Human in vitro modeling identifies adjuvant combinations that unlock antigen cross-presentation and promote T-helper 1 development in newborns, adults and elders

Abstract

Adjuvants are vaccine components that can boost the type, magnitude, breadth, and durability of an immune response. We have previously demonstrated that certain adjuvant combinations can act synergistically to enhance and shape immunogenicity including promotion of Th1 and cytotoxic T-cell development. These combinations also promoted protective immunity in vulnerable populations such as newborns. In this study, we employed combined antigen-specific human in vitro models to identify adjuvant combinations that could synergistically promote the expansion of vaccine-specific CD4⁺ cells, induce cross-presentation on MHC class I, resulting in antigen-specific activation of CD8⁺ cells, and direct the balance of immune response to favor the production of Th1-promoting cytokines. A screen of 78 adjuvant combinations identified several T cell-potentiating adjuvant combinations. Remarkably, a combination of TLR9 and STING agonists (CpG + 2,3-cGAMP) promoted influenza-specific CD4⁺ and CD8⁺ T cell activation and selectively favored production of Th1-polarizing cytokines TNF and IL-12p70 over co-regulated cytokines IL-6 and IL-12p40, respectively. Phenotypic reprogramming towards cDC1-type dendritic cells by CpG + 2,3-cGAMP was also observed. Finally, we characterized the molecular mechanism of this adjuvant combination including the ability of 2,3-cGAMP to enhance DC expression of TLR9 and the dependency of antigen-presenting cell activation on the Sec22b protein important to endoplasmic reticulum-Golgi vesicle trafficking. The identification of the adjuvant combination CpG + 2,3-cGAMP may therefore prove key to the future development of vaccines against respiratory viral infections tailored for the functionally distinct immune systems of vulnerable populations such as older adults and newborns.

Graphical Abstract

Introduction:

Vaccines effectively prevent and control transmission of infections [1]. Challenges remain, however, in vaccine development for newborns, older adults, and immunocompromised individuals due to the complexity of the functional differences in their immune systems [2–4]. Impaired production of T-helper (Th) 1-polarizing cytokines, TNFα and IL-12p70, in response to commonly used immunostimulators in newborns can diminish their ability to induce a protective response towards intracellular pathogens such as viruses [5–8]. Infants develop a Th2-biased response with low-affinity non-neutralizing antibodies against viruses such as respiratory syncytial virus (RSV), resulting from functional distinctions in molecular signaling following Toll-like receptor (TLR) activation [9–11]. In comparison, older adults have a reduced antigen-specific CD4⁺ and CD8⁺ T cell response and produce fewer CXCR3⁺ circulating follicular T cells and Th1 cells [12].

While protein subunit-based vaccines have a track record of safety and can be immunogenic, they generally elicit relatively low Th1 and cytotoxic T lymphocyte responses if not appropriately adjuvanted, limiting their effectiveness against intracellular pathogens [2,13]. However, these responses may be increased by effective adjuvants. Reduced CD8⁺ T cell activation may reflect impaired antigen cross-presentation via the major histocompatibility complex (MHC) class I pathway [13–15]. Adjuvantation can increase vaccine immunogenicity and improve responsiveness in populations at high risk for infections by potentiating the levels of cellular and humoral responses [16]. Use of adjuvants in the development of non-live vaccines aims at the induction of robust, safe, and long-lasting response, ideally with a single dose [2]. Many adjuvants stimulate the innate immune system by activating pattern-recognition receptors (PRRs) such as TLRs on antigen-presenting cells including dendritic cells (DCs), monocytes, neutrophils, and macrophages, enhancing vaccine immunogenicity by inducing cytokine and chemokine production following the detection of pathogen-associated molecular patterns (PAMPs) or host damage-associated molecular patterns (DAMPs) [2]. Formulation of protein subunit vaccines with a PRR agonist can thus improve both efficacy and safety [9].

We and others have previously demonstrated that adjuvant combinations can overcome age-specific impairments in type and magnitude of immunity [9,17–32]. A combination of the adjuvants aluminum hydroxide and monophosphoryl lipid A (MPLA), a TLR4 agonist, showed a prolonged cytokine response at the site of injection, resulting in increased numbers of activated DCs and monocytes in the draining lymph nodes [33]. A combination of 3M-052, a TLR7/8 agonist, and trehalose-6,6-dibehenate (TDB), a Mincle agonist, was able to induce protective immunity against RSV in newborn mice by reprogramming their DCs to promote the development of Th1 over Th2 immunity [34]. This combination acted synergistically in inducing alternative NF-kB signaling pathways to overcome an impairment in newborns to activate NF-κB [11]. We also demonstrated the ability of this combination of adjuvants to induce antigen cross-presentation, a key requirement for the induction of CD8-mediated immunity by a protein subunit vaccine. To confirm these observations, we developed a fully autologous human co-culture assay, the DC:T cell interface (DTI) assay, where monocyte-derived dendritic cells (MoDCs) are stimulated with a vaccine antigen in the presence or absence of candidate adjuvants, and the ability of these adjuvants to induce cross-presentation is evaluated by the ability of the MoDCs to activate antigen-specific CD8⁺ T cells [31,34].

Here, we hypothesize that identification of additional adjuvant combinations that may induce cross-presentation will widen availability of adjuvant formulations for preclinical evaluation and improve our understanding of adjuvant mechanism of action. By high throughput discovery of adjuvant combinations in the DTI assay, combined with comprehensive immunophenotyping of MoDCs, we identified combinations that preferentially induce a desired Th1 response in young adults, and studied their age- and population-specificity. We screened various commercially available PRR-activating adjuvants including TLR agonists, C-type lectin receptor (CLR) agonists, nucleotide-binding and oligomerization domain (NOD)-like receptor (NLR) agonists, stimulator of interferon genes (STING) agonists, and retinoic acid-inducible gene I (RIG-I) agonists to investigate their ability to promote antigen-specific CD4⁺ and CD8⁺ T cell activation and selectively uncouple co-regulated cytokines IL-12p40 and IL-12p70, and TNF and IL-6. Adjuvant combinations of interest were further evaluated in vitro in an additional cohort of young adults to confirm the initial observations and test their ability to stimulate innate immune responses in cells from newborns and older adults.

Remarkably, we found that the combination of CpG ODN 2395 (a TLR9 agonist) and 2,3-cGAMP (a STING agonist) successfully induced CD4⁺ and CD8⁺ T cell activation, driving polarization towards a Th1 phenotype and inducing Sec22b-dependent antigen cross-presentation. This adjuvant combination was further evaluated in vitro using cells from different age groups. In addition to acting synergistically in early life, it promoted the production of Th1-favoring IL-12p70/IL-12p40 and TNF/IL-6 ratios. We found that although MoDCs express very little TLR9, required for activation using CpG, 2,3-cGAMP was able to increase TLR9 expression, thus providing a potential explanation for the efficacy of this adjuvant combination. Our characterization of the action of novel adjuvant combinations, particularly CpG + 2,3-cGAMP, will inform development of adjuvanted vaccines against respiratory viral infections that can be tailored to vulnerable populations.

Materials and Methods:

Immunomodulators:

We sourced the following immunomodulators from InvivoGen for screening: Pam₃CSK₄ (TLR1/2 agonist), Poly I:C HMW (TLR3 agonist), MPLA (TLR4 agonist), CL264 (TLR7 agonist), TL8–506 (TLR8 agonist), R848 (TLR7/8 agonist), CpG ODN 2395 (TLR9 agonist), 2,3-cGAMP (STING agonist), TDB and BGP (CLR agonists), MDP and AlPO/Adju-Phos (NLR agonists), and 3p-hpRNA (RIG-I agonist). All stimulants, except for MPLA, were tested for the absence of endotoxin using the limulus amoebocyte lysate (LAL) assay.

Isolation of peripheral blood mononuclear cells, monocytes and T lymphocytes, and generation of monocyte-derived dendritic cells:

Peripheral blood samples from healthy young and older adults who received their annual immunization against influenza were collected after obtaining written informed consent per the protocols approved by the Institutional Review Boards of Boston Children’s Hospital (X07-05-0223) and Brigham and Women’s Hospital (P00013867). Cord blood was obtained following scheduled cesarean section, with informed consent by the parents and approval from the Ethics Committee of the Beth Israel Deaconess Medical Center (protocol number: 2011P-000118).

De-identified, heparinized blood samples were processed within 2–4 hours to isolate mononuclear cells as previously described [35]. Following centrifugation of whole blood at 500 g for 10 minutes, platelet-rich plasma was removed and centrifuged at 3000 g for 10 minutes to obtain platelet-poor plasma for use in downstream cell culture. We reconstituted blood with Dulbecco’s phosphate-buffered saline (DPBS) and layered it on Ficoll-Paque (Cytiva, Marlborough, MA, USA) for density gradient centrifugation for 30 minutes at 1000 g. We then collected the mononuclear cell fraction, washed it twice with DPBS, and isolated monocytes from this fraction through positive selection with magnetic CD14 MicroBeads (Miltenyi Biotec, Auburn, CA, USA) in accordance with the manufacturer’s instructions. CD8⁺ and CD4⁺ T cells were sequentially isolated from the remaining cell fraction using CD8⁺ and CD4⁺ MicroBeads, and cryopreserved in RPMI/10% fetal bovine serum (FBS) (Gibco, Ward Hill, MA, USA)/10% dimethyl sulfoxide (DMSO) at −80°C for five days until the start of the DTI assay. CD14⁺ monocytes were cultured in a humidified incubator at 5% CO₂ and 37°C in 75 cm² tissue culture flasks at a concentration of 10⁶ cells/mL of RPMI supplemented with 1% penicillin-streptomycin (Gibco, Ward Hill, MA, USA), 10% FBS, 50 IU/mL IL-4, and 100 IU/mL granulocyte-macrophage colony-stimulating factor (GM-CSF) (Miltenyi Biotec, Bergisch Gladbach, Germany) for five days. Following this, immature MoDCs were harvested for use in the MoDC and DTI assays.

MoDC assay:

MoDCs were resuspended in RPMI and autologous plasma at a concentration of 1.39 million cells/mL, and a volume of 180 μL (250,000 cells) was plated into each well of a 96-well plate containing 20 μL of freshly prepared adjuvant formulations. Suspensions were mixed gently, and cells were cultured for 24 hours in a humidified incubator at 37°C and 5% CO₂. Following culture, cells were centrifuged for 3 minutes at 500 g and supernatants were harvested and stored at −80°C for cytokine multiplexing. Cells were washed with DPBS and fixated in 1% paraformaldehyde (PFA) for flow cytometry.

MoDC-T cell interface (DTI) assay:

We performed the DTI assay as previously described [31]. MoDCs from young and older adults who were vaccinated against influenza were suspended in RPMI supplemented with 1% penicillin-streptomycin, 10% FBS, 50 IU/mL IL-4, and 100 IU/mL GM-CSF. To a 96-well plate containing 10 μL of adjuvant formulations, we added 90 μL (25,000 cells) of MoDCs per well in the presence or absence of 1 μg/mL influenza hemagglutinin (HA) protein (Flublok quadrivalent, season 2021/22, Sanofi Pasteur) or 1 μg/mL HIV envelope glycoprotein, gp120. Suspensions were gently mixed and cultured for 24 hours. On the same day, autologous CD4⁺ and CD8⁺ T cells were thawed, resuspended in RPMI with 10% FBS and 1% penicillin-streptomycin, and cultured for 24 hours. 100 μL of either CD4⁺ or CD8⁺ T cells was added to each well of the respective DTI plate in a 1:10 (MoDC:T cell) ratio, and the co-culture was incubated for four days. Control conditions containing T cells alone without MoDCs were used to confirm that the adjuvants were not directly acting on T cells (not shown). Following this, cells were centrifuged for 3 minutes at 500 g, washed with DPBS, fixated in 1% PFA, and stored at 4°C for flow cytometry. As pre-existing immunity to the HA antigen was required to quantify the reactivation of antigen-specific T cells, we did not include cord blood in this assay.

For further investigation into the actual ability of tested adjuvant combinations to induce Th1 polarization, we modified the DTI assay to include only naïve (CD45RA⁺) T cells. Following a 24-hour incubation of MoDCs with CpG, cGAMP, 3p-hpRNA, CpG + cGAMP, or CpG + 3p-hpRNA, cells were co-cultured for four days with autologous, naïve T cells in a 1:10 (MoDC:naïve T cell) ratio and were provided antigen-specific stimulation with anti-CD3 + anti-CD28 beads. After four days of incubation, naïve T cells were fixated in 1% PFA, permeabilized with a perm/wash buffer (BD Biosciences) for intracellular staining of transcription factors T-bet and GATA3 (markers for Th1 and Th2 cells, respectively), and stored for flow cytometry at 4°C.

Flow cytometry:

We utilized flow cytometry (LSRFortessa flow cytometer, Becton Dickinson) to identify and characterize cell sub-populations following MoDC and DTI assays. Data were analyzed using the FlowJo software version 10. Cells fixated in 1% PFA were stained for 30 minutes in the dark with panels which used the following antibodies – MoDCs: anti-CCR7/CD197-BUV395, anti-HLA-A/B/C-V450, anti-HLA-DR-BV605, anti-XCR1-FITC, anti-CD86-PerCP-Cy5.5, anti-CD287-PE, anti-CD289-APC, and anti-TLR9-APC; DTI: anti-CD4-V450 or anti-CD8-V450, anti-CD25-Alexa488, anti-CD134-PE, anti-CD154-APC, anti-T-bet-A700, and anti-GATA3-PE-CF594 (BD Biosciences, East Rutherford, NJ). Following staining with antibodies, cells were washed and resuspended in DPBS prior to data acquisition. Representative gating strategies for all experiments are shown in Supplementary Figure 4.

Cytokine multiplex:

We analyzed cytokine profiles of supernatants from the MoDC assay using a multianalyte fluorescent bead-based array (Luminex Corp., Austin, TX, USA). Quantification was performed using custom-selected cytokines from the Milliplex HCYTA-60K Human Cytokine/Chemokine/Growth Factor Panel A kit (Millipore, Merk, Darmstadt, Germany): CXCL10, IFNα-2, IL-1β, IL-6, IL-10, IL-12p40, IL-12p70, and TNF. Sample fluorescence data was collected using a Flexmap 3D analyzer running xPONENT software version 4.2, and results were fit to a 5-point log curve and converted into pg/mL values using manufacturer-provided standard solutions and Milliplex Analyst software version 5.1.

Statistical analyses:

We used GraphPad Prism version 10.0.3 for statistical analysis. Pearson correlation and regression analyses were performed to predict the degree of correlation between the analytes induced by the tested adjuvant combinations in comparison to the control using Wilcoxon rank-sum test. Concentration-response curves and bar plots (with mean ± SD) were compared using two-way repeated measures ANOVA with Bonferroni post-test correction. Synergy between adjuvant combinations was calculated using the Loewe method of additivity, and D values less than 1 were considered synergistic [35]. Flower plots were generated using Grapher v15.2.311. We indicate statistical significance by stars in the figures (* p < 0.05, ** p = 0.01, *** p = 0.001). N is 3 for newborns, 9 for young adults, and 8 for older adults.

Results:

MoDC and DTI assays identify novel adjuvant combinations.

We evaluated PRR-activating adjuvants of different classes, including Pam3CSK4 (TLR1/2 agonist), Poly I:C HMW (TLR3 agonist), MPLA (TLR4 agonist), CL264 (TLR7 agonist), TL8–506 (TLR8 agonist), R848 (TLR7/8 agonist), CpG ODN 2395 (TLR9 agonist), 2,3-cGAMP (STING agonist), TDB and BGP (CLR agonists), MDP and AlPO/Adju-Phos (NLR agonists), and 3p-hpRNA (RIG-I agonist), in distinct and complementary human in vitro platforms. The MoDC assay modeled the ability of the adjuvant combinations to promote DC maturation and secretion of T cell-polarizing cytokines [11,34,36], and the DTI assay measured enhanced antigen uptake and presentation on MHC to induce reactivation of autologous antigen-specific CD4⁺ and CD8⁺ T cells [31,34]. Many adjuvant combinations were able to boost the frequency of activated antigen-specific CD4⁺ or CD8⁺ T cells (represented as percentage of total T cells), induce the upregulation of costimulatory receptors (represented as percentage of positive cells), and promote the secretion of pro-inflammatory cytokines (quantified in pg/mL) in MoDCs (Figure 1A). A Pearson correlation matrix (Figure 1B) illustrates which endpoints correlate strongest with each other, suggesting their co-regulation, and reveals MoDC-derived endpoints that correlate positively with CD4⁺ or CD8⁺ T cell activation in the DTI assay. We found that only IL-6 and IFNα from the MoDC assay correlated very modestly with influenza HA-specific CD4⁺ and CD8⁺ T cell activation, respectively, in the DTI assay. Although we observed expression of the chemokine receptor XCR1, an in vivo marker for professional cross-presenting DCs [37,38], there was no correlation with CD8⁺ T cell induction, suggesting that in vitro induction of XCR1 expression on cDCs may not be sufficient for antigen cross-presentation. We also do not observe any positive correlation between CD4 T cell induction and HLA-DR expression, or between CD8 T cell induction and CD86 expression. No individual MoDC endpoint from flow cytometry or cytokine multiplex analysis had a strong correlation with the magnitude of activated antigen-specific CD4⁺ or CD8⁺ T cells from the DTI assay. This observation emphasizes the novelty of using the DTI assay as a screening method and shows that the DTI assay is necessary for accurately predicting T cell activation induced by the tested adjuvants. In general, most adjuvant combinations enhanced antigen-specific CD4⁺ T cell activation. Only combinations such as CpG + 2,3-cGAMP, CpG + 3p-hpRNA, CpG + TL8–506, R848 + TDB, CL264 + 3p-hpRNA, CL264 + MDP, Poly I:C + CL264, Poly I:C + BGP, and Poly I:C + 3p-hpRNA induced antigen-specific cytotoxic T cell activation. Overall, the strongest correlation was observed in the production of TNF, IL-12p70, IL-6, IL-12p40, CXCL10, and IL-10 with several adjuvant combinations. Combinations that included TLR8, TLR7/8, and STING agonists selectively induced IL-12p70. Thus, the utilization of both MoDC and DTI assays for screening of adjuvants identified combinations that promote antigen-specific T cell activation and selectively uncouple typically co-regulated cytokines IL-12p40 and IL-12p70, as well as TNF and IL-6.

Figure 1. Adjuvant screening, including antigen-specific recall, identified combinations that promote CD4 and CD8 immunity in MoDC and DTI assays.

A) Heat map representation of the screen of 78 adjuvant combinations in healthy young adults (n = 5, Wilcoxon rank-sum test, * p < 0.05, ** p = 0.01, *** p = 0.001). Intensities represent the mean of all biological replicates tested in DTI (top), supplemented with MoDC immunophenotyping (bottom). Units of measurement for CD4 and CD8 panels (purple) is the percentage of total antigen-specific T cells activated in response to adjuvant stimulation, for receptor expression (red) is the percentage of positive cells, and for cytokine levels (blue) is pg/mL. B) Pearson correlation analysis demonstrates the degree of correlation between cytokines produced by MoDCs, costimulatory receptors expressed on these MoDCs, and magnitude of T cell activation (n = 5). The color bar indicates the Pearson correlation coefficient value and stars represent statistical significance (* p < 0.05, ** p = 0.01, *** p = 0.001).

TLR7- and TLR8-activating immunostimulators differ in their ability to induce CD4⁺ and CD8⁺ T cell activation.

In addition to R848, an imidazoquinoline able to activate immune cells through both TLR7 and TLR8, the screen also included a TLR7-specific agonist (CL264) and a TLR8-specific agonist (TL8–506). Even though TLR7/8 function in adjuvant design and DC activation is considered interchangeable, we observed distinct differences between these TLR agonists. Some adjuvant combinations significantly induced CD4⁺ or CD8⁺ T cell activation. Regression analysis combined with Pearson correlation analysis between CD4 and CD8 induction of all combinations tested in this study, including TLR7, TLR7/8 and TLR8 combinations, suggests an overall weak correlation (Figure 2A). However, the relative ratio obtained from the frequencies of activated CD4⁺ and CD8⁺ T cells following stimulation with these TLR7, TLR7/8, and TLR8 agonists shows greater, statistically significant CD8/CD4 ratio for the TLR7 adjuvant, suggesting a higher degree of antigen-specific CD8 induction than CD4 induction (Figure 2B). In contrast, combinations containing the TLR8 adjuvant have a below average CD8/CD4 ratio. One interesting observation is that the combination of CL264 (TLR7 agonist) with Poly I:C (TLR3 agonist) (Figure 2 C–E) significantly induced both influenza HA-specific CD4⁺ and CD8⁺ T cell recall (Figure 2C). The production of CXCL10, a key chemokine secreted in response to a viral exposure, was increased when CL264 was combined with Poly I:C, along with enhanced response of pro-inflammatory cytokines, TNF, IL-12p70, and IL-6 (Figure 2 D–E). Production of these cytokines was also induced by CL264 alone and Poly I:C alone in newborns and older adults. Although Poly I:C + 3p-hpRNA induced statistically significant amounts of CD4 and CD8 activation, we did not observe any significant production of Th1 cytokines with this combination. MoDCs from newborns (cord blood, NB) and older adults (OA) were stimulated with Poly I:C and CL264 to assess age-specific differences in type and magnitude of adjuvantation (Figure 2 D–E). Notably, the magnitude of induction of most cytokines that were evaluated was greatest in newborns and decreased with age. However, the production of both IL12p40 and IL-12p70 was markedly increased in older adults compared to young adults. Despite differences with age, a preferential secretion of IL-12p70 and TNF over that of IL-12p40 and IL-6 was not observed. We hypothesized that such preferential induction of IL-12p70 and TNF over IL-12p40 and IL-6 would be of interest for further investigation, as it would likely result in a stronger promotion of Th1 immunity. Thus, we next interrogated the screening results in search for an adjuvant combination with potent CD4 and CD8 induction as well as potentially favorable TNF/IL-6 and IL12p70/IL-12p40 ratios.

Figure 2. TLR7- and TLR8-stimulating adjuvants differ in their ability to promote CD4 and CD8 immunity in the DTI assay.

A) Pearson correlation analysis demonstrates the degree of correlation between CD4⁺ and CD8⁺ T cell induction by all adjuvants and combinations tested (grey) (n = 5). CL264 (TLR7)-containing combinations (all biological replicates) are highlighted in red, R848 (TLR7/8)-containing combinations (all biological replicates) in purple, and TL8–506 (TLR8)-containing combinations (all biological replicates) in blue. N = 5. B) CD8/CD4 ratio of all TLR7, TLR8, or TLR7/8-containing combinations (all biological replicates). N = 5, Wilcoxon rank-sum test. C-E) CL264 + Poly I:C was identified as an adjuvant combination with significant induction of HA-specific CD4⁺ as well as CD8⁺ T cells. N = 5 for (C). Analysis of cytokines secreted by MoDCs showed age dependency in magnitude of production of TNF, IL-1β, IL-6, and IL-10 (reduced with age), as well as IL-12p70 (increased with age) (D/E). Heat maps represent mean expression (percent positive) of MoDC costimulatory receptors. N = 9 young adult (YA), n = 8 older adult (OA), and n = 3 newborn (NB) participants; Wilcoxon rank-sum test; * p < 0.05, ** p = 0.01, *** p = 0.001.

Two novel TLR9-activating combinations induce age-indiscriminate T cell activation as well as Th1-promoting cytokine induction.

Correlation analysis of CD4 and CD8 induction also led to the identification of two adjuvant combinations involving the TLR9 agonist CpG (ODN 2395). Correlation of TNF and IL-6, as well as IL-12p70 and IL-12p40 (Figure 3A), further revealed that one of these two combinations, CpG + 2,3-cGAMP, preferentially promoted production of the Th1-polarizing cytokines TNF and IL-12p70 over IL-6 and IL-12p40. Confirmation analysis in five additional young adult study participants showed that both CpG-containing combinations enhanced the activation of antigen-specific CD4⁺ and CD8⁺ T cells in a concentration-dependent manner, and the adjuvants in each combination were mathematically determined to act synergistically, using an adaptation of Loewe’s definition of additivity (Figure 3B–D) [11]. CpG + 3p-hpRNA did not shift the TNF/IL-6 or IL12p70/p40 balance but did induce a statistically significant amount of IFNα secretion as well as XCR1 expression in newborns and young adults, both hallmark features of professional cross-presenting DCs in vivo (Figure 3C). Further investigation into the age-specific ability to activate MoDCs revealed a distinct age-specific immune activation profile, with the greatest magnitude of cytokine induction in newborns and almost none in older adults. The combination of CpG and 2,3-cGAMP induced preferential production of TNF and IL-12p70, in addition to XCR1 expression and induction of greater levels of IFNα, IL-1β, IL-6, and IL-10 in newborns and older adults (Figure 3E). Among older adults, CpG + cGAMP induced significant production of TNF and IFNα. Because TLR9 expression in MoDCs is typically low, we quantified the expression of TLR9 by flow cytometry after stimulation. Indeed, TLR9 expression in untreated MoDCs was low by comparison, but was significantly increased by stimulation with either 3p-hpRNA or 2,3-cGAMP (Supplementary Figure 1), potentially contributing to the efficacy of these two CpG-containing adjuvant combinations.

Figure 3. CpG + 2,3-cGAMP induces CD4⁺ and CD8⁺ immunity and preferential induction of Th1-promoting cytokines in MoDC and DTI assays.

A) Pearson correlation analysis demonstrates the degree of correlation between CD4 and CD8 induction (left), MoDC IL-12p70 and IL-12p40 induction (middle), and TNF and IL-6 production (right) by all adjuvants and combinations tested (grey) (n = 5). Highlighted in blue are CpG + 2,3-cGAMP combination (all biological replicates) and in red are CpG + 3p-hpRNA combination (all biological replicates). B) CpG + 3p-hpRNA was identified as an adjuvant combination with significant induction of HA-specific CD4⁺ as well as CD8⁺ T cells. C) Analysis of cytokines and chemokines produced by MoDCs showed age dependency in magnitude of secretion of TNF, IL-6, IL-10, and IL-12p40 (decreased with age). Heat maps represent mean expression (percent positive) of MoDC costimulatory receptors. D) CpG + 2,3-cGAMP was identified as an adjuvant combination with significant induction of HA-specific CD4⁺ as well as CD8⁺ T cells. E) Analysis of cytokines and chemokines produced by MoDCs showed combination dependency in magnitude of secretion of TNF, IL-12p70, IFNα, IL-1β, and IL-12p40. Heat maps represent mean expression (percent positive) of MoDC costimulatory receptors. B,E) n = 9 young adult (YA), n = 8 older adult (OA), and n = 3 newborn (NB) participants; Wilcoxon rank-sum test; * p < 0.05, ** p = 0.01, *** p = 0.001. Synergy value of D<1 indicates synergy.

To further confirm induction of Th1 immunity by these TLR9-activating adjuvant combinations, we modified our DTI assay to quantify the percentage of total activated naïve (CD45RA⁺) CD4⁺ T cells in response to anti-CD3 and anti-CD28 stimulation. Indeed, upon stimulation with CpG + cGAMP and CpG + 3p-hpRNA, we obtained a greater frequency of T-bet-expressing Th1 cells than GATA3-expressing Th2 cells (Supplementary Figure 2). Untreated MoDCs did not show significant levels of differentiation into Th1 or Th2 cells following anti-CD3/28 stimulation. Following the addition of blocking antibodies against Th1-polarizing cytokines IL-12p70 and TNF at a concentration of 1 μg/mL, we observe higher frequency of GATA3-expressing Th2 cells. Although this finding was not statistically significant, the trend we observe may further support the ability of these CpG-containing combinations to induce the favorable Th1 response upon antigen exposure.

To test the specificity of T cell activation, we also stimulated the DC:T (1:10 ratio) cell co-culture with the HIV envelope protein, gp120, to which our study participants were immunologically naïve. In response to stimulation with CpG + cGAMP and CpG + 3p-hpRNA, we observe reactivation of only HA-specific CD4⁺ and CD8⁺ T cells and not gp120-specific CD4⁺ and CD8⁺ T cells (Supplementary Figure 3).

Antigen cross-presentation induced by TLR9-activating adjuvant combinations is Sec22b-dependent.

To confirm that both CpG-containing combinations identified in Figure 3 induced the differentiation of in vivo-like XCR1⁺ cross-presenting dendritic cells, we evaluated the role of Sec22b in the activation of antigen-specific CD8⁺ T cells. Sec22b is a SNARE protein residing in the endoplasmic reticulum-Golgi intermediate compartment, previously shown by us and others to be important for antigen cross-presentation on MHC I [34,39,40]. The role of Sec22b in CD8⁺ T cell activation was investigated by siRNA-mediated silencing in MoDCs from young adults prior to stimulation with adjuvants/HA antigen. After silencing the Sec22b gene in cells stimulated with CpG + 2,3-cGAMP and CpG + 3p-hpRNA, there was a statistically significant reduction in activation of influenza HA-specific CD8⁺ T cells (Figure 4), suggesting that Sec22b plays a critical role in priming a cytotoxic antiviral immune response. The percentage of CD25/CD154-expressing antigen-specific CD8⁺ T cells induced by CpG + 3p-hpRNA or CpG + 2,3-cGAMP was comparatively higher in untransfected MoDCs or MoDCs treated with mock siRNA, than in MoDCs treated with Sec22b siRNA.

Figure 4. Induction of cross-presentation by CpG + 3p-hpRNA (A) and CpG + 2,3-cGAMP (B) requires Sec22b in modified DTI assay.

Human monocytes were untreated, or treated with mock siRNA pool, or Sec22b-specific siRNA pool for 24 hours during MoDC generation before stimulation with CpG, 3p-hpRNA, 2,3-cGAMP, or combinations as indicated in the presence of HA protein antigen. After 24 hours of stimulation, MoDCs were co-cultured with autologous CD4⁺ or CD8⁺ T cells for five days. Magnitude of antigen-specific T cell activation was quantified as in Figures 1–3 (n = 5 young adults; Wilcoxon rank-sum test; * p < 0.05, ** p = 0.01, *** p = 0.001).

Discussion:

Herein, we employed for the first time in vitro assays that (a) can assess the extent of cross-presentation by quantification of the amount of CD8⁺ T cells activated after treatment of MoDCs with a protein antigen in the presence of candidate adjuvants, and (b) immunophenotype MoDCs to identify adjuvant combinations capable of inducing CD4⁺ as well as CD8⁺ immunity. We identified several adjuvant combinations that have not been previously described that possess the ability to significantly enhance the activation of antigen-specific CD4⁺ and CD8⁺ T cells.

Adjuvants can be critical tools which help protein-based vaccines elicit strong immunogenic response by directing the type and magnitude of the immune response. Thus, adjuvants can not only address the challenge of limited immunogenicity, but also offer a distinct advantage as a form of precision medicine, allowing for the tailoring of vaccines to the immunological needs of specific vulnerable populations [11,34,41–45]. In response to a foreign antigen, adjuvanted vaccines may activate PRRs that provide a repertoire to DCs to sense pathogen-derived ligands, inducing a DAMP-mediated pro-inflammatory cytokine and chemokine response at the intramuscular injection site [28,46]. This creates an immunocompetent environment locally, resulting in the recruitment of innate immune cells such as monocytes, macrophages, and neutrophils that process the antigen, transport it to the draining lymph node, and present it to T lymphocytes to activate the adaptive immune response [47,48]. An additional challenge for successful induction of antiviral immunity using protein-based vaccines is the priming of a CD8⁺ T cell response. For antigens from protein-based vaccines to be presented to CD8⁺ T cells on MHC I, they must undergo cross-presentation, a process which requires maturation of conventional DCs (cDCs) [49,50]. In vivo, certain sites such as the lungs and spleen are populated with XCR1⁺ cDC1 cells which specialize in cross-presentation [37,38,51]. Most vaccines, however, especially those commonly given to children, are administered intramuscularly where they predominantly encounter cDC2 cells which are poor cross-presenters, like the MoDCs used in this study.

A limitation of our study is that the MoDCs we isolated from cord blood to investigate the mechanism of action of adjuvant combinations in neonates may not be representative of post-natal immunity to infections [52]. However, we have previously identified a combination of adjuvants that can potently induce cross-presentation in MoDCs and induce protective CD8 immunity against RSV in newborn mice [34]. We have also shown that 3M-052, a TLR7/8 agonist, acted synergistically with pneumococcal conjugate vaccine (PCV) to stimulate the production of Th1-polarizing cytokines TNF, IFNγ, IL-12p70, and IL-6 from cord blood-derived leukocytes [30]. This 3M-052-adjuvanted PCV formulation induced a robust antibody response against Streptococcus pneumoniae in non-human primates. Another limitation of cord blood was that we could not include it in the DTI assay as newborns lack pre-existing immunity to influenza which was necessary for us to demonstrate the reactivation of influenza-specific T cells.

In addition to the identification of several candidate adjuvant combinations, we also observed several notable patterns across our screening dataset. Firstly, many adjuvant combinations induced the expression of XCR1 receptor on MoDCs, but only a select subset of these adjuvant combinations showed effective CD8⁺ T cell induction – suggesting that although XCR1 serves as a marker for cross-presenting cDC1 DCs in vivo, in vitro expression of this receptor does not guarantee cross-presentation. Secondly, we observed a difference in preference for CD4 and CD8 activation between combinations containing a TLR7- or a TLR8-activating adjuvant. This challenges the notion that TLR7- and TLR8-stimulating adjuvants can be used interchangeably. Since the screen included only one TLR7-specific and one TLR8-specific adjuvant, these results will need to be confirmed in a follow-up study that includes other adjuvants of these types. Thirdly, we observed a strong co-regulation of several cytokines and chemokines produced by MoDCs. As these included cytokines both desirable and undesirable for Th1 induction, we sought to identify adjuvant combinations capable of selectively uncoupling these cytokines in favor of those that promote Th1 development.

From our screening of different combinations of adjuvants, we found that the combination of CpG ODN 2395 (TLR9 agonist) and 2,3-cGAMP (activator of the STING pathway) was most effective in inducing the secretion of cytokines mediating Th1 and antigen-specific CD8 immune responses that are critical for defense against intracellular pathogens such as viruses. We also found that the knockdown of Sec22b inhibited the endoplasmic reticulum-phagosome pathway for antigen cross-presentation consistent with prior studies [40,53]. In addition to acting synergistically in vulnerable age groups such as newborns, this adjuvant combination also successfully uncoupled co-regulated cytokines such as TNF and IL-6. This new mechanistic insight into the use of combination adjuvants at the extremes of age when infection proneness is high may inform development of antiviral vaccines. However, our observations are limited to a small cohort of participants across the different age groups (n = 3 for newborns, n = 9 for young adults, and n = 8 for older adults) for the initial screen, and an additional n=5–8 for confirmation experiments in Figures 2 and 3.

Given that the adjuvant combination of CpG + 2,3-cGAMP can induce cross-presentation in cDC2-like MoDCs, future studies may include immunization of animals to investigate whether CpG + 2,3-cGAMP can safely and effectively induce antiviral immunity through intramuscular immunization and achieve similar cDC1-driven immunity as can be achieved through intranasal administration [54–57] – the route of immunization which is currently not recommended for populations highly susceptible to infections such as children younger than two years of age, adults older than 50 years of age, people with asthma, and pregnant women [58].

In summary, screening of key PRR-activating adjuvant combinations using human in vitro modeling has identified several novel adjuvant combinations with the potential to act synergistically for the induction of Th1- and CD8-driven immunity. The identification of these adjuvant combinations, particularly CpG + 2,3-cGAMP, provides important insights into the mechanisms underlying adjuvant combination and may inform future development of vaccine formulations tailored for vulnerable populations such as newborns and older adults.

Highlights.

1. Human in vitro modeling identifies adjuvants that activate CD4⁺ and CD8⁺ T cells.

2. Several adjuvant combinations induced varying immunophenotypic responses.

3. TLR7/8-stimulating adjuvants differed in their propensity to induce T cells.

4. CpG + 2,3-cGAMP was most potent in inducing cDC1-like dendritic cells.

5. CpG + 2,3-cGAMP is a promising adjuvant candidate for vaccine development tailored towards vulnerable populations.