Aluminum adjuvants differentially induce IL-1β release in vitro yet share NLRP3 inflammasome-independent adjuvant effects in vivo

Abstract

Aluminum adjuvants are the most common adjuvants in human vaccines. The two types of aluminum adjuvants, aluminum hydroxide adjuvant (AH) and aluminum phosphate adjuvant (AP), differ in physical and chemical characteristics, but little is known about possible biological differences. While previous work demonstrated that AH and AP induce the secretion of IL-1β in an NLRP3-dependent manner, the role of NLRP3 in the stimulation of the immune response by aluminum adjuvants is controversial. Here, we report that AP induces more IL-1β in human and mouse macrophages and dendritic cells than AH. This effect is caused by increased NLRP3-dependent proteolysis of pro-IL-1β. In addition, AP caused a greater degree of cell damage than AH, resulting in the release of lactate dehydrogenase (LDH) and pro-IL-1β. The cell damage caused by aluminum adjuvants was partially dependent on NLRP3, suggesting that pyroptosis and other mechanisms of cell death are involved. In spite of these differences in NLRP3-dependent production of IL-1β, the ability of both aluminum adjuvants to enhance the antibody response in two different mouse models was not affected by deletion or inhibition of NLRP3. These results support the concept that aluminum adjuvants elicit redundant innate immune mechanisms that result in an enhanced adaptive immune response.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-34660-6.

Introduction

The discovery that bacterial toxoids precipitated with alum, a solution of aluminum potassium sulfate, induced a stronger antibody response than soluble toxoids¹ led to the use of aluminum-containing adjuvants in vaccines. The method of antigen precipitation by alum has largely been replaced by adding pre-made aluminum gels in the form of aluminum hydroxide adjuvant (AH) or aluminum phosphate adjuvant (AP) to vaccine formulations². In the past 25 years, vaccines have been approved with other types of adjuvants, namely the oil-in-water emulsions MF-59³ and AS03⁴, liposome-based combination adjuvants AS01⁵ and Matrix-M⁶, and the TLR9 agonist CpG1018⁷, but aluminum adjuvants remain the most commonly used adjuvants in human vaccines. More recently, aluminum adjuvants have been used as the basis for the development of novel adjuvants by combining AH or AP with other immunostimulatory molecules such as Toll-like receptor agonists^8–10.

Knowledge of the mechanisms by which aluminum adjuvants enhance the immune response has evolved over the past century but remains incomplete. Observations that alum-precipitated toxoids were more slowly removed from the injection site than soluble toxoids led to the theory that aluminum adjuvants form a depot from which antigen is slowly released, resulting in a lasting effect on the immune response^11,12. The depot theory was challenged by experiments that showed that removal of the injection site after 7 days did not interfere with the development of an antibody response to subcutaneous injection of alum-precipitated diphtheria toxoid¹³. However, removal of the injection site within 4 days diminished the antibody response. A more recent study showed that removal of the injection site within 2 h after injection into the ear pinna of mice did not interfere with the antibody response, suggesting that under certain conditions aluminum adjuvants can enhance the immune response independent of a depot effect¹⁴. Other mechanisms by which aluminum adjuvants can enhance the immune response include enhancement of antigen uptake by antigen-presenting cells¹⁵; inducing local release of danger signals such as uric acid¹⁶, IL-33¹⁷, and DNA^18,19; activation of the complement system^20,21; and activation of the NLRP3 inflammasome in macrophages and dendritic cells resulting in the release of the inflammatory cytokines IL-1β and IL-18^22–24.

While studies have consistently shown that aluminum adjuvants induce the release of IL-1β from mouse and human macrophages and dendritic cells in an NLRP3-dependent manner in vitro, contradictory data have been published on the role of NLRP3 in generating antibody responses to aluminum adjuvanted vaccines in mouse models. Intraperitoneal or subcutaneous injection of aluminum-adjuvanted vaccines in NLRP3-deficient mice was reported to reduce antigen-specific IgG1 by some groups^25,26, but not by others^27–30. Similarly, one study reported an increase of IgG2c by aluminum adjuvants in NLRP3-deficient mice²⁹, but this was not observed in other research^25,27,30. These contradictory results have been attributed to differences in experimental factors such as mouse strains, type and dose of aluminum adjuvant, the nature and dose of antigen, and the route of injection^23,31. It should be noted that these studies were generally conducted with AH or with Imject™ Alum, which consists of aluminum hydroxycarbonate and magnesium hydroxide³² and is not used in licensed vaccines, and they did not address the role of NLRP3 following intramuscular immunization, the most common route of injection of human vaccines.

The physical and chemical properties of AH and AP are quite different³³. AH is composed of needle-shaped primary nanoparticles and has a positive surface charge at neutral pH, whereas AP consists of plate-like primary nanoparticles and has a negative surface charge at neutral pH. The different surface charges affect the adsorption of antigens via electrostatic mechanisms and are an important consideration in selecting an adjuvant for a vaccine formulation². However, little is known about potential biological differences between AH and AP. Although about an equal number of licensed vaccine products contain AH and AP³⁴, most preclinical research with aluminum adjuvants has focused on AH. In earlier studies, we demonstrated that AH was more efficient than AP in antigen presentation of ovalbumin to T cells in vitro³⁵. This was supported by a recent proteomic analysis of human monocytes, which showed that AH, but not AP, enhanced pathways associated with antigen processing and presentation³⁶. It has also been reported that AP was more cytotoxic than AH for human THP-1 cells in vitro³⁷, but whether this results in differences in the immunological function of these adjuvants was not determined. It has been suggested that only positively charged particulates, including AH, can activate the NLRP3 inflammasome through destabilization of lysosomes and release of cathepsin B³⁸, although previous studies have shown that both positively charged AH and negatively charged AP stimulate the release of IL-1β from dendritic cells^35,39.

While NLRP3 is expressed in various immune cells and epithelial cells^40–42, its function has mostly been studied in myeloid cells, especially macrophages and dendritic cells. Dendritic cells play an important role in the immune response to aluminum-adjuvanted vaccines administered via intraperitoneal or intramuscular injection^16,43. The role of macrophages is less clear as intraperitoneal injection of aluminum adjuvants results in rapid depletion of peritoneal macrophages^16,28. Here, we report on investigations into the effects of AH (Rehydragel HPA) and AP (AdjuPhos) on dendritic cells and macrophages in vitro and on the role of NLRP3 in the function of AH and AP in vitro and in vivo. AP induced a significantly greater release of IL-1β from human and mouse antigen-presenting cells than AH, while the IL-1β production was dependent on NLRP3 for both adjuvants. We further confirm that AP was more cytotoxic than AH and that the adjuvant-induced cell death was partially inhibited by preventing Ninjurin-1 (NINJ1)-driven cell membrane rupture. However, using two genetically different mouse models and the intramuscular route of immunization, we show that the adjuvant effect of both AH and AP does not require a functional NLRP3.

Results

AP induces significantly greater production of IL-1β than AH and the release of IL-1β is dependent on NLRP3

Initial experiments were performed with human THP-1 macrophages and mouse bone marrow-derived macrophages (BMDMs) to assess IL-1β secretion induced by AH and AP. The production of this cytokine requires two signals^44,45. A priming step with lipopolysaccharide (LPS) enhances the expression of NLRP3 and pro-IL-1β. This reaction prepares the cell for activation of the NLRP3 inflammasome and release of IL-1β following a second signal (AH or AP). Incubation of human and mouse LPS-primed macrophages and dendritic cells with AP for 24 h induced significantly greater IL-1β secretion than AH (Fig. 1). Pre-treatment of cells with MCC950, a specific NLRP3 inhibitor⁴⁶, significantly reduced IL-1β levels (Fig. 1a and b), demonstrating the essential role of NLRP3 activation in the maturation and secretion of this inflammatory cytokine upon stimulation with aluminum adjuvants under in vitro conditions. There was no significant IL-1β release in BMDMs that were not primed with LPS upon incubation with aluminum adjuvants (Suppl. Fig. 1b). In THP-1 macrophages, there was a modest increase of IL-1β following treatment with LPS alone or AP alone (Suppl. Fig. 1a). This is likely caused by the increased expression of NLRP3 and pro-IL-1β by the PMA treatment of THP-1 cells as described previously⁴⁷. The difference in IL-1β secretion between AP and AH persisted when cells were incubated for 48 h (Suppl. Fig. 2) indicating that the larger IL-1β concentration in supernatants of cells stimulated with AP was not the result of different kinetics.

Fig. 1.

IL-1β secretion by human and murine macrophages induced by AP is greater compared to AH and is dependent on NLRP3 activation. Human THP-1 macrophages (a), mouse bone marrow-derived macrophages (BMDMs) (b and c) and mouse bone marrow-derived dendritic cells (BMDCs) (d) were primed with lipopolysaccharide (LPS, 100 ng/mL, 3 h), pretreated with the NLRP3-specific inhibitor MCC950 (10 µM, 30 min) where indicated (a, b), and subsequently stimulated with aluminum hydroxide (AH) or aluminum phosphate (AP) at 100 µg Al³⁺/mL for 24 h. The concentration of IL-1β in the culture supernatants was measured by ELISA. Bars represent mean ± SEM of triplicate wells. Statistical analysis was performed using two-way ANOVA followed by Tukey’s post hoc test. p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), p < 0.0001 (****). Data are representative of three independent experiments with similar results.

To confirm the essential role of NLRP3, we performed additional experiments with BMDMs and bone marrow-derived dendritic cells (BMDCs) from Nlrp3^-/- mice. AP induced significantly greater IL-1β secretion than AH in C57BL/6 wild type (WT) mice, but this response was abrogated in cells derived from Nlrp3^-/- mice (Fig. 1c and d). Deletion of NLRP3 did not affect the LPS-induced secretion of tumor necrosis factor (TNF) from BMDMs (Suppl. Fig. 2). The aluminum adjuvants did not induce secretion of TNF and had no effect on the LPS-induced TNF secretion (Suppl. Fig. 3).

Lysates and supernatants of BMDMs and BMDCs were evaluated by western blots to determine the levels of the mature form of IL-1β (17 kD) and pro-IL-1β (31 kD) (Fig. 2). Priming of cells with LPS increased the amount of pro-IL-1β in cell lysates but did not induce the mature form of IL-1β. Stimulation of LPS-primed cells from WT mice with AP resulted in greater levels of mature IL-1β in the supernatants compared to AH, consistent with the ELISA results. However, pro-IL-1β was also present in the supernatants following stimulation with AH and AP, although at smaller amounts than the mature form of IL-1β. Mature IL-1β was undetectable in the supernatants of LPS-primed Nlrp3^-/- cells, but there was an increase of pro-IL-1β in the supernatants of Nlrp3^-/- cells following adjuvant stimulation.

Fig. 2.

Aluminum adjuvants induce the release of inflammasome components in wild-type and Nlrp3^-/- cells. Western blot analysis of cell lysates (Lys) and supernatants (Sup) from BMDMs (a) and BMDCs (b) isolated from WT and Nlrp3^-/- mice. Cells were untreated (control), primed with LPS alone, or primed with LPS followed by stimulation for 24 h with AH or AP. Additional groups included unprimed cells stimulated with either AH or AP alone. Blots were probed for mature IL-1β (~ 17 kDa), pro-IL-1β (~ 31 kDa), caspase-1 (pro-form, ~ 45 kDa; cleaved form, ~ 20 kDa), and ASC monomers (~ 22 kDa). Ponceau S staining was used to confirm equal protein loading prior to antibody probing. Data shown are representative of two independent experiments.

The release of pro-IL-1β in the supernatants suggested cell membrane damage, which may also result in the release of other large molecules. Pro-caspase-1 is an inactive enzyme that is recruited to the inflammasome and cleaved into the active form caspase-1, which in turn cleaves pro-IL-1β into IL-1β. There was an increase of pro-caspase 1 (~ 45 kD) in the supernatants of BMDMs and BMDCs of both WT and Nlrp3^-/- mice following incubation with AP with or without priming with LPS (Fig. 2). The active form of caspase-1 (~ 20 kD) was only detected in the supernatants of WT BMDMs primed with LPS following stimulation with AP. In contrast, incubation with AH resulted in minimal detection of pro-caspase-1 and an absence of active caspase-1 in both BMDMs and BMDCs regardless of LPS priming or genotype (Fig. 2).

Another component of the NLRP3 inflammasome is the adaptor protein ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain). ASC monomers (~ 22 kDa) were detected in supernatants of both BMDMs and BMDCs incubated in medium only. However, incubation with AP consistently increased extracellular ASC release in both BMDMs and BMDCs, regardless of LPS priming or NLRP3 genotype. Incubation of cells with AH triggered ASC release in BMDMs, while ASC levels in AH-treated BMDCs were comparable to unstimulated controls. These findings indicate that AP induced more ASC release than AH in both cell types in an NLRP3-independent manner.

The ASC monomer is an adaptor protein that links pattern recognition receptors such as NLRP3 with caspase-1 as an effector molecule upon detection of pathogens or cell damage, resulting in the assembly of a multimolecular complex, the inflammasome. In addition, the ASC monomers can polymerize upon NLRP3 activation into large particles that are visible upon fluorescence microscopy as specks. To determine if the ASC monomers observed in the supernatants of cells incubated with aluminum adjuvants form specks, we performed immunofluorescence staining of BMDCs derived from WT and Nlrp3^-/- mice, cultured in chamber slides and stimulated with AH or AP, with or without LPS priming (Fig. 3). ASC-positive specks were observed, followed by AH or AP stimulation in WT LPS-primed BMDCs. These specks were visible as discrete, intense puncta and often appeared to be present extracellularly. ASC specks were also observed in non-primed WT BMDCs stimulated with AP alone, suggesting that AP may be sufficient to trigger partial inflammasome activation even in the absence of LPS priming. In contrast, ASC speck formation was not detected in Nlrp3^-/- BMDCs under any condition. Instead, Nlrp3^-/- cells had a diffuse cytoplasmic ASC labeling pattern (Fig. 3).

Fig. 3.

Immunofluorescence visualization of ASC speck formation in BMDCs from WT and Nlrp3^-/- mice following adjuvant stimulation. BMDCs derived from WT and Nlrp3^-/- mice were cultured in chamber slides (250,000 cells/well) and stimulated with aluminum hydroxide (AH) or aluminum phosphate (AP) (50 µg Al³⁺/mL) for 24 h, with or without prior LPS priming. Nuclei were stained with DAPI (blue), adjuvants were labeled with lumogallion (red), and ASC was detected using an anti-ASC antibody followed by an AF488-conjugated secondary antibody (green). In WT BMDCs (upper panel), ASC formed specks (arrows), particularly in LPS-primed cells stimulated with AH or AP. ASC specks were also detected in non-primed WT cells stimulated with AP alone. In contrast, a diffuse cytoplasmic labeling of ASC without speck formation was observed in Nlrp3^-/- BMDCs (lower panel). Fluorescence and corresponding phase contrast images were acquired at 200 × magnification. Scale bars: 20 μm. These images are representative of two independent experiments. The same images with phase contrast overlays are provided as Suppl. Fig. 4.

AP induces more cell damage and IL-1β release than AH in BMDMs and BMDCs derived from CD-1 mice

The previous experiments showed that incubation of BMDMs and BMDCs with AP induced more IL-1β release than AH and caused more cell death as indicated by an increase of pro-IL-1β, pro-caspase 1, and ASC in the supernatants. These experiments were conducted with a relatively high dose of 100 µg Al³⁺/mL. To examine whether the inflammasome activation and cytotoxicity were dose-dependent, we assessed IL-1β secretion and cell damage in LPS-primed BMDMs and BMDCs exposed to increasing concentrations (11, 33, and 100 µg Al³⁺/mL) of AH and AP. AP induced greater dose-dependent cytotoxicity than AH, as indicated by LDH release in the supernatant (Fig. 4a). Similarly, AP induced IL-1β secretion in a dose-dependent manner (Fig. 4b). The concentration of IL-1β exceeded that induced by AH even at the lowest dose of AP. Western blot analysis confirmed these findings and showed that the majority of IL-1β in the supernatant was present as the mature form with a small amount of pro-IL-1β (Fig. 4c). A similar trend was observed in BMDCs, but the magnitude of the response was generally smaller compared to BMDMs. LDH release was higher with AP than with AH, although the overall concentration of LDH was lower than what we observed with BMDMs (Fig. 4d). Notably, even the lowest concentration of AP (11 µg/mL) triggered detectable IL-1β secretion in BMDC supernatants (Fig. 4e), an effect not observed with AH at this dose. Western blot analysis (Fig. 4f) confirmed the stronger response to AP compared with AH. Together, these data demonstrate that AP induces more cytokine release and cell damage than AH in both mouse macrophages and dendritic cells.

Fig. 4.

Dose-dependent inflammasome activation and cytotoxicity in LPS-primed mouse macrophages and dendritic cells stimulated with AH or AP. BMDMs (a–c) and BMDCs (d–f) derived from CD-1 mice were incubated in medium only (M) or primed with LPS and stimulated with increasing concentrations of AH or AP for 24 h. (a, d). Cytotoxicity was assessed by measuring lactate dehydrogenase (LDH) release. (b, e) IL-1β secretion was quantified by ELISA. (c, f) Western blot analysis for IL-1β of supernatants (Sup) and lysates (Lys). Ten µg of total protein was loaded per lane for BMDMs (c) and 20 µg for BMDCs (f). Bars represent the mean ± SEM of triplicate wells. Statistical analysis was performed using two-way ANOVA followed by Tukey’s post hoc test. p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), p < 0.0001 (****).

Rapid production of IL-1β by AP and the role of cell membrane rupture in the release of pro-IL-1β

To investigate the kinetics of IL-1β processing and release in response to adjuvant stimulation, we compared early (2 h) and late (24 h) responses in LPS-primed BMDMs. At 2 h, there was minimal cell damage as indicated by the lack of LDH release (Fig. 5a). However, there was more IL-1β in the supernatant of cells treated with AP compared with AH (Fig. 5b). The concentration of IL-1β further increased in 24 h supernatants from AP-treated cells, and this was associated with increased release of LDH, indicating rupture of the cell membranes of BMDMs.

Fig. 5.

IL-1β release in macrophages stimulated with aluminum adjuvants for 2 and 24 h and effect of membrane-protective agents. LPS-primed BMDMs were pre-treated for 30 min with glycine or muscimol prior to stimulation with AH or AP at 100 µg Al³⁺/mL. Cell lysates (Lys) and supernatants (Sup) were collected at 2 h and 24 h post-stimulation to assess LDH release by colorimetric assay (a) and the concentration of IL-1β by ELISA (b) and immunoblotting (c,d). Statistical analysis was performed using one-way ANOVA followed by Tukey’s post hoc test. p < 0.001 (***), p < 0.0001 (****).

Immunoblot analysis of lysates revealed uniform strong expression of pro-IL-1β at 2 h, confirming successful LPS priming (Fig. 5c). The mature form of IL-1β was also detected in the lysate of cells incubated with AP, but not in cells incubated with AH, while only faint bands of IL-1β were present in the supernatants, suggesting a lag between the processing of pro-IL-1β into IL-1β and its release from the cell. At 24 h after the addition of aluminum adjuvants, pro-IL-1β levels in lysates had decreased. The decrease was greater in cells exposed to AP than to AH, consistent with the greater release of both pro-IL-1β and IL-1β in the supernatant induced by AP (Fig. 5b).

The presence of pro-IL-1β as well as pro-caspase-1 and ASC in the supernatants of cells incubated with aluminum adjuvants, especially AP, is suggestive of cell membrane rupture. Cell membrane rupture as the final step in programmed cell death can be orchestrated by oligomerization of NINJ1⁴⁸. Incubation of BMDMs with the membrane protective agent glycine⁴⁹ and the NINJ1 inhibitor muscimol⁵⁰ prior to the incubation with AP inhibited LDH release after 24 h (Fig. 5a). The treatment caused a decrease of pro-IL-1β in the supernatants and an increase of pro-IL-1β in the lysates, while the presence of the mature form of IL-1β was not affected (Fig. 5c and d). This experiment shows that aluminum adjuvants cause cell death by NINJ1-induced cell membrane rupture and that maintaining membrane integrity helps to retain pro-IL-1β in the cell.

The antibody response to vaccines formulated with AH or AP is affected by antigen adsorption to the adjuvant and is independent of NLRP3

Previous studies have shown conflicting results on the requirement of the NLRP3 inflammasome in the antibody response to proteins formulated with AH or Imject™ Alum, a surrogate for AH^25–29. The much greater IL-1β production induced by AP in both macrophages and dendritic cells compared with AH led us to revisit the role of the NLRP3 inflammasome in the immune response to aluminum-adjuvanted vaccines. Immunization with OVA and AH resulted in high levels of anti-OVA IgG1 and low anti-OVA IgG2c with no difference between WT and Nlrp3^-/- mice (Fig. 6). Immunization with AP induced weak antibody responses, regardless of the genotype (Fig. 6). The lower response to OVA and AP than to OVA and AH likely reflects more efficient adsorption of OVA (isoelectric point 4.6) to AH than to AP through electrostatic and ligand exchange interactions⁵¹.

Fig. 6.

NLRP3-deficiency does not impair the antibody response to OVA. C57BL/6 wild-type and Nlrp3^-/- mice were immunized intramuscularly with ovalbumin (OVA; 10 µg) alone or adjuvanted with aluminum hydroxide (AH; 100 µg Al³⁺) or aluminum phosphate (AP; 100 µg Al³⁺). Serum samples were collected 32 days post-immunization, and antigen-specific IgG1 and IgG2c responses were measured by ELISA. The symbols represent individual mice with mean ± SE of 2 males and 2 females per group (n = 4). Statistical analysis was performed using two-way ANOVA followed by Tukey’s post hoc test. p < 0.01 (**), p < 0.001 (***), p < 0.0001 (****).

We repeated the experiments with two different antigens, OVA to evaluate the antibody response to AH and the nucleoprotein (NP) of SARS-CoV-2 to assess the response to AP. NP has an isoelectric point of 10.07⁵² resulting in a positive charge at neutral pH, allowing for adsorption to AP by electrostatic interactions. AH and AP both induced a robust antigen-specific IgG1 response in inbred C57BL/6J and CD-1 mice (Fig. 7). The role of NLRP3 in the antibody response was evaluated by comparing the response between WT C57BL/6J and Nlrp3^-/- mice and treatment with the NLRP3 inhibitor MCC950 in CD-1 mice. No differences were observed in the antibody responses (Fig. 7), indicating that a functional NLRP3 is not required for the immunostimulatory activity of aluminum adjuvants in vivo.

Fig. 7.

The antibody response to protein antigens injected with aluminum adjuvants does not require functional NLRP3. (A, B) C57BL/6J WT and Nlrp3^-/- mice were injected twice with OVA and AH (A) or nucleoprotein of SARS-CoV-2 (NP) and AP (B). The graphs show the mean ± SE of four mice per group. (C, D) CD-1 mice were treated with MCC950 or untreated and then injected twice with OVA and AH (C) or nucleoprotein of SARS-CoV-2 (NP) and AP (D). The graphs show data from individual mice and the mean ± SE of 4–5 mice per group. ns – not significant (Student’s t-test, p > 0.05).

Discussion

In vitro experiments with LPS-primed human THP-1 macrophages and mouse BMDMs and BMDCs demonstrated that AP induced significantly higher levels of IL-1β release into the supernatant than AH. AP has an approximately two-fold greater mass per mg Al³⁺ than AH⁵³, but the release of IL-1β was greater even with a nine-fold lower dose of AP than AH, suggesting that this is an inherent property of AP rather than a mass effect. The release of IL-1β induced by both AH and AP was nearly completely abolished by genetic deletion of NLRP3 or by treatment with the NLRP3 inhibitor MCC950. There is general agreement in the literature that IL-1β release induced by AH is NLRP3-dependent, and our experiments show that this is also the case for AP. However, the role of NLRP3 in the immunostimulatory effect of aluminum adjuvants in vivo is controversial, with some reports indicating an essential or partial role^25,26,29 while other studies failed to support a role of NLRP3^27,28,30. The much greater release of IL-1β by AP than AH from cells in vitro motivated us to reinvestigate the role of NLRP3 in the adjuvant effect of aluminum adjuvants in vivo. Using inbred mice with a genetic deletion of Nlrp3 and outbred mice treated with the NLRP3 inhibitor MCC950, we demonstrate that the ability of both aluminum adjuvants, AH and AP, to enhance the antibody response to protein antigens following intramuscular injection is independent of NLRP3.

Priming of macrophages and dendritic cells with TLR agonists like LPS or with tumor necrosis factor induces the synthesis of NLRP3, pro-caspase-1, and pro-IL-1β^54–56. The subsequent exposure to activators of NLRP3, such as ATP, nigericin, and aluminum adjuvants, triggers the oligomerization of NLRP3 followed by recruitment of ASC and pro-caspase-1, resulting in assembly of a multimolecular complex, the NLRP3 inflammasome^54–56. The inflammasome acts as a scaffold for the polymerization of ASC, creating large protein aggregates known as “specks”. Autoproteolytic processing of pro-caspase-1 causes activation of caspase-1, which cleaves pro-IL-1β into IL-1β. Caspase-1 also cleaves gasdermin D molecules, releasing the N-terminal fragments, which assemble into 18 nm pores in the cell membrane, enabling the selective release of IL-1β^54–56. This is followed by osmotic swelling of the cell and rupture of the cell membrane, resulting in a form of programmed cell death called pyroptosis. Recent studies revealed that cell membrane rupture is mediated by the oligomerization of NINJ1⁵⁷. The rupture allows for the release of larger molecules such as LDH and, as shown here, pro-caspase-1, ASC, and pro-IL-1β.

The ELISA used to measure the IL-1β concentration in the supernatants does not distinguish between IL-1β and pro-IL-1β. Western blot analysis showed that there is more pro-IL-1β and mature IL-1β in the supernatants of cells exposed to AP compared with AH, indicating that the greater concentration of IL-1β in the supernatants of AP-treated cells is caused by a combination of increased processing of pro-IL-1β by the NLRP3 inflammasome (resulting in the release of more IL-1β) and increased cytotoxicity (resulting in the release of more pro-IL-1β). This is further supported by the finding that inhibition of NLRP3 with MCC950 or deletion of Nlrp3, which prevents the generation of the mature form of IL-1β, did not completely eliminate the difference in IL-1β secretion induced by AP versus AH.

The greater cytotoxicity of AP compared with AH is likely caused in part by more efficient pyroptosis, which is consistent with greater activation of the NLRP3 inflammasome by AP than by AH. Inhibition of cell membrane rupture by treatment of cells with glycine or muscimol, both inhibitors of NINJ1 oligomerization^49,50, reduced the release of LDH and pro-IL-1β, but did not affect the release of IL-1β. However, AP, and to a lesser extent AH, also induced the release of pro-IL-1β, pro-caspase-1, and ASC in LPS-primed macrophages from Nlrp3^-/- mice, suggesting that other cell death pathways are involved in the cytotoxic effect of these adjuvants.

Incubation of BMDMs and BMDCs with AP and AH induced the release of ASC and caspase-1 in the supernatant. Microscopy revealed extracellular ASC-positive specks in cultures of WT cells. Although ASC and pro-caspase-1 were also released by Nlrp3^-/- cells, ASC staining in Nlrp3^-/- cells remained diffuse and cytoplasmic, indicating the absence of canonical speck assembly. Previous studies have shown the release of specks from LPS-primed macrophages stimulated with ATP and nigericin^58,59. It has been shown that these extracellular particles can be taken up by macrophages and induce NLRP3 inflammasome activation, extending the inflammatory reaction. However, ASC-deficiency did not affect the recruitment of neutrophils after injection of AH⁶⁰ nor did it affect the antibody response to vaccines^30,61.

Our mouse studies conclusively demonstrate that the ability of aluminum adjuvants to enhance the antibody response to protein antigens after intramuscular immunization is independent of the NLRP3 inflammasome. To account for possible genetic influences on the response to aluminum adjuvants, both inbred C57BL/6 mice and outbred CD-1 mice were used. While in vitro data point to a central role for NLRP3 in IL-1β processing by macrophages and dendritic cells, the in vivo findings emphasize the robustness of inflammatory responses, where multiple cell types and proteases can converge to ensure cytokine maturation and adjuvant function even in the absence of NLRP3. Extracellular pro-IL-1β can be cleaved by other proteases such as neutrophil serine proteases, resulting in the generation of biologically active IL-1β^62–65. Moreover, damage-associated signals generated through cell injury and protease activity could further help create an immunostimulatory environment.

In conclusion, our data confirm the importance of NLRP3 activation in the release of IL-1β from macrophages and dendritic cells induced by aluminum adjuvants and show that the IL-1β release is much greater following exposure to AP than to AH. In spite of this difference, the antibody responses to vaccines formulated with AP or AH do not depend on a functional NLRP3. Similar results have been reported for the saponin adjuvant QS-21. The adjuvant induced IL-1β release in an NLRP3-dependent manner, but Nlrp3^-/- mice generated a greater antibody response to a vaccine formulated with QS-21 compared with WT mice⁶⁶. These results underscore the built-in redundancy of the innate immune system equipped with multiple alternative sensors that can respond to cell damage and danger signals⁶⁷, thereby sustaining and amplifying adjuvant-induced inflammation in the absence of NLRP3. The fact that the adjuvanticity of AH and AP is mediated by a multifactorial network of mechanisms that collectively contribute to their immunostimulatory effect may underlie the safety and general responsiveness to aluminum-adjuvanted vaccines.

Methods

Mice

Female Hsd:ICR (CD-1®) (CD-1) mice were purchased from Inotiv (Indianapolis, IN). A colony of B6.129S6-Nlrp3^tm1Bhk/J (The Jackson Laboratory; stock #021,302; Nlrp3^−/−) mice was established. Male and female offspring were used for the experiments. Age and sex-matched C57BL/6J wild-type (WT) mice were purchased from The Jackson Laboratory (stock #000,664) to serve as controls. All mice were housed in ventilated racks with free access to food and water and were acclimated for a week before any procedures. The mice were 6 to 8 weeks old (20–25 g body weight) at the start of the experiments. Mice were euthanized by intraperitoneal injection of ketamine (90 mg/kg) and xylazine (10 mg/kg) followed by terminal bleeding and cervical dislocation. All procedures were approved by the Institutional Animal Care and Use Committee of Purdue University and conducted in accord with the Guide for the Care and Use of Laboratory Animals and ARRIVE guidelines.

Generation of macrophages and dendritic cells

Human THP-1 cells (TIB-202) were purchased from the American Type Culture Collection (Manassas, VA). The cells were cultured in complete RPMI-1640, composed of RPMI-1640 (ThermoFisher Scientific, Waltham, MA) supplemented with 25 mM HEPES, 2 mM L-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, 0.25 µg/ml amphotericin B, 50 mM 2-mercaptoethanol (2-ME), and 10% fetal bovine serum. Note that 2-ME was excluded from the stimulation media, as its redox-modulating properties could potentially influence inflammasome activation in vitro. Cells were differentiated into macrophages by stimulation with 50 ng/mL phorbol 12-myristate 13-acetate for 24 h followed by 24 h of rest.

Bone marrow cells were collected from tibias and femurs of mice and washed twice with PBS. Red blood cells were lysed using BD Pharm Lyse (BD Biosciences, San Jose, CA).

To generate bone marrow-derived macrophages (BMDMs), cells were seeded and incubated with 50 ng/mL M-CSF in complete RPMI media. Cells were incubated overnight at 37 °C with 5% CO₂. Culture media was replaced every 2 days with complete RPMI and 25 ng/mL recombinant mouse M-CSF (ThermoFisher Scientific). On day 6, cells displayed macrophage-like morphology, and samples were immunophenotyped to confirm differentiation into macrophages, with cultures containing > 90% F4/80⁺ cells (Suppl. Fig. 4). Bone marrow-derived dendritic cells (BMDCs) were generated as previously described with minor modifications³⁵. Briefly, bone marrow cells were seeded in complete RPMI-1640 supplemented with 10 ng/mL recombinant mouse GM-CSF (ThermoFisher Scientific) and maintained at 37 °C with 5% CO₂. On day 3, non-adherent cells were removed, and fresh medium was added. On days 5, 7, and 9, half of the culture medium was replaced with fresh GM-CSF containing medium to maintain a final concentration of 10 ng/mL. On day 11, cells were harvested, washed, and resuspended in GM-CSF-free medium. The cultures contained at least 85% BMDCs based on labeling for CD11c and less than 5% F4/80⁺ macrophages (Suppl. Fig. 5).

Cell culture

For in vitro experiments comparing Nlrp3^-/- and WT cell cultures, BMDMs or BMDCs derived from C57BL/6J mice were initially seeded in 24-well plates at 0.3 × 10⁶ cells/mL in 0.5 mL final medium per well (n = 3 wells per treatment). For dose response experiments, BMDMs or BMDCs from CD-1 mice were seeded in 48-well plates at 0.2 × 10⁶ cells/mL in 0.25 mL final medium per well (n = 4 wells per treatment) and stimulated for 24 h with AH (Rehydragel HPA, Chemtrade, Berkely Heights, NJ) or AP (Adju-Phos, Invivogen, San Diego, CA) at 11, 33, or 100 µg Al³⁺/mL. For the kinetics experiments, BMDMs from CD-1 mice were seeded in 12-well plates at 0.6 × 10⁶ cells/well in 1 mL final medium per well; n = 3 wells per treatment) and exposed to AH or AP, with supernatants collected at 2 h and 24 h.

BMDMs and BMDCs were primed with 100 ng/mL ultrapure lipopolysaccharide (LPS from E. Coli 0111:B4; InvivoGen) or left untreated for 3 h, followed by incubation with aluminum hydroxide (AH) or aluminum phosphate (AP) at the indicated concentrations for 24 h (or 2 h, where specified) at 37 °°C with 5% CO₂. All treatments within each independent experiment were carried out in triplicate (minimum) under identical culture conditions. In some experiments, cells were treated with 10 µM MCC950 (Selleck Chemicals, Houston, TX), 1 mM muscimol (Tocris Biosciences) or 5 mM glycine (ThermoFisher Scientific) for 30 min prior to the addition of adjuvants. At the end of the stimulation period, culture supernatants were carefully collected, centrifuged at 300 × g for 5 min to remove cell debris, and stored at − 80 °C until cytokine quantification and LDH measurement (CyQUANT™ LDH Cytotoxicity Assay, ThermoFisher Scientific).

IL-1β and TNF ELISA

The concentration of IL-1β and TNF in the supernatants was determined using mouse IL-1β and mouse TNF ELISA kits (R&D Systems, Minneapolis, MN), following the manufacturer’s protocol. Optical density was measured at 450 nm using a microplate reader, and IL-1β and TNF concentrations were interpolated from standard curves using four-parameter logistic regression.

Western blot

Western blotting was performed on lysates and culture supernatants from pooled triplicates of stimulated cells. Supernatants were collected after stimulation and centrifuged at 300 × g for 5 min to remove cell debris. Supernatants were stored at − 80 °C until analysis. Cells were washed with cold PBS and lysed in Cell Lysis Buffer II (ThermoFisher Scientific) supplemented with protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). Cell lysates were incubated on ice for 30 min with periodic vortexing and centrifuged at 14,000 × g for 15 min at 4 °C. Supernatants containing soluble proteins were collected and stored at − 80 °C until use. Protein concentrations were determined using the BCA assay (ThermoFisher Scientific), and equal amounts of protein (10–20 µg per sample) from cell lysates or supernatants were mixed with NuPAGE™ LDS Sample Buffer (4X) (ThermoFisher Scientific) and NuPAGE™ Sample Reducing Agent (10X) (ThermoFisher Scientific) according to the manufacturer’s instructions. Samples were denatured by heating at 70 °C for 10 min and loaded onto 4–12% Bis–Tris SDS-PAGE gels (ThermoFisher Scientific). Electrophoresis was conducted under reducing conditions, and proteins were transferred onto PVDF membranes (Millipore Sigma Aldrich) using a semi-dry transfer system. Following protein transfer, membranes were stained with Ponceau S solution (0.1% in 5% acetic acid) to confirm uniform protein loading and efficient transfer. After destaining, membranes were incubated in SuperBlock™ T20 (TBS) Blocking Buffer (ThermoFisher Scientific) for 1 h at room temperature with gentle agitation to minimize nonspecific binding. After blocking, membranes were incubated overnight at 4 °C with the following primary antibodies diluted in TBST (10 mM Tris-buffered saline with 0.05% Tween® 20) containing 5% bovine serum albumin (BSA): Biotinylated anti-mouse IL-1β/IL-1F2 (R&D Systems, Minneapolis, MN, cat. #BAF401; 1:1000), rabbit monoclonal anti-mouse ASC/TMS1 (D2W8U) (Cell Signaling Technology, Danvers, MA, cat. #67,824; 1:1000), mouse monoclonal anti-mouse caspase-1 (clone Casper-1, Adipogen, San Diego, CA, cat. #AG-20B-0042-C100; 1:1000), or anti- β-actin (Cell Signaling Technology, cat.#58169T). Membranes were then washed five times in TBST and incubated for 1 h at room temperature with either streptavidin-HRP (R&D Systems, cat. #DY998; 1:2000) for IL-1β detection, or HRP-conjugated anti-rabbit IgG for ASC and HRP-conjugated anti-mouse IgG for caspase-1. Protein bands were visualized using enhanced chemiluminescence (Amersham ECL, Cytiva, Marlborough MA; Cat# 45,002,401) and imaged using the Azure c280 Imaging System. The relative amount of the target proteins normalized to the Ponceau S staining was determined by densitometry as presented in Supplementary Materials.

Mouse immunization

Endotoxin-free ovalbumin (OVA) was purchased from InvivoGen and the SARS-CoV-2 Nucleocapsid protein (NP) was purchased from Acro Biosystems (Newark, DE). Formulations were prepared by mixing antigen and adjuvant for 1 h at room temperature. Vaccine formulations consisted of 50 or 100 µg/mL OVA with 1 mg Al³⁺/mL AH or AP, and 50 μg/mL N-protein with 1 mg Al³⁺/ml AP in 10 mM Tris-saline buffer, pH 7.4. Mice were injected intramuscularly with 50 µL vaccine in each hindleg on Days 0 and 21. Blood samples were collected on Days 15 and 32.

Mice assigned to MCC950 treatment received an intraperitoneal (i.p.) injection of MCC950 (25 mg/kg) 1 h prior to intramuscular immunization (Day 0). A second i.p. dose of 10 mg/kg was administered 24 h later (Day 1). Thereafter, mice received i.p. injections of MCC950 (10 mg/kg) every other day until the day before euthanasia.

Serum antibody analysis by ELISA

Serum samples were collected 15 days post primary injection and 11 days after the second injection. OVA-specific and NP-specific IgG1 and IgG2a or IgG2c were analyzed by ELISA in 96-well plates coated overnight at 4 ˚C with 1 µg/mL OVA or 1 µg/mL NP in coating buffer (15 mmol/L Na₂CO₃, 35 mmol/L NaHCO₃, 7.7 mmol/L NaN₃, pH 9.6). Wells were washed four times with 200 μl/well TBST, blocked with 200 μl/well of TBST/2% BSA and incubated at 37 °C for 1 h. The wells were incubated with 100 µl of diluted serum samples (1:400 for IgG1; 1:50 for IgG2a/c) in duplicate for 1 h, followed by washing and incubation with 100 µl of peroxidase-conjugated goat IgG1 (1073–05), IgG2a (1080–05) or IgG2c (1079–05; all from SouthernBiotech, Birmingham, AL) for 2 h. After washing the plate, wells received 100 µL 3,3’,5,5’ tetramethylbenzidine substrate solution (Neogen, Lexington, KY) and allowed to react in the dark at room temperature for 5–10 min. The reaction was terminated with 50 µl of 2 M sulfuric acid, and absorbance at 450 nm (OD 450) was measured in a microplate reader (BioTek Instruments, Winooski, VT). We measured IgG2a antibodies in CD-1 mice, but these mice may also express IgG2c as they are outbred.

Immunofluorescence staining and microscopy

BMDMs and BMDCs were cultured in chamber slides and stimulated as indicated. Aluminum adjuvants were pre-labeled with lumogallion (30 μM; ThermoFisher) by overnight incubation at 2 mg Al³⁺/mL in 150 mM NaCl, pelleted (13,000 × g, 10 min), washed once, and re-suspended to 2 mg Al³⁺/mL before addition to cultures at 100 μg/mL. Cells were fixed in 4% paraformaldehyde in PBS for 15 min, washed, and permeabilized/blocked in PBS containing 5% normal goat serum and 0.3% Triton X-100 for 60 min. Samples were incubated overnight at 4 °C with anti-ASC/TMS1 diluted in PBS with 1% BSA and 0.3% Triton X-100, washed, and incubated for 2 h at room temperature in the dark with Alexa Fluor® 488 goat anti-rabbit IgG (H + L) secondary antibody (Jackson ImmunoResearch, West Grove, PA, cat. #111–545-045). Nuclei were counterstained with DAPI using ProLong™ Gold Antifade Mountant (Thermofisher Scientific), and slides were cured overnight and stored at 4 °C. Images were acquired on a ZEISS Imager A.2 microscope equipped with an Axiocam 503 mono camera using ZEN software, with identical acquisition settings applied across experimental and control conditions.

Statistical analysis

The statistical significance of differences in IL-1β and in OVA-specific antibody levels between experimental groups was determined by one-way or two-way analysis of variance (ANOVA) test as indicated in the figure legends followed by a Tukey’s multiple comparisons test. Student’s t-test was used to analyze the effect of NLRP3 inhibition on AH and AP enhanced antibody responses. Statistical analyses were performed with GraphPad Prism version 10.4.1 (GraphPad Software, San Diego, CA) for Windows.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: D.C.P., H.H.; Methodology: D.C.P., H.H.; Investigation: D.C.P., J.F.H-F.; Formal Analysis: D.C.P., H.H.; Data curation: D.C.P.; Writing—original draft preparation: D.C.P., H.H.; Writing—review and editing: D.C.P., J.F.H-F., H.H.; Supervision: H.H.; Project administration: H.H.; Funding acquisition: H.H.

Funding

This research was supported in part by Hatch formula funds from USDA-NIFA (Project No. IND90022882) and funds from the Purdue University College of Veterinary Medicine.

Data availability

Data will be made available upon request to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.