Research ArticleImmunology

Select hyperactivating NLRP3 ligands enhance the TH1- and TH17-inducing potential of human type 2 conventional dendritic cells

See allHide authors and affiliations

Science Signaling  27 Apr 2021:
Vol. 14, Issue 680, eabe1757
DOI: 10.1126/scisignal.abe1757

Dendritic cells in overdrive

Upon detection of danger signals, such as molecules associated with infectious pathogens, dying cells, or tumor cells, antigen-presenting cells called dendritic cells initiate adaptive immune responses. This response is enhanced through inflammasome signaling that induces cytokine secretion but usually leads to death of the activated cells. Hatscher et al. analyzed human primary DCs from the blood, spleen, and thymus and found that type 2 conventional dendritic cells (cDC2) uniquely did not die after inflammasome-induced cytokine secretion. Instead, these cells entered a “hyperactive” state that enhanced the stimulation of certain T helper cell subsets. The findings suggest that cDC2 could be critical to the efficacy of vaccines and immunotherapies as well as for therapeutically controlling inflammation.

Abstract

The detection of microorganisms and danger signals by pattern recognition receptors on dendritic cells (DCs) and the consequent formation of inflammasomes are pivotal for initiating protective immune responses. Although the activation of inflammasomes leading to secretion of the cytokine IL-1β is typically accompanied by pyroptosis (an inflammatory form of lytic programmed cell death), some cells can survive and exist in a state of hyperactivation. Here, we found that the conventional type 2 DC (cDC2) subset is the major human DC subset that is transcriptionally and functionally poised for inflammasome formation and response without pyroptosis. When cDC2 were stimulated with ligands that relatively weakly activated the inflammasome, the cells did not enter pyroptosis but instead secreted IL-12 family cytokines and IL-1β. These cytokines induced prominent T helper type 1 (TH1) and TH17 responses that were superior to those seen in response to Toll-like receptor (TLR) stimulation alone or to stronger, classical inflammasome ligands. These findings not only define the human cDC2 subpopulation as a prime target for the treatment of inflammasome-dependent inflammatory diseases but may also inform new approaches for adjuvant and vaccine development.

INTRODUCTION

Dendritic cells (DCs) are potent antigen-presenting cells and required for the initiation of strong adaptive immune responses (1). Upon patrolling in peripheral tissues and continuously taking up antigens from their surroundings, DCs migrate into lymphoid tissues, where they present the processed antigens to T cells through the formation of peptide:major histocompatibility (MHC) complexes. Depending on the presence or absence of additional danger signals—such as pathogen-associated molecular patterns (PAMPs) or danger-associated molecular patterns (DAMPs)—during antigen uptake, DCs either induce adaptive immune responses or maintain peripheral tolerance. The first process is supported by a high abundance of costimulatory molecules and activation markers on the DC surface and the secretion of cytokines shaping the immune response (26).

DCs can be subdivided into plasmacytoid DCs (pDCs) and conventional DCs (cDCs) (2, 7). pDCs are specialized in responding to viral infections through rapid secretion of type I interferons (811). cDCs can be further separated into migratory and/or tissue-resident cDC1 and cDC2 (12). Former classification of DCs solely occurred on surface marker expression. Now, combined phenotypic and transcriptomic analysis allows redefining murine CD11bCD8+ and CD11bCD103+ as interferon regulatory factor 8 (IRF8)– and basic leucine zipper ATF-like transcription factor 3 (BATF3)–dependent cDC1. Equivalently, murine CD11b+CD8 as well as CD11b+CD103 and CD11b+CD103+ DCs can be described as IRF4-dependent cDC2 (2, 7, 12, 13). cDC2 are further composed of a Notch2+ and a Krüppel-like factor 4+ subset (1416). Murine cDC1 play a central role in immunity against intracellular pathogens and cancer by preferably inducing CD8+ and T helper type 1 (TH1) CD4+ T cell responses, whereas murine cDC2 preferentially induce TH2 and TH17 responses (2, 13, 1723).

Human cDC1 share transcriptional similarities with murine cDC1 (4, 24, 25) and are characterized by the expression of the surface markers CD141, C-type lectin domain containing 9A (CLEC9A), and X-C motif chemokine receptor (XCR1), whereas human cDC2 are transcriptionally related to murine cDC2 and are positive for the surface markers CD1c and CLEC10A (2630). Human cDC1, believed to be the counterpart of murine CD11bCD8+ cDC1, are thought to be specialized in cross-presentation and, thus, the induction of CD8+ T cells (3136). They further produce type III interferons and are resistant to viral infection (32, 34, 37, 38). In contrast, human cDC2 produce interleukin-12 (IL-12) and preferentially induce TH17-skewed CD4+ T cells in response to extracellular pathogens, combined bacterial and viral Toll-like receptor (TLR) ligands, or bacterial toxins, therefore most probably resembling murine CD8CD11b+ cDC2 (22, 25, 3945).

DCs sense DAMPs and PAMPs by pattern recognition receptors (PRRs), which mediate an essential step in DC activation and DC-mediated immunity (46). Nucleotide-binding oligomerization domain-like receptors (NOD)–like receptors (NLRs) are important cytosolic PRRs whose activation leads to formation of the inflammasome (47). Inflammasomes are multiprotein platforms consisting of a sensor molecule (mostly an NLR) that is connected to caspase-1 through an adaptor molecule called apoptosis-associated speck-like protein containing a CARD (ASC) [also known as PYD and CARD domain containing (PYCARD)] (48). Inflammasome formation requires a TLR-dependent priming step and a second activation step leading to caspase-1 activation and the subsequent cleavage of pro–IL-1β and pro–IL-18 into their mature forms (49). The priming step induces transcription and translation of the inflammasome-dependent cytokines IL-1β and IL-18 and different components of the inflammasome, such as sensor proteins and caspase-1 (50). The subsequent activation step is typically initiated by ligands targeting specific NLRs, such as NLRP3 (NOD-, leucine-rich repeat– and pyrin domain–containing protein 3) by adenosine triphosphate (ATP), monosodium urate (MSU) crystals, nigericin, or muramyl dipeptide (MDP), NLRP1 by Bacillus anthracis lethal toxin, or NLRC4 by cytosolic flagellin (51). Activation of the inflammasome generally leads to pyroptosis, an inflammatory form of programmed cell death caused by caspase-1–mediated cleavage of gasdermin D (52, 53). Pyroptosis has also been described as required for the secretion of active IL-1β (54). This is in contrast to studies indicating that certain stimuli such as lipopolysaccharide (LPS), peptidoglycan, oxidized phospholipids, or Salmonella typhimurium infection induce IL-1β release from living cells, which therein included murine bone marrow–derived DCs (BMDCs), murine bone marrow–derived macrophages (BMDMs), human (CD14+) and murine monocytes, murine bone marrow–derived lymphocyte antigen 6 complex, locus G (Ly6G)+ neutrophils, murine Ly6C+/Ly6G+ neutrophils, and mouse embryonic fibroblasts (5559). Moreover, IL-1β is secreted along with other cytokines, such as IL-6 and tumor necrosis factor α (TNFα), in a cellular state termed “hyperactivation” (52, 60). These findings were consistent with the fact that these hyperactive BMDCs were still capable of inducing T cell responses. Whether a cell initiates pyroptosis or switches into the state of hyperactivation is a matter of current research. Two mechanisms for the induction of hyperactivation have been proposed: Evavold et al. (61) show that gasdermin D–mediated pore formation is necessary for IL-1β release from living macrophages, whereas Carty et al. (62) demonstrate that hyperactivating ligands do not activate the Toll–IL-1R protein sterile α and HEAT armadillo motif-containing protein, which induces pyroptosis through mitochondrial depolarization .

There is compelling evidence that inflammasome activation in DCs contributes to the adjuvanticity of vaccines, resistance to infections, and antitumor responses (6367). However, the role of single human DC subsets during inflammasome activation has not been determined yet. By investigating the ability of primary human DC subsets to respond to inflammasome stimulation, we found that inflammasome activation and hyperactivation are special features of the cDC2 subset, thus identifying them as a new possible target for specific DC-based vaccination protocols and for anti-inflammatory treatment options.

RESULTS

cDC2 produce inflammasome-dependent IL-1β in response to extracellular PAMPs

To investigate the capability of human primary DCs to respond to inflammatory stimuli, we analyzed cytokine and chemokine secretion upon stimulation with a variety of PAMPs and heat-killed bacteria. After negative enrichment, we used fluorescence-activated cell sorting (FACS) to sort living, single, nonautofluorescent, lineage (CD3, CD19, CD20, CD56, and CD335)–negative (Lin), human leukocyte antigen–DR isotype (HLA-DR)+ fresh human cDC1, cDC2, and pDCs from human peripheral blood, spleen, and thymus organ donations (fig. S1). Next, we stimulated the sorted DCs with select TLR agonists, namely, LPS, Pam3CSK4, flagellin, polyinosinic:polycytidylic acid (pIC), R848, CL264, single-stranded polyuridylic acid (ssPolyU), and CpG, as well as heat-killed extracellular pathogens (Escherichia coli and Staphylococcus aureus) and the yeast component zymosan and measured IL-1β and IL-8 secretion (Fig. 1A). Our data indicate that cDC2 were superior in IL-1β secretion upon stimulation with heat-killed S. aureus or E. coli, as well as with zymosan. Other proinflammatory cytokines, such as IL-8, displayed a subset- and tissue-independent secretion (Fig. 1B). Thus, these data indicate that primary cDC2 are the only DC subset capable of secreting IL-1β upon incubation with pathogens containing several distinct PAMPs.

Fig. 1 Among primary human DCs, only cDC2 produce IL-1β in response to extracellular PAMPs.

(A to C) cDC1, cDC2, and pDCs were sorted from PBMC, splenic, or thymic single-cell suspensions of otherwise healthy donors and treated with LPS (1 μg/ml), Pam3CSK4 (0.5 μg/ml), flagellin (2.5 μg/ml), pIC (5 μg/ml), R848 (5 μg/ml), CL264 (5 μg/ml), ssPolyU (5 μg/ml), CpG (5 μg/ml), heat-killed E. coli (50 μg/ml), heat-killed S. aureus (50 μg/ml), zymosan (25 μg/ml) particles (A and B), or R848 (5 μg/ml) (C) for 3 hours, followed by L18-MDP (0.5 μg/ml), 5 mM ATP, or poly(dA:dT) (5 μM/ml). As a control, cells were cultured with medium. After 12 hours, supernatants were analyzed for IL-1β (A and C) or IL-8 (B) by CBA Flex Set (BD Biosciences). Bar graphs show means ± SD for three individual donors.

IL-1β secretion is dependent on inflammasome formation. Because zymosan and heat-killed S. aureus are known inducers of inflammasome-dependent IL-1β secretion in murine BMDCs and human monocyte-like THP-1 cells (68, 69), we next investigated whether specific inflammasome activators are capable of inducing IL-1β secretion in primary DCs. Because inflammasome assembly is dependent on two subsequent triggers, we first primed peripheral blood cDC1, cDC2, and pDCs with the TLR7 and TLR8 ligand R848 that we and others (30, 70, 71) identified as potent stimulator of all three human DC subpopulations, followed by activation of the inflammasome with commercially available inflammasome ligands. These included L18-MDP (targeting NLRP1 and/or NLRP3), ATP (targeting NLRP3), and poly(deoxyadenylic-deoxythymidylic) acid poly(dA:dT) [targeting the interferon-inducible protein absent in melanoma 2 (AIM2)] (Fig. 1C). Consistent with our results described above, we found that the IL-1β secretion into the supernatant was fully restricted to cDC2 and excelled after treatment with ATP. Our data further revealed that cDC1 released only very small amounts of IL-1β after stimulation with the AIM2 ligand poly(dA:dT), whereas pDCs did not secrete IL-1β at all (Fig. 1C). To exclude that unresponsiveness of cDC1 and pDCs to inflammasome stimulation was due to improper priming of the cells, we checked the expression of TLRs in a published microarray dataset of human primary DCs [Gene Expression Omnibus (GEO) accession number GSE77671]. Both cDC1 and pDCs showed the expression of TLR8 and TLR7, respectively, which should render them responsive to R848 treatment (fig. S2A). cDC1 further showed high mRNA expression of TLR3, whereas pDCs expressed high levels of TLR9 mRNA (fig. S2A). We therefore stimulated cDC1 and pDCs with different ligands for intracellular TLRs and analyzed the expression of costimulatory molecules (CD40 and CD86) and the lymph node–homing chemokine receptor CCR7 by flow cytometry. As expected from the microarray data, cDC1 reacted to TLR3 (pIC) and TLR8 (R848 and ssPolyU) stimulation with enhanced expression of CCR7, CD40, and CD86 (fig. S2B), whereas pDCs responded to TLR7 (R848 and CL264) and TLR9 (CpG) stimulation (fig. S2C). To test whether stimulation with another TLR ligand would induce inflammasome responses, we primed cDC1 and pDCs with pIC and CpG, respectively, followed by activation of the inflammasome with L18-MDP, ATP, or poly(dA:dT) (fig. S2, D and E). Comparable to priming with R848, neither cDC1 (fig. S2D) nor pDCs (fig. S2E) secreted substantial amounts of IL-1β. Thus, our data suggest that, compared to cDC1 and pDCs, cDC2 represent the superior DC population capable of secreting IL-1β upon combined TLR and inflammasome stimulation.

Inflammasome subunits are predominantly expressed in cDC2

To understand in more detail why only the cDC2 subset has the capacity to secrete IL-1β upon inflammasome stimulation, we next sought to elucidate the expression of different genes of the inflammasome pathway among the three human DC subsets (fig. S3A). Therefore, previously published transcriptome data of steady-state cDC1, cDC2, and pDCs sorted from the human peripheral blood, spleen, and thymus were reanalyzed in closer detail for molecules important for inflammasome formation (4) (GEO accession number GSE77671). In a first step, the data were normalized, and only genes expressed in at least one group (present probes) were further analyzed. Then, key genes of the different inflammasome subtypes were selected and displayed in a heatmap analysis. In line with the results described above (Fig. 1), the heatmap illustrates the expression differences of the investigated inflammasome-related genes of all three DC subsets, which were expressed largely independent of their tissue origin (fig. S3A). These data clearly demonstrate that cDC2 are the major DC subset expressing genes encoding the inflammasome sensor proteins NLR family apoptosis inhibitory protein (NAIP), NLRP1, NLRP3, NLRP12, NLRC4, and pyrin (gene MEFV) as well as the effector caspase-1, caspase-4, and caspase-5, which are needed for the cleavage of pro–IL-1β into active IL-1β. We further found that NOD2 showed a specific mRNA expression in cDC2, whereas NOD1 mRNA was also present in pDCs (fig. S3A). Consistent with the capability to react to the AIM2 inflammasome ligand poly(dA:dT), our data revealed that AIM2 was the only inflammasome component gene that was increased in cDC1. Moreover, we found that the mRNA of IL-1β was already expressed in steady-state cDC2 but not in the other DC subpopulations (fig. S3A). Thus, our data indicate a predominant and preformed expression of inflammasome components in primary cDC2.

Because inflammasome formation and activation usually require a TLR priming step leading to up-regulation of genes involved in inflammasome pathways, we next aimed to identify whether genes involved in inflammasome formation or cytokine secretion in human primary DCs might be transcriptionally regulated in response to TLR stimulation. Because pDCs were incapable of secreting IL-1β in response to the two-step inflammasome activation (Fig. 1), we excluded them from further analyses. Thus, we FACS-sorted peripheral blood cDC1 and cDC2 (from three blood donors each) and either incubated them with R848 or left them untreated. The transcriptional profiles of 800 genes (770 genes from the nCounter Human Myeloid Innate Immunity V2 Panel and 30 DC- and inflammasome-specific genes) were analyzed from cDC1 and cDC2 RNAs by NanoString technology. After a quality control step, the NanoString data were analyzed and exported using nSolver software (version 4.0). Then, the remove unwanted variation (RUVg) method was applied to remove the batch effect. Afterward, data were normalized using NanoStringNorm according to 14 control genes and 40 housekeeping spikes genes. The result was examined using a principal components analysis plot, revealing four distinct clusters resembling either cDC2 or cDC1 in steady state or after priming with R848 (Fig. 2A). Here, the principal component 1 (PC1) and PC2 distinguished the DC subpopulations and separated steady state from R848-stimulated cells (Fig. 2A).

Fig. 2 Inflammasome subunits are predominantly expressed in cDC2.

(A to C) cDC1 and cDC2 were sorted from PBMCs of three donors and were either stimulated with R848 (5 μg/ml) for 3 hours or left in medium as control. mRNA expression of 800 genes was determined using NanoString technology (770 genes from the nCounter Human Myeloid Innate Immunity V2 Panel and 30 DC- and inflammasome-specific genes). (A) Twelve samples of 800 gene features were normalized by RUVg and clustered using principal components analysis. (B and C) mRNA data were then analyzed using edgeR with a threshold of a FDR-adjusted P value of 0.01 and a log2FC of 1.5. The volcano plots (B) show all relative expression of 800 genes either between steady-state cDC1 and cDC2 (top graph) or R848-primed cDC1 and cDC2 (bottom graph). Genes differentially expressed were highlighted by colors (enriched in cDC1, orange; enriched in cDC2, red), and key inflammasome genes were labeled dependent on their expression (significantly enriched in cDC1, orange; significantly enriched in cDC2, red; not significantly, gray). The heatmap (C) shows expression of key inflammasome genes for human cDC1 and cDC2 under steady-state conditions and after priming with R848.

In a first step of comparison, we identified differentially expressed genes (DEGs) between cDC1 and cDC2, both in steady state and after stimulation with R848, as well as expression of genes that were specifically enhanced or attenuated in cDC1 and cDC2 after stimulation with R848 (fig. S4A). Our heatmap analysis revealed that, in total, 343 of the 800 genes were differentially expressed. In both steady state and after stimulation with R848, cDC1 showed selective expression of transcripts of cDC1 signature genes, such as XCR1, CADM1, C1ORF54, ZNF366, CLNK, BTLA, and IRF8, whereas cDC2 expressed cDC2 signature genes, such as FCER1A, FCGR2B, CLEC10A, and CD1C (fig. S4, A and B). Stimulation with R848 induced a common activation program in cDC1 and cDC2 with enhanced transcription of genes coding for TLR and nuclear factor κB (NF-κB) signaling, including cytokines (such as encoded by IL6, IL1B, IL23A, IL12B, IL15, and TNF), chemokines (as encoded by CXCL1, CXCL8, and CCL3 to CCL5), and activation markers or costimulatory molecules (as encoded by CD40, CD80, CD83, CD86, and CCR7) (fig. S4). The expression of only a few genes was attenuated after stimulation with R848, including those encoding the inhibitory receptors hepatitis A virus cellular receptor 2 (HAVCR2) [also known as T cell, immunoglobulin, mucin domain-containing molecules (TIM-3)] and V-set immunoregulatory receptor (VSIR) [also known as V-domain immunoglobulin suppressor of T cell activation (VISTA)], which are known to negatively influence T cell responses (fig. S4A). Thus, both cDC1 and cDC2 were activated upon stimulation with the TLR agonist R848.

Because we were interested whether inflammasome-related genes were enriched in either of the DC subpopulations and whether their expression would change under inflammatory conditions, we plotted all 800 genes in a volcano plot comparing cDC1 and cDC2 in steady state and after R848 stimulation and marked key inflammasome genes (Fig. 2B). As shown in Fig. 2B, we found that under both steady-state and inflammatory conditions (the latter referring to the presence of R848), mRNA transcripts encoding most sensor proteins—namely, PYRIN (MEFV), NLRP3, NLRP12, and NOD2, as well as the effector caspases caspase-1, caspase-4, and caspase-5 and the inflammasome-dependent cytokines IL1B and IL18—were selectively enriched in cDC2 (Fig. 2B). Only AIM2 was present among the identified DEGs that were more abundantly expressed in cDC1 under steady-state conditions (Fig. 2B). We next analyzed the expression dynamics of these key inflammasome genes by heatmap to evaluate whether these molecules were transcriptionally induced in cDC2 or cDC1, respectively, upon priming with the TLR ligand R848. Our data revealed that although the expression of IL1B was enhanced after stimulation with R848, that of sensor proteins or effector caspases was not in cDC1 (Fig. 2C). In contrast to this, R848 treatment of cDC2 resulted in the transcriptional induction of PYRIN (MEFV), NLRP3, IL1B, IL18, the mRNA encoding adaptor protein ASC (PYCARD), and that encoding the pore-forming protein gasdermin D (GSDMD). Our data further revealed that the expression of NLRC4 and NAIP was the only two among inflammatory sensor–encoding genes that were transcriptionally attenuated upon R848 stimulation in cDC2 (Fig. 2C). We validated our mRNA data by reanalyzing published microarray data of cDC1 and cDC2 that were sorted from R848- or control-treated humanized mice (GSEE99666) (70). In accordance with our data, both cDC1 and cDC2 exhibited increased expression of IL1B. Although cDC2 displayed increased expression of sensor proteins (encoded by NLRP3, NOD2, and MEFV) and gasdermin D, cDC1 did not transcriptionally promote sensor protein or effector caspase expression under stimulatory conditions (fig. S3B). In conclusion, we found that many inflammasome sensor proteins, effector caspases, and target cytokines were differentially enriched in cDC2 at the transcriptional level, both under steady-state and inflammatory conditions. Moreover, expression was further enhanced after TLR priming with R848 in cDC2, but not in cDC1, thus suggesting that cDC2 are the superior DC subset capable of responding to invading pathogens by a preformed and further promotable expression of inflammasome-related proteins.

Bona fide cDC2 induce inflammasome responses

Human CD1c+ DCs are shown to consist of bona fide cDC2 and DC3, which are transcriptionally related to monocytes (72, 73). Because mounting of inflammasome responses is a feature associated with monocytes (74), we wondered whether our as cDC2 defined cell population might contain DC3 that contribute to the observed IL-1β production. Therefore, we reanalyzed published single-cell RNA sequencing (scRNA-seq) data of human blood DCs and monocytes for the expression of inflammasome genes (GEO dataset GSE94820). CD1c+ DCs identified in the scRNA-seq data consisted of two clusters (fig. S5A). Heatmap analysis of known genes enriched in either cDC2 or DC3 revealed that the cluster represented cDC2 and DC3 (fig. S5B). We then compared the expression of key inflammasome genes between cDC2 and DC3 (fig. S5, C to D). Both cDC2 and DC3 showed a comparable expression of inflammasome genes, which was validated using a gene expression correlation test (fig. S5, C to D). Although the expression of NLRP12 and IL1B was slightly enriched in DC3 and MEFV, NLRC4, and IL18 in cDC2, the other inflammasome genes showed a linear correlation (fig. S5D). To exclude that our sorted cDC2 contain DC3, we analyzed sorted cDC2 for surface marker shown to be expressed on either cDC2 (CD5) or DC3 (CD163) (72, 73). Flow cytometric analysis revealed that our sorted cDC2 showed the expression of CD5 on the majority of the cells, whereas CD163 was absent (fig. S5, E to F). We conclude that the sorted cDC2 in our experimental system do not contain DC3, thereby supporting that bona fide cDC2 are able to induce inflammasome responses.

The nonclassical inflammasome inducer L18-MDP provokes a cDC2 cytokine profile that resembles a state of hyperactivation

Inflammasome activation is not necessarily accompanied by pyroptosis. Therefore, cells can retain their viability upon inflammasome stimulation and IL-1β secretion. This specific state is termed hyperactivation (60, 61). Because hyperactivation might improve immune responses as described for murine BMDMs and BMDCs, we wondered whether human cDC2 can enter a state of hyperactivation (59, 60). To identify inflammasome ligands that lead to DC hyperactivation and, therefore, to an augmented DC-mediated immune response, we performed inflammasome activation assays. For this, we primed FACS-sorted peripheral blood cDC2 with the potent TLR7 and TLR8 ligand R848 and incubated the cells with select inflammasome ligands that had been described by others (47, 75). These include crystalline structures (MSU) and ionophores (nigericin) (Fig. 3). To determine whether hyperactive viable cells are able to secrete other proinflammatory cytokines along with IL-1β, we measured various cytokines using cytometric bead array (CBA) in the harvested supernatants 12 hours after inflammasome stimulation. Our analyses revealed that stimulation with the classical inflammasome activators ATP, nigericin, and MSU led to the production of massive amounts of IL-1β. Furthermore, IL-18 secretion was mainly observed after stimulation with ATP and nigericin and seemed to be independent of TLR priming as shown before (Fig. 3) (76). The inflammasome ligands ATP and nigericin rendered cDC2 incapable of producing T cell–polarizing cytokines such as IL-12p40, IL-12p70, and IL-23 (Fig. 3). In contrast to classical inflammasome ligands, L18-MDP, a lipidated form of MDP that improves uptake of the molecule, did not only induce a robust IL-1β response but also enhanced the secretion of the T cell–polarizing cytokines IL-12p40 and IL-23 (Fig. 3). However, the secretion of other IL-1 family members, such as IL-1α and IL-18, was not affected. Incubation with the NLRP3 inflammasome ligand MSU provoked a cytokine profile that laid between classical inflammasome (ATP and nigericin) and L18-MDP stimulation, because MSU triggered strong IL-1β secretion comparable to ATP and nigericin while retaining the ability to secrete T cell–polarizing cytokines such as IL-12p40 and IL-23. However, in contrast to L18-MDP, MSU did not enhance the secretion of IL-12p40 and IL-23 compared to R848 stimulation alone. Together, these data suggest that L18-MDP and MSU, in contrast to ATP or nigericin, might induce hyperactivation of cDC2, because IL-1β production of cDC2 was accompanied by an enhanced (L18-MDP) or at least unaltered (MSU) secretion of TH1- and TH17-polarizing cytokines.

Fig. 3 The nonclassical inflammasome inducer L18-MDP provokes a cDC2 cytokine profile that resembles a state of hyperactivation.

cDC2 were sorted from PBMCs of healthy donors and primed with R848 (5 μg/ml) or left untreated as control. After 3 hours, cells were either stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, or 5 μM nigericin or left untreated in medium. A further 12 hours later, supernatants were analyzed for IL-1β, IL-1α, IL-18, IL-12p40, IL-12p70, and IL-23 by LEGENDplex Human Cytokine Panel 2 (BioLegend). Bar graphs show means ± SD for three to eight individual donors. *P < 0.05 and **P < 0.01 by Mann-Whitney U test in GraphPad Prism.

Inflammasome-dependent cytokine secretion induced by L18-MDP in human cDC2 is mediated by caspase-1 and NLRP3

The mechanism of inflammasome activation by MDP is still not fully understood. Several pathways have been proposed, including the activation of NLRP3 alone, a combined activation of NLRP3 and NOD2, or the activation of a complex of NLRP1 and NOD2 (7779). However, two studies suggest that MDP, as well as L18-MDP–mediated inflammasome activation and IL-1β production, does not depend on NLRP1 but rather depends on NLRP3 (80, 81). Besides inflammasome induction, MDP-mediated NOD2 activation is well defined (82). To identify the cellular mechanism by which L18-MDP induces inflammasome formation in cDC2, we performed inflammasome activation assays in the presence or absence of specific inhibitors for caspase-1 and NLRP3. We therefore treated R848-primed cDC2 with the caspase-1 inhibitor ac-yvad-cmk (83) or the NLRP3 inhibitor MCC950 (8486) before and while activating the inflammasome with L18-MDP, MSU, or ATP. Afterward, we compared the results of L18-MDP with the known NLRP3 inducer ATP and MSU. Our data revealed that the IL-1β secretion by all three ligands was strongly diminished by both inhibitors (Fig. 4A). L18-MDP–induced IL-1β production was almost completely abolished after inhibition of NLRP3. In addition, IL-18 production by ATP was also inhibited effectively (Fig. 4A). Because the activation of the inflammasome using ATP, in contrast to L18-MDP and MSU, did not induce any other cytokine than the inflammasome-dependent cytokines IL-1β and IL-18 (Fig. 3), we wondered whether blocking the NLRP3 inflammasome, and thereby potentially pyroptosis, might enable cDC2 to regain this capacity. We found that the inhibition of the NLRP3 inflammasome did not restore secretion of proinflammatory, T cell–polarizing cytokines by cells treated with ATP (Fig. 4A). In addition, in case of L18-MDP and MSU stimulation, other non–inflammasome-dependent cytokines, such as IL-1α, IL-12p40, IL-12p70, and IL-23, were not affected by either caspase-1 (ac-yvad-cmk) or NLRP3 (MCC950) inhibition (Fig. 4A). Thus, our data suggest that inflammasome activation and subsequent IL-1β and IL-18 secretion by L18-MDP, MSU, and ATP in cDC2 require NLRP3 and caspase-1, whereas other proinflammatory cytokines do not.

Fig. 4 Inflammasome-dependent cytokine secretion by L18-MDP in human cDC2 is predominantly mediated by caspase-1 and NLRP3.

(A and B) cDC2 were sorted from PBMCs of healthy donors and pretreated with (A) ac-yvad-cmk (300 μg/ml) or 10 μM MCC950 or (B) 15 μM GSK717 or medium containing DMSO. After 1 hour, cells were either primed with R848 (5 μg/ml) or left unprimed as control and were subsequently stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), or 5 mM ATP or left untreated in medium in the presence or absence of (A) ac-yvad-cmk (300 μg/ml) or 10 μM MCC950 or (B) 15 μM GSK717 or medium containing DMSO. After 12 hours, supernatants were analyzed for IL-1β, IL-1α, IL-18, and IL12p40 by LEGENDplex Human Cytokine Panel 2 (BioLegend). Bar graphs show means ± SD for four to five individual donors. *P < 0.05 by Mann-Whitney U test in GraphPad Prism. n.s., not significant.

Next, we wanted to determine whether NOD2 activation is also important for inflammasome activation and subsequent IL-1β secretion in cDC2. For this, we made use of the NOD2-specific inhibitor GSK717 (80). We found that NOD2 inhibition did not affect the secretion of IL-1β, which contests the role of NOD2 in L18-MDP–induced inflammasome activation (Fig. 4B). In summary, our data suggest that NLRP3 and caspase-1 play a pivotal role in L18-MDP–mediated inflammasome activation in human cDC2.

Human cDC2 simultaneously express IL-1β and IL-12 and retain their viability and costimulatory capacity upon L18-MDP–induced inflammasome activation

Lactatdehydrogenase (LDH) is an enzyme that is rapidly released into the supernatant after pyroptosis and can be used for quantification of cytotoxicity of any given substance or stimulus (87). Thus, LDH release assays are applied to detect cell death in the context of hyperactivation (5961). To confirm whether L18-MDP and MSU stimulation truly induce hyperactivation of cDC2, we performed LDH release assays. Therefore, we primed cDC2 with R848 and then treated primed DCs with the inflammasome ligands L18-MDP, MSU, ATP, or nigericin. We found that cDC2 retained their viability after stimulation with L18-MDP, because only a marginal release of LDH into the supernatant was detectable, whereas ATP and nigericin induced prominent cell death (Fig. 5A). Unexpectedly, MSU, although defined as a classical strong NLRP3 inducer, caused only moderate LDH release in cDC2 comparable to R848 stimulation alone (Fig. 5A). Thus, our data indicate that L18-MDP and MSU treatment does not induce significant cell death in contrast to classical inflammasome ligands such as ATP or nigericin.

Fig. 5 Human cDC2 simultaneously express IL-1β and IL-12 and retain their viability and costimulatory capacity upon L18-MDP–induced inflammasome activation.

(A and B) cDC2 were sorted from PBMCs of healthy donors and primed with R848 (5 μg/ml) or were left unprimed as control. After 3 hours, cells were stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, or 5 μM nigericin or left untreated in medium. After 12 hours, (A) cell viability was assessed by measuring LDH release into the supernatant (A), or cDC2 were analyzed for their expression of the costimulatory molecules CD40 and CD86 by flow cytometry (B). Bar graphs show means ± SD for (A) six and (B) four to five individual donors. (C) cDC2 sorted from PBMCs of healthy donors were primed with R848 (1 μg/ml) for 3 hours and then stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, or 5 μM nigericin or left untreated in medium in the presence of brefeldin A. After a further 6 hours, cells were fixed and stained for IL-1β and IL-12p40. Top: Dot plots show the gating for IL-1β and/or IL-12p40–positive cells for a representative donor. Bottom: The results are summarized and represented in scatter plots showing the mean of four different donors. *P < 0.05, **P < 0.01, and ***P < 0.001 by one-way ANOVA in GraphPad Prism. MFI, mean fluorescence intensity.

Because costimulatory molecule expression is a prerequisite for the induction of T cell responses by DCs, we further examined whether cDC2 showed increased expression of CD40 and CD86 after stimulation of the NLRP3 inflammasome. Therefore, cDC2 were primed with R848 and were subsequently stimulated with ligands targeting the NLRP3 inflammasome. After 12 hours, surface expression of the costimulatory molecules CD40 and CD86 was determined using flow cytometry (Fig. 5B and fig. S6A). Our data revealed that the expression of CD40 excelled after stimulation with L18-MDP and R848 (fig. S6A), although the difference to TLR simulation alone was not significant (Fig. 5B). Treating cDC2 with MSU seemed to have only minor, nonstatistically significant effects on the expression of CD40 (Fig. 5B). In contrast to L18-MDP and MSU, only low numbers of living DCs could be recovered in the presence of ATP or nigericin, and the remaining cells did not show increased CD40 surface expression (Fig. 5B). Regarding the effect of the different ligands, CD86 surface expression was similar to CD40 (fig. S6A), although CD86 expression was less strongly influenced by administration of R848 (Fig. 5B). Thus, we conclude that, in contrast to the TLR priming step, inflammasome activation itself has no additional activation effect on the surface expression of costimulatory molecules.

Because hyperactivation is defined by the secretion of IL-1β along with other proinflammatory cytokines from the same living cell, we wanted to exclude that IL-1β is secreted by few cells undergoing pyroptosis, whereas the remaining living cDC2 secreted the other proinflammatory cytokines. We therefore primed cDC2 with R848 for 3 hours and activated the cells with inflammasome ligands for a further 6 hours in the presence of brefeldin A to block secretion of produced cytokines. Next, we performed intracellular cytokine staining for IL-1β and IL-12p40 and performed intracellular flow cytometry analyses (Fig. 5C). Consistent with the data from the LDH assay, cDC2 stimulated with L18-MDP in addition to R848 priming showed the same morphology and live/dead profile as cells that were primed with R848 alone (fig. S6B). In contrast to that, cDC2 stimulated with ATP showed a change in morphology and a higher percentage of dead cells, whereas nigericin seemed to induce even faster cell death as markedly fewer cells could be recovered for analysis (fig. S6B). When we then analyzed the living cDC2 for production of cytokines, we found that living cDC2 were generally capable of simultaneously expressing IL-1β and IL-12p40 (Fig. 5C). Stimulation with L18-MDP induced significant higher amounts of IL-1β and IL-12p40 double-positive cDC2 compared to every other condition (Fig. 5C). In accordance to the previous CBA assays (Fig. 3), stimulation with the classical inflammasome ligands ATP and nigericin predominantly induced IL-1β single-positive cDC2 (Fig. 5C). This potent effect of L18-MDP was further displayed by the almost complete absence of IL-1β and IL-12p40 double-negative cells (Fig. 5C). However, stimulation with MSU resulted in only a moderate number of IL-1β and IL-12p40 double-positive cells comparable to ATP stimulation (Fig. 5C).

To test whether the concentration of classical inflammasome ligands, such as ATP, has an influence on the decision between hyperactivation and pyroptosis, we performed a titration experiment. We stimulated sorted, R848-primed cDC2 with increasing concentrations of ATP and measured LDH release, costimulatory molecule expression, and cytokine secretion. Although the highest concentrations (2.5 and 5 mM) induced comparable induction of LDH release and thus pyroptosis, lower concentrations (10 μM, 100 μM, and 1 mM) induced nearly no cell death (fig. S7A). Next, we analyzed the expression of costimulatory molecules on ATP-stimulated cDC2. Although the recovered cells under pyroptosis-inducing conditions showed lower expression of CD40, cDC2 stimulated with low concentrations of ATP showed a comparable expression of costimulatory molecules as TLR stimulation alone (fig. S7B). This was also the case for cytokine secretion: Although high concentration of ATP induced massive secretion of IL-1β, other proinflammatory cytokines were reduced compared to TLR stimulation alone (fig. S7C). In contrast, low concentrations of ATP—hardly inducing cell death—enabled cDC2 to secrete comparable concentrations of IL-12 family cytokines as TLR stimulation alone (fig. S7C), whereas IL-1β was strongly reduced (fig. S7C). Given that 1 mM ATP induced slightly enhanced IL-1β secretion and comparable amounts of other proinflammatory cytokines by simultaneously reducing cytotoxicity (fig. S7, A and C), we were interested whether this also resembled hyperactivation. Therefore, we primed sorted cDC2 with R848 followed by stimulation with different concentrations of ATP and measured active caspase-1 using FAM FLICA. Although we could observe some FLICA+ cDC2 when we stimulated the cells with 1 mM ATP, we found that these few cells were in addition 4′,6-diamidino-2-phenylindole (DAPI)+ and thus undergoing pyroptosis (fig. S7D). Therefore, the low levels of IL-1β observed using low concentrations of ATP seemed to be derived from pyroptotic cDC2, whereas the other proinflammatory cytokines are produced by non–inflammasome-activated cDC2. Thus, our data indicate that IL-1β and IL-12 are secreted from the same living cDC2 when stimulated with R848 in combination with the inflammasome ligand L18-MDP, but not MSU or pyroptosis-inducing NLRP3 ligands such as ATP, suggesting a predominant hyperactivation potential of cDC2 by treatment with L18-MDP.

Hyperactive cDC2 induce an enhanced TH1 and TH17 response in vitro

To determine the impact of inflammasome induction and hyperactivation of cDC2 on immune responses, we performed CD4+ T cell proliferation and differentiation assays. In a first set of experiments, we cocultured unprimed, R848-primed, or fully inflammasome-stimulated cDC2 with allogeneic sorted naϊve CD4+ T cells in mixed leukocyte reactions with a low DC:T cell ratio (1:100). We found that R848-primed cDC2 treated with L18-MDP induced significantly increased T cell proliferation compared to DCs that had been treated with classical inflammasome ligands, such as ATP or nigericin (fig. S8A). The polarization into interferon-γ (IFNγ)–secreting TH1 cells was enhanced when cDC2 were treated with L18-MDP compared to other inflammasome ligands (fig. S8B). However, we could hardly detect any IL-17+ TH17 cells (fig. S8C), and treatment with L18-MDP did not alter the induction of IL-4–secreting TH2 cells compared to stimulation with R848 alone (fig. S8D). We therefore made use of an antigen-specific assay by coculturing unprimed, R848-primed, or fully inflammasome-stimulated cDC2 with autologous carboxyfluorescein succinimidyl ester (CFSE)–labeled CD4+ T cells in the presence of MHC-II peptides consisting of a pool of 68 known MHC-II–restricted epitopes from different frequent pathogens [such as human herpesviruses 1 to 6 (including Epstein-Barr virus and cytomegalovirus), influenza A, or Toxoplasma gondii]. After 6 days, T cells were restimulated with phorbol 12-myristate 13-acetate (PMA) and ionomycin in the presence of brefeldin A and analyzed for proliferation and CD4+ T cell differentiation by staining for T cell activation marker and typical intracellular cytokines and by measuring the dilution of CFSE as an estimation of cell proliferation (fig. S9). TLR stimulation followed by L18-MDP treatment induced a better T cell proliferation capacity compared to classical inflammasome ligands or TLR stimulation alone (Fig. 6A and fig. S9). Thus, we conclude that hyperactive cDC2 exhibit a better T cell–proliferating capacity compared to classical inflammasome ligands or TLR stimulation alone.

Fig. 6 Hyperactive cDC2 induce an enhanced TH1 and TH17 response in vitro.

(A to E) cDC2 were sorted from PBMCs of healthy donors and primed with R848 (1 μg/ml) or were left unprimed as control. After 3 hours, cells were stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, and 5 μM nigericin or left untreated in medium. After further 3 hours, 1 × 104 DCs were cocultured with 1 × 105 CFSE-labeled autologous negatively enriched CD4+ T cells in the presence of MHC-II peptides (CEFX Ultra SuperStim Pool MHC-II peptides, JPT Innovative Peptide Solutions). After 6 days, CD4+ T cells were restimulated with PMA (100 ng/ml), ionomycin (1 μg/ml), and brefeldin A for 6 hours. See fig. S8 for further T cell analysis. Bar graphs show means ± SD over control stimulation with medium alone for five different donors. *P < 0.05, **P < 0.01, and ***P < 0.001 by one-way ANOVA in GraphPad Prism.

To further address the effect of inflammasome activation and hyperactivation of cDC2 on different CD4+ T cell subsets, we performed FACS analyses to understand CD4+ T cell differentiation upon coculture with activated cDC2. Therefore, we performed the same experiments as above (Fig. 6A and fig. S9), applying flow cytometry antibodies for extracellular definition of CD4+ T cells and typical intracellular cytokines associated with CD4+ T cell differentiation, such as IFNγ, IL-4, and IL-17 (Fig. 6, B to E). In line with the cytokine profile measured before (Fig. 3), our FACS analyses revealed that L18-MDP–stimulated cDC2 induced an enhanced TH1 and TH17 response displayed by a higher frequency of IFNγ and IL-17–producing CD4+ T cells compared to classical inflammasome ligands or TLR stimulation alone (Fig. 6, C and E). In contrast, administrating ATP or nigericin did not enhance TH1 or TH17 responses and even reduced the numbers of differentiated T cells (Fig. 6, B to E). Moreover, MSU-treated cDC2 retained the ability to induce TH1 and TH17 polarization (Fig. 6, B to E), although this effect was not statistically significant when compared to medium control. Similar effects were observed for IFNγ and IL-17 double-positive T cells (Fig. 6E). In line with the cytokine data showing predominant secretion of TH1- and TH17-polarizing cytokines (Fig. 3), the TH2 response was not significantly affected by stimulating TLR-primed cDC2 with L18-MDP or MSU (Fig. 6D). Thus, our data suggest that autologous CD4+ T cells shift toward TH1 and TH17 CD4+ T cell responses by administrating a combination of R848 and the inflammasome ligand L18-MDP. In summary, our cumulative data prompt us to conclude that cDC2 represent the only bona fide human DC subset transcriptionally and functionally licensed to induce inflammasome-dependent immune responses.

DISCUSSION

Activation of inflammasomes is pivotal for the induction of DC-mediated immunity against not only infectious agents but also cancer (6367). However, inflammasome activation usually results in pyroptosis leading to the secretion of IL-1β and subsequent death of the activated cell (53, 54). In this study, we could not only demonstrate that primary human cDC2 (CD1c+ DCs) display a unique specialization among human primary DC subsets for inflammasome activation but also demonstrate that primary cDC2 can enter a state of hyperactivation that lacks pyroptosis induction and is accompanied by enhanced TH1 and TH17 CD4+ T cell responses (Fig. 7) (57, 5961).

Fig. 7 Summary of the findings.

Among human DCs, cDC2 are uniquely specialized for inflammasome activation and, in response to certain inflammasome activators, can enter a state of hyperactivation that lacks pyroptosis induction and enhances TH1 and TH17 CD4+ T cell responses. Teff, effector T cell; Tm, memory T cell; Tn, naive T cell. The cellular images are provided and adapted from Servier Medical Art (smart.servier.com). The images are licensed under a Creative Commons Attribution 3.0 Unported License (creativecommons.org/licenses/by/3.0/).

In contrast to mice, our current knowledge about the role and function of primary DCs in men remains limited (88, 89). In line with a proteomic study of primary human blood DCs of Worah and colleagues (90), our data revealed a tissue-independent superior expression of most inflammasome-related molecules in cDC2 in lymphohematopoietic tissues under steady-state conditions. This implies that cDC2 do not completely rely on transcriptional induction of inflammasome genes by TLR priming to respond to inflammasome ligands. Moreover, transcriptional profiling of cDC1 and cDC2 stimulated with R848 revealed an increased expression of inflammatory cytokines and costimulatory molecules, whereas the signature gene expression profile was unchanged, which was similar to data found by transcriptional profiling of murine DC subsets upon cytomegalovirus infection (91, 92). On top of that, we found that TLR stimulation enhanced the preformed expression of inflammasome-related sensors or receptors such as NLRP3, MEFV, NOD2, and AIM2, whereas the expression of NAIP and NLRC4 was attenuated in cDC2. In conjunction with the potent inhibition mediated by the NLRP3 inhibitor MCC950, this outlines a pivotal role of the NLRP3 inflammasome in human cDC2.

Now, our understanding about the transcriptional control of most inflammasome sensor proteins is restricted (93). In murine studies, Bauernfeind and colleagues (94) described a positive regulation of NLRP3 expression by NF-κB signaling, whereas Karki et al. (95) found IRF8 to dictate basal expression and activation of NAIPs and NLRC4. However, McDaniel et al. (96) reported an IRF8- or IRF4-dependent restriction of inflammasome expression in murine cDC1 and cDC2, respectively. Because IRF8 is a signature gene for cDC1 (2, 4), it might explain the limited expression of inflammasome sensor genes in human cDC1 under steady-state conditions. In contrast, the dominant expression of IRF4 in cDC2 (4) seems not to limit inflammasome activity we observed in human cDC2. This might be attributed to differences between murine and human cDC2. Further, posttranscriptional modifications might regulate inflammasome activity and thus could change the cDC2-specific inflammasome response (97).

Inflammasome activation was described to be restricted to the monocyte and macrophage lineage because only murine BMDMs, but not BMDCs, showed inflammasome activity and IL-1β release (74, 98). Previous data suggest that human cDC2 (identified as CD1c+ DCs) include an additional subpopulation, designated as DC3. Depending on the published study, these subpopulations of CD1c+ DCs are discriminated either by expression of CD14 (99, 100), CD5 (100, 101), CD36 (72), CD163 (72, 100), or CLEC10A (27). Although different marker combinations were used to identify subpopulations of CD1c+ DCs, in all studies, a subpopulation with monocytic features was described, either functionally or transcriptionally. Our analysis of scRNA-seq data of human CD1c+ DCs showed that cDC2 and DC3 can be discriminated on the basis of their transcriptome. However, both subsets did not differ in their inflammasome expression profile. Moreover, Bourdely et al. demonstrated that DC3 display a rather monocytic behavior, which is indicated by the secretion of large amounts of IL-1β in response to TLR priming without specific inflammasome ligands (73). In contrast, our data demonstrated that cDC2 secreted IL-1β only after additional inflammasome stimulation. Thus, our data suggest that bona fide human DCs are also able to mount inflammasome responses.

Whether a ligand causes hyperactivation in addition to inflammasome activation seems to depend on the ligand used and the cell type as shown for murine BMDCs and BMDMs stimulated with the same hyperactivating ligand (59, 60). In our analyses, we identified L18-MDP as weaker and MSU as stronger inducer of NLRP3- and caspase-1–dependent inflammasome activation for human cDC2 leading to IL-1β secretion. Strong inflammasome activators are typically associated with the induction of pyroptosis (52). Neither L18-MDP nor MSU induced significant cell death after TLR priming. In our study, we further found that L18-MDP upon priming allowed concomitant and enhanced secretion of IL-12 and IL-23, suggesting L18-MDP as strong hyperactivating ligand. This was in line with our CD4+ T cell stimulation analyses, in which treatment with L18-MDP together with R848 priming induced a pronounced TH1 and TH17 polarization. Because gasdermin D–mediated pore formation allows for IL-1β release from living cells, Evavold et al. (61) suggested that the activation of gasdermin D is a prerequisite for hyperactivation. In accordance, we found slightly enhanced gasdermin D expression in primary human cDC2 after TLR priming. In addition, it was reported that hyperactivation occurs when weak inflammasome signaling leads to low numbers of gasdermin D pores that can be repaired by the endosomal sorting complexes required for transport ESCRT machinery (61, 102, 103). This might explain why L18-MDP, as a rather weak NLRP3 ligand, seems to preferentially induce hyperactivation. Overall, our findings suggest that human cDC2 have a specialized role under inflammatory conditions because cDC2 might be the only cell type able to produce both DC (IL-12 family) and monocytic (IL-1 family) cytokines that enhance TH1 and TH17 CD4+ T cell responses. Whether DC3 are also able to enter a state of hyperactivation has yet to be determined.

An inflammasome ligand seems to direct the fate of the cell to either enter pyroptosis or hyperactivation. Thus, hyperactivation-inducing inflammasome ligands seem to be promising candidates for future vaccine development (66, 104), as indicated by a study of Ghiringhelli et al. (63), in which inflammasome activation in DCs played a pivotal role for the immune response against tumors. Moreover, inflammasomes and TH17 cells are critical players in many inflammatory disorders (105108). Immune modulation of inflammasome responses in cDC2 may be even suitable for anti-inflammatory approaches.

MATERIALS AND METHODS

Human blood preparation

Leukocyte reduction system (LRS) cones were obtained from thrombocytapharesis of healthy adult donors and approved by the local ethical committee (Ethikkommission der Friedrich-Alexander-Universität Erlangen-Nürnberg). Human blood was processed as previously described in (4). Briefly, LRS cones were diluted with RPMI 1640 up to 40 ml, and a density gradient centrifugation was performed using Pancoll (ρ = 1.077 g/ml; PAN-Biotech). For this, 14 ml of Pancoll was overlaid with 20 ml of the diluted human LRS cone. After 20 min of centrifugation (deceleration set to 0) at 525g, the interphase containing the peripheral blood mononuclear cells (PBMCs) was isolated by pipetting, washed with RPMI 1640, and then used for further experiments.

Human tissue preparation

All tissue samples were freshly isolated, immediately processed, and enzymatically digested in phosphate-buffered saline (PBS) with 2% mixed human serum type AB (Lonza), collagenase D (400 U/ml; SERVA), and 100 μg (spleen) to 300 μg (thymus) deoxyribonuclease I (Sigma-Aldrich). Mechanical disruption was conducted in C tubes (Miltenyi Biotec) using the gentleMACS tissue dissociator (Miltenyi Biotec). The tissue samples were filtered twice. Single-cell suspensions were diluted with RPMI 1640 medium, followed by gradient centrifugation as described before as a standard procedure for human blood preparation.

Human DC enrichment

Before performing cell sorts and further experiments, human DCs were enriched using the EasySep Human Pan-DC Pre-Enrichment Kit (STEMCELL Technologies) following the manufacturer’s instructions with slight adaptations. Briefly, PBMCs were diluted in PBS containing 2% human serum and 1 mM EDTA to a concentration of 1 × 108 cells/ml. Cells were subsequently stained with anti-human CD32 [Fc fragment of IgG receptor II (FcγR II)] block (0.8 μg/ml; STEMCELL Technologies) and 15 μl/ml each of the pan-DC enrichment cocktail components A + B (STEMCELL Technologies ) for 30 min at room temperature. After this, magnetic Dextran RapidSpheres (30 μl/ml; STEMCELL Technologies ) were added. After 10 min at room temperature, cells were put inside the EasySep magnet (STEMCELL Technologies) for another 5 min. The nondepleted DCs were poured off into a fresh tube and stained for cell sorting.

Cell sorting of human DC subsets

For the cell sorting, DC-enriched suspensions were labeled with the following antibodies: V450-CD19 (HIB19, BD Biosciences), e450-CD20 (2H7, eBioscience), BV421-CD56 (5.1H11, BioLegend), BV421-NKp46 (9E2, BioLegend), phycoerythrin (PE)/Cy5-CD11b (M1/70, BioLegend), A700-CD14 (HCD14, BioLegend), PE-CF594–HLA-DR (G46-6, BD Biosciences), allophycocyanin (APC)/Cy7-CD1c (L161, BioLegend), PE/Cy-CD11c (3.9, BioLegend), BV605-CD123 (6H6, BioLegend), BV711-CD141 (1A4, BD Biosciences), and PerCP/Cy5.5-CD303 (201A, BioLegend). To ensure the purity of sorted DC subpopulations, B cells were labeled with antibodies against both CD19 and CD20, coupled to fluorochromes with same excitation and emission characteristics to increase the separation of B cells and cDC2, because B cells share the expression of HLA-DR and CD1c with cDC2. Because natural killer (NK) cells consist of subsets with low and high expression of CD56, an antibody against NKp46 coupled to the same fluorochrome was added to ensure exclusion of all NK cells. After 30 min of incubation on ice, cells were washed and resuspended in PBS with 2% human serum and 0.01% DAPI and sorted on a FACSAria II (BD Biosciences). cDC2 were sorted as CD3CD11bCD14CD19CD20CD56CD123DAPINKp46HLA-DR+CD1c+CD11c+, cDC1 were sorted as CD1cCD3CD11bCD14CD19CD20CD56CD123DAPINKp46HLA-DR+CD11c+CD141+, and pDCs were sorted as CD1cCD3CD11bCD14CD19CD20CD56DAPINKp46HLA-DR+CD123+CD303+. Purity obtained after reanalysis was generally above 95%. To analyze subpopulation composition of cDC2, cells were stained with PE-CD5 (UCHT2, BioLegend) and A647-CD163 (GHI/61, BioLegend). Acquisition of samples was performed using a BD LSRFortessa 2, and data were analyzed by FlowJo.

TLR and inflammasome stimulation of human cDC2

After cell sorting, cDC2 were resuspended in DC medium [RPMI 1640 plus 5% panexin NTA, 5% panexin NTS, 5% human serum, and 1% penicillin/streptomycin + 1% sodium pyruvate + 1% glutamine] to reach a final concentration of 2 × 105 cells/ml. Cells were seeded in a 96-well plate with a maximum of 4 × 104 cells per well. For stimulation with TLR ligands and extracellular pathogens, cDC2 were either incubated with LPS (1 μg/ml), Pam3CSK4 (0.5 μg/ml), flagellin (2.5 μg/ml), pIC (5 μg/ml), R848 (5 μg/ml), CL264 (5 μg/ml), ssPolyU (5 μg/ml), CpG (5 μg/ml), heat-killed E. coli (50 μg/ml), heat-killed S. aureus (50 μg/ml), or zymosan particles (25 μg/ml) or kept in medium as control for 12 hours. For inflammasome activation, cDC2 were primed with R848 (5 μg/ml) for 3 hours. Subsequently, cells were washed and stimulated with either L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, 5 μM nigericin, or poly(dA:dT) (5 μ/ml) or kept in medium. After 12 hours of stimulation, cells were analyzed for the activation markers CD40 and CD86 [CD40-PE, clone: 5C3, BioLegend; CD86-PE-CF594, 2331(FUN-1), BD Biosciences] using flow cytometry. At the same time, supernatants were harvested and stored at −80°C until further analysis by CBA using the LEGENDplex Human Cytokine Panel 2 (BioLegend). Acquisition of samples was performed using a BD LSRFortessa 2, and data were analyzed by FlowJo or LEGENDplex software (V8.0, VigeneTech). TLR and inflammasome ligands (InvivoGen) were reconstituted in either water or PBS.

Cytometric bead array

After stimulation of sorted DCs with TLR and inflammasome ligands, supernatants were stored at −80°C. For analysis of cytokine secretion by the stimulated DCs, supernatants were thawed, and concentration of cytokines was determined using the LEGENDplex Human Cytokine Panel 2 (BioLegend) as suggested by the manufacturer. To determine exact concentration of the cytokines, a standard curve was performed using a standard cocktail included in the assay. For measuring of cytokine concentrations, 10 μl of each standard and the supernatants of the stimulated DCs were diluted with 10 μl of assay buffer (BioLegend) in a 96-well plate (V bottom). Then, 10 μl each of premixed beads and biotin-coupled detection antibodies [thymic stromal lymphopoietin, IL-1α, IL-1β, granulocyte-macrophage colony-stimulating factor, IFNα2, IL-23, IL-12p40, IL-12p70, IL-15, IL-18, IL-11, IL-27, and IL-33] were added and incubated for 2 hours on a plate shaker. Next, 10 μl of streptavidin-PE was added to each well, and the plate was incubated for further 30 min on a plate shaker. One hundred fifty microliters of 1× wash buffer was added, and the plate was centrifuged at 1000g for 5 min. After another washing step using 100 μl of 1× wash buffer, beads were acquired using a BD LSRFortessa 2, and cytokine concentrations were analyzed using the LEGENDplex software (V8.0, VigineTech).

T cell assay

After human blood preparation was performed, cells were counted, and 1.5 × 108 leukocytes were used for CD4+ T cell isolation using an EasySep Human CD4+ T cell enrichment kit (STEMCELL Technologies) following the manufacturer’s instructions. Briefly, PBMCs were diluted in PBS containing 2% human serum and 1 mM EDTA to a concentration of 5 × 107 cells/ml. Cells were subsequently stained with human CD4+ T cell enrichment cocktail (25 μl/ml) and incubated for 5 min at room temperature. Afterward, magnetic Dextran RapidSpheres (25 μl/ml) were added. The tube was subsequently put into a magnet, and the nondepleted CD4+ T cells were poured off into a fresh tube. About 3 × 106 isolated CD4+ T cells were then labeled with 5 μM CFSE (Invitrogen, Thermo Fisher Scientific). The remaining PBMCs were used for DC enrichment as described above.

After DC enrichment, 1 × 104 cDC2 per well were either stimulated with R848 (1 μg/ml) or kept in medium as control. After 3 hours, cDC2 were either stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, or 5 μM nigericin or kept in medium. A further 3 hours later, cells were washed, pulsed with CEFX Ultra SuperStim Pool MHC-II peptides (1 μg/ml; JPT Innovative Peptide Solutions) and cocultured with 2 × 105 autologous CFSE+ T cells. After 6 days, T cells were restimulated with PMA (100 ng/ml; InvivoGen), ionomycin (500 ng/ml; InvivoGen), and 0.1% brefeldin A (BioLegend) for 6 hours. T cells were stained with Zombie UV Fixable Viability Dye (1:200; BioLegend). Intracellular cytokines were analyzed using the Fixation/Permeabilization Solution Kit (BD Biosciences) and antibodies to CD3-BV570 (UCHT1, BioLegend), CD4-BV510 (RPA-T4, BioLegend), CD25-PE/Cy7 (BC96, BioLegend), IFNγ-BV421 (4S.B3, BioLegend), IL-4–A647 (8D4-8, BioLegend), IL-10–PE (JES3-19FI, BioLegend), and IL-17A–BV605 (BL168, BioLegend). As an isotype control, PE-labeled rat immunoglobulin G2a (IgG2a) (RTK2758, BioLegend) and BV421-, BV605-, or A647-labeled mouse IgG1 (MOPC-21, BioLegend) antibodies were used. Flow cytometry was performed using a BD LSR Fortessa, and data were analyzed by FlowJo (FlowJo LLC).

Mixed leukocyte reaction

After human blood preparation was performed, cells were counted, and 1.5 × 108 leukocytes were used for naïve CD4+ T cell isolation using an EasySep human naϊve CD4+ T cell enrichment kit (STEMCELL Technologies) following the manufacturer’s instructions. Briefly, PBMCs were diluted in PBS containing 2% human serum and 1 mM EDTA to a concentration of 5 × 107 cells/ml. Cells were subsequently stained with human naϊve CD4+ T cell enrichment cocktail (25 μl/ml) and incubated for 5 min at room temperature. Afterward, magnetic Dextran RapidSpheres (25 μl/ml) were added. The tube was subsequently put into a magnet, and the nondepleted naϊve CD4+ T cells were poured off into a fresh tube. About 3× 106 isolated T cells were then labeled with 5 μM CFSE (Invitrogen, Thermo Fisher Scientific). The remaining PBMCs were used for DC enrichment as described above.

After DC enrichment, 1 × 103 cDC2 per well were stimulated with R848 (1 μg/ml) or kept in medium as control. After 3 hours, cDC2 were stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, and 5 μM nigericin or kept in medium. A further 3 hours later, cells were washed and cocultured with 1 × 105 allogeneic CFSE+ T cells. After 6 days, T cells were restimulated with PMA (100 ng/ml; InvivoGen), ionomycin (500 ng/ml; InvivoGen), and 0.1% brefeldin A (BioLegend) for 6 hours. T cells were then stained with Zombie UV Fixable Viability Dye (1:200; BioLegend). T cells were subsequently stained for intracellular cytokines using the Fixation/Permeabilization Solution Kit (BD Biosciences) and antibodies to CD3-BV570 (UCHT1, BioLegend), CD4-BV510 (RPA-T4, BioLegend), CD25-PE/Cy7 (BC96, BioLegend), IFNγ-BV421 (4S.B3, BioLegend), IL-4–A647 (8D4-8, BioLegend), IL-10–PE (JES3-19FI, BioLegend), and IL-17A–BV605 (BL168, BioLegend). As an isotype control, PE-labeled rat IgG2a (RTK2758, BioLegend) and BV421-, BV605-, or A647-labeled mouse IgG1 (MOPC-21, BioLegend) antibodies were used. Flow cytometry was performed using a BD LSRFortessa 2, and data were analyzed by FlowJo (FlowJo, LLC).

Inflammasome and caspase inhibition assay

After cell sorting, cDC2 were treated with ac-yvad-cmk (30 μg/ml; InvivoGen), 10 μM MCC950 (InvivoGen), or 15 μM GSK717 (Sigma-Aldrich) or kept in medium containing dimethyl sulfoxide (DMSO). After 1 hour, cells were either stimulated with R848 (5 μg/ml) or kept in medium as control for 3 hours. Subsequently, cells were either incubated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), or 5 mM ATP or kept in medium in the presence or absence of ac-yvad-cmk (30 μg/ml), 10 μM MCC950, and 15 μM GSK717 or medium containing DMSO. After another 12 hours, supernatants were harvested and stored at −80°C until further analysis by CBA assay using the LEGENDplex Human Cytokine Panel 2 (BioLegend). Acquisition of samples was performed using a BD LSR Fortessa 2, and data were analyzed by LEGENDplex software (V8.0, VigeneTech).

Microarray analysis

Published microarray data were analyzed for relative expression of AIM2, CASP1, CASP4, CASP5, GSDMD, IL1B, IL18, MEFV, NAIP, NLRC4, NLRP1, NLRP3, NLRP12, NOD1, NOD2, and PYCARD (4). Microarray data are available in the GEO database (www.ncbi.nlm.nih.gov/gds) under the accession number GSE77671. Transcriptome data of Whole Human Genome Oligo Microarray (Agilent) of human cDC1, cDC2, and pDCs from three blood, spleen, and thymus donors were used. Raw values generated by automated feature extraction have been robust multi-array analysis (RMA) background-corrected and quantile-normalized using R (Windows, x64, 3.3.1) (109). Relative expression values were plotted using Qlucore Omics Explorer 3.6.

NanoString analysis

cDC1 and cDC2 were sorted from PBMCs, and 1.5 × 105 cells were subsequently stimulated with R848 (5 μg/ml) or kept in medium as control. After 3 hours, cells were lysed using one-third RLT buffer, and sample mRNAs were stored at −80°C until further analysis. NanoString assay with the nCounter Human Myeloid Innate Immunity V2 Panel (NanoString Technologies) with 30 additional genes (CASP4, CASP8, NLRP1, NLRP12, NLRC4, NAIP, AIM2, MEFV, GSDMD, ANPEP, IFNL1, IFNL2, IFNL3, CADM1, NLRC5, RAB15, ZNF366, CD226, TEAD4, TCF7L2, EPAS1, LHX6, TCF7, CLNK, C1ORF54, ABCA1, ZBTB32, GZMK, AXL, and SIGLEC6) was performed using the manufacturer’s instructions. Briefly, samples were thawed and put into a thermal cycler for hybridization with reporter code sets (target-specific reporter and capture probes) for 24 hours. Reporter code sets consist of many RNA capture probes targeting specific mRNAs. Each probe is labeled with a unique combination of fluorochromes (molecular barcode). Therefore, specific mRNAs can be detected and counted after hybridization using the nCounter System. Data were subsequently analyzed and exported using the nSolver software (version 4.0, NanoString Technologies). Data were further normalized with RUVg (110) to remove unwanted batch effects. Eventually, they were analyzed in R using NanoStringNorm (111) and edgeR (112), and DEGs were determined with a threshold of false discovery rate (FDR)–adjusted P value of 0.01 and a log2 fold change (FC) of 1.5.

Analysis of scRNA-seq data

scRNA-seq data of human blood DCs, monocyte, and progenitors were obtained from GEO database (accession number GSE94820). The R software package Seurat 3 (113) was applied for data preprocessing; a total of 15,424 gene features and 1123 cells were obtained. Cells with a low number of expressed genes and the gene features of abnormal presence (lower than 0.5% in all cells) were removed from the expression matrix. The data were then normalized with SCTransform for downstream analyses. Cell clusters were visualized with a Uniform Manifold Approximation and Projection method. Differential gene expression analysis was performed with Model-Based Analysis of Single-Cell Transcriptomics (114), and the top gene markers of DC3 and cDC2 were displayed as heatmap. In addition, key inflammasome genes were compared and visualized as heatmap, followed by a gene expression correlation test. The Pearson correlation coefficient and the P value were calculated.

LDH assay

After cell sorting, 2 × 106 cDC2/ml was plated in a 96-well plate forming three different groups: high control, low control, and test group. The test group was stimulated as described above for TLR and inflammasome stimulation. The low control group was kept in medium. The high control group was kept in medium for 11.5 hours until the lysis buffer was added for 30 min. After a total of 12 hours, lysis was blocked, and cells were centrifuged at 250g for 2 min. Supernatants were harvested and analyzed using the LDH-Cytox Assay Kit (BioLegends). Briefly, 50 μl of working solution containing a tetrazolium salt was added to 50 μl of supernatant. After 30 min, a stop solution was added, and the absorbance was measured at 490 nm by a VersaMax enzyme-linked immunosorbent assay plate reader (Molecular Devices). Cytotoxicity was subsequently calculated using the manufacturer’s formulaCytotoxicity (%)=Test substanceLow ControlHigh ControlLow Control×100

Intracellular cytokine staining of inflammasome-stimulated cDC2

After cell sorting, cDC2 were seeded at a final concentration of 5 × 105 cells/ml and primed with R848 (1 μg/ml) for 3 hours. Subsequently, the inflammasome was stimulated with L18-MDP (0.5 μg/ml), MSU (300 μg/ml), 5 mM ATP, or 5 μM nigericin in the presence of 0.1% brefeldin A. After 6 hours of inflammasome activation, cDC2 were stained with Zombie UV Fixable Viability Dye (1:200; BioLegend). Intracellular cytokines were analyzed after fixation and permeabilization using the Fixation/Permeabilization Solution Kit (BD Biosciences) as described in the manufacturer’s instruction. Intracellular staining was performed in 1× Perm/Wash buffer (BD Biosciences) using Alexa Fluor 647–coupled antibody to IL-1β (clone: H1b-98) and PE-coupled antibody to IL-12p40 (clone: C11.5) or Alexa Fluor 647–coupled and PE-coupled mouse IgG1 as the respective isotype controls. Flow cytometry was performed using a BD LSRFortessa 2, and data were analyzed by FlowJo (FlowJo LLC).

FAM FLICA assay

After cell sorting, cDC2 were seeded at a final concentration of 5 × 105 cells/ml and primed with R848 (1 μg/ml) for 3 hours. Cells were subsequently stimulated with increasing concentrations of ATP (10 μM, 100 μM, 1 mM, 2.5 mM, and 5 mM) for another 3 hours in the presence of the FAM FLICA reagent (ImmunoChemistry). Cells were then washed with 1× apoptosis wash buffer, stained with DAPI (1:10,000; AppliChem), and directly analyzed at a BD LSRFortessa 2 cytometer. FAM FLICA expression was then measured in the fluorescein isothiocyanate channel. Data were analyzed using FlowJo.

Statistics

Mann-Whitney U test or one-way analysis of variance (ANOVA) with Bonferroni post hoc test was performed using Prism 5.0 (GraphPad). Results were displayed as means ± SD. DEGs in the NanoString analysis were determined with a threshold of a FDR-adjusted P value of 0.01 and a log2FC of 1.5.

Ethics statement

The study was carried out in accordance with the recommendations of the local ethical committee (Ethikkommission der Friedrich-Alexander-Universität Erlangen-Nürnberg) with written informed consent from all individuals. All individuals gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the local ethical committee (Ethikkommission der Friedrich-Alexander-Universität Erlangen-Nürnberg).

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/14/680/eabe1757/DC1

Fig. S1. Gating strategy used for cell sorting of human DC subpopulations.

Fig. S2. TLR expression and responsiveness of human DC subpopulations.

Fig. S3. Microarray analysis reveals tissue-independent superior expression of inflammasome-related molecules in human cDC2 under steady-state conditions.

Fig. S4. Priming with R848 induces strong transcriptional changes in cDC1 and cDC2 but does not change the identity of the cells.

Fig. S5. The CD1c+ DC subsets cDC2 and DC3 harbor a comparable expression of inflammasome genes.

Fig. S6. Comparable costimulatory molecule expression and morphological characteristics of control and L18-MDP–treated cDC2.

Fig. S7. ATP induces IL-1β secretion from pyroptotic cDC2 independently of the ATP concentration.

Fig. S8. Mixed leukocyte reaction shows similar, but weaker, T cell polarization profiles.

Fig. S9. Gating strategy for the restimulation of antigen-specific T cells with hyperactive cDC2.

REFERENCES AND NOTES

Acknowledgments: We thank D. Schönhöfer and M. Mroz (Core Unit Cell Sorting and Immunomonitoring) for cell-sorting support and the Department of Transfusion Medicine of the University Hospital Erlangen. We also thank the members of the Dudziak laboratory for critical comments and N. Eissing and S. Dörfler for technical support. Funding: This work was partly supported by grants from the German Research Foundation [Deutsche Forschungsgemeinschaft (DFG)] to D.D. (CRC1181-TPA7 and DU548/5-1) and a proposal funded by the Agence Nationale de la Recherche (ANR) and the DFG (DU548/6-1). D.D. received support from Interdisziplinäres Zentrum für Klinische Forschung (IZKF) (IZKF-A80). L. Hatscher’s medical thesis was supported by RTG1660. L. Heger was supported by Erlanger Leistungsbezogene Anschubfinanzierung und Nachwuchsförderung (ELAN) (DE-17-09-15-1-Heger). D.D. was funded by the Bavarian State of Ministry of Science and Art, Bayresq.Net. O.G. was supported by the German Research Foundation (DFG) through SFB 1160, SFB/TRR 167, SFB 1425, GRK 2606, and (under the Excellence Strategy of the German Federal and State Governments) through CIBSS (EXC-2189, project ID 390939984). F.N. was supported by grants from the German Research Foundation (CRC1181-TPA7 and FOR 2886-B2). M.K. received Era-Net grant 01KT1801 of the German Federal Ministry of Education and Research (BMBF). C.L. acknowledges funding by the Land Bavaria (contribution to SFB TR221/INF 324392634). H.B. was supported by Wilhelm-Sander Foundation. Author contributions: L. Hatscher performed the experiments with participation of L. Heger and C.H.K.L. L. Heger and C.H.K.L. analyzed the microarray data. L. Heger, C.L., and M.K. analyzed the NanoString data. A.H., A.P., and R.C. ensured the human tissue sample supply. C.O. and I.I.-B. provided the facility and support for inflammasome ligand preparation. L. Hatscher, L. Heger, and D.D. contributed to the data analysis. O.G., F.N., and H.B. contributed to the design of the study and discussion of the data. L. Heger and D.D. supervised and designed the study. L. Hatscher, L. Heger, and D.D. wrote the manuscript. The present work was performed in fulfillment of the requirements for obtaining the degree “Dr. med.” by L. Hatscher. Competing interests: The authors declare that they have no competing financial interests. Data and materials availability: The NanoString data are deposited in GEO, number GSE162407.

Stay Connected to Science Signaling

Navigate This Article