is capable of successfully eliciting the most potent multifaceted Ag-specific CD4 T cell response in the draining CLNs in response to intradermal vaccination.

Because DCs are abundant at sites of immunization and are largely responsible for the initiation of the CD4 T cell response, we sought to determine how dmLT, MPL-A, or the combination adjuvant effects gene expression in DCs to gain insight into how these adjuvants might modulate the subsequent adaptive immune response (33). To do this, BMDCs were treated for 24 h in vitro with each individual adjuvant or their combination and then adjuvant-induced changes to the transcriptome were determined using RNA-seq. The number of genes differentially regulated in BMDCs treated with dmLT was 753; with MPL-A it was 279, and in the combination it was 896 genes (Fig. 3A). Using Qiagen IPA, we identified relationships between gene signatures and predicted the activation state of various pathways in the DCs. The MIF regulation of innate immunity, Th1/2/17, IFN, TLR, PKA, inflammasome, and NF-κB were among the top activated pathways (Fig. 3B). Among the genes that were significantly upregulated in the combination adjuvant treatment were those associated with cell survival and T cell immunity (PRDX1, TNFSF11, TNFRSF1B, CD80, CD86, CD40), TLR signaling (TLR4, CD14, TRAFD1, TRAF6, IRF1), inflammasome signaling (IL1R1, IL1A, IL1B, Casp1, NLRP3), Th cell polarization (IL12A, IL6, Tgfb1/2), and cAMP and PG signaling (PRKAR2A, PDE4C, PTGES, PTGS2, PTGER2) (Fig. 3C). Using the IPA upstream regulator analysis, we identified IL-1β as a key transcriptional regulator within our gene expression dataset (Fig. 3D). These data suggested that the combination of dmLT and MPL-A acts through TLR4 and PKA in DCs to initiate a multifunctional, hyperactivated response that is potentially mediated by the canonical NLRP3 inflammasome followed by the subsequent release of IL-1β.

Identifying that the inflammasome pathway was activated in DCs after a 24-h treatment with the combination adjuvant and finding that IL-1β was predicted to be a major contributor to transcriptional regulation, we next sought to understand the temporal expression of inflammasome genes in DCs. Cells were treated for 1, 2, 3, 6, 12, and 24 h with single or combination adjuvants and then caspase-1, NLRP3, IL-1β, and GSDMD gene expression was measured using RT-qPCR and fold change was normalized to GAPDH. We found that NLRP3 and caspase-1 peak expression occurred at 6 h poststimulation with the combination adjuvant, that IL-1β expression peaked by 12 h, but that expression of GSDMD was minimal during the entire time course (Fig. 3E), suggesting that the secretion of IL-1β may be independent of classical inflammasome pyroptosis. Furthermore, using the secreted alkaline phosphatase (SEAP) reporter system, we found that both dmLT and MPL-A independently caused activation and translocation of NF-κB during the span of 72 h and that the combination does this to a significantly greater extent (Supplemental Fig. 2A). Taken together, these data demonstrated that DCs undergo a priming event to initiate the expression of NF-κB and the canonical inflammasome genes NLRP3, caspase-1, and IL-1β upon stimulation with the combination adjuvant and they exhibited the greatest potential for inflammasome activity and release of active IL-1β compared with either adjuvant alone.

To verify that observed gene expression was reflected in protein synthesis, we next sought to determine whether the combination adjuvant could drive the production of the NLRP3 protein and both the inactive pro- and active cleaved forms of caspase-1 and IL-1β by Western blot (Fig. 4A). We found that NLRP3 was upregulated after a 24-h stimulation with each adjuvant but was highest in BMDCs treated with MPL-A, either alone or in combination with dmLT (Fig. 4B). We also found the greatest level of both pro- and active caspase-1 in BMDCs treated with the combination adjuvant (Fig. 4C, 4D), suggesting that the NLRP3 inflammasome complex is assembled and has converted pro–caspase-1 to functionally active caspase-1 after exposure to the combination adjuvant. Intracellular pro–IL-1β (Fig 4E) was produced at low levels in cells treated with dmLT or MPL-A alone, with the greatest response occurring upon treatment with the combination adjuvant; however, fully functional cleaved IL-1β (Fig. 4F) was absent in cells treated with dmLT and was very modest after MPL-A treatment. In contrast, active IL-1β was greatest after treatment with the combination adjuvant, suggesting that DCs treated with the combination adjuvant produce the inflammasome proteins NLRP3, pro–caspase-1, and pro–IL-1β upon stimulation, and within 24 h they assemble the NLRP3 inflammasome to convert pro–caspase-1 to active caspase-1, which subsequently converts pro–IL-1β to active IL-1β. Based on these results, these cells were either able to continue the production of both pro- and active IL-1β and/or sequester it in the cytosol during a 24-h period, which runs counter to canonical inflammasome pyroptosis and release of cleaved IL-1β. Because it is known that cAMP has an inhibitory effect on the NLRP3 inflammasome and because dmLT relies on the induction of low levels of cAMP for adjuvanticity, we next sought to determine how the addition of MPL-A would affect cAMP production. BMDCs were treated as before with each adjuvant individually or in combination and cAMP was assessed. We found that the combination adjuvant induced the highest amount of cAMP at 24 h compared with either adjuvant alone (Fig. 4G). Because the addition of MPL-A to dmLT increased intracellular cAMP production, we hypothesized that the eicosanoid PGE₂ and the enzyme cyclooxygenase-II (COX-II), known to be upregulated by TLR4 ligation, contribute to downstream cAMP-PKA signaling. As expected, both intracellular COX-II (Fig. 4H) and secreted PGE₂ levels (Fig. 4I) were highest in cells stimulated for 24 h with the combination adjuvant, and for PGE₂ this level showed a significant interaction between the two adjuvants that could not be explained by adding the results from each alone (Table I). Because inducible type I IFN genes were upregulated in our RNA-seq analysis, we next sought to determine how the adjuvants affected secretion of IFN-β, another cytokine shown to negatively regulate inflammasome activity. We found that MPL-A significantly increased secretion both alone and in combination with dmLT during the course of 6 h (Supplemental Fig. 2B). These findings contradict published observations that signaling by IFN-β, cAMP, and, more specifically, PKA negatively regulates the NLRP3 inflammasome and downstream processing of pro–IL-1β to the active form.

To understand how single and combination adjuvants affect the phenotype of DCs and whether the NLRP3 inflammasome is required for full activation of these cells, we treated BMDCs derived from WT C57BL/6, NLRP3 KO, ASC KO, caspase-1 KO, and GSDMD KO mice with each adjuvant alone or in combination for 24 h and subsequently analyzed a variety of activation-induced surface markers (Fig. 5A). We found that the combination adjuvant imparted the greatest surface expression of CD40, CD86, and MHC-II in WT cells (Fig. 5B) and that these elevated levels were maintained in caspase-1 KO (Fig. 5C) cells but were slightly attenuated in ASC KO cells (Fig. 5D). In contrast, there was no significant increase in CD40 or CD86 and only a slight increase in MHC-II expression in NLRP3 KO cells (Fig. 5E), demonstrating that NLRP3 is likely required for DCs to either become activated or maintain an activation phenotype for 24 h poststimulation. Notably, even the single-adjuvant treatment effects were attenuated in NLRP3 KO cells, indicating that this outcome was broad. As the canonical inflammasome generally results in GSDMD-mediated pyroptosis and the subsequent release of intracellular cleaved IL-1β, we expected BMDCs derived from GSDMD KO animals to display an activation defect. Surprisingly, we found that GSDMD is ultimately dispensable for combination adjuvant induction of DC activation (Fig. 5F), showing that the combination adjuvant is most effective at activating DCs and that this activation is NLRP3- dependent but is caspase-1– and GSDMD-independent.

Because RNA expression indicated that pathways associated with Th1-, Th2-, and Th17-polarizing cytokines were activated, we next sought to determine whether DCs derived from inflammasome KO animals would have a defect in their ability to produce and secrete the hallmark Th-polarizing cytokines IL-1β, IL-6, IL-12, and TNF after stimulation with single or combined adjuvants. We found that WT DCs secreted IL-6, IL-12, and TNF in response to MPL-A regardless of whether it was combined with dmLT. This trend held true in DCs derived from NLRP3 KO, ASC KO, caspase-1 KO, and GSDMD KO animals (Fig. 6A). As expected, IL-1β secretion in WT DCs (Fig. 6B) displayed a multiplicative effect after treatment with the combination adjuvant (Table I). Moreover, IL-1β was completely ablated in NLRP3 KO, ASC KO, and caspase-1 KO DCs (Fig. 6C); however, this was not true for DCs derived from GSDMD KO animals, where IL-1β secretion was unaffected by the absence of GSDMD. Because classical inflammasome activation typically results in pyroptosis, we next sought to determine whether the increase in IL-1β secretion after treatment with the combination adjuvant was due to increased cell death. We found that neither dmLT, MPL-A, nor the combination induced significant cell death in DCs after treatment for up to 24 h when compared with untreated controls (Fig. 6D). These data indicate that the loss of the NLRP3 inflammasome complex does not affect the ability of the combination adjuvant to induce production of typical proinflammatory cytokines such as IL-6, IL-12, and TNF, but that it is essential for IL-1β secretion. Additionally, GSDMD is not required for the secretion of IL-1β, indicating that the inflammasome is operating independently of classical pyroptosis and subsequent cell death. Because we observed increased inflammasome activity in our combination adjuvant–treated BMDCs, we sought to determine the role of dmLT in the two-step process of inflammasome signaling. It is well established that LPS and its derivative MPL-A prime the canonical inflammasome, and we found that DCs primed with dmLT and subsequently treated with ATP secrete IL-1β, indicating that dmLT may in fact also be contributing to the priming (signal 1) step (Supplemental Fig. 3A). To ensure that our observations were not due to residual LPS contamination in our dmLT stock, we rigorously removed endotoxin such that the level was 0.02 pg/µg protein. Because it has been shown that cholera toxin and LT can modulate TLR signaling, we wanted to test whether the secretion of IL-1β in our combination adjuvant–treated BMDCs was dose-dependent (16). Additionally, because LPS can activate the inflammasome independent of TLR signaling (34), we wanted to verify that the loss of TLR4, using TLR4 KO BMDCs, would ablate cytokine secretion. We found the optimal dose of dmLT to be 10 μg/ml, where additional dosing of MPL-A impacted the magnitude of IL-1β secretion, with the optimal dose being 1 μg/ml (Supplemental Fig. 3B), and that TLR4 ligation is essential for our combination adjuvant to elicit cytokine responses (Supplemental Fig. 4). This suggests that the extent of IL-1β secretion is TLR4-dependent and can be controlled with the addition of MPL-A in a dose-dependent fashion.

Infectious diseases remain one of the leading causes of death worldwide, and with the increasing threat of newly emerging pathogens, there is a need for novel and improved vaccines, particularly those that can induce multipotent mucosal immunity. Adjuvants are essential for the development of many effective vaccines, yet only five adjuvants have been used in humans in approved vaccine formulations since the discovery of alum in the 1920s, and they are limited in their ability to induce a protective cellular or mucosal immune response. For example, MF59, an oil-in-water emulsion containing squalene, has been successful in increasing Ab titers to coadministered influenza Ag in the Fluad vaccine (35). Both alum and MF59 function by inducing the secretion of chemokines at the immunization site, which results in the recruitment and activation of APCs, increased migration to the draining lymph nodes, and Ag presentation to the adaptive arm of immunity (6, 35, 36). Although these adjuvants are effective at recruiting innate immune cells to the site of immunization via local chemokine secretion in the tissue, they alone do not have a substantial direct immunological effect on DCs and macrophages and often result in a nonbiased Th response (MF59) or Th2-skewed polarization (alum) (37, 38). Alternatively, the synthetic bacterial and viral DNA analog CPG 1018, used in the hepatitis B vaccine Heplisav-B, is capable of polarizing Th1 responses and driving cellular immunity to coadministered vaccine Ags (39, 40). CPG 1018 has been shown to directly modulate DC and B cell functionality through TLR9 ligation, enhancing the activation and Ag presentation capacity of DCs, which results in the generation of Ag-specific Th1 CD4 T cells and CTLs, as well as the proliferation and polyclonal activation of B cells and the generation of plasma cells (39, 41). TLR engagement is a highly effective method of cellular activation because the downstream signaling processes are reflective of an active infection. Because the discovery of new adjuvants that are both safe and effective has been difficult, some newer vaccines are exploring combining existing adjuvants to elicit potent responses. For example, the shingles vaccine Shingrix consists of zoster Ag adjuvanted with AS01_B, which is a combination of MPL-A and QS-21 (42). Because MPL-A mimics endotoxin found on the surface of Gram-negative bacteria, it can initiate a cascade of proinflammatory events through ligation with TLR4 on a variety of cell types and predominantly polarizes Th1 immunity. Unlike MPL-A, adjuvanticity by the saponin QS-21 is not mediated through cellular receptors but is simultaneously endocytosed with Ag and promotes Th1 immunity, although its use has been limited due to local reactogenicity (43, 44). Another combination adjuvant, AS04, consisting of coadsorbed alum and MPL-A, is used in the human papillomavirus and hepatitis B virus vaccines to improve both humoral and cell-mediated immunity. Although alum is poor at eliciting a Th1 cellular immunity, the addition of MPL-A aids in inducing a mixed Th1 and Th2 response. AS01_B and AS04 are prime examples of how combination adjuvants can exploit characteristics of each individual adjuvant, whether through TLR ligation or receptor-independent mechanisms, to generate a more potent and tailored immune response.

In this study, we show that the combination of the adjuvants dmLT and MPL-A acts in a multiplicative manner, causing a significant interaction between the two that cannot be explained by simply adding the effects of the two adjuvants together to increase vaccine-specific CD4 T cell immunity and NLRP3-dependent, GSDMD-independent activation and secretion of IL-1β in BMDCs. Although not yet used in licensed human vaccines, dmLT is currently in the clinical trial pipeline and has not only been demonstrated to be safe and efficacious but, unlike many other adjuvants, can induce both systemic and mucosal Ag-specific humoral and cellular immune response even when administered parenterally. We have previously shown that this occurs when dmLT engages skin CD103⁺ DCs, inducing upregulation of the gut-homing integrin α₄β₇ on CD4 T cells (15). The ability of AB₅ ADP-ribosylating proteins such as dmLT to activate DCs to polarize a Th17 cellular response has been shown to require both cAMP as well as IL-1 receptor signaling (17, 18). In our study, dmLT alone only slightly induces the expression of NLRP3, caspase-1, and IL-1β and is unable to induce DCs to process and secrete/release active IL-1β. DCs primed with dmLT and subsequently treated with ATP secrete IL-1β, indicating that dmLT may in fact be contributing to the priming step (signal 1) in the inflammasome pathway. Previous studies have shown that other ADP-ribosylating, enzymatically active AB₅ proteins such as cholera toxin and LT do not elicit cytokine secretion in murine BMDCs on their own but are capable of modulating responses to TLR agonists such as LPS (16, 45).

Furthermore, we observed that treatment of BMDCs with the combination adjuvant increased the production of cAMP, which may be attributed to activation of the arachidonic acid pathway. We found that there was an increase in production of COX-II, which is required for the conversion of arachidonic acid into the functional messenger PGE₂. One of the ways that PGE₂ can signal is through the PG surface receptors EP₂ and EP₄, and it has been shown that PGE₂ can both suppress the ability of DCs to make proinflammatory cytokines but can also synergize with TNF to upregulate synthesis of IL-1β (46). Similar to dmLT, PGE₂ results in the ribosylation of Gsα, the irreversible activation of adenylate cyclase, and the accumulation of intracellular cAMP (46–50). Interestingly, cAMP has been shown to be necessary for dmLT to function as a vaccine adjuvant but has also been shown to negatively regulate inflammasome activity by both directly binding to NLRP3 to inhibit inflammasome assembly and through downstream PKA activity (51). Contrary to these findings, we observed that the increase in cAMP in our combination adjuvant–treated BMDCs coincided with an increase in NLRP3 inflammasome-related gene signatures, protein synthesis, and secretion of the effector proinflammatory cytokine IL-1β. Additionally, the combination adjuvant increased both intracellular and secreted cAMP in BMDCs, the latter of which is likely converted to AMP in the extracellular milieu by CD39 and then ultimately metabolized to adenosine by CD73 (52, 53). Both CD39 and CD73 are ectophosphodiesterases that were found to be upregulated in our study via RNA-seq. AMP has been shown to increase DNA synthesis, mitogenesis, F-actin polymerization, chemotaxis, and CD80/86 surface expression, whereas adenosine has been shown to have regulatory/immunosuppressive effects. It is possible that dmLT induces enough secreted cAMP that it can initiate the AMP pathway via CD39, but then is quickly converted to adenosine by CD73, upon which this activity is lost; however, because the combination of dmLT and MPL-A results in much higher levels of cAMP compared with dmLT alone, this potentially allows the pathway to remain active longer. We speculate that the AMP pathway may be contributing to the robust proinflammatory innate immune responses observed in BMDCs in our study, whereas the CD39/73 axis imparts a balanced Th1/2/17-polarizing phenotype on these cells (53).

Additionally, both MPL-A alone as well as the combination adjuvant stimulate BMDCs to secrete the type I IFN IFN-β, another factor that has been shown to negatively regulate NLRP3 inflammasome activity. Despite the increases in cAMP and IFN-β, the combination adjuvant increases NLRP3 inflammasome activity and the secretion of active IL-1β compared with unstimulated controls; however, the levels are substantially lower than BMDCs stimulated with the canonical inflammasome activators LPS and ATP, which rely on GSDMD-mediated pyroptosis. It is possible that the combination adjuvant induces low-level inflammasome activity that is more tightly regulated by cAMP and PKA signaling, type-1 IFNs, and the self-limiting properties of caspase-1 (54), which leads to longer duration of cytokine secretion, albeit with less cytotoxicity. We further observed that the combination adjuvant-mediated activation of BMDCs was partially dependent on NLRP3, less so on ASC, and completely independent of caspase-1 and GSDMD. This is intriguing considering that we also showed that IL-1β secretion is completely ablated in BMDCs lacking NLRP3, caspase-1, or ASC, thus identifying a disconnect between cellular activation and IL-1β secretion. This is intriguing given that it has been shown that, in macrophages, caspase-1–dependent cleaved IL-1β relocates from the cytosol to the plasma membrane where it associates with the phospholipid phosphatidylinositol 4,5-bisphosphate (PIP2) and is subsequently released in the active form from the cell (55). This can occur in a GSDMD-independent manner and may partly explain what we observe in the present study, although this requires further investigation. Most surprising was the finding that expression of GSDMD was minimal during the entire time course, suggesting that the secretion of IL-1β may be independent of classical inflammasome pyroptosis. The nonpyroptotic inflammasome has only recently been identified and attributed to a specialized subset of phagocytes termed “hyperactivated” (56, 57). These cells are unique in that they have increased capacity for Ag uptake and presentation, are hypermigratory via CCR7, are long-lived despite inflammasome activity, and their phenotype is NLRP3-, caspase-11–, and GSDMD-dependent (58). In this case, GSDMD activity is substantial enough for IL-1β release, but minimal enough for cells to repair the cell membrane and maintain viability (59). The implications of our finding are significant because the combination adjuvant is upregulating expression of genes associated with the NLRP3-caspase-1 inflammasome, but secretion of IL-1β appears to be completely independent of GSDMD (58). Interestingly, PGE₂ has been shown to be a key factor for CCR7-mediated migration of DCs from the skin to the lymph nodes and increases their T cell stimulatory potential and Th1 polarization (60). Our BMDC RNA-seq data show that the combination adjuvant increased CCR7 expression by nearly 3.5-fold, and it upregulated genes and pathways associated with costimulation and migration capacity. This finding, along with the fact that treatment of BMDCs with the combination adjuvant increases both PGE₂ and cAMP and that immunization with the combination adjuvant results in significant expansion and activation of Ag-specific CD4 T cells, shows that the combination adjuvant potentially imparts a hyperactivated phenotype on DCs at the immunization site. These hyperactivated DCs may then be primed to activate vaccine-specific CD4 T cells via increased Ag presentation and costimulation. One potential advantage for the ability of the combination adjuvant to activate the inflammasome in DCs in the absence of pyroptosis is that this likely contributes to the robust Ag-specific CD4 T cell responses seen after immunization, as this allows for DCs to maintain the ability to present vaccine Ags to T cells without being eliminated as APCs. This advantage extends to the ability for DCs to continue to produce T cell–polarizing cytokines such as IL-1β that offers sustained activation of these T cells. This likely contributes to the overall increased expansion of vaccine-specific T cells in our study over and above what would be predicted by simply adding together the T cell numbers from either dmLT or MPL-A alone. Perhaps this is not surprising, as neither of these adjuvants on alone has the capacity to activate the nonpyroptotic inflammasome and thus cannot sustain the requisite inflammatory cytokine production. As a vaccine, increasing T cell numbers is likely to offer increased and sustained protection against a variety of pathogens, which is also advantageous. Indeed, this concept has already been proposed for increasing T cell responses to combat tumors and can also be applied in this case (58).

Although we believe that this study adds to a growing body of knowledge regarding how the nonpyroptotic inflammasome can be activated, it is not without caveats. One of the weaknesses of the current study is the relatively small sample size in some experiments, although this is somewhat counterbalanced by repeated measures. Additionally, although mice serve as an excellent tool for assessing immune responses at a basic mechanistic level, it is not always clear how well such experiments might translate into human trials, as it is known that preclinical vaccine models are not always predictive of human outcomes. A strength of our study is the demonstration of in vitro inflammasome involvement using multiple approaches including RNA-seq to identify inflammasome-related gene expression in the combination adjuvant–treated DCs, the results of which were confirmed using several inflammasome pathway knockouts. Taken together, we found that combining two potent, yet mechanistically different adjuvants together induces a more potent and tailored innate immune response that can then initiate multifaceted CD4 T cell immunity. Because of the mucosal homing properties following parenteral immunization with dmLT and the increased expansion and activation of Ag-specific CD4 T cells with the addition of MPL-A, our combination adjuvant could be a great candidate for use in the next generation of respiratory vaccines such as those directed against SARS-CoV-2. Furthermore, this combination has the potential to combat a variety of infections across multiple mucosal sites of entry. In addition to driving vaccine-specific immunity to the mucosa, the combination adjuvant could also potentially allow for dose sparing of vaccine Ag, which is often a limiting reagent in vaccine production. As we begin to understand how these adjuvants work together, we can use them as a tool to manipulate the immune response in the most effective manner against a wide array of existing and emerging pathogens.

The authors have no financial conflicts of interest.

We thank C. Flemington, Alanna G. Wanek, J.K. Kolls, and the Tulane Center for Translational Research in Infection and Inflammation for help with the RNA-seq studies. We thank Dr. L.A. Morici for thoughtful reading of the manuscript, and Dr. I. Brodsky at the University of Pennsylvania for scientific guidance and for providing bone marrow from NLRP3, ASC, caspase-1/11, and GSDMD KO animals for the inflammasome studies.