Lipid nanoparticles as a tool to dissect dendritic cell maturation pathways

Summary

Depending on how antigens are being decoded by dendritic cells (DCs), their acquisition will induce a homeostatic or immunogenic maturation program. This determines how antigens are being presented and whether DCs instruct T cells to induce tolerance or immunity. So far, the field lacks proper tools to distinguish the two maturation states. By using a lipid nanoparticle (LNP)-based approach and cellular indexing of transcriptomes and epitopes sequencing analysis, we designed a flow cytometry panel and transcriptional profiling tools to reliably annotate the two DC maturation states. The data corroborate that uptake of empty (or peptide-containing) LNPs induces homeostatic maturation in DCs, while uptake of Toll-like receptor ligand-adjuvanted (or mRNA-containing) LNPs induces immunogenic maturation, yielding distinct T cell outputs. This reveals that LNPs are not decoded as “dangerous” by DCs, and that the cargo is essential to provide adjuvant activity, which is highly relevant for the targeted design of LNP-based therapies.

Graphical abstract

Highlights

• •Flow cytometry and transcriptional profiling discern homeostatic from immunogenic mature DCs
• •pIC alone induces a transient immunogenic state in cDC1s, in contrast to pIC coupled to LNPs
• •The cargo, not the lipids, determines adjuvanticity of lipid nanoparticles
• •Meta-analysis supports a unified nomenclature of homeostatic and immunogenic mature DCs

Rennen et al. provide methods to distinguish homeostatic from immunogenic mature cDC1s, based on lipid nanoparticle (LNP)-induced maturation. Only the uptake of TLR ligand- or mRNA-encapsulated LNPs induces immunogenic maturation in DCs, while empty LNPs induce homeostatic maturation, revealing that empty LNPs are not sufficient to provide adjuvant activity.

Introduction

Dendritic cells (DCs) are crucial gatekeepers that maintain the balance between immunity and tolerance.¹ Conventional DCs (cDCs) are classified into two subsets, X-C motif chemokine receptor 1 (XCR1)-expressing cDC1s and signal regulatory protein alpha (SIRP)α-expressing cDC2s. They are derived from a progenitor pre-cDC in the bone marrow and have different functionalities, especially in vivo.^2,3,4 cDC1s excel in cross-presentation, a process in which exogenously acquired antigens are presented on major histocompatibility complex class I (MHC class I) molecules to CD8 T cells. cDC2s mainly present exogenously acquired antigens on the MHC class II molecules to CD4 T cells.⁵ In the immature state, cDCs act as sentinel cells that migrate through peripheral tissues and sample the environment for the presence of foreign (or self) antigens. Antigen acquisition induces their maturation and leads to upregulation of the chemokine receptor C-C chemokine receptor 7 (CCR7), which will guide them toward the T cell zone in the draining lymph nodes (dLN) to present the antigen to naive T cells.⁶ During maturation, DCs lose their phagocytic capacity and acquire antigen-presenting functionalities.⁷ At the same time, they start expressing numerous cell surface molecules and chemokines that allow them to communicate efficiently with a variety of cell types, not only T cells, but also other immune cells such as natural killer cells or monocytes, or non-immune cells such as stromal cells in the lymph node.¹ The discovery of pattern recognition receptors—decoding the presence of pathogens as “danger”—led to a better understanding of the pathways controlling the immunogenic maturation program of DCs.⁸ However, how DCs perceive “self” and remain immunologically silent remains poorly understood. We and others recently found that engulfment of apoptotic cells and influx of cholesterol drives the homeostatic maturation program of cDC1s.^{9,10,11,12,13} The process can be mimicked by the uptake of empty, non-adjuvanted lipid nanoparticles (LNPs).⁹

LNPs recently emerged as a safe and powerful tool to deliver mRNA vaccines in vivo.¹⁴ They marked a critical turning point during the COVID-19 pandemic to drive worldwide immunity against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). LNPs consist of four different types of lipids: cholesterol, polyethylene glycol (PEG)ylated lipids, phospholipids, and an ionizable lipid.¹⁵ The use of ionizable lipids with a pKa below 7 appeared to be a game-changer in the field. Due to their low surface charge, ionizable cationic LNPs (iLNPs, labeled as LNPs throughout the manuscript) exhibit lower toxicity in vivo. Furthermore, the non-bilayer lipid structure of the ionizable lipid has been proposed to ensure endosomal escape and efficient delivery of the mRNA cargo to the cytosol.¹⁵ The recent success of mRNA-LNPs pushed the LNP technology to the forefront of medicine and launched worldwide interest in their potential as a non-viral delivery system for applications such as vaccination, gene-editing, and protein replacement therapy. Still, many questions remain, and there is an urgent need to understand the molecular principles behind their mode of action. One of the most remarkable properties of mRNA-LNPs is that they do not require additional adjuvants to induce potent immune responses, in contrast to most classical vaccines.¹⁴ Whether the lipid components of the LNPs can serve as a stand-alone driver of adjuvanticity of mRNA-based vaccines without the need for mRNA components remains a conundrum. Previous data suggested that this is indeed the case based on the induction of a humoral response by LNPs co-injected with proteins.¹⁶ However, our data revealed that empty LNPs (eLNPs)—in contrast to polyinosinic:polycytidylic (pIC) acid-adjuvanted LNPs (pIC-LNPs)—induce homeostatic and non-immunogenic DC maturation in vivo,⁹ suggesting that the LNP itself may not be sufficient to induce a potent T cell immune response.

Therefore, we decided to dissect in more detail how different types of LNPs (encapsulating Toll-like receptor [TLR] ligands or empty) steer DC maturation pathways in vivo. Cellular indexing of transcriptomes and epitopes sequencing (CITE-seq) analysis allowed us to identify appropriate markers to design a flow cytometry panel to dissect homeostatic from immunogenic DC maturation states in vivo. The panels were further validated in an infection and a tumor model. The phenotypic analysis was complemented with different functional assays in which we addressed how the uptake of LNPs by DCs affects their cytokine/chemokine profile and modulates downstream T cell outcomes. Our data reveal that empty, non-cargo containing LNPs are not sufficient to induce immunogenic DC activation in contrast to mRNA-LNP or TLR-adjuvanted LNPs. This suggests that the observed adjuvant activity in mRNA-LNPs is derived from the mRNA component rather than from the lipids. This indicates that the scope of therapeutic applications of LNPs might be broader than originally anticipated and that they may have potential as a vehicle to induce tolerance against non-immunogenic cargo.

Results

Uptake of empty LNPs drives activation of a homeostatic cDC1 maturation program

Our lab previously noted that uptake of empty versus pIC-coupled LNPs can be used to direct DC maturation pathways in vivo, as monitored by the induction of selected gene signatures.⁹ To assess in detail how LNPs induce DC maturation pathways both early and late after uptake, we decided to apply an unbiased approach. To this end, we performed a CITE-seq analysis on splenic cDC1s sorted from mice at steady state (SS) or 2 and 8 h after intravenous injection (post-injection, p.i.) with eLNPs, pIC-LNPs, cytosine-phosphate-guanine (CpG)-encapsulated LNPs (CpG-LNPs), or pIC alone (Figure 1A). pIC is a trigger for TLR3 and the retinoic acid inducible gene-I-like receptors (RLRs),¹⁷ while CpG is a trigger for TLR9.¹⁸ The LNPs were formulated with the ionizable lipid ALC-0315 (see STAR Methods) which is the ionizable lipid consequently used throughout the paper. However, the effects were similar when LNPs were formulated with other ionizable lipids (see Figure 5B). The LNPs were labeled with a Cy5-fluorochrome covalently coupled to 1,2-dioleoylphosphatidyleathnolamine (DOPE), and only the Cy5⁺ fraction of the cDC1s was sorted to focus specifically on those cells that had engulfed LNPs. In SS conditions, immature CCR7⁻ and mature CCR7⁺ cDC1s were sorted in a 3:1 ratio to increase the percentage of mature cDC1s per sample. In the pIC condition, all cDC1s were sorted (Figure 1A). Uniform manifold approximation and projection (UMAP) dimensionality reduction was performed to visualize the cDC1s on a two-dimensional plot, visualized with all groups together (Figure 1B, top left) and split per group (Figure 1B). Harmony normalization was subsequently performed to properly compare the different maturation stages upon different treatment conditions (Figure 1D and Tables S1 and S2). Annotation was done based on earlier identification of DC clusters,⁹ which revealed three major maturation clusters: immature, early mature (EM), and late mature (LM). In addition, a cluster of proliferating cDC1s could be distinguished. The heatmap in Figure S1A shows the marker genes used per cluster, based on our previous annotations.⁹ The differences in gene expression between immunogenic and homeostatic conditions will be discussed later, in the next section. Harmony normalization was used to annotate the different maturation stages in SS mice and confirmed that the majority of cDC1 was in the immature cluster, while a minority was in the EM and LM clusters (Figure 1C). The comparison of the merged and split UMAP plots of the different conditions (Figure 1B) revealed that the transcriptional profiles of cDC1s that engulfed eLNPs at 2 h (light green) and 8 h p.i. (dark green) had a large overlap with the transcriptional profiles of early and late homeostatic mature cDC1s in the SS condition, respectively (Figure 1C). cDC1s that engulfed LNPs coupled to the TLR ligands pIC or CpG clustered separately. Of note, cDC1s that matured due to injection of pIC (light purple) or pIC-LNP (pink) clustered closely together at 2 h p.i., while at 8 h p.i., both conditions diverged. cDC1s that were activated by pIC (dark purple) moved toward the cluster of homeostatic mature cDC1s, while cDC1s that engulfed pIC-LNPs (red) kept a distinct antiviral transcriptional profile (Table S3). In line, Gene Ontology (GO) terms such as “inflammatory response,” “response to virus,” and “response to interferon-beta” were upregulated in the pIC-LNP condition compared to the pIC condition at 8 h p.i. (Figure S1B). Independent of the upstream maturation trigger, the majority of the cDC1s could be found in the EM cluster at the 2-h time point, while at 8 h p.i., most of the cDC1s were found in the LM cluster (Figure S1C). For further gene expression analysis and proper comparison between the different treatments, the EM population was chosen as the reference population at 2 h p.i. and the LM population as the reference population at 8 h p.i.. A relative similarity in gene expression can be indicated by the distance between the pseudobulked samples on a multidimensional scaling (MDS) plot (Figure 1E). Supporting our earlier observations, EM cDC1 samples 2 h p.i. with eLNPs clustered closely together with SS EM cDC1, while EM cDC1 samples obtained 2 h p.i. with pIC, pIC-LNPs, or CpG-LNPs clustered together at a distinct spot (Figure 1E), revealing their dissimilarity. At a later stage, both the LM SS and LM eLNP samples remained clustered together, as well as the LM pIC-LNP and LM CpG-LNP samples. On the contrary, LM cDC1 samples in the pIC condition moved toward the homeostatic condition samples. This again suggests that pIC alone may not be sufficient to induce a sustained immunogenic antiviral response in cDC1s, while pIC embedded within a lipid environment is in line with previous studies.^19,20 Taken together, unbiased transcriptional profiling reveals that the uptake of non-adjuvanted eLNPs induces a homeostatic maturation program in cDC1s that is comparable to the transcriptional profile in mature cDC1 in SS conditions, due to uptake of apoptotic cells.⁹ The uptake of pIC-LNPs or CpG-LNPs induces immunogenic cDC1 maturation, while treatment with pIC results in a transient immunogenic response.

Figure 1.

Uptake of non-cargo-loaded LNPs drives activation of a homeostatic cDC1 maturation program

(A) Experimental set-up. For details, see materials and methods.

(B) Non-integrated UMAP plot featuring the combined treatment and time point annotation of the nine conditions together. Subsequently, the UMAP is split per condition.

(C) Non-integrated UMAP plot featuring the SS condition with cDC1 maturation stage annotation.

(D) Harmony UMAP plot featuring the cDC1 maturation stage annotation of the nine conditions integrated together. We refer to the material and methods section for more information on the “other cDC1” clusters.

(E) MDS plot from the muscat pseudobulk DS analysis showing the level of similarity between pseudobulk cluster-sample instances of the CITE-seq. Only the EM cluster was included for the 2-h samples and the LM cluster for the 8-h samples. For the SS sample, three clusters were included: immature, EM, and LM. n = 4 for all conditions, except CpG-LNP at 8 h (n = 3). EM: early mature; LM: late mature.

Figure 5.

Homeostatic and immunogenic mature DCs provoke distinct T cell outputs

(A) Histogram plots showing CXCL16, CXCL9, and CCL5 expression of splenic CCR7⁻ and/or CCR7⁺ cDC1 populations of representative samples in SS, 8 h after injections with eLNP, or pIC-LNP. MFI plots show the expression of the CCR7⁻ and/or CCR7⁺ populations of the three conditions (n = 4 for all). One-way ANOVA corrected for multiple testing by Tukey’s multiple comparisons test with a single pooled variance (for CXCL9 and CCL5). Brown-Forsythe and Welch ANOVA corrected for multiple testing by Dunnett’s T3 multiple comparisons test with individual variances (for CXCL16).

(B) MFI of CD80 of CCR7⁻ and/or CCR7⁺ cDC1s in SS (n = 4) or 12 h after injection with eLNPs made with the ionizable lipid ALC-0315 (n = 5), eLNPs made with the ionizable lipid SM-102 (n = 5), LNPs containing OTI and OTII peptide (OTI/OTIIpeptide-LNPs) (n = 5), mRNA-LNPs (n = 3), or pIC-LNPs (n = 5). All LNPs were made with the ionizable lipid ALC-0315 except otherwise indicated. Kruskal-Wallis test corrected for multiple testing with Dunn’s multiple comparison test.

(C) Experimental set-up for data shown in (D), (E), and (F): CD45.2 acceptor mice were injected with 0.5 million CD45.1.2 OTII cells. Two days later, the mice were injected with different LNPs, and at two time points, the OTII cells were checked in the blood by flow cytometry.

(D) Percentage of OTII cells of CD4 T cells in the blood at days 7 and 12 of control mice (n = 5) or after injection of eLNPs (n = 5), OTIIpeptide-LNPs (n = 5), and OTIIpeptide-pIC-LNPs (n = 4). Two-way ANOVA, Tukey’s multiple comparisons test with single pooled variance.

(E) Percentage of Foxp3+ OTII cells in the blood 7 and 12 days after injection of OTII peptide-LNPs and OTII-peptide-pIC-LNPs.

(F) Percentage of CD44⁺CD62L⁺ and CD44⁺CD62L⁻ OTII cells in the blood at day 7 (n = 5 for OTII peptide-LNPs and n = 4 for OTII peptide-pIC-LNPs) and day 12 (n = 4 for OTII peptide-LNPs and n = 3 for OTII peptide-pIC-LNPs) after LNP injections. Mixed-effects model corrected for multiple testing by Sídák’s multiple comparisons test with a single pooled variance. Representative contour plots (right) showing the expression of CD44 versus CD62L, gated on OTII cells. For (E and F), two datapoints were omitted on day 12 due to low counts of OTII cells.

(G) Experimental set-up for data shown in (H), (I), and (J), indicating that CD45.2 acceptor mice were injected with 0.3 million CD45.1.2 OTI cells. One day later, the mice were injected with different LNPs. Three days later, the OTI cells were checked in the blood by flow cytometry. At day 9, the mice were injected with SIINFEKL-lentivirus, and at day 13, the blood was checked again, and an ex vivo cytotoxicity assay was performed.

(H) Percentage of OTI cells of CD8 T cells in the blood at days 3 and 13 (left) of control mice (n = 4) or after injection of eLNPs (n = 5), OTI peptide-LNPs (n = 6) and OTI peptide-pIC-LNPs (n = 6). Two-way ANOVA, corrected for multiple testing by Tukey’s multiple comparisons test with a single pooled variance. Fold-change ratio (day13/day 3) of OTI cells in the blood (right). One-way ANOVA, corrected for multiple testing by Tukey’s multiple comparisons test with a single pooled variance.

(I) Percentage of CD44⁺CD62L⁺, CD44⁺CD62L⁻, and CD44⁻CD62L⁻ OTI cells in the blood at day 13 (n = 6 for both OTII peptide-LNPs and OTII peptide-pIC-LNPs). Two-way ANOVA with Geisser-Greenhouse correction, corrected for multiple testing by Sídák’s multiple comparisons test with individual variances.

(J) OTI cytotoxicity was measured as the ratio of live CTV-labeled (WT thymocytes) and CTR-labeled (OVA thymocytes) after co-culture with splenocytes. Kruskal-Wallis test, corrected for multiple testing by Dunn’s multiple comparisons test. The mean ± SEM is shown in (A), (B), (D), (E), (F), (H), (I), and (J). Representative of two experiments for (A), (B), (C), (D), (E), (F), (G), (H), and (I). For (J), n = 1. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001; not significant is not shown.

The homeostatic cDC1 maturation signature is conserved across tissues and subsets

To further evaluate the transcriptional differences between SS immature cDC1s and LM cDC1s 8 h p.i. of eLNPs or pIC-LNPs, a pseudobulk differential state (DS) analysis was performed. This enabled us to take advantage of the replicates present in each group, perform a robust differential expression (DE) analysis and visualize the results on a heatmap (Figure S1A) and triwise plot (Figure 2A and Table S4). Genes that are commonly upregulated or downregulated during late maturation, both during homeostatic and immunogenic conditions, can be found in the triwise plot on the left side of the horizontal axis or on the right side, respectively. Genes that uniquely belong to the homeostatic or immunogenic program can be found enriched according to the eLNP or pIC-LNP axis, respectively. This largely confirms the previous observations made by the Malissen lab, which described the so-called MAT^OFF and MAT^ON genes that are turned OFF or ON during maturation.²⁰ MAT^OFF genes are highly expressed in immature cDC1s and include genes like Itgae (encoding CD103), Cd207 (Langerin), Tlr12, Clec9a, and Cd36 (both linked with engulfment^9,21) or March1 (implicated in the regulation of MHC class II²²), which is in line with the increased phagocytotic capacity and decreased antigen presentation capacity of immature cDC1s. Genes shared between homeostatic and immunogenic LM cDC1s include Ccr7, Fscn1 (involved in DC migration²³), Cd63, or Mreg (both linked to vesicular transport^24,25,26). Genes uniquely associated with homeostatic LM cDC1s include the cholesterol efflux transporter Abcg1, the apolipoproteins Apol7c and Apol10b, Mical3, Slco5a1, and H2-M2, most of them still with unknown function. Finally, genes uniquely associated with the immunogenic maturation program include chemokines Cxcl11 or Cxcl10 or interferon (IFN)-stimulated genes, such as Isg20, Gbp7, and Ifnb1. In line with previous studies,^27,28,29 immunogenic maturation programs are tailored to specific types of pathogens, which can be appreciated when comparing the transcriptomic profiles in CpG-LNP vs. pIC-LNP-induced LM cDC1s (Figure 2B and Table S5). Genes enriched in cDC1s that engulfed pIC-LNP include Card19, Ifnl3, and Ifnl2, all associated with the IFN response or the induction of an antiviral state,^30,31 but also Pcp4 (involved in calcium binding³²), Rbms2 (DNA/RNA binding³³), and Sema4f (known for migration of neurons³⁴). Genes specifically associated with LM cDC1s upon engulfment of CpG-LNPs include Il12a, Ccl3, and the A20-binding protein Tnip3.³⁵ Of note, genes that were previously associated with so-called tolerogenic DCs, such as Btla,³⁶ Cd274¹¹ (encoding CD274 or PD-L1), and Ido1,³⁷ were expressed to a much higher extent in immunogenic compared to homeostatic mature cDC1s (Figure 2B red font). This suggests that they may have an immunoregulatory role involved in dampening hyperinflammation, rather than having a role in SS conditions, which is in line with previous observations.^20,37,38 Despite the overall similarity in transcriptomic profile between SS mature cDC1s and cDC1s that matured due to uptake of eLNPs, several genes came out as DE (Figure S2A and Table S6). The differential gene expression could be linked to the difference in cargo (apoptotic cells versus influx of lipids) or the difference in timing (LM eLNP-induced cDC1s were all sorted 8 h p.i., while SS mature cDC1s may represent a more heterogeneous population regarding the time since maturation started). Injection of LNPs has previously been shown to cause activation of a pro-inflammatory innate immune response in many cell types.^{16,39,40,41,42} Still, when analyzing the transcription profile of cDC1s that engulfed eLNPs, this does not seem to be the case. While some Ifit genes appeared slightly skewed toward the eLNP condition on the triwise plot, their expression was much more prominent in DCs that took up pIC-LNPs. In addition, we did not detect major induction of cytokines or chemokines by eLNPs (see also Homeostatic and immunogenic mature DCs provoke distinct T cell responses and Figure S6A and B). Therefore, we conclude that uptake of eLNPs does not induce a prominent inflammatory signature in cDC1s.

Figure 2.

Common and distinct transcriptomic profiles mark homeostatic and immunogenic mature cDC1s

(A and B) Triwise plots featuring the relative expression of genes in three separate conditions. (A) Comparison between immature cDC1s in SS and LM cDC1s 8 h after injection with eLNP or pIC-LNP. (B) Comparison between LM cDC1s 8 h after injection with eLNP injection, pIC-LNP, or CpG-LNP. For all details, see materials and methods.

(C) Plot showing the average mRNA expression (Seurat module score) for a selection of ISGs.

(D) Scatterplot indicating the type of DC maturation of various gene lists available in the literature. The maturation type, plotted on the x axis, is a score calculated as the difference in overlap between each literature gene list and the top 200 uniquely upregulated immunogenic and homeostatic cDC1 maturation genes, according to adjusted p value, respectively, in our CITE-seq data (for details on the calculation of the score, see Figure S2C). The log10 transformation of the literature gene list length determines the position on the y axis, indicative of the number of genes represented in the gene list. Literature gene lists with a score < −5 are colored in red (immunogenic), between −5 and 5 are colored in gray (undefined), and > 5 are colored in green (homeostatic). The gene lists include the names used in the original papers and are labeled with basic information about the type of DC maturation, the manuscript, the sequencing technology, and the tissue. BMDC: bone marrow-derived dendritic cell; EM: early mature; LM: late mature, SI PP: small intestinal Peyer’s patches.

To validate the dataset and compare it to previous gene expression studies of homeostatic versus immunogenic mature DCs, the mRNA expression of several genes based on the clusters of the study of Ardouin et al.²⁰ is visualized in Figure S2B. This confirmed the expected expression of common, homeostatic, and immunogenic signatures in the different conditions, and in addition, showed that LM cDC1s in pIC-stimulated conditions started expressing genes associated with the homeostatic cluster, while they lost genes associated with an immunogenic signature. IFN-stimulated genes (ISGs) were largely absent in homeostatic mature cDC1s in SS conditions or after uptake of eLNPs and highly expressed in all immunogenic conditions in which TLR ligands were present (Figure 2C). Of note, while ISGs were expressed to a similar extent after uptake of pIC-LNPs or pIC in EM cDC1s, they were lost upon stimulation with pIC in LM cDC1s.

Over the past few years, several studies have been published describing matured/activated DCs using different terminologies, such as immunoregulatory DC (mregDCs), CD103^INT cDC1s (INT, intermediate), or DC3.^11,13,43 Unfortunately, the name cDC3s has been used as well to classify a distinct DC subset with a distinct ontogeny.⁴⁴ This has created some confusion in the field as the relationship between all these “subsets” is not always obvious. To resolve the confusion, we performed a meta-analysis in which we compared different maturation signatures from the literature to the homeostatic and immunogenic maturation signatures obtained in this study (for a detailed set-up of the analysis, see Figure S2C). In this way, we assessed how well our signatures were conserved across different DC subsets and tissues, and in addition, tried to reveal the connection and overlap between this growing assortment of cDC populations^{10,11,13,20,38,43,45,46,47,48} (Figure 2D and Table S7). A “maturation type” score was calculated for each DC maturation gene list extracted from various types of sequencing studies by comparing signatures of related mature DCs and immature DCs. This was done by determining the overlap between different MAT^ON gene lists taken from the literature, with the top 200 upregulated homeostatic or immunogenic cDC1 maturation genes obtained from our CITE-seq dataset. Subsequently, we calculated the maturation score by subtracting the immunogenic overlap from the homeostatic overlap (see Figure S2C and materials/methods). Positive scores > 5 were annotated as homeostatic (green color) and scores < −5 as immunogenic (red color). This maturation score was then plotted against the size of the DC maturation gene list (on the y axis) to visualize the effect of the gene list size. As shown in Figure 2D, murine cytomegalovirus infection (derived from a study of Ardouin et al.²⁰) yielded the highest immunogenic maturation score, while migratory DCs in the mediastinal LN in SS conditions (derived from a study of Miller et al.³⁸) showed a high homeostatic maturation score. On the other hand, 18-h treatment with pIC (derived from a study of Ardouin et al.²⁰) induced mature cDC1s with a maturation type score around 0, which is in line with our findings that at late time points, upon pIC stimulation, the cells lose immunogenic and gain homeostatic maturation genes, yielding a net score of 0. It can be noted that, independent of the tissue/LN and independent of the subset (cDC1s or cDC2s), mature DCs isolated in SS conditions uniformly clustered in the homeostatic mature group, while almost all DCs treated with TLR ligands ended up in the immunogenic mature group. Also, mature DCs derived from the tumor (annotated as “mregDC”) clustered within the homeostatic mature group, which is in line with the original description.¹¹ Similarly, the signatures derived from neonatal CD103^INT DCs¹³ overlapped well with the signatures from homeostatic mature cDC1s in the spleen, indicating that they all refer to the same maturation state. On the contrary, confinement-induced DC maturation, which was reported as a physiological trigger for homeostatic DC maturation⁴⁹ showed a higher overlap with the immunogenic maturation signature. This is in line with previous reports showing that mechanical stress induces pro-inflammatory pathways in DCs.⁵⁰ Finally, a recently published study showed transcriptional signatures in DCs in the muscle after uptake of eLNP versus mRNA-LNPs, respectively.⁴² The study revealed specific induction of an ISG cluster only in DCs that took up mRNA-LNP, not eLNPs. We reanalyzed the DC clusters in their dataset and found that, similar to what we show here, the uptake of eLNPs in muscle DCs induces a homeostatic maturation program. We added their study to the meta-analysis shown in Figure 2D. To assess whether certain conditions induce homeostatic or immunogenic DC maturation programs, we generated a web-based interactive tool for the meta-analysis, where gene lists of mature DCs (MAT^ON genes) can be uploaded and plotted against the maturation score (https://www.single-cell.be/Spleen_cDC1_LNP_induced_Homeostatic_and_Immunogenic_maturation).

In conclusion, engulfment of LNPs induces DC maturation in vivo. While eLNPs induce a homeostatic maturation program, uptake of TLR-adjuvanted LNPs induces an immunogenic program. This indicates that cDCs do not decode LNPs as “danger,” even though injection of empty LNPs has been shown to induce a proinflammatory response in many other cell types. Of note, pIC by itself is not sufficient as a trigger to induce a sustained IFN response, suggesting that the induction of a proper antiviral immune response might require the combined detection of viral pathogen-associated molecular patterns (PAMPs) (such as pIC) within the context of a cholesterol-rich membrane, boosting their adjuvanticity. A meta-analysis confirmed that many of the previously annotated DC “subsets,” such as mregDCs or CD103^INT DCs, refer to the homeostatic mature state rather than constituting separate DC subsets. Therefore, we propose to use a more unified nomenclature based on the recognition of mature DCs as a DC state and classifying them as homeostatic or immunogenic mature cDC1s/cDC2s rather than classifying them as a distinct subset.

Different transcriptional programs drive the homeostatic and immunogenic maturation programs

To determine differential gene regulatory networks during homeostatic and immunogenic maturation of cDC1s, the generated CITE-seq dataset was analyzed by using the DoRothEA database and VIPER statistical testing, which deduces transcription factor (TF) activity based on the expression of TF target genes. This allowed us to identify TFs with specific expression patterns across the different conditions and maturation stages and visualize them with heatmaps. The “common on” group includes TFs for which the activity was higher in all LM clusters independent of the treatment as compared to the immature cluster of the SS condition (Figure 3A); the “common off” group includes TFs for which the activity was lower in LM clusters (Figure 3B); the homeostatic group lists the top TFs for which the activity was higher in LM clusters of SS and eLNP conditions (Figure 3C); and the immunogenic group lists the top TFs for which the activity was higher in LM clusters of pIC-LNP and CpG-LNP conditions (Figure 3D, further specifications of the defined groups can be found in the materials and methods section). TFs that are commonly induced in homeostatic and immunogenic conditions include the estrogen receptor alpha (known from in vitro data to be implicated in DC development⁵¹), RFX2, which may be implicated in survival of immune cells,⁵² c-REL, STAT4, and STAT6 (Figure 3A). TFs that are commonly turned off during maturation include the cell cycle regulator MYC, which is in line with previous findings,⁸ EGR1, a negative regulator of DC immunogenicity,⁵³ ZEB1, a zinc-finger E homeobox-binding TF with yet unknown function in DCs⁵⁴ and, RFX5, a known regulator of MHC class II expression^55,56 (Figure 3B). Amongst the TFs that appear uniquely linked to the induction of the homeostatic maturation program are the cholesterol efflux and cholesterol metabolism regulators LXRα/LXRβ (NR1H2/3) and SREBP1/SREBP2 (SREBF1/2), in line with previous findings.^9,20,57,58 In addition, BACH2, which is critically needed for Treg induction,⁵⁹ SMAD3, known to be essential for the activation of the TGFβ1 promoter in DCs,⁶⁰ and the AP1 members c-JUN and JUNB, came out as homeostasis-specific (Figure 3C). Finally, TFs that are uniquely linked to the immunogenic maturation program are members of the IRF family: IRF1, IRF2, IRF3, IRF7, and IRF9, members of the STAT family that are specifically activated by IFN signaling or IL6⁶¹—STAT1, STAT2, and STAT3—and members of the NF-κB family: NF-κB1, RelA, and RelB (Figure 3D). These data show a clearly distinct pattern of gene regulatory networks being induced in homeostatic versus immunogenic conditions. Previous studies revealed that the canonical IκB kinase IKK2 is involved in homeostatic DC maturation and migration of skin DCs to dLNs in an Fscn1-dependent manner.^62,63 In this regard, we were surprised to note that none of the NF-κB subunits were found to be specific to the homeostatic maturation program. Therefore, we decided to validate the role of IKK2 in homeostatic cDC1 maturation. Deletion of the Ikk2 allele in the cDC compartment by using CD11c-cre showed signs of autoimmunity, as reflected by increased presence of monocytes, activated B cells, and neutrophils (data not shown). To avoid this, we crossed the Ikk2^fl/fl mice to Xcr1-Cre mice (IKK2ΔcDC1), generating a conditional loss of the Ikk2 allele in the cDC1s only. This did not affect the myeloid or lymphoid compartment (Figure 3E). In line with previous data,^62,63 we noticed a slight decrease in mature cDC1s. Closer inspection revealed that this decrease was caused by a reduction of the LM CD63⁺CCR7⁺ stage, while the EM CD63⁻CCR7⁺ stage appeared unaffected by the loss of IKK2 (Figures 3F and 3G). This made us wonder whether IKK2 was needed to drive the onset of the maturation program, and we tested the contribution of IKK2 to the expression of several known maturation genes. None of them, including Fscn1, appeared affected by the loss of IKK2 (Figure 3H), which contrasted with data obtained in DCs in the skin.⁶² In that study, the qPCR analysis was performed on total cDC1s. Hence, the decrease in expression of maturation genes might have reflected the loss of mature cells rather than reflecting a direct effect of IKK2 on the induction of the maturation program. Our data suggest that in homeostatic conditions, the IKK2/NF-κB pathway may be needed at later stages of the maturation process to ensure survival of mature DCs rather than driving the onset of the homeostatic transcriptional program.⁶⁴ We conclude that even though TF analysis predicts large differences in gene regulatory networks and specific TFs involved in common, homeostatic, and immunogenic cDC1 maturation, more work is needed to identify the TF network that orchestrates the onset of the homeostatic maturation program.

Figure 3.

Different transcriptional programs drive the homeostatic and immunogenic maturation programs

(A–D) Heatmap of TF activity during cDC1 maturation analyzed using DoRothEA. The color scale of the heatmap represents the scaled TF activity (inferred from mRNA expression of TF targets). The columns represent the clusters from the conditions as indicated. For all details, see materials and methods.

(E) Percentage of myeloid cells and lymphoid cells in the spleen of IKK2^fl/fl (n = 4) and IKK2 ^fl/flXcr1-Cre (IKK2ΔcDC1 mice) (n = 5) mice.

(F) Percentage of CCR7⁺ cDC1s and cDC2s in the spleen of IKK2^fl/fl and IKK2ΔcDC1 mice (left). Percentage of CD63⁻CCR7⁺ (EM) and CD63⁺CCR7⁺ (LM) cDC1s in the spleen of IKK2^fl/fl and IKK2ΔcDC1 mice (right). Two-way ANOVA corrected for multiple testing by Sídák’s multiple comparisons test with a single pooled variance.

(G) Cell number of CCR7⁺ cDC1s and cDC2s in the spleen of IKK2^fl/fl and IKK2ΔcDC1 mice (left). Cell number of CD63⁻CCR7⁺ (EM) and CD63⁺CCR7⁺ (LM) cDC1s in the spleen of IKK2^fl/fl and IKK2ΔcDC1 mice (right). Two-way ANOVA corrected for multiple testing by Sídák’s multiple comparisons test with a single pooled variance.

(H) Relative expression of Ikk2, Ccr7, Fscn1, Ccl22, Cd40, Cd80, and Cd86 in sorted CCR7⁻ and CCR7⁺ cDC1s from the spleen of IKK2^fl/fl and IKK2ΔcDC1 mice (n = 5), measured by RT-qPCR. The expression was normalized to housekeeping genes Sdha and Ywhaz. Two-way ANOVA corrected for multiple testing by Sídák’s multiple comparisons test with a single pooled variance. The mean ± SEM is shown in (E–H). Representative of two experiments for (F) and (G). ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗and p < 0.0001; not significant is not shown. EM: early mature; LM: late mature.

Different cell surface markers can be used to distinguish the homeostatic from the immunogenic maturation state in cDCs

Though the presence of a homeostatic and immunogenic maturation state in DCs has been recognized for a long time,^20,65,66 the field still lacks good markers to carefully distinguish the two maturation states. Previous studies have led to the identification of the so-called regulatory or tolerogenic markers, such as CD274 (PD-L1), CD200, or IDO1, but close comparison between the two maturation states revealed that these are not uniquely associated with the homeostatic mature condition and are even more highly expressed during immunogenic maturation (Figure 2B). Therefore, we analyzed the antibody-derived tags (ADTs), derived from a library of more than 150 antibodies that were sequenced along with the transcriptomes in the CITE-seq analysis to identify appropriate cell surface markers to distinguish the maturation states (Table S8). Heatmaps were constructed based on DE surface markers in homeostatic versus immunogenic conditions, both at the early and at the late maturation stage (Figure S3A and Tables S9 and S10). Only three surface markers came out as DE at the early stage, with CD69 and especially CD274 showing prominent expression in immunogenic mature cDC1s, while CD62P was particularly enriched in the early homeostatic mature cDC1s (Figure S3A). At the LM stage, more DE surface markers were identified, which were all plotted on a triwise plot for better visualization (Figure 4A). In line with previous data,⁹ CD80 is highly enriched in immunogenic mature DCs versus homeostatic mature DCs. Similar results were obtained for other markers such as CD201 (described to be necessary for responsiveness to therapy with activated protein C in cDC1⁶⁷), CD150 (a self-ligand receptor inhibiting CD40-induced signaling and capable of mediating measle virus entry in monocyte-derived DCs^68,69), Ly6A-Ly6E (a marker that so far has mostly been associated with activated pDCs in response to endogenous type I IFN⁷⁰), CD14 (a marker used to describe the inflammatory “cDC3 subset” or monocyte-like cDC2s^71,72), and CD69 (induced on DCs after injection with TLR triggers⁷³) (Figure 4A). Markers such as CD86 and CD200 (associated with the mregDC signature and known to increase immunosuppression¹¹) appeared slightly skewed toward the pIC-LNP axis, while TIM4 (phosphatidylserine receptor involved in the engulfment of tumor cells by cDC1s⁷⁴) was uniquely associated with CpG-induced maturation. On the other hand, at the homeostatic side of the spectrum, specific expression of proteins, such as CD63, a tetraspanin associated with cholesterol distribution through endosomes and exosomes,²⁶ CD83, a regulator of MHC class II,⁷⁵ CD278 (also named ICOS), and the glycolipid antigen presentation molecule CD1d,⁷⁶ was found. The expression of a selected list of marker proteins was validated by flow cytometry (Figures 4B and S4A). For technical reasons, the CCR7⁺ population was analyzed after LNP-driven maturation, rather than the LNP+ population, due to the instability of Cy5 at later time points, as shown in the gating strategy (Figure S3B). At early time points, CD274 and CD69 were expressed at a higher level on immunogenic mature cDC1s (Figure S4A), which was in line with the CITE-seq data (Figure S3A). On the contrary, we could not confirm the homeostatic role of CD62P as it also appeared more expressed in immunogenic conditions. Due to the limited differences between EM cDC1s, we focused on markers that could be used to annotate the LM stage (Figure 4B). Several markers, such as CD80, CD86, CD274, CD200, CD69, or CD201 were suited to identify immunogenic mature DCs, while CD63 and CD1d appeared good candidate markers to label homeostatic mature DCs (Figure 4B). Based on their expression levels, we selected a range of pairwise combinations of surface markers to evaluate how well they were suited to deconvolute homeostatic from immunogenic mature DCs. In the left panel in Figure 4C, the commonly used maturation markers CCR7 and MHCII are shown, which separate mature DCs from immature DCs, but cannot distinguish the homeostatic from immunogenic mature state. Different pairwise combinations with CD80, CD86, CD200, or CD274 as immunogenic markers and CD63 or CD1d as homeostatic markers yielded a clear separation of the homeostatic and immunogenic mature populations (Figure 4C). Of note, some of these markers only become DE upon interaction with other cell types. A prominent example is CD80, which is known to be downregulated on homeostatic mature DCs in a CTLA4-dependent manner upon interaction with Tregs after migration to the T cell area.^9,77 This can be observed when we adopt a bone marrow chimeric approach in which 50% of the transferred bone marrow is wild-type (WT) and 50% CCR7-deficient (labeled with different congenic markers). Only CCR7^+/+ homeostatic mature DCs that migrate to the white pulp lose their expression of CD80, while CCR7^−/− homeostatic mature DCs (gated as CD63⁺) that cannot migrate, retain increased CD80 levels on the surface (Figure S4B). Similarly, when adopting a similar gating strategy on in vitro cultivated cDC1s (by using the coculture system with cDC1s grown on an OP9-DLL1 feeder cells⁷⁸), it is clear that CD80 and CD200 remain high in homeostatic conditions, while CD1d is not downregulated in immunogenic conditions (Figure S4C). The only marker that can be used to discriminate homeostatic from immunogenic mature DCs in vitro appears to be CD63. This implies that the use of CD80 and CD86 as sole markers to monitor immunogenic activation of DCs may be useful in vivo, but not in vitro. Since mature cDC1s and cDC2s converge in the mature state, both at the transcriptional and at the cell surface protein level,^11,43,79 we assessed whether the markers that we identified on cDC1s could be used to dissect homeostatic from immunogenic mature cDC2s as well. To this end, we investigated whether we could use LNPs to induce cDC2 maturation. As shown before,⁹ cDC2s did not engulf eLNPs substantially, and as a consequence, eLNP injection did not result in a clear induction of cDC2 maturation. On the other hand, injection of pIC-LNPs did induce cDC2 maturation, possibly also as a result of bystander-induced maturation (Figures S5A and S5B). Nevertheless, similar expression patterns could be observed as we had seen before for cDC1s on the mature cDC2s (Figures S5C and S5D). CD80, CD86, CD274, CD200, and CD201 were highly expressed in immunogenic mature cDC2s, while CD1d and CD63 were highly expressed in the homeostatic conditions, although the results for CD63 were less outspoken (Figure S5D). Based on these data, a high-parameter flow cytometry panel was designed with a selection of immunogenic and homeostatic markers that together could properly distinguish the two DC maturation states (Figure S3C). Once we obtained this “DC maturation state” panel, we decided to validate it in different conditions. First, we used the immunogenic Toxoplasma gondii (T. gondii) model and harvested the spleen 6 days p.i. with 500 tachyzoites (Figure 4D). Using CCR7 to distinguish homeostatic from immunogenic matured cDC1s would be ineffective since the percentage of CCR7⁺ cDC1s was similar with or without T. gondii infection (Figure 4E). However, applying the DC maturation panel, such as the one presented in Figure S3C, allowed us to annotate the mature cDC1s in the T. gondii-infected mice as immunogenic by high expression of CD80, CD86, CD274, CD200, CD69, or CD201 and low expression of CD63 and CD1d (Figures 4F and S4D). For cDC2s, the immunogenic markers showed increased expression in the T. gondii-infected mice, but CD63 and CD1d did not show higher expression on mature cDC2s in control mice compared to T. gondii-infected mice (Figure S5E). Finally, we used the DC maturation panel to monitor the maturation state of tumor-infiltrating DCs in a model of subcutaneous MC38 colon carcinoma in the tumor and tumor-draining lymph node (tdLN) after intratumoral injection of eLNPs, pIC-LNPs, or no injection of LNPs as control (Figure 4G). In non-injected mice, CD80 was highly expressed in mature CCR7⁺ cDC1s in the tumor but became downregulated in mature CCR7⁺ cDC1s in the tdLN, showing that the tumor cDC1s matured in a homeostatic manner, as reflected by loss of CD80 expression once DCs reached the LN.⁹ Of note, 8 h upon single intratumoral injection with pIC-LNPs, but not eLNPs, CD80 expression levels were increased in tdLN CCR7⁺ cDC1s (Figure 4H). Similarly, CD63, as a representative homeostatic marker, was high in CCR7⁺ cDC1s in non-injected mice, both in the tumor and in the tdLN, reflecting their homeostatic mature state. Intratumoral injection with pIC-LNPs led to a downregulation of CD63 in the tumor and tdLN CCR7⁺ cDC1s, indicating the potential of adjuvanted LNPs to redirect the maturation state of tumor-infiltrating DCs from homeostatic mature into immunogenic mature DCs (Figure 4H). Also for cDC2s, the expression patterns of CD80 and CD63 in tumor and tdLN of non-injected mice indicated a homeostatic maturation pattern of tumor-infiltrating cDC2s (Figure S5F). Note that there was also a striking contrast in CD80 expression levels on homeostatic mature cDC2s in the tumor compared to the tdLN, showing again that CD80 can only be used as a discriminating marker in the lymph node or in the spleen. Similar to what we observed for cDC1s, intratumoral injection with pIC-LNPs but not eLNPs was able to increase the expression of CD80 in CCR7⁺ cDC2s in the tdLN, indicating its potential to redirect the tumor DC maturation program toward the immunogenic state. Still, both for the CD80 levels in the tumor and the CD63 levels in the tdLN after pIC-LNP injection, the results differed from what we observed for cDC1s, showing that some markers may behave differently in cDC1s in comparison to cDC2s.

Figure 4.

Different cell surface markers can be used to distinguish the homeostatic from the immunogenic maturation state in cDCs

(A) Triwise plot of the CITE-seq data featuring the relative expression of surface proteins (ADT) in LM cDC1s 8 h after injection with eLNP pIC-LNP and CpG-LNP. For details, see materials and methods.

(B) Mean fluorescence intensity (MFI) of a selection of markers in the CCR7⁻ and CCR7⁺ populations of splenic cDC1s in SS (n = 3), cDC1s 12 h after eLNP injection (n = 4), and cDC1s 12 h after pIC-LNP injection (n = 4). The statistics are shown only for the comparison between eLNP and pIC-LNP groups. Two-way ANOVA corrected for multiple testing by Sídák’s multiple comparisons test with a single pooled variance.

(C) Representative contour plots showing the expression of MHC class II versus CCR7 before the dashed line (left) and the expression of selected immunogenic markers on the y axis versus selected homeostatic markers on the x axis.

(D) Experimental set-up of the T. gondii experiment showing that 500 T. gondii tachyzoites were injected intraperitoneally 6 days before analysis of the spleen.

(E) Percentage of CCR7⁺ cDC1s in control mice (n = 5) or mice infected with T. gondii (n = 5).

(F) MFI of a selection of markers showing the CCR7⁻ and CCR7⁺ populations of cDC1s of control mice (n = 5) and mice infected with T. gondii (n = 5). Two-way ANOVA corrected for multiple testing by Sídák’s multiple comparisons test with a single pooled variance.

(G) Experimental set-up of the MC38 colon carcinoma model showing that MC38 cells were subcutaneously injected 12 days before intratumoral treatment with either eLNPs, pIC-LNPs, or control (not injected). After 8 h, the tumor and tdLN were analyzed.

(H) MFI of CD80 and CD63 of CCR7⁻ and CCR7⁺ cDC1s in the tumor and tdLN. Two-way ANOVA corrected for multiple testing by Tukey’s multiple comparisons test with a single pooled variance. The mean ± SEM is shown in (B), (E), (F), and (H). Representative of two experiments for (B), (D), (E), (F), (G), and (H). ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001; not significant is not shown.

Taken together, based on the CITE-seq analysis of cDC1s matured after uptake of different types of LNPs, we generated a DC maturation flow cytometry panel that includes several surface markers that together will reliably dissect homeostatic from immunogenic matured cDC1s in vivo.

Homeostatic and immunogenic mature DCs provoke distinct T cell outputs

To determine functional differences between cDC1s after injection of either eLNPs or pIC-LNPs, DE genes belonging to the GO term cytokine activity (GO: 0005125) were highlighted on triwise plots featuring the relative expression of genes in immature SS cDC1s versus LM cDC1s 8 h p.i. of eLNPs and pIC-LNPs (Figure S6A and Table S4) or in LM cDC1s 8 h p.i. of eLNPs, pIC-LNPs, and CpG-LNPs (Figure S6B and Table S5). Several cytokines and chemokines with preferential expression in homeostatic versus immunogenic conditions could be discerned, such as Cxcl16 and Ltb after eLNP injection or Ccl5 and Cxcl9 after pIC-LNP injection. Intracellular flow cytometry staining validated these findings (Figure 5A) and showed that intracellular levels of CXCL16 were higher in mature cDC1s both in SS and upon engulfment of eLNPs, while CXCL9 and CCL5 reached higher levels in mature cDC1s that had engulfed pIC-LNPs (Figure 5A). Similarly, mature cDC2s showed higher intracellular levels of CXCL16 in SS, while CXCL9 and CCL5 levels were higher after injection with pIC-LNPs (Figure S6C). To assess whether uptake of eLNPs versus pIC-LNPs in cDC1s influenced downstream T cell responses, we monitored the effect of incorporating antigenic cargo inside LNPs. First, we assessed the role of the ionizable lipid. We noticed that the induction of the homeostatic maturation program in cDCs by uptake of empty LNPs appeared independent of the type of ionizable lipid used, as formulation of eLNPs with the more immunostimulatory SM-102 lipid⁴⁰ yielded similar effects (Figure 5B). Secondly, we assessed the role of the cargo and compared the immunogenicity of peptide-loaded versus mRNA-loaded LNPs. Twelve hours p.i. with eLNPs, OTI/II peptide-loaded, or ovalbumin (OVA) mRNA-loaded LNPs, DCs were isolated from the spleen and immunophenotyped with different maturation markers. While uptake of mRNA-loaded LNPs led to immunogenic DC maturation, incorporation of peptide-loaded LNPs did not, as monitored by the surface expression levels of CD80 (Figure 5B) or contour plots combining CD80 and CD63 (Figure S6D). This indicates that the cargo is the determining factor in the immunogenicity of LNP-based mRNA vaccines in DCs, which is in line with a recent study in the muscle.^14,42 Next, we assessed the T cell responses after injection of peptide-loaded LNPs versus peptide-pIC-LNPs. CD45.2 WT mice were injected with 500,000 sorted naive CD45.1.2 OVA-specific CD4⁺ T cells (OTII) (Figure 5C). Two days later, the acceptor mice were injected with either eLNPs, LNPs encapsulating OVA_323-339 peptide (OTII peptide-LNP) (coupled to an E10 tag to increase the net negative charge^80,81), LNPs encapsulating both the peptide and pIC (OTII peptide-pIC-LNP), or no injection as a control. At 7 and 12 days p.i., the number of OTII cells in the blood was monitored. In the mice that received OTII peptide-pIC-LNPs, the percentage of OTII cells remained high over time, as also reflected by a continued increase in the percentage of Ki67⁺ OTII cells over time (Figures 5D and S6E). On the other hand, in the mice that received OTII peptide-LNPs, the percentage of OTII cells initially increased at day 7 but then dropped at day 12 in comparison to day 7 (Figure 5D), which was also reflected by a marked reduction in the percentage of Ki67⁺ OTII cells at day 12 (Figure S6E). As migratory DCs have been described to promote the development of antigen-specific Tregs,^82,83 the expression of FOXP3 in the OTII cells was assessed. No induction of FOXP3⁺ OTII cells could be observed after injection of OTII peptide-LNPs or OTII peptide-pIC-LNPs (Figure 5E). Overall, the expression of FOXP3 appeared very low (Figure S6F), suggesting that conversion of naive CD4 T cells to peripheral Tregs was very rare in the spleen. However, further phenotyping of the OTII cells revealed a distinct fate, with central memory CD62L⁺CD44⁺ OTII cells being induced upon injection with OTII peptide-LNPs, while effector memory CD62L⁻CD44⁺ OTII cells were induced upon injection with OTII peptide-pIC-LNPs (Figure 5F). This suggests that homeostatic or immunogenic mature DCs may differentially affect the localization and homing of CD4⁺ T cells, causing a preferential accumulation of central memory T cells in the LN in homeostatic conditions while favoring tissue recirculation of effector memory T cells in immunogenic conditions, respectively. In a second experimental set-up, CD45.2 WT mice were injected with 300,000 sorted naive CD45.1.2 OVA-specific CD8⁺ T cells (OTI) (Figure 5G). One day later, the acceptor mice were injected with either eLNPs, LNPs encapsulating OVA_257-265 peptide (OTI peptide-LNP) (coupled to the E10 tag), LNPs encapsulating the OVA_257-265 peptide and pIC (OTI peptide-pIC-LNP), or no injection as a control. Three days after LNP injection, the percentage of OTI cells in the blood was very similar, independent of whether OTI peptide-LNP or OTI peptide-pIC-LNPs were injected (Figure S6G). Nine days after LNP injection, the mice were re-challenged with a lentivirus expressing the OVA_257-265 peptide SIINFEKL to induce CD8⁺ T cell-mediated immune cell responses. Four days after rechallenging, we noted an extensive increase in OTI cells in the blood of mice that received OTI peptide-pIC-LNP injection in comparison to the mice that received OTI peptide-LNP (Figure 5H, left). This was especially prominent when calculating the ratio of OTI cells before and after rechallenge, which showed a strongly reduced increase in OTI cells in the mice that received OTI peptide-LNP in comparison to the other conditions (Figure 5H, right). The percentage of central memory CD62L⁺CD44⁺ OTI cells was comparable, while there was a slightly higher percentage of effector memory CD62L⁻CD44⁺ OTI cells upon OTI peptide-LNPs injection and a higher percentage of effector CD62L⁻CD44⁻ OTI cells upon OTI peptide-pIC-LNPs injection (Figure 5I). To evaluate the cytotoxicity of the OTI cells, splenocytes were co-cultured with a 1:1 mixture of CTV-labeled WT thymocytes and CTR-labeled OVA-expressing thymocytes (Figure 5G). The ratio of WT to OVA live cells was monitored to assess specific killing of OVA-expressing cells (which would result in a higher ratio). OTI cells isolated from mice that received OTI peptide-pIC-LNPs were more potent in killing OVA-specific target cells compared to OTI cells isolated from mice that received OTI peptide-LNPs (Figure 5J).

In conclusion, homeostatic or immunogenic mature cDC1s produce a different cytokine and chemokine spectrum, which may allow them to fine-tune their communication with different cell types. While we could not detect de novo Treg induction by homeostatic mature cDC1s, we did notice differential T cell outcomes when activated by DCs that matured in homeostatic versus immunogenic conditions. CD4⁺ T cells show a central memory phenotype after injection with peptide-LNPs, while CD4⁺ T cells expressed an effector memory phenotype after injection with peptide-pIC-LNPs, suggesting that the maturation state of DCs may influence the spatial distribution of T cell subsets. On the other hand, only injection of peptide-pIC-LNPs induced cytotoxic potential in CD8⁺ T cells, while antigen-specific CD8⁺ T cells induced after injection of peptide-LNPs appear deleted or rendered anergic over time. In conclusion, these data demonstrate that the packaging of non-immunogenic cargo in LNPs is not sufficient to induce a protective cellular immune response. Either the cargo must be immunogenic (by being provided as non-purified mRNA) or the LNP needs to be adjuvanted.

Discussion

The introduction of single-cell technologies in the immunology field allowed researchers to unlock the heterogeneity within immune subsets with an unprecedented resolution, leading to an increased understanding of their biology and functioning. Unfortunately, this went hand in hand with an overflow of annotations, and often, the same immune cell subset was assigned different names, creating confusion rather than clarity. The DC field has witnessed a growing number of DC clusters being annotated as mDC, migratory DC, mregDC, DC3, CD63⁺, or CD103^INT, without much knowledge on their mutual relationship. The goal of the present study was to identify a set of markers to classify them and clarify the signatures currently associated with mature DCs in the literature. In addition, the present study provided us with more insights into how LNPs are being decoded by DCs.

In general, DCs can be found in an immature (also called resting) state, where they scavenge the environment for the presence of antigens and “context” information or in a mature (also called activated) state, where they migrate to the T cell zone, present their antigen cargo to T cells and instruct them to generate an appropriate downstream output. Depending on the context in which an antigen is perceived at the time of engulfment, DCs can mature in a homeostatic or immunogenic manner, which represents the two endpoints on a maturation spectrum. Homeostatic mature DCs likely contribute to tolerance (although this remains to be formally proven), while immunogenic mature DCs induce protective immunity.^1,8,20,66,84 Most commonly, mature DCs are identified by expression of CCR7 and MHC class II, two markers that hallmark all mature DCs, irrespective of the maturation type. In this study, we provide a set of flow cytometry markers and transcriptional signatures that can be applied to unambiguously deconvolute homeostatic from immunogenic mature DCs across DC subsets and tissues.

To identify these markers, we used an LNP-based approach, leveraging our insights from a previous study where we found that uptake of empty LNPs induces a homeostatic maturation program in cDC1s in vivo, while the uptake of TLR-ligand adjuvanted LNPs induces an immunogenic maturation program.⁹ At the transcriptional level, the most prominent features that distinguish immunogenic from homeostatically matured DCs, are the presence of an IFN-stimulated gene signature in immunogenic mature DCs and the presence of an SREBP signature in homeostatic mature DCs, in line with previous findings.^9,20,57,58 By using CITE-seq technology, we identified DE ADTs that represent surface proteins that were selective for the two maturation types. A panel incorporating any combination of CD200, CD274, CD80, CD86, and CD201 as immunogenic markers and CD63 or CD1d as a homeostatic marker, turned out to be a potent tool to distinguish homeostatic from immunogenic mature DCs. The panel is universal, containing markers that can be applied for both the cDC1 and the cDC2 subsets, and was validated in a T. gondii infection model and in a tumor model. In addition, we used the maturation panel to monitor whether a tumor therapy (in this case, intratumoral injection with pIC-coupled LNPs) elicited the desired effect and steered DC maturation toward an immunogenic profile. Therefore, we believe that the DC maturation panel will be a highly valuable tool for assessing DC maturation status in the tumor and tdLN.

Several observations raised awareness on some commonly accepted markers of homeostatic mature DCs. Previous studies proposed that high expression of CD274 (PD-L1) can be used as a marker to delineate the so-called mregDCs in the tumor.¹¹ Our meta-analysis confirmed that mregDCs fall within the homeostatic mature DC cluster and, hence, will not be able to establish protective anti-tumor immunity. Still, CD274 expression does not typify them; on the contrary, high expression of CD274 appears to be a powerful marker to identify immunogenic mature DCs, just like CD200. This observation is not new,²⁰ but we want to highlight it again, as it raises concerns about the suitability of CD274 or CD200 as a marker for “regulatory” DCs. Along these lines, immunogenic mature DCs show high expression of other well-established immunoregulatory proteins, such as IDO1, which regulates the kynurenine pathway,^85,86,87 or BTLA, which is needed for the induction of Tregs and which is presently proposed as a marker to delineate “tolerogenic” DCs.⁸⁸ The elevated expression of these molecules specifically on immunogenic mature DCs might indicate an important role in dampening inflammation in highly immunogenic conditions.⁸⁴ In this regard, both IDO1 and BTLA have been shown to be instrumental for protection in experimental autoimmune encephalitis (EAE), a highly inflammatory autoimmune model in which autoantigens are presented to DCs in the presence of complete Freud’s adjuvant.^36,37 Due to the presence of strong adjuvants in the EAE model, DC maturation may occur in an immunogenic manner, and the concomitant expression of T cell regulators like IDO1 and BTLA may prevent overt T cell-mediated inflammation and pathology. On the contrary, in homeostatic conditions, when self-antigens are being presented by DCs in the absence of any infectious agents, these regulatory molecules may not be at play. Secondly, the induction of costimulatory molecules like CD80 and CD86 is often used in the immunogenic cell death field to “prove” the adjuvanticity of certain danger-associated molecular patterns or DAMPs. However, both homeostatic and immunogenic mature DCs will show an initial increase in expression of these costimulatory molecules, and only upon interaction with Tregs in the lymph node and subsequent CTLA4-dependent transendocytosis, CD80 and CD86 expressions become downregulated on homeostatic mature DCs. This implies that these markers cannot be used as discriminating markers in vitro to show the immunogenicity of certain compounds, as has been suggested previously.^89,90,91

What drives the onset of the homeostatic maturation program has remained enigmatic for a long time. Recent studies suggested a role for the uptake of apoptotic cells as an essential upstream trigger to elicit the homeostatic maturation process in cDC1s in SS.^{9,10,11,12,13} In addition, confinement of the DCs during migration in the tissue has recently been proposed as a potential upstream trigger of homeostatic maturation. However, the maturation program induced by confinement showed little to no overlap with the maturation program observed in SS and the presence of an IFN-stimulated gene signature in confinement-induced mature state showed that it clustered within the immunogenic mature DCs. Previous studies have shown that mechanical confinement of DCs is associated with transient nuclear rupture,^92,93 which may cause leakage of DNA or other nuclear components to the cytosol, potentially triggering cytosolic danger receptors in the DCs. Alternatively, the activation of DNA damage pathways may also trigger the activation of NF-κB.⁹⁴

TF activity inference studies detected several TFs that were uniquely associated with the homeostatic maturation process, such as Jun, JunB, BACH2, and several known nuclear receptors, such as LXRα and LXRβ, Nr4a1, or RARα, previously associated with the suppression of inflammatory responses.^59,95 Still, at this point, for none of them, we could assign a clear role as the driver of the maturation process, nor could we confirm their unique expression in homeostatically maturing DCs. This could be linked to technical issues associated with low expression levels and/or the lack of good antibodies. In addition, we noticed that a mere decrease in the number of mature DCs in the absence of a certain TF is not sufficient to claim its role as a driver of the maturation process. As an example, we noted that the absence of IKK2 leads to a slight reduction in the number of mature cDC1s, as previously reported.^62,63 However, closer inspection revealed that this was due to a reduction in the LM CD63⁺CCR7⁺ cells rather than in the EM CD63⁻CCR7⁺ pool, suggesting that the onset of the maturation program was not affected by the lack of IKK2. This was further supported by qPCR experiments performed on the immature versus mature DC stages, which showed that none of the maturation genes previously proposed to be regulated by IKK2 were affected by the loss of IKK2. This suggests that the maturation defects that were previously reported on bulk DC pools may reflect the heterogeneity in population (change in the ratio of immature versus mature DCs) rather than reflecting a direct transcriptional effect. As the NF-κB pathway is well known to sustain cell survival,⁹⁶ the observed loss in LM DCs in the absence of IKK2 could also be explained by a cell survival deficit specifically in the mature stage. In conclusion, at this point, the upstream drivers of the homeostatic maturation program remain largely undefined. It remains possible that only TFs associated with the common program, such as c-Rel or STAT4/6, are driving the onset of the maturation program, while TFs that are uniquely associated with one of the two maturation types may be needed to tweak the maturation type toward a homeostatic or immunogenic direction. Along these lines, we previously showed that the loss of LXR signaling leads to a loss of repression of type I IFN genes in homeostatic conditions.⁹

Finally, our study indicates that the engulfment of empty non-adjuvanted LNPs, constructed with the ionizable lipid ALC-0315 (used in the BNT162b2 COVID-19 vaccine⁹⁷) or the ionizable lipid SM-102 (used in SpikeVax⁹⁸), induces a homeostatic cDC1 maturation program with a transcriptional profile that is highly similar to the one observed in SS mature DCs. This implies that the adjuvant activity of these mRNA-LNPs is derived from the nucleic acid cargo rather than from the presence of the ionizable lipid, contrary to what has been suggested before.^{16,39,99,100,101} Originally, it was believed that the pseudouridine modifications of the mRNA component rendered the mRNA component immunosilent.¹⁰² However, studies with the BNT162b2 vaccine from Pfizer revealed that upon mRNA vaccine injection, a strong induction of a type I IFN response could be observed.^41,103 Furthermore, the cell-mediated immune responses appeared dependent on the presence of Ifnar and Mda5.⁴¹ Similarly, the Tfh response to mRNA-LNP was shown to depend on MyD88, while this was not observed after injection of protein/LNP formulations.¹⁶ The presence of the ionizable lipid has been proposed to mediate efficient antigen delivery to the cytosol,¹⁴ which is essential for the induction of both cellular immunity and a humoral response. In this regard, the ionizable lipid may contribute to the antigenicity, rather than the adjuvanticity, of the LNP-mediated response. This aligns well with our observations that only in the presence of extra added adjuvants, such as TLR ligands or mRNA complexed to LNPs, the uptake of LNPs does induce an immunogenic maturation program in DCs. On the other hand, uptake of eLNPs or peptide-containing LNPs is sufficient to drive homeostatic DC maturation. Since this will induce the upregulation of several surface markers that are shared with immunogenic mature DCs, LNP-induced DC maturation may have been misinterpreted as proper (immunogenic) DC activation, especially when studying DC activation in vitro. Our data are in line with a recent study that showed the presence of a so-called ISG cluster (mDC_ISG) only upon injection of mRNA-LNPs, but not upon injection of eLNPs.⁴² Of note, while the presence of this cluster appeared essential for triggering antigen-specific IFNg⁺ T cells, IFN signaling appeared dispensable for the induction of neutralizing antibodies.⁴² This suggested that the humoral response may be independent of immunogenic DC maturation and IFN-signaling, which could also explain why previous studies had found that proteins complexed with LNPs can trigger a proper Tfh cell response.¹⁶

DCs maturing upon uptake of non-adjuvanted versus adjuvanted LNPs induce a divergent cytokine and chemokine spectrum. In homeostatic conditions, we found prominent expression of the chemokine CXCL16, previously associated with the induction of tissue residency through the CXCR6 axis.^104,105,106 In immunogenic conditions, high expression of the chemokines CXCL9, involved in T cell recruitment to tumors,^107,108,109 and CCL5, secreted by migratory DCs to guide CCR5⁺ monocytes to the dLN,¹¹⁰ could be noted. The differential production of cytokines and chemokines indicates that DCs might instruct different T cell responses dependent on their maturation status. This is further highlighted by the adoptive transfer of OTI and OTII cells before injections of LNPs with encapsulated OVA peptide in the presence or absence of adjuvants. While injection of non-adjuvanted OVA-containing LNPs resulted in decreased proliferation of OTI and OTII T cells over time, injection of adjuvanted LNPs encapsulating OVA peptide resulted in increased T cell proliferation, which—in the case of OTI cells—could be further boosted upon rechallenge and led to cytotoxicity. Of note, for OTII cells, we observed a preferential induction of T_RM cells upon injection of OVA-peptide LNPs, while injection of adjuvanted LNPs led to the induction of T_EM cells. This was in line with the specific induction of CXCL16 in homeostatic mature DCs and confirmed previous reports showing that migratory DCs in SS conditions precondition naive T cells for the induction of tissue-resident memory fate.¹¹¹ Finally, we did not observe any Foxp3 expression in OTII cells, showing the absence of de novo peripheral Treg induction in the spleen.

In conclusion, we have shown that homeostatic and immunogenic mature cDC1s differ in terms of transcriptomic profile, TFs driving these divergent gene signatures, surface markers expressed, cytokines produced, and T cell responses instructed. The injection of empty LNPs or LNPs encapsulating peptides leads to homeostatic DC maturation and, hence, could have potential as tolerogenic vaccines. So far, LNP-based vaccination has mainly gained interest in the treatment of infectious diseases or in improving anti-tumor immunity.^112,113 This and other studies¹¹⁴ suggest that their therapeutic potential could be extended to the treatment of allergies or autoimmune diseases, as tolerogenic vaccines.

Limitations of the study

The major limitation of the study is that all the data presented in the manuscript are based on experiments in mice. The markers proposed for the “DC maturation” flow cytometry panel are conserved in human and can be perfectly used in human DCs as well to annotate their maturation state. However, we did not include any human data in the paper, since all human DCs that we tested so far (isolated from spleen obtained via the transplantation center at UZ Ghent or from tonsils obtained via the Department of Otorhinolaryngology at UZ Ghent) appeared to show the expression of immunogenic markers as well. This implies that they are probably not in a homeostatic condition, as may be expected, and we judged that, without comparison to a proper “homeostatic” reference sample, we could not accurately validate whether our maturation panel would be able to discriminate the two human DC maturation states in vivo. Nevertheless, we believe the panel may be valuable for comparing mature DC states in different tumor conditions (with or without treatment). Secondly, while we propose a reliable strategy to discriminate and annotate the two DC maturation states in this resource paper, we do not provide any additional mechanistic insights into what drives the two maturation programs. However, we felt that at this stage it was more important to provide the field with a useful toolbox to discriminate the two maturation states and hopefully move to a more uniform nomenclature, which would be an essential first step to address more functionally related questions later.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Sophie Janssens (sophie.janssens@irc.vib-ugent.be).

Materials availability

This study did not generate new, unique reagents.

Data and code availability

• •All CITE-seq data enclosed in this publication have been deposited in NCBI’s Gene Expression Omnibus¹¹⁵ and are accessible through GEO series accession number GEO: GSE279232 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE279232). The single-cell datasets provided in this manuscript can be accessed via our online tool: https://www.single-cell.be/Spleen_cDC1_LNP_induced_Homeostatic_and_Immunogenic_maturation.
• •The code used for analyzing the CITE-seq data is deposited in a GitHub repository that is publicly accessible at https://github.com/JanssensLab/Spleen_cDC1_LNP_induced_Homeostatic_and_Immunogenic_maturation. An archive of the GitHub repository can be downloaded from Zenodo: https://doi.org/10.5281/zenodo.15771963.¹¹⁶
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank the members of the Janssens lab for helpful discussions regarding the manuscript. We thank the VIB Flow Core, VIB Single-Cell Core, and the IRC Animal House Facility and IRC Cell Culture Facility. We would like to give a special thanks to Arne Soete from the IT department of IRC for facilitating the web-based tools. We would also like to thank M. Pasparakis (CECAD, Cologne) for sharing the IKK2^fl/fl mice, B. Malissen (CIML, Marseille) for sharing the XCR1-cre, J. Maelfait for providing the plasmids for lentivirus production, J. Borst (Leiden University Medical Center) for providing the MC38 cell line, and S. Marion (Université de Lille) for help with the Toxoplasma experiments. This work was supported by an ERC Consolidator grant (DCRIDDLE-819314), ERC PoC (LNP-DECODE-101138227), FWO program grants (3G063218 and 3G050622), FWO EOS (EOS 30837538), GOA Ghent University (01G01524), and a BOF Research Project Ghent University (BOF/24J/2023/150). S.R. was supported by a BOF grant (Ghent University, 01D02419) and V.B. by an FWO PhD Grant (3F002418).

Author contributions

Study and experimental design: S.R., V.B., and S.J.; experiments: S.R., V.B., S.M., J.V., G.W., J.S., E.V.D.V., F.F., K.V.L., R.R., and N.V.; specific resources: K.B., S.C.D.S., S.V., E.H., P.G., R.V., I.L., and B.G.D.G.; data analysis: S.R.; bio-informatics analysis: C.D.N. and K.V.; figures: S.R. and C.D.N.; writing: S.R., C.D.N., V.B., and S.J.; and supervision and funding: S.J..

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-CD3 (clone 17A2), biotin conjugated	Thermo Fisher Scientific	Cat# 13-0032-82; RRID: AB_2572762
Anti-CD3e (clone 145-2C11), biotin conjugated	Thermo Fisher Scientific	Cat# 13-0031-82; RRID: AB_466319
Anti-CD3e (clone 145-2C11), BV510 conjugated	BD Biosciences	Cat# 563024; RRID: AB_2737959
Anti-CD3e (clone 145-2C11), BV605 conjugated	BioLegend	Cat# 100351; RRID: AB_2565842
Anti-TCRb (clone H57-597), biotin conjugated	Thermo Fisher Scientific	Cat# 13-5961-82; RRID: AB_466819
Anti-TCRb (clone H57-597), BV785 conjugated	BioLegend	Cat# 109249; RRID: AB_2810347
Anti-CD19 (clone eBio1D3 (1D3)), biotin conjugated	Thermo Fisher Scientific	Cat#13-0193-82; RRID: AB_657656
Anti-CD19 (clone 1D3), BUV563 conjugated	BD Biosciences	Cat#749028; RRID: AB_2873425
Anti-CD19 (clone 6D5), BV605 conjugated	BioLegend	Cat#115539; RRID: AB_11203538
Anti-CD19 (clone eBio1D3 (1D3)), PE-Cy5 conjugated	Thermo Fisher Scientific	Cat#15-0193-82; RRID: AB_657672
Anti-CD19 (clone 1D3), APC-Cy5 conjugated	BD Biosciences	Cat#557655; RRID: AB_396770
Anti-CD64 (clone ×54-5/7.1), biotin conjugated	BioLegend	Cat#139318; RRID: AB_2566557
Anti-CD64 (clone ×54-5/7.1), BV421 conjugated	BioLegend	Cat#139309; RRID: AB_2562694
Anti-CD64 (clone ×54-5/7.1), BV711 conjugated	BioLegend	Cat#139311; RRID: AB_2563846
Anti-CD49b (clone DX5), biotin conjugated	BioLegend	Cat#108904; RRID: AB_313411
Anti-TER-119 (clone TER-119), biotin conjugated	Thermo Fisher Scientific	Cat#13-5921-82; RRID: AB_466797
Anti-Ly-6G (clone 1A8), biotin conjugated	BioLegend	Cat#127604; RRID: AB_1186108
Anti-Siglec-F (clone E50-2440), biotin conjugated	BD Biosciences	Cat#567600; RRID:
Anti-CD11c (clone N418), BUV496 conjugated	BD Biosciences	Cat#750450; RRID: AB_2874611
Anti-CD11c (clone N418), eFluor450 conjugated	Thermo Fisher Scientific	Cat#48-0114-82; RRID: AB_1548654
Anti-CD11c (clone N418), BV750 conjugated	BioLegend	Cat#117357; RRID: AB_281035
Anti-CD11c (clone N418), eFluor660 conjugated	Thermo Fisher Scientific	Cat#50-0114-82; RRID: AB_11151507
Anti-CD11c (clone HL3), FITC conjugated	BD Biosciences	Cat#553801; RRID: AB_396683
Anti-CD11c (clone N418), APC-eFluor780 conjugated	Thermo Fisher Scientific	Cat#47-0114-82; RRID: AB_1548652
Anti-I-A/I-E (clone M5/114.15.2), FITC conjugated	Thermo Fisher Scientific	Cat#11-5321-82; RRID: AB_465232
Anti-I-A/I-E (clone M5/114.15.2), BV480 conjugated	BD Biosciences	Cat# 566088; RRID: AB_2869739
Anti-I-A/I-E (clone M5/114.15.2), BUV805 conjugated	BD Biosciences	Cat# 748844; RRID: AB_2873247
Anti-I-A/I-E (clone M5/114.15.2), APC-eFluor780 conjugated	Thermo Fisher Scientific	Cat# 47-5321-82; RRID: AB_1548783
Anti-XCR1 (clone ZET), PE conjugated	BioLegend	Cat#148204; RRID: AB_2563842
Anti-XCR1 (clone ZET), BV421 conjugated	BioLegend	Cat#148216; RRID: AB_2565230
Anti-XCR1 (clone ZET), BV605 conjugated	BioLegend	Cat#148222; RRID: AB_2927815
Anti-XCR1 (clone ZET), BV650 conjugated	BioLegend	Cat#148220; RRID: AB_2566410
Anti-CD172a (clone P84), PerCP-eFluor710 conjugated	Thermo Fisher Scientific	Cat#46-1721-82; RRID: AB_10804639
Anti-CD172a (clone P84), PE-Cy7 conjugated	BioLegend	Cat#144008; RRID: AB_2563545
Anti-CCR7 (clone 4B12), biotin conjugated	Thermo Fisher Scientific	Cat#13-1971-82; RRID: AB_466642
Anti-CCR7 (clone 4B12), PE-eFluor610 conjugated	Thermo Fisher Scientific	Cat# 61-1971-82; RRID: AB_2802387
Anti-CD11b (clone M1/70), BUV395 conjugated	BD Biosciences	Cat# 565976; RRID: AB_2738276
Anti-CD11b (clone M1/70), PE-Cy5 conjugated	BioLegend	Cat# 564985; RRID: AB_2739033
Anti-CD11b (clone M1/70), APC-R700 conjugated	BD Biosciences	Cat# 101210; RRID: AB_312793
Anti-CD63 (clone NVG-2), PE-Cy7 conjugated	Thermo Fisher Scientific	Cat#25-0631-82; RRID: AB_2573356
Anti-CD63 (clone NVG-2), AF647 conjugated	BioLegend	Cat#143921; RRID: AB_2832513
Anti-P65/RELA (clone F-6), AF647 conjugated	Santa Cruz Biotechnology	Cat#sc-8008;
Anti-JUNB (clone C-11), PE conjugated	Santa Cruz Biotechnology	Cat#sc-8051;
Anti-CD45 (clone 30-F11), BUV496 conjugated	BD Biosciences	Cat#749889; RRID: AB_2874129
Anti-CD45 (clone 30-F11), BV480 conjugated	BD Biosciences	Cat#566095; RRID: AB_2739565
Anti-CD45.1 (A20), APC conjugated	Thermo Fisher Scientific	Cat#17-0453-82; RRID: AB_469398
Anti-CD45.1 (A20), FITC conjugated	Thermo Fisher Scientific	Cat#11-0453-82; RRID: AB_465058
Anti-CD45.2 (clone 104), BUV737 conjugated	BD Biosciences	Cat# 612778; RRID: AB_2870107
Anti-F4/80 (clone BM8), BV785 conjugated	BioLegend	Cat#123141; RRID: AB_2563667
Anti-Phospho-STAT1 (clone 4a), PE conjugated	BD Biosciences	Cat#612564; RRID: AB_399855
Anti-Phospho-STAT4 (clone 38/p-Stat4), PerCP-Cy5.5 conjugated	BD Biosciences	Cat#561217; RRID: AB_10643004
Anti-Phospho-c-Jun (clone D47G9), PE conjugated	Cell Signaling Technology	Cat#8752S; RRID:
Anti-Ly-6C (clone HK1.1), BV570 conjugated	BioLegend	Cat#128029; RRID: AB_10896061
Anti-Ly-6C (clone HK1.4), PerCP-Cy5.5 conjugated	Thermo Fisher Scientific	Cat#45-5932-82; RRID: AB_2723343
Anti-CD80 (clone 16-10A1), BUV737 conjugated	BD Biosciences	Cat#612773; RRID:
Anti-CD106 (clone 429 (MVCAM.A)), FITC conjugated	BioLegend	Cat#105705; RRID: AB_313206
Anti-ICOSL (clone HK5.3), biotin conjugated	BioLegend	Cat# 107403; RRID: AB_345259
Anti-CD86 (clone GL1), BUV737 conjugated	BD Biosciences	Cat# 741737; RRID: AB_2871107
Anti-CD274 (clone 10F.9G2), BV421 conjugated	BioLegend	Cat# 124315; RRID: AB_10897097
Anti-CD274 (clone 10F.9G2), AF488 conjugated	BD Biosciences	Cat# 568303; RRID:
Anti-CD83 (clone Michel-19), PE-Cy7 conjugated	BioLegend	Cat# 121517; RRID: AB_2566123
Anti-Ly-6A/E (clone D7), Pacific Blue conjugated	BioLegend	Cat# 108120; RRID: AB_493273
Anti-CD117 (clone c-Kit), FITC conjugated	Thermo Fisher Scientific	Cat# 11-1171-82; RRID: AB_465186
Anti-CD1d (clone 1B1), PE conjugated	BioLegend	Cat# 123509; RRID: AB_1236547
Anti-CD366 (clone RMT3-23), PE-Cy7 conjugated	BioLegend	Cat# 119716; RRID: AB_2571933
Anti-CD14 (clone Sa2-8), biotin conjugated	Thermo Fisher Scientific	Cat# 13-0141-82; RRID: AB_466370
Anti-CD200 (clone OX-90), BV421 conjugated	BD Biosciences	Cat# 565547; RRID: AB_2739289
Anti-CD200 (clone OX-90), FITC conjugated	Thermo Fisher Scientific	Cat# MA5-17980; RRID: AB_2539364
Anti-CD107a (clone 1D4B), PE conjugated	BD Biosciences	Cat# 558661; RRID: AB_1645247
Anti-CD150 (clone TC15-12F12.2), biotin conjugated	BioLegend	Cat# 115908; RRID: AB_345278
Anti-CD31 (clone 390), BV421 conjugated	BioLegend	Cat# 102423; RRID: AB_2562186
Anti-CD201 (clone RCR-16), PE conjugated	BioLegend	Cat# 141503; RRID: AB_10899579
Anti-CD62P (clone RMP-1), PE-Cy7 conjugated	BioLegend	Cat# 148309; RRID: AB_2565985
Anti-CD278 (clone C398.4A), biotin conjugated	BioLegend	Cat# 313504; RRID: AB_416328
Anti-TIM-4 (clone 54 (RMT4-54)), PE conjugated	Thermo Fisher Scientific	Cat# 12-5866-82; RRID: AB_1257163
Anti-CD95 (clone Jo2), PE-Cy7 conjugated	BD Biosciences	Cat# 557653; RRID: AB_396768
Anti-CD39 (clone 24DMS1), SB702 conjugated	Thermo Fisher Scientific	Cat# 67-0391-82; RRID: AB_2717143
Anti-CD69 (clone H1.2F3), PE-Cy7 conjugated	BD Biosciences	Cat# 561930; RRID: AB_394508
Anti-CD103 (clone M290), BUV395 conjugated	BD Biosciences	Cat#568715; RRID:
Anti-CD103 (clone 2E7), Pacific Blue conjugated	BioLegend	Cat#121418; RRID: AB_2128619
Anti-CD26 (clone H194-112), APC conjugated	BioLegend	Cat#137807; RRID: AB_10663403
Anti-CXCL9 (clone MIG-2F5.5), PE conjugated	BioLegend	Cat#515603; RRID: AB_2245490
Anti-CXCL9 (clone MIG-2F5.5), AF647 conjugated	BioLegend	Cat#515606; RRID: AB_1877135
Anti-IL-12/IL-23 (clone C17.8), PE-Cy7 conjugated	Thermo Fisher Scientific	Cat#25-7123-82; RRID: AB_2573528
Anti-CXCL16 (clone 12–81), PE conjugated	BD Biosciences	Cat#566740; RRID: AB_2869842
Anti-CCL5 (clone 2E9/CCL5), PE-Cy7 conjugated	BioLegend	Cat#149106; RRID: AB_2860706
Anti-CD4 (clone RM4-5), BUV805 conjugated	BD Biosciences	Cat# 612900; RRID: AB_2739008
Anti-CD4 (clone RM4-5), BV605 conjugated	BD Biosciences	Cat# 563151; RRID: AB_2687549
Anti-CD4 (clone RM4-5), AF700 conjugated	Thermo Fisher Scientific	Cat# 56-0042-82; RRID: AB_494000
Anti-CD62L (clone MEL-14), BV421 conjugated	Biolegend	Cat# 104436; RRID: AB_2562560
Anti-CD62L (clone MEL-14), FITC conjugated	Thermo Fisher Scientific	Cat# 11-0621-82; RRID: AB_465109
Anti-CD62L (clone MEL-14), PE-Cy5 conjugated	BioLegend	Cat# 104410; RRID: AB_313097
Anti-CD8a (clone 53–6.7), PerCP-Cy5.5 conjugated	Thermo Fisher Scientific	Cat#45-0081-82; RRID: AB_1107004
Anti-CD8a (clone 53–6.7), BUV395 conjugated	BD Biosciences	Cat#565968; RRID: AB_2732919
Anti-CD8a (clone 53–6.7), BUV496 conjugated	BD Biosciences	Cat#750024; RRID: AB_2874242
Anti-CD8a (clone 53–6.7), PE-Cy7 conjugated	Thermo Fisher Scientific	Cat#25-0081-82; RRID: AB_469584
Anti-CD44 (clone IM7), BV605 conjugated	BD Biosciences	Cat#563058; RRID: AB_2737979
Anti-CD44 (clone IM7), redFluor710 conjugated	Tonbo Biosciences	Cat#TONB80-0441-U100
Anti-FOXP3 (clone FJK-16s), FITC conjugated	Thermo Fisher Scientific	Cat#11-5773-82; RRID: AB_465243
Anti-CD25 (clone PC61.5), PE conjugated	Thermo Fisher Scientific	Cat#12-0251-82; RRID: AB_465607
Anti-CD161 (clone PK136), PE-CF594 conjugated	BD Biosciences	Cat#562864; RRID: AB_2737850
Anti-CD16/CD32 (clone 2.4G2), unconjugated (Fc block)	Bioceros	N/A
TruStain FcX Block	BioLegend	Cat#101320
Bacterial and virus strains
Tachyzoites of T. gondii type II Pru Tomato SAG1-Ova parasites	Sabrina Marion; Poncet et al.117	N/A
Chemicals, peptides, and recombinant proteins
Lipofectamine	Thermo Fisher Scientific	Cat#11668030
Fetal calf serum	Greiner	N/A
GlutaMAX	Thermo Fisher Scientific	Cat# 35050038
ALC-0315	Broadpharm	Cat# BP-25498
SM-102	Broadpharm	Cat#BP-25499
Cholesterol	Sigma-Aldrich	Cat# C8667
1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol	Avanti	Cat#880151P
dioleoyl phosphatidylethanolamine	Avanti	Cat#850725P
1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(Cyanine 5)	Avanti	Cat#810335C
1,2-distearoyl-sn-glycero-3-phosphocholine	Avanti	Cat#850365P
1,2-dioleoyl-sn-glycero-3-phosphocholine-N-(Cyanine 5)	Avanti	Cat#850483C
pIC; Poly(I:C)	Invivogen	Cat#tlrl-picw
CpG B	Invivogen	Cat# tlrl-1826
OTI peptide (EEEEEEE EEESSSIINFEKL)	Peptide and Tetramer Facility Immunology, Leiden University Medical Center	N/A
OTII peptide (EEEEEEEEEES SSQAVHAAHAEINEAGR)	Peptide and Tetramer Facility Immunology, Leiden University Medical Center	N/A
Triton X-100	Sigma	Cat# ×100-5ML
Liberase TM	Roche	Cat#05 401 127 001
recombinant DNase I	Roche	Cat#04 536 282 001
Collagenase A	Roche	Cat#11088793001
Acridine orange/propidium iodide	Logos Biosystems	Cat#LB F23001
β-Mercaptoethanol	Sigma	Cat# M3148
L-alanyl-L-glutamine dipeptide	Thermo Fisher Scientific	Cat# J66996.14
Flt3L	VIB Protein Core	N/A
Gentamicin	Thermo Fisher Scientific	Cat#15710-049
streptavidin-PE-CF594	BD Biosciences	Cat#562284
streptavidin-PE-Cy5	BD Biosciences	Cat#554062
Fixable Viability Dye eFluor 506	Thermo Fisher Scientific	Cat#65-0866-14
Fixable Viability Dye eFluor 780	Thermo Fisher Scientific	Cat#65-0865-14
Brefeldin A	BioLegend	Cat#420601
Phosflow Lyse/Fix	BD Biosciences	Cat#558049
Phosflow PermIII	BD Biosciences	Cat#558050
Cell Proliferation Dye eFluor670	Thermo Fisher Scientific	Cat#65-0840-85
Cell Proliferation Dye eFluor450	Thermo Fisher Scientific	Cat#65-0842-90
Penicillin-streptomycin	Gibco	Cat# 15140122
MEM Non-Essential Amino Acids Solution	Gibco	Cat# 11140050
Sodium pyruvate	Gibco	Cat# 11360070
HEPES	Gibco	Cat# 15630080
Critical commercial assays
Quant-iT RiboGreen RNA Assay	Thermo Fisher Scientific	Cat# R11490
Cytofix/Cytoperm fixation/permeabilization kit	BD Biosciences	Cat# 554714
Foxp3 Transcription Factor Staining Buffer Set	Thermo Fisher Scientific	Cat# 00-5523-00
RNeasy Plus Micro Kit	Qiagen	Cat# 74034
Ovation PicoSL WTA system V2 Kit	TECAN	Cat# 3302-60
MinElute PCR purification kit	Qiagen	Cat# 28204
SensiFast SYBR No-ROX kit	Bioline	Cat# BIO-98020
Deposited data
CITE-seq splenic cDC1s after LNP treatment	This paper	GEO: GSE279232
Interactive CITE-seq visualization website	This paper	https://www.single-cell.be/Spleen_cDC1_LNP_induced_Homeostatic_and_Immunogenic_maturation
Spleen_cDC1_LNP_induced_Homeostatic_and_Immunogenic_maturation (Zenodo archival snapshot of GitHub repository)	Zenodo	https://doi.org/10.5281/zenodo.15771963116 (Zenodo archive of release v2.0.0)
Experimental models: Cell lines
HEK293T	In-house, STR profiled by the VIB-IRC Cell Core Facility	N/A
OP9-DLL1	Tom Taghon, UGent	N/A
Human Foreskin Fibroblasts	ATCC	Cat#CRL-2429TM
MC38	Jannie Borst, LUMC	N/A
Experimental models: Organisms/strains
Mouse: C57/BL6J	Janvier	N/A
Mouse: C57BL/6-Tg(TcraTcrb)1100Mjb/Crl	Charles River	N/A
Mouse: C57BL/6-Tg(TcraTcrb)425Cbn/Crl	Charles River	N/A
Mouse: B6.SJL-PtprcaPepcb/BoyJ	In-house breeding	N/A
Mouse: C57BL/6-Tg(CAG-OVAL)916Jen/J	Jackson Laboratory	N/A
Mouse: Ikbkbtm1.1Cgn	Manolis Pasparakis, CECAD Cologne	N/A
Mouse: TgItgax-cre1−1Reiz	In-house breeding	N/A
Mouse: XCR1tm1Ciphe	Bernard Malissen, CIML, Marseille	N/A
Mouse: B6.129P2(C)-Ccr7tm1Rfor/J	Reinhold Forster et al.6	N/A
Oligonucleotides
Ikk2 F: 5′-ACAGCCAGGAGATGGTACG-3′	IDT	N/A
Ikk2 R: 5′-CAGGGTGACTGAG\ TCGAGAC-3′	IDT	N/A
Ccr7 F: 5′-TGTACGAGT CGGTGTGCTTC-3′	IDT	N/A
Ccr7 R: 5′-GGTAGGTATC CGTCATGGTCTTG-3′	IDT	N/A
Fscn1 F: 5′-GACTGCG AAGGTCGCTACC-3′	IDT	N/A
Fscn1 R: 5′-CTGATCGGTCTCT TCATCCTGA-3′	IDT	N/A
Ccl22 F: 5′-TCTTGCTGTGGC AATTCAGA-3′	IDT	N/A
Ccl22 R: 5′-GAGGGTGACGGATGTAGTCC -3′	IDT	N/A
Cd40 F: 5′-TTGTTGACAGCGGTCCATCTA-3′	IDT	N/A
Cd40 R: 5′-GCCATCGTGGAGGTACTGTTT-3′	IDT	N/A
Cd80 F: 5′-GCAGGATACACCACTCCTCAA-3′	IDT	N/A
Cd80 R: 5′-AAAGACGAATCAGCAGCACAA-3′	IDT	N/A
Cd86 F: 5′-TGTTTCCGTGGAGACGCAAG-3′	IDT	N/A
Cd86 R: 5′-TTGAGCCTTTGTAAATGGGCA-3′	IDT	N/A
Sdha F: 5′-AACTACAAGGGACAGGTGCTG-3′	IDT	N/A
Sdha R: 5′-CTCCCCACAGGCATACAGAC-3′	IDT	N/A
Ywhaz F: 5′-CTCTTGGCAGCTAATGGGCTT-3′	IDT	N/A
Ywhaz R: 5′-GGAGGTGGCTGAGGATGGA-3′	IDT	N/A
Recombinant DNA
plasmids pMD2.G (VSV-G)	Maelfait et al.118	N/A
pNL4-3 deltaE-GFP SL8 (SIINFEKL inserted)	Maelfait et al.118	N/A
Software and algorithms
FACSDiva	BD Biosciences	RRID: SCR_001456
Flowjo	BD Biosciences	www.flowjo.com; RRID: SCR_008520
qbase+	Biogazelle/Cellcarta	RRID: SCR_003370
Cell Ranger pipeline v6.0.0	10× Genomics	RRID: SCR_017344
Spleen_cDC1_LNP_induced_Homeostatic_and_Immunogenic_maturation (code repository)	GitHub	https://github.com/JanssensLab/Spleen_cDC1_LNP_induced_Homeostatic_and_Immunogenic_maturation
scater R package (v1.18.6)	McCarthy et al.119	RRID: SCR_015954
Seurat R package (v3.1.4, v4.0.5)	Hao et al.120	RRID: SCR_016341
DoubletFinder (v2.0.3)	McGinnis et al.121	RRID: SCR_018771
Harmony (v0.1.0)	Korsunsky et al.122	RRID:SCR_022206
Muscat (v1.4.0)	Crowell et al.123	https://github.com/HelenaLC/muscat
DESeq2 (v1.30.1)	Love et al.124	RRID: SCR_015687
Triwise (v0.99.5)	van de Laar et al.125	https://zouter.github.io/triwise/index.html
ClusterProfiler (v3.18.1)	Yu et al.126	RRID: SCR_016884
DoRothEA (v1.2.2)	Garcia-Alonso et al.127	https://github.com/saezlab/dorothea
VIPER (v1.24.0)	Alvarez et al.128	https://doi.org/10.18129/B9.bioc.viper
Limma (v3.42.2)	Ritchie et al.129	RRID: SCR_010943
EdgeR (v3.28.1)	Robinson et al.130	RRID: SCR_012802
GraphPad Prism 10	GraphPad	https://www.graphpad.com/
Other
DMEM	Gibco	Cat#41965-039
Slide-A-Lyzer dialysis cassettes	Thermo Fisher Scientific	Cat#66107
Amicon Ultra 10K Centrifugal Filters	Millipore	Cat#UFC910024
RPMI 1640	Thermo Fisher Scientific	Cat#21875-059
70 mm cell strainer	Falcon	Cat#734-0003
MEMα	Thermo Fisher Scientific	Cat#12571063
UltraComp eBeads	Thermo Fisher Scientific	Cat#01-2222-42
MagniSort Streptavidin Negative Selection Beads	Thermo Fisher Scientific	Cat#MSNB-6002-74
OVA encoding, N1-methylpseudourine modified mRNA	Produced by Breckpot lab; de Mey et al.131	N/A

Experimental model and study participant details

In vivo animal studies

Male and female WT mice C57/BL6J mice (6–12 weeks old) were purchased from Janvier (France) or bred at Ghent University in specific pathogen-free conditions. OTI [C57BL/6-Tg(TcraTcrb)1100Mjb/Crl] and OTII [C57BL/6-Tg(TcraTcrb)425Cbn/Crl] mice originate from Charles River (France) and were crossed in-house to CD45.1 (B6.SJL-Ptprc^aPepc^b/BoyJ) mice. ActmOVA [C57BL/6-Tg(CAG-OVAL)916Jen/J] mice were purchased from Jackson Laboratory (USA) and bred in-house. IKK2^fl/fl (Ikbkbtm1.1Cgn, gift from Prof. Manolis Pasparakis, CECAD Cologne) mice were crossed to Itgax-cre (Tg^{Itgax-cre1−1Reiz}, CD11c-Cre) to generate DC-specific knockouts (IKK2DDC) or crossed to Xcr1-cre (XCR1^tm1Ciphe, XCR1-cre, gift from Prof. Bernard Malissen, CIML, Marseille) to generate cDC1-specific knockouts (IKK2DcDC1). Bone marrow cells from CCR7^−/− mice (B6.129P2(C)-Ccr7^tm1Rfor/J)(generated by Prof. Reinhold Forster¹³²) were used to generate chimeric mice. All mouse strains were kept on a C57/BL6J background. All animal experiments were approved and performed in accordance with institutional guidelines for animal care of the Ethical Committee at the VIB site Ghent–Ghent University Faculty of Science.

Method details

LNP formulation

The ionizable lipid ALC-0315 was purchased from BroadPharm (BP-25498) and used for all formulations of LNPs, except for Figure 5B. In Figure 5B, an additional LNP was formulated with the ionizable lipid SM-102, purchased from BroadPharm (BP-25499). Cholesterol was purchased from Sigma-Aldrich (C8667). The other lipids; 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol (DMG-PEG; PEG length 2 kDa) (880151P), dioleoyl phosphatidylethanolamine (DOPE; 850725P), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(Cyanine 5) (PE-Cy5; 810335C), 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC, 850365P) and 1,2-dioleoyl-sn-glycero-3-phosphocholine-N-(Cyanine 5) (PC-Cy5; 850483C) were purchased from Avanti. pIC (tlrl-picw) and CpG B (tlrl-1826) were purchased from Invivogen. OTI peptide (EEEEEEEEEESSSIINFEKL) and OTII peptide (EEEEEEEEEESSSQAVHAAHAEINEAGR) conjugated to polyglutamic acid were provided by Peptide and Tetramer Facility Immunology, Leiden University Medical Center. OVA encoding, N1-methylpseudourine modified mRNA was produced from an in-house developed pLMCT plasmid as previously described.¹³¹ This OVA construct contains the code for a truncated and non-secreted variant of OVA (tOVA) flanked by the trans-membrane and cytosolic domains of the DC-LAMP protein. The inclusion of this sequence on mRNAs has been demonstrated to improve the presentation of antigens by HLA class II molecules, as it directs the synthesis of the polypeptide chain to the endoplasmic reticulum while also improving HLA class I presentation. mRNA transcribed from the pLMCT-sig-tOVA-DCLAMP plasmid was co-transcriptionally capped with Cleancap AG and modified with N1-methylpseudouridine-5′-triphosphate.

LNPs (composition shown in the table below) were formulated via the ethanol dilution principle by vortex mixing or by microfluidic mixing. For the vortex mixing method, LNPs were formulated by mixing an aqueous solution (either empty, containing pIC, containing CpG, containing peptide or containing peptide and pIC) and an ethanol solution in a 2:1 ratio. The aqueous solution contained 6.66 mL 5 mM acetate buffer (pH 4) in which pIC, CpG and/or peptides were dissolved according to the molar ratios as specified in the table below. Ethanol solutions (3.33 mL) consisted of the lipid fraction as specified in the table below. LNPs were formed by solvent displacement. Hereto, the ethanol solution containing all lipids was added to the aqueous solution under vigorous mixing with a vortex mixer. To remove ethanol, the formed LNP suspensions were dialyzed for 6 h against phosphate buffered saline using Slide-A-Lyzer dialysis cassettes (cut-off 3.5 kDa) (Thermo Fisher Scientific; 66107). Subsequently, the dialyzed LNP suspensions were concentrated 10 times using Amicon Ultra 10K Centrifugal Filters (Millipore; UFC910024). For most experiments shown and unless otherwise stated, LNPs were administered intravenously at a volume dose of 100μL per mouse, and equivalent to a 25μg dose of TLR agonist and 15mg of peptide. For the MC38 experiment, 20μL of either eLNPs or pIC-LNPs were injected intratumorally. For the CITE-seq experiment, shown in all figures related to the CITE-seq, a volume dose of 100mL was administered per mouse intravenously, equivalent to a 50mg dose of TLR agonist. pIC alone was administered intravenously at a dose of 50μg. For the T cell readout assays, the amount of LNPs injected (eLNP, peptide-LNPs, peptide-pIC-LNPs) was 25μL per mouse, equivalent to 15mg peptide per mouse and 10mg of pIC.

The LNP batches (eLNPs, pIC-LNP, mRNA-LNP and OTI/OTII peptide-LNP) used in Figure 5B were prepared using an automated T-junction mixing device as reported by Meulewaeter et al.¹³³ While the molar ratios of the lipids remained the same, the helper lipid DOPE and PE-Cy5 were replaced by DSPC and PC-Cy5, respectively. The LNPs were subjected to a size and zeta potential quality control using a Malvern Zetasizer nano-ZS (Malvern Instruments Ltd.). The Quant-iT RiboGreen RNA Assay was used to determine mRNA encapsulation and concentration according to manufacturer’s protocols (ThermoFisher). In order to release and detect the encapsulated mRNA content, mRNA particles were diluted in TE buffer containing 1% (v/v) Triton X-100 (Sigma) and incubated for 10 min at 37°C, while the free (not-encapsulated) mRNA content was directly measured after particle dilution in TE buffer. mRNA-LNPs were administered at an encapsulated mRNA dose of 20μg per mouse, while the other LNPs were administered at an equivalent lipid dose (as determined by an in-house developed UPLC-CAD method for lipid quantification (Vanquish Flex UPLC 1000 bar system, Thermo Scientific)).

Table. Composition of LNPs

Cargo (wt/wt of total lipids to cargo)	eLNP	pIC-LNP CpG-LNP		Peptide-LNP	Peptide-pIC-LNP	mRNA-LNP
Peptide or mRNA	/	/	/	64:1	64:1	23:1
TLR agonist	/	38.5:1	38.5:1	/	38.5:1	/
Lipids (mol%)
ALC-0315 or SM-102	50	50	50	50	50	50
Cholesterol	38.5	38.5	38.5	38.5	38.5	38.5
DMG-PEG	1.5	1.5	1.5	1.5	1.5	1.5
DOPE or DSPC	9.87	9.87	9.87	9.87	9.87	9.87
DOPE-Cy5 or PC-Cy5	0.13	0.13	0.13	0.13	0.13	0.13

Tissue preparation

Cervical dislocation or CO₂ asphyxiation was used to sacrifice the mice. Spleens and lymph nodes were minced manually with scissors before digestion in RPMI 1640 (Thermo Fisher Scientific, 21875-059) containing Liberase TM (0.02 mg/mL; Roche, 05 401 127 001) and recombinant DNase I (10U/mL; Roche, 04 536 282 001) for 30 min for the spleen and 20 min for the lymph node at 37°C. Red blood cells were removed by osmotic lysis (homemade buffer) (only for spleen), followed by passage through a 70 mm cell strainer (Falcon, 734-0003).

For the use as target cells in the cytotoxicity assay, thymocytes were obtained by smashing the tissue over a 70 mm cell strainer, followed bfy osmotic lysis.

Tumors were minced manually and digested in DMEM with Collagenase A (2mg/mL, Roche) and DNAse I (50 μg/mL, Roche) for 30 min at 37°C. Red blood cells were removed by osmotic lysis (homemade buffer), followed by passage through a 70 mm cell strainer (Falcon, 734-0003).

Prior to antibody staining, live cells were counted by staining with acridine orange/propidium iodide (Logos Biosystems, LB F23001) and using the LUNA-FX7 (Logos Biosystems).

Bone-marrow derived Flt3 notch in vitro DCs

Flt3 Notch DCs were generated as described by Kirkling et al.⁷⁸ In brief, bone marrow was isolated from tibia and femur. Red blood cells were removed by osmotic lysis. The cells were differentiated in tissue culture medium (TCM: RPMI-1640 medium (Thermo Fisher Scientic) containing 10% fetal calf serum (FCS, Gibco), 1.1 mg/mL β-Mercaptoethanol (Sigma-Aldrich), 2 mM L-alanyl-L-glutamine dipeptide (Thermo Fisher Scientic) and 56 μg/mL Gentamicin (Thermo Fisher Scientic)) supplemented with 250 ng/mL Flt3L (PSF, VIB Protein Core). Cells were cultured at 1 × 10⁶ cells/mL in 6-well plates at a volume of 8mL at 37°C and 5% CO2 for 3 days. OP9-DLL1 cells (gift from Prof. Tom Taghon, UGent) were cultured in OP9 culture medium (MEMα medium (Thermo Fisher Scientic) containing 20% FCS (Bodinco), 2 mM L-alanyl-L-glutamine dipeptide (Thermo Fisher Scientic) and 56 μg/mL Gentamicin (Thermo Fisher Scientic)). On day 3 of differentiation, half of the volume of bone marrow cells was transferred to a well containing OP9 cells which contained a volume of 4 mL fresh medium and was supplemented with 250 ng/mL Flt3L. On day 8 of differentiation, half of the medium was refreshed, supplemented with 250 ng/mL Flt3L. On day 10, the cells were incubated with either 10 μL of eLNPs or 10 μL of pIC-LNPs. Six hours later, the cells were mechanically dislodged from the plate and further processed for flow cytometry.

Flow cytometry and FACS

For conventional flow cytometry analysis, 4-5x10⁶ cells were stained by several antibody staining steps with fluorochrome- or biotin-conjugated antibodies. An initial staining step consisted of Fc block (Polpharma Biologics), CD64-BV711 (Biolegend) and biotinylated antibody of choice with incubation of 45 min at 4°C. A second staining step included all other antibodies to stain the remaining surface proteins with incubation of 30 min at 4°C. Biotinylated antibodies were conjugated to streptavidin-PE-CF594 (BD Biosciences, 562284) or streptavidin-PE-Cy5 (BD Biosciences, 554062). Viability was assessed by use of Fixable Viability Dye eFluor 506 (Thermo Fisher Scientific, 65-0866-14) or eFluor 780 (Thermo Fisher Scientific, 65-0865-14). To check cytokine production of cDC1s, splenocytes were cultured ex vivo with 5 μg/mL Brefeldin A (BioLegend, 420601) in RPMI 1640 supplemented with 10% fetal calf serum (Gibco) for 3.5h at 37°C. Staining of intracellular markers was performed by fixation and permeabilization by use of either the BD Cytofix/Cytoperm fixation/permeabilization kit (BD, 554714) for cytoplasmic proteins (CXCL16-PE (BD), IL12/IL23p40-PE-Cy7 (Thermo Fisher Scientific), CXCL9-PE (Biolegend) and Ccl5-PE-Cy7 (Biolegend)) or the Foxp3 Transcription Factor Staining Buffer Set (Thermo Fisher Scientific, 00-5523-00) for nuclear proteins (FOXP3-FITC (Thermo Fisher Scientific). Phosflow Lyse/Fix (BD, 558049) and Phosflow PermIII buffers (BD, 558050) were used for intracellular staining of phosphorylated proteins (p-STAT1-PE (BD), p-cJUN-PE (Cell Signaling Technology) and p-STAT4 (BD)). Flow cytometry was performed on FACSymphony A5 and A3 (BD Biosciences) using FACSDiva software (BD Biosciences). To adjust photomultiplier tube voltages and to calculate the compensation matrix, single stained cells and UltraComp eBeads (Thermo Fisher Scientific, 01-2222-42) were used. A list of antibodies can be found in the key resources table.

Prior to fluorescence-activated cell sorting of DCs, splenocytes were enriched by negative selection. The total single cell suspensions were incubated with a biotinylated antibody mix containing antibodies for CD3e (Thermo Fisher Scientific), CD2 (BioLegend), CD19 (Thermo Fisher Scientific), CD49b (BioLegend), TER119 (Thermo Fisher Scientific), Ly-6G (BioLegend) and SiglecF (BD Biosciences). For the CITE-seq experiment, biotinylated antibodies for CD3 (Thermo Fisher Scientific), TCRb (Thermo Fisher Scientific) and CD64 (BioLegend) were also included. Next, the cells were incubated with MagniSort Streptavidin Negative Selection Beads (Thermo Fisher Scientific, MSNB-6002-74) and separated by use of magnets. After antibody staining, cell sorting was performed on FACS Aria II, FACS Aria III or FACSymphony S6 cell sorter (all BD Biosciences). Data was analyzed using Flowjo software (BD Biosciences).

T cell read outs

For the OTII readout, splenocytes were isolated from CD45.1.2 OTII mice by pressing the spleen over a 70 mm cell strainer and red blood cells were removed by osmotic lysis. OTII cells were sorted with FACS Aria II and FACS Aria III (both BD Biosciences) sorters, purified on live cells, MHCII-, CD19⁻, CD11c-, CD8a-, CD4⁺, CD62L+. By intravenous injection, 500.000 OTII cells were adoptively transferred in CD45.2 acceptor mice. Two days later, these mice were intravenously injected with either eLNPs, OTIIpeptide-LNPs or OTIIpeptide-pIC-LNPs. OTII cells in the blood were checked through flow cytometry 7 days and 12 days later.

For the OTI readout, splenocytes were isolated from CD45.1.2 and CD45.1 OTI mice by the same protocol as described above. OTI cells were sorted with FACS Aria II and FACS Aria III (both BD Biosciences) sorters, purified on live cells, MHCII-, CD19⁻, CD11c-, CD4⁻, CD8a+, CD62L+. By intravenous injection, 300.000 OTI cells were adoptively transferred in CD45.2 acceptor mice. One day later, these mice were intravenously injected with either eLNPs, OTIpeptide-LNPs or OTIpeptide-pIC-LNPs. OTI cells in the blood were checked through flow cytometry 3 days later. After 6 more days, the mice received SIINFEKL-lentivirus (production method mentioned below) through intraperitoneal injection. After another 4 days, OTI cells in the blood were checked through flow cytometry. Additionally, cytotoxicity was tested by isolating splenocytes and co-culturing these cells ex vivo with 100.000 cells of a 1:1 mixture of CTR-labeled thymocytes of ActmOVA mice and CTV-labeled thymocytes of WT mice. Thymocytes from ActmOVA mice and WT mice were isolated as described above and labeled with Cell Proliferation Dye eFluor670 (CTR, Thermo Fisher Scientific, 65-0840-85) and Cell Proliferation Dye eFluor450 (CTV, Thermo Fisher Scientific, 65-0842-90), respectively, according to manufacturer’s instructions. OTI cytotoxicity was measured 1 day later by measuring the ratio of CTV-labeled (WT) versusvs. CTR-labeled (OVA) cells through flow cytometry.

SIINFEKL-lentivirus production and infection

SIINFEKL-Lentivirus was produced as described in a study by Maelfait et al.¹¹⁸ In short, HEK293T cells were transfected using Lipofectamine (Thermo Fisher Scientific, 11668030) with plasmids pMD2.G (VSV-G) and pNL4-3 deltaE-GFP SL8 (SIINFEKL inserted). One day after transfection, medium containing DMEM (Gibco, 41965-039), fetal calf serum (Greiner) and Glutamax (Thermo Fisher Scientific, 35050038) was refreshed and 2 days later supernatant was collected and filtered through a 0.45 mm polyethersulfone filter (Merck). Virus aliquots were kept at −80°C until use. Mice were injected with 100 mL of the produced virus through intraperitoneal injection.

RNA extraction and RT-qPCR

DC subsets (maximum 30.000 cells) were sorted directly in RLT buffer (RNeasy Plus Micro Kit, Qiagen, 74034) supplemented with b-mercaptoethanol (diluted 1:100, Sigma, M3148) and stored at −80°C until further processing. RNA was extracted using the RNeasy Plus Micro Kit (Qiagen, 74034) according to the manufacturer’s instructions. cDNA was amplified using the Ovation PicoSL WTA system V2 Kit (TECAN, 3302-60) following the manufacturer’s instructions. Afterward, pollutants were removed using the MinElute PCR purification kit (Qiagen, 28204) according to manufacturer’s instructions before SybrGreen-based RT-qPCR with the SensiFast SYBR No-ROX kit (Bioline, BIO-98020) using a LightCycler 480 (Roche). mRNA expression levels were analyzed using qbase+ software (Biogazelle). A list of primer sequences can be found in the key resources table.

Toxoplasma gondii infection

Tachyzoites of T. gondii type II Pru Tomato SAG1-Ova¹³⁴ parasites were grown as described in the study of Poncet et al.¹¹⁷ In short, tachyzoites were grown in vitro on Human Foreskin Fibroblasts (CCD-1112Sk (ATCC, CRL-2429TM)) in DMEM (Gibco) supplemented with fetal calf serum (Gibco) and 1% penicillin-streptomycin (Gibco) and were extracted by sequential passages through 17-gauge and 26-gauge needles followed by filtration with a 3 mm polycarbonate membrane filter. Mice were infected by intraperitoneal injection of 500 tachyzoites. Splenocytes were investigated 6 days later with flow cytometry.

MC38 injections

MC38 cell line was provided by Prof. Dr. Jannie Borst and Dr. Ferry Ossendorp (LUMC) and has not been authenticated in our lab. Cells were cultured under standard conditions at 37°C, 5% CO2 in DMEM supplemented with 10% fetal bovine serum (FBS, Bodinco), non-essential amino acids (ThermoFisher Scientific), 1 mM sodium pyruvate (ThermoFisher Scientific) and 10 mM HEPES (Gibco), and are routinely tested for mycoplasma contamination. Tumor cell lines were harvested, washed with PBS, and resuspended in a final injection volume of 50μL PBS. 1 × 10⁶ MC38 cells were injected subcutaneously in the right flank of shaved mice. Tumor growth was followed by two orthogonal measurements with a caliper. After 12 days, 20 μL of either eLNPs or pIC-LNPs were injected intratumorally and the tumor and tumor-draining lymph node (inguinal and axillary) were dissected 8h later.

Bone marrow chimeras

CD45.1.2 recipient animals were sublethally irradiated with 2 rounds of 400cG, with a 4-h interval in between the irradiations. Four hours after the second irradiation cycle, recipient mice received an intravenous injection with a 1:1 mixture of 4 × 10⁶ bone marrow cells derived from CD45.1 and wild type (CD45.2) mice on the one hand or CD45.1 and CCR7^−/− (CD45.2) mice on the other hand. Mice were used at least 8 weeks after reconstitution.

CITE-sequencing

Sorting and library prep

DCs were enriched as described above and 4–8 million cells were stained with FACS antibodies, TruStain FcX Block (BioLegend, 101320) and the mouse cell surface protein antibody panel containing 155 oligo-conjugated antibodies (TotalSeq-A, BioLegend), 9 TotalSeq-A isotype controls and 4 different hashing antibodies. Each biological replicate was stained with a different hashing antibody, 20.000 cells were sorted per biological replicate and pooled together. For the steady state condition, 15.000 CCR7-cDC1s and 5.000 CCR7+ cDC1s were sorted. For the LNP conditions, 20.000 LNP+ cDC1s were sorted based on the Cy5 fluorochrome which was incorporated in the LNPs. For the pIC alone condition, 20.000 total cDC1s were sorted. Sorting was done on a FACS Symphony S6 (BD). cDC1s were gated on LD-, CD3e-, CD19⁻, CD64⁻, F4/80dim, CD11c+, MHCII+, XCR1+, CD172-and only the steady state condition was sorted based on CCR7.

Sorted single-cell suspensions were resuspended at an estimated final concentration of 1000 cells/μL and loaded on a Chromium GemCode Single Cell Instrument (10× Genomics) to generate single-cell gel beads-in-emulsion (GEM). The scRNA-Seq libraries were prepared using the GemCode Single Cell 3′ Gel Bead and Library kit, version NextGEM 3.1 (10× Genomics) according to the manufacturer’s instructions with the addition of amplification primer (3 nM, 5′CCTTGGCACCCGAGAATT^∗C^∗C) during cDNA amplification to enrich the TotalSeq-A cell surface protein oligos. Size selection with SPRIselect Reagent Kit (Beckman Coulter) was used to separate amplified cDNA molecules for 3′ gene expression and cell surface protein construction. TotalSeq-A protein library construction including sample index PCR using Illumina’s Truseq Small RNA primer sets and SPRIselect size selection was performed according to the manufacturer’s instructions. The cDNA content of pre-fragmentation and post-sample index PCR samples was analyzed using the 2100 BioAnalyzer (Agilent).

CITE-sequencing analysis

Sequencing libraries were loaded on an Illumina NovaSeq 6000 flow cell at the VIB Nucleomics core with sequencing settings according to the recommendations of 10× Genomics, pooled in a 70:20:10 ratio for the combined 3′ gene expression, cell surface protein ADT expression and HashTag-Oligo (HTO) data, respectively.

The Cell Ranger pipeline (10× Genomics, v6.0.0) was used to perform sample demultiplexing and to generate FASTQ files for read 1 and read 2 for the gene expression and cell surface protein libraries. Read 2 of the gene expression libraries was mapped to the reference genome (mouse mm10) using STAR. Subsequent barcode processing, Unique Molecular Identifier (UMI) filtering and gene counting was performed using the Cell Ranger suite. CITE-seq reads were quantified using the feature-barcoding functionality. The mean reads per cell across all sample libraries was 20,854 RNA reads, and 333 ADT reads respectively, with an average sequencing saturation of 38.6% and 46.1%, as calculated by Cell Ranger. 9 individual single-cell libraries were analyzed for this experiment, totaling 160,745 cells. After individual analysis, the cDC1s of the resulting Seurat objects were merged into one large cDC1 object. This object was further investigated.

Pre-processing data

The data of each sample was pre-processed using the scater R package (v1.18.6) according to the workflow proposed by McCarthy and colleagues.¹¹⁹ QC metrics were calculated for all the single cells of each sample. Outlier cells were identified based on three calculated metrics: library size, number of expressed genes and mitochondrial proportion. Cells were tagged as outliers if one of their three metrics was a minimum of 3 median absolute deviations (MADs) removed from the respective metric median for each sample.

Log-transformed normalized expression values were then computed from each count matrix. Subsequently, the Seurat R package (v4.0.5)¹²⁰ was used to create a Seurat object for each sample with both the raw counts and log2 transformed counts. SCTransform was performed on the raw count data to normalize the UMI counts by regularized negative binomial regression. Additionally, highly variable features were found and scaling was performed on the log-transformed count data as a back-up for SCT. Principal component analysis (PCA), clustering and Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction were performed.

DoubletFinder (v2.0.3) was performed to predict any remaining doublets in each sample.¹²¹

The ADT UMI matrices did not have extensive pre-processing performed. The same cells were filtered for each sample as during the RNA pre-processing to keep the columns of the matrices equal. The ADT expression data was processed using the Seurat pipeline, with Centered Log-Ratio (CLR) normalization and scaling of the data performed using the default parameters. PCA, clustering and UMAP dimensionality reduction were performed too.

The HTO assay was transformed via CLR normalization. Biological replicates in each sample were demultiplexed by running the MULTIseqDemux function included in Seurat. It is based on the classification method from MULTI-seq.¹³⁵ The autoTresh parameter of the probability density function (PDF) was set to True.

Finally, a multimodal analysis was also performed based on the weighted nearest neighbor workflow (WNN) from Seurat. The weighted information from the SCT and ADT assay was used to create a WNN graph integrating both modalities for each sample.

Marker genes and surface markers per identified subpopulation were found using the FindAllMarkers function of the Seurat pipeline and this informed the annotation process of the clusters.

The DoubletFinder and MULTI-Seq classification together with ADT metrics and information on mitochondrial proportion, UMI count and gene count were used to filter doublets, contaminating cells and low-quality clusters.

After analyzing all individual samples, all the cDC1s were merged into one large object. Subsequently a Seurat pipeline was performed with slight modifications compared to above.

SCTransform was omitted from this analysis and the RNA assay was used as the default assay. Log normalization of the raw counts was performed with the NormalizeData function from Seurat with the scale.factor set to 10000. Highly variable gene detection was completed with the FindVariableFeatures function from Seurat according to the vst selection method. The number of features was set to 2000. Subsequently, the RNA assay was scaled with the ScaleData function from Seurat. Then PCA was run with the RunPCA function from Seurat to identify the top 150 Principal Components (PCs). Finally, various number of PCs for UMAP dimensionality reduction and resolutions for unsupervised clustering were attempted. This was guided by running the maxLikGlobalDimEst function from the intrinsicDimension package (v1.2.0).

Thirty-two RNA PCA dimensions and a resolution of 0.8 were used to create the RNA UMAP dimensionality reduction plot and clustering.

The ADT assay of the merged object was analyzed with the same workflow as for the individual samples. Ten ADT PCA dimensions and a resolution of 0.8 were used to create the ADT UMAP dimensionality reduction plot and clustering.

Due to the strong differences between the biological conditions of the different samples, the same cDC1 maturation stages in the different samples did not cluster together and this complicated the detailed annotation of the clusters. To mitigate this complication, Harmony (v0.1.0)¹²² batch correction was performed by running the RunHarmony function. This was completed on both the RNA assay and the ADT assay with the sample identifier supplied to the group.by.vars parameter and a default theta value of 2. For the RNA assay the top 40 PCs were used whereas only the top 10 PCs for the ADT assay were used for integration. This resulted in a good mixing of the cDC1s from the respective samples, so the data was deemed sufficiently integrated with Harmony.

Twenty Harmony corrected RNA PCA dimensions and a resolution of 0.8 were used to create the final RNA Harmony UMAP plot and clustering. Fifteen Harmony corrected ADT PCA dimensions and a resolution of 0.8 were used to create the final ADT Harmony UMAP plot and clustering.

Additionally, a multimodal analysis was also performed based on the WNN workflow from Seurat. The weighted information from the harmony corrected RNA and ADT PCA dimensions were used to create a WNN graph integrating both modalities. Various resolutions for clustering were attempted.

The RNA Harmony UMAP and clustering was used for further downstream analysis. Marker genes and surface markers per identified subpopulation were found using the FindAllMarkers function of the Seurat pipeline and this informed the annotation process of the clusters. Both a detailed annotation (based on unsupervised clustering) and a fundamental annotation (based on cDC1 maturation stages) were created. The “Other cDC1s 1” cluster contains a sizable fraction of the cDC1s from 1 mouse of the CpG-LNP 8h sample which appears to not have fully reacted to the treatment and thus some of its cDC1s have an aberrant gene signature. The “Other cDC1s 2” cluster contains some strange cDC1s which have hallmarks of both immature and mature cDC1s. These could possibly be doublets, but they were not identified as such in the preprocessing pipeline and therefore were not removed.

The final object contains 116,588 cells, 18 detailed and 6 fundamentally annotated clusters.

Plot showing the average mRNA expression (Seurat module score) for a selection of IFN-stimulated genes (ISGs) was depicted in Figure 2B. This selection of ISGs is based on the marker genes of the EM cDC1 cluster of our previously published work,⁹ using Cxcl10, Cxcl9, Ifi47, Cd40, Gbp2, Gbp5, Isg15, Nfkbia, Ifi204, Ifi211, Irf7, with a few extra ISGs manually added in, namely Ifit1, Ifit2, Ifit3.

Differential state analysis

Various comparisons between cDC1 maturation stages and conditions were performed using the R package muscat (v1.4.0).¹²³ It allows for Differential State (DS) analysis in multi-sample, multi-group, multi-(cell-)subpopulation CITE-seq data. The RNA and ADT data for three fundamental cDC1 clusters (Immature, Early Mature and Late Mature cDC1s) were aggregated according to condition and timepoint and various “pseudobulk” DS analyses were performed. The performed DS analyses utilized the DESeq2 method.

In this manuscript, ADT and RNA DS results are featured for 1) the Early Mature cDC1 pairwise comparison between 2h samples for the eLNP, pIC-LNP and CpG-LNP condition; 2) the Late Mature cDC1 pairwise comparison between 8h samples for the eLNP, pIC-LNP and CpG-LNP condition; 3) the pairwise comparison between Steady State (SS) Immature cDC1s, 8h eLNP Late mature cDC1s and 8h pIC-LNP Late mature cDC1s.

The DS results contained a very large number of DE genes and surface proteins, so strict filtering was performed to retain the most significant results: global adjusted p-value <0.05, |logFC| > 1 and baseMean >50.

Overview cluster-sample average expression heatmaps of DE surface proteins were created with Seurat to visualize the commonalities and differences between CpG-LNPs and pIC-LNPs compared to eLNPs at the 2h Early mature cDC1 stage and also between CpG-LNPs and pIC-LNPs compared to eLNPs at the 8h Late mature cDC1 stage. This was done by determining the intersecting and distinct DE surface proteins between the respective pairwise DS lists.

Triwise analysis

To be able to create good visualizations comparing three conditions, triwise analysis was performed for the second and third analysis described in the DS analysis section above.

Expression matrices for the RNA and ADT pseudobulk data were created including all samples from the DS analyses. This was done by normalizing the pseudobulk count data with the DESeq2 (v1.30.1)¹²⁴ function varianceStabilizingTransformation.

For each described DS analysis, all the expressed genes and/or surface proteins for the respective samples were plotted out on a hexagonal diagram called a triwise plot. This was achieved by subsetting the expression matrix to only include the relevant samples/conditions and then converting the subsetted matrix to barycentric coordinates with the transformBarycentric function from the triwise package (v0.99.5).¹²⁵ This process reduced the data from 3 dimensions to 2 dimensions, but the differential expression information was preserved. The angle on the triwise plot shows in which condition the feature is upregulated and the distance (r) from the center shows the strength of the upregulation. Each hexagonal grid line represents a fold change of 2 (logFC = 1). For the Triwise plots in which DE genes are represented (Figures 2A, 2B, and S2A), the fold changes from inner to outer grid lines correspond to 2-, 4-, 8-, 16-, 32-, 64-, 128- and 256--fold change. Each expressed gene is plotted as a dot on the hexagonal triwise plot. Genes that are not DE in any muscat comparison between the featured conditions cluster in the gray center and not significant DE genes are represented as gray dots. Significant DE genes are plotted as black dots. For some figures, genes highlighted in the text are indicated in red. For the Triwise plot in which DE surface proteins (ADT markers) are represented (Figure 4A), the fold changes from inner to outer grid lines correspond to 2-, 4-, 8-, 16-, 32--fold change. Proteins that are not significantly DE in any muscat comparison between the featured conditions are represented as gray dots. Significant DE proteins are plotted as red dots and are also labeled.

Functional annotation analyses

Gene ontology enrichment analysis was performed using the clusterProfiler R package (v3.18.1).¹²⁶ Three ontologies (“Biological Pathway”, “Cellular Compartment” and “Molecular Function”) were included and a p-value cut-off of 0.05 was utilized. Different analyses were conducted on various DS gene sets for the 2hr EM and 8hr LM cDC1s obtained via muscat. All the genes expressed in cDC1s or a respective subcluster of cDC1s were used as the background genes depending on the type of analysis. Each DS geneset was used in its entirety and also split up into negative and positive logFC genes. Subsequently, three separate GO enrichment analyses were run. A dotplot is featured with the top 10 significantly enriched Biological Pathway GO terms linked to the upregulated genes in pIC-LNP LM cDC1s compared to pIC-alone LM cDC1s. The adjusted p-value is displayed as the color of the dot and the size of the dot is determined by the GeneRatio parameter, which is the ratio of the input DE gene set annotated in the respective GO term. The most significant BP GO categories are ordered in the plot according to GeneRatio.

Gene regulatory network analysis

We used DoRothEA (v1.2.2)¹²⁷ to perform gene regulatory network analysis. The tool provides a collection of mouse regulons which consist of transcription factors (TFs) with their transcriptional targets. It combines various data resources to attain this database of signed TF - target gene interactions and assigns varying confidence levels depending on the weight of the evidence found. This resource can be combined with a statistical method, VIPER (v1.24.0),¹²⁸ to infer TF activity from the RNA expression of its transcriptional targets.

The mouse regulon data was pre-filtered to only contain the high confidence interactions (levels A/B/C). VIPER was then run on the downsampled cDC1 scRNA-seq expression data (2000 cells per condition) and the resulting inferred TF activity data was then stored inside the seurat object as a separate data assay.

Various pseudobulk heatmaps were created displaying the activity of the TFs which are variable across the different conditions and cDC1 maturation stages. The columns indicate the cDC1 subclusters from the different samples and are ordered according to maturation and condition. The rows indicate the TFs and are ordered alphabetically. These TFs were linked to certain types of cDC1 maturation by requiring minimal scaled TF activity differences between specific conditions. Homeostatic maturation TFs needed a scaled TF activity 0.5 higher in Late Mature SS and eLNP cDC1s compared to SS Immature cDC1s, with a scaled TF activity higher than 0 in Late Mature SS and eLNP cDC1s and lower than 0 in Late Mature CpG-LNP and pIC-LNP cDC1s. Immunogenic maturation TFs needed a scaled TF activity 0.5 higher in Late Mature CpG-LNP and pIC-LNP cDC1s compared to SS Immature cDC1s, with a scaled TF activity higher than 0 in Late Mature CpG-LNP and pIC-LNP cDC1s and lower than 0 in Late Mature SS and eLNP cDC1s. To determine Common maturation TFs, which turned on in Late Mature cDC1s across the conditions, the TFs needed a scaled TF activity above 0 which was at least 0.5 higher in Late Mature cDC1s in the 4 conditions (pIC alone excluded) compared to SS Immature cDC1s. Common maturation TFs, which turned off in Late Mature cDC1s across the conditions, needed the exact opposite, i.e., a scaled TF activity below 0 and at least 0.5 lower.

Meta-analysis

Gene list construction

For all gene lists only the DE genes linked to DC maturation were collected, i.e., only the positive logFC DE genes when comparing mature to immature DCs.

Data Alraies et al.⁴⁹: The Bulk RNA-seq count table was collected from GSE207653 which included all the samples. The count table was filtered to the WT bone marrow derived DC samples. Two DE analyses were performed: 1) Confinement-induced DC maturation: mature confined DCs (labeled as immature confined in GSE207653) vs. immature non-confined DCs; 2) LPS-induced DC maturation: LPS-activated non-confined DCs vs. immature non-confined DCs. Significant DE genes linked to these forms of DC maturation were collected. These cut-offs were used for both the gene lists: adj. p-value <0.05 and logFC >1.

Data Ardouin et al.²⁰: The quantile normalized Bulk RNA-seq count table was collected from GSE71171 which included all the samples. Various DE analyses were performed between the different cDC1 conditions with the limma R package (v3.42.2)¹²⁹: 1) Homeostatic cDC1 maturation in mediastinal lymph node (medLN): mh_MedLN_XCR1+_CCR7+_Mig vs. im_L_XCR1+_Int (GSM1828816 + GSM1828817 vs. GSM1828813 + GSM1828814 + GSM1828815); 2) Homeostatic cDC1 maturation in spleen: mh_S_XCR1+_CCR7+ vs. im_S_XCR1+_CCR7- (GSM1828785 + GSM1828786 + GSM1828787 vs. GSM1828782 + GSM1828783 + GSM1828784); 3) Homeostatic cDC1 maturation in thymus: mh_T_XCR1+_CCR7+ vs. im_T_XCR1+_CCR7- (GSM1828779 + GSM1828780 + GSM1828781 vs. GSM1828776 + GSM1828777 + GSM1828778); 4) MCMV-induced cDC1 maturation in spleen: mtlr_MCMV_S_XCR1+ vs. im_S_XCR1+_CCR7- (GSM1828803 + GSM1828804 + GSM1828809 + GSM1828810 vs. GSM1828782 + GSM1828783 + GSM1828784); 5) 18hr pIC-induced cDC1 maturation in spleen: mtlr_polyI:C_18hrs_S_XCR1+_CCR7+ vs. im_S_XCR1+_CCR7- (GSM1828788 + GSM1828789 + GSM1828790 vs. GSM1828782 + GSM1828783 + GSM1828784); 6) STAg-induced cDC1 maturation in spleen: mtlr_STAg_S_XCR1+CCR7+ (labeled as mtlr_STAg_S_XCR1+ in GSE71171) vs. im_S_XCR1+_CCR7- (GSM1828799 + GSM1828800 vs. GSM1828782 + GSM1828783 + GSM1828784). Significant DE genes linked to these various forms of cDC1 maturation were collected. These cut-offs were used for each of these gene lists: adj. p-value <0.05 and logFC >1.

Data Bosteels et al.⁹: Both Bulk RNA-seq and CITE-seq data from our previously published research on homeostatic dendritic cell maturation was used. This data is available from GSE228523 and GSE228544. For the Bulk RNA-seq data, one DE analysis was performed between steady state mature and immature cDC1s using a limma-edgeR workflow (v3.42.2)¹²⁹ (v3.28.1).¹³⁰ This gene list corresponds to homeostatic cDC1 maturation in spleen: mature cDC1s vs. immature cDC1s (GSM7123373-76 + GSM7123381-84 vs. GSM7123377-80 + GSM7123385-88). These cut-offs were used: adj. p-value <0.05 and logFC >1. For the CITE-seq data, both the cDC1 and cDC2 CITE-seq Seurat objects were used, which are available from GSE228544. In each object a Seurat (v3.1.4)¹³⁶ FindMarkers analysis was performed: 1) Late mature cDC1s vs. Early and Late immature cDC1 clusters combined; 2) Mature cDC2s vs. ESAM- and ESAM+ immature cDC2 clusters combined. These gene lists correspond to 1) homeostatic cDC1 maturation and 2) homeostatic cDC2 maturation. Default FindMarkers cut-offs were used for both gene lists: adj. p-value <0.01 and logFC >0.25.

Data Cui et al.⁴⁸: Rds objects for cDC1s and migratory DCs were downloaded from the download page of the Immune Dictionary interactive web portal accessible at www.immune-dictionary.org. The data was loaded into R and the two objects were merged with Seurat. The “cluster_manual” and “sample” metadata columns of the merged object were concatenated to form a cell type annotation for DEG analysis. A Seurat (v4.0.5)¹²⁰ FindMarker analysis was performed between MigDC_2_PBS (clearest cluster of migratory cDC1s based on marker expression) and cDC1_1_PBS. This gene list corresponds to the gene signature of migratory cDC1s. Default FindMarkers cut-offs were used for filtering both gene lists: adj. p-value <0.01 and logFC >0.25.

Data Cummings et al.¹⁰: Micro array data was analyzed through GEO2R (https://www.ncbi.nlm.nih.gov/geo/geo2r/). DE analysis was performed between GFPplus CD103 and GFPminus CD103 intestinal dendritic cells. GSE85682: GSM2281283 + GSM2281284 + GSM2281285 vs. GSM2281292 + GSM2281293 + GSM2281294. Significant DE genes were collected linked to an apoptotic cell-induced cDC1 signature. These cut-offs were used: p-value <0.01 and logFC >1.

Data Kim et al.⁴²: The h5ad object containing processed scRNA-seq data of the muscle injection site was downloaded from their figshare repository (https://doi.org/10.6084/m9.figshare.24547210.v2). The object was converted to a Seurat object with help from the SeuratDisk package. The “annotation” and “treat” metadata columns of the Seurat object were concatenated to form a more detailed annotation for DEG analysis. Four Seurat (v4.0.5)¹²⁰ FindMarkers analyses were performed: 1) between DC-migratory_LNP and DC-cDC1_LNP; 2) between DC-migratory_LNP+mRNA and DC-cDC1_LNP; 3) between DC-migratory_LNP and DC-cDC2_LNP; 4) between DC-migratory_LNP+mRNA and DC-cDC2_LNP. Default FindMarkers cut-offs were used for all the gene lists: adj. p-value <0.01 and logFC >0.25.

Data Maier et al.¹¹: Table S2 was downloaded from their published manuscript which contained various gene lists. The mouse mRegDC gene list was used for this analysis. No further filtering was performed. DCs from tumor-bearing lungs were sequenced.

Data Miller et al.³⁸: The raw Bulk RNA-seq count table was collected from GSE122108 which included all the samples. The count table was filtered to the relevant lung and lung lymph node cDC1 and cDC2 samples. Two DE analyses were performed between mature and immature cDC1s and cDC2s using a limma-edgeR workflow (v3.42.2 and v3.28.1)^129,130: 1) Migratory CD103+ DCs in lung-draining LN: DC.mig.103p.LuLN vs. DC.103p11bnsiglecFn.Lu (GSM3455312 + GSM3455313 + GSM3455314 vs. GSM3455303 + GSM3455304 + GSM3455305); 2) Migratory CD11b+ DCs in lung-draining LN: DC.mig.11bp.LuLN vs. DC.103n11bpsiglecFn.Lu (GSM3455315 + GSM3455316 vs. GSM3455306 + GSM3455307 + GSM3455308). Significant DE genes linked to these migratory DC subsets were collected. These cut-offs were used for both the gene lists: adj. p-value <0.05 and logFC >1.

Data Silva-Sanchez et al.¹³: One scRNA-seq Seurat object was downloaded from GSE224178, GSE224178_mouse_mregDC_cleaned_061622.rds.gz. The data was subsetted to remove the neonatal data and only focus on the adult data. The annotation from figure 6E of their manuscript was used with one slight modification. Based on marker expression, the mature DC clusters were split up into Mature cDC1s (clusters 1, 11 and 18) and Mature cDC2s (clusters 7 and 8). Then two Seurat (v4.0.5)¹²⁰ FindMarkers analyses were performed: 1) between Mature cDC1s and all other cDC1 clusters; 2) between Mature cDC2s and all other cDC2 clusters. These gene lists correspond to 1) homeostatic cDC1 maturation and 2) homeostatic cDC2 maturation. Default FindMarkers cut-offs were used for both of the gene lists: adj. p-value <0.01 and logFC >0.25. Additionally, the Bulk RNA-seq DESeq2¹²⁴ analysis results from the same manuscript were downloaded from GSE112270, GSE112270_SilvaSanchez_differential_expression_180322.xlsx, containing multiple comparisons. The comparison between CD103int XCR1+ and CD103Hi XCR1+ cDC1s was used to acquire another gene list corresponding to an activated cDC1 subset in the neonatal lung that spontaneously migrates to the LN. These cut-offs were used: FDR <0.05 and logFC >1.

Data Tamoutounour et al.⁴⁷: Micro array data was analyzed through GEO2R (https://www.ncbi.nlm.nih.gov/geo/geo2r/). DE analysis was performed between WT CLN (cutaneous lymph node) CD11b+ migDC (steady-state) and WT Dermis CD11b+ DC (steady-state). GSE49358: GSM1198102 + GSM1198103 + GSM1198104 vs. GSM1198091 + GSM1198092 + GSM1198093. Significant DE genes were collected linked to these migratory CD11b+ DCs. These cut-offs were used: adj p-value <0.05 and logFC >1.

Data Torow et al.⁴⁵: One scRNA-seq Seurat object containing Small Intestinal Peyer’s Patch (SI PP) Mononuclear Phagocytes (MNP) was collected from the research group behind the published paper with the final annotation used in their manuscript. The data was subsetted to remove the neonatal data and only focus on the adult data. The annotation from figure 6H in their manuscript was used. Then four Seurat (v4.0.5)¹²⁰ FindMarkers analyses were performed on the RNA assay: 1) between actDC1_PBS (activated) and qDC1_PBS (quiescent); 2) between actDC2_PBS and qDC2_PBS; 3) between actDC1_R848 and qDC1_PBS; 4) between actDC2_R848 and qDC2_PBS. Significant DE genes linked to these four activated DC subsets were collected. Default FindMarkers cut-offs were used for all these gene lists: adj. p-value <0.01 and logFC >0.25.

Data Weckel et al.⁴⁶: The scRNA-seq RDS object NeoSkin_Ctvsbact.rds was downloaded from Mendeley (https://doi.org/10.17632/j4kt7nmtfm.1). The annotation from the manuscript was used and the same comparisons were performed. Zsgreen is used in the manuscript to determine if the cDC2s have taken up S. epidermidis. Zsgreen negative cells do not contain bacterial antigen whereas Zsgreen positive cells do contain bacterial antigens (S.epidermidis-zsgreen), which in the study was linked to a regulatory activated DC phenotype. This object is from the skin of D10 neonatal mice. It was subsetted to only contain CD301b positive cDC2s and then a Seurat (v3.1.4)¹³⁶ FindMarkers analysis was performed between Zsgreen+ and Zsgreen- CD301b+ cDC2s. This gene list corresponds to the gene signature of S. epidermidis+ cDC2s. Default FindMarkers cut-offs were used for all these gene lists: adj. p-value <0.01 and logFC >0.25.

Data Zilionis et al.⁴³: The normalized mouse scRNA-seq data, mouse gene names and metadata were downloaded from GSE127465. Additionally, the human data html file with python code was used as a guide to explore the mouse scRNA-seq data in python and to unnormalize the data to acquire the raw counts. The analysis was continued in R to process the raw counts via the basic Seurat (v4.0.5)¹²⁰ workflow. The metadata of the paper was used for the annotation of the cell types. Then two Seurat (v4.0.5)¹²⁰ FindMarkers analyses were performed on the mouse data: 1) between tumor DC3 and tumor DC1; 2) between tumor DC3 and tumor DC2. These gene lists correspond to lung tumor DC3 gene signatures. Default FindMarkers cut-offs were used for these gene lists: adj. p-value <0.01 and logFC >0.25.

Data Rennen et al. (this study): CITE-seq data from this study on cDC1 maturation induced by various LNP set-ups was also included. In the merged Seurat object containing all set-ups, the RNA data for three fundamental cDC1 clusters (Immature, Early Mature and Late Mature cDC1s) was aggregated according to condition and timepoint. Subsequently, a pseudobulk Differential State analysis using the R package muscat (v1.4.0)¹²³ was performed to compare the various cDC1 maturation stages and conditions to each other. The DS analysis performed utilized the DESeq2 method. The comparisons included in this meta-analysis are: 1) Steady State (SS) Late Mature cDC1s vs. SS Immature cDC1s; 2) Late Mature cDC1s from the pIC-LNP 8hr condition vs. SS Immature cDC1s; 3) Late Mature cDC1s from the CpG-LNP 8hr condition vs. SS Immature cDC1s; 4) Late Mature cDC1s from the pIC alone 8hr condition vs. SS Immature cDC1s; 5) Late Mature cDC1s from the eLNP 8hr condition vs. SS Immature cDC1s. These gene lists correspond to 1) homeostatic cDC1 maturation, 2) pIC-LNP-induced cDC1 maturation, 3) CpG-LNP induced cDC1 maturation, 4) pIC-induced cDC1 maturation and 5) eLNP-induced cDC1 maturation. Due to the large size of the gene lists, strict filtering was performed: global adjusted p-value <0.01, logFC >1 and baseMean >50. The SS, pIC-LNP and CpG-LNP gene lists (1, 2 and 3 as listed above) from this study were further analyzed and served to create specific Common, Homeostatic and Immunogenic cDC1 maturation gene signatures. The pIC-LNP and CpG-LNP gene lists were combined to combine both aspects of immunogenic maturation into one gene list. This combined immunogenic gene list was then intersected with the SS maturation gene list and the asymmetric difference was also determined to create the three distinct lists of cDC1 maturation: 1) common genes between both lists = common cDC1 maturation; 2) SS specific genes = homeostatic cDC1 maturation; 3) immunogenic specific genes = immunogenic cDC1 maturation. To avoid the varying gene list length of these 3 core gene lists having an influence on the subsequent intersection analysis, these 3 core gene lists were limited to their top 200 genes according to adjusted p-value in the relevant comparisons. This value of 200 was chosen based on the length of the shortest list, the homeostatic cDC1 maturation list. These 3 top 200 core gene lists and all the other gene lists discussed above were combined into a table (Table S7).

Intersection analysis

The intersection of the homeostatic and immunogenic core gene list was determined with the other gene lists. The immunogenic intersection size was subtracted from the homeostatic intersection size to determine a maturation type score. The maturation type score and log₁₀ of the gene list length of the various gene lists were plotted out as the x and y axis of a scatterplot (gene lists 34–36 not shown because used for generating the maturation signatures 37–39 (also not shown)). The color of the dot is determined by the type of maturation. The immunogenic gene lists (maturation type score < −5) are colored red and the homeostatic gene lists (maturation type score >5) are colored green. Gene lists with a maturation type score between −5 and 5 are colored gray and labeled as Undefined. The shape of the dot is determined by the DC subtype.

Quantification and statistical analysis

All statistical tests and data representations were performed using Prism (Graphpad). Data are presented as biological replicates and as means with error bars representing the standard error of the mean (SEM). Data distribution and variances were taken into account for deciding on appropriate statistical tests. Performed tests are indicated in figure legends. Unpaired t tests or Mann-Whitney tests were performed when comparing two groups. One-way analysis of variance (ANOVA) with correction for multiple testing was performed when comparing two or more groups. two-way ANOVA with correction for multiple testing was performed when comparing two or more groups with two independent variables. Statistical significance was defined as p < 0.05 with the levels of significance indicated as ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. The statistical significance is not shown in all graphs or between all groups to ensure clarity of the figures. The investigators were not blinded to allocation during experiments. Sample size was calculated by power analysis or based on availability.

Differential expression analysis to determine the cluster markers in the CITE-seq analysis was performed using the Wilcoxon Rank-Sum test through the Seurat functions FindAllMarkers and FindMarkers. P-value adjustment was accomplished with Bonferroni correction. RNA and ADT markers for the annotated clusters were determined with these cutoffs: min.pct = 0.10, logfc.threshold = 0.30 and return.thresh (adj. P-value) = 0.01. Only positive markers were evaluated. An extra “score” column was calculated to rank the importance of the genes as markers. It was calculated with this function: “pct.1/(pct.2 + 0.01)^∗avg_logFC”. The markers are ordered according to this score. Differential State analysis to determine the differential markers between cDC1 maturation stages and conditions was performed using DESeq2 within muscat. P-value adjustment was performed using Benjamini-Hochberg correction at a global level. Significant differentially expressed (DE) genes and surface proteins were determined by using these cut-offs: global adjusted p-value <0.05, |logFC| > 1 and baseMean >50.

Published: August 18, 2025

Supplemental information

Table S1. RNA markers maturation stage annotation CITE-seq

RNA markers to annotate the different maturation stage clusters in the RNA Harmony UMAP CITE-seq object. The marker genes were determined by DE analysis for each cluster versus all other clusters. This was performed by running the Seurat function FindAllMarkers, which utilizes the Wilcoxon rank-sum test. The markers are ordered according to the appended score column (score = “pct.1/(pct.2 + 0.01) × avg_logFC”).

Table S2. ADT markers fundamental annotation CITE-seq

Surface protein markers for the different maturation stage clusters in the RNA Harmony UMAP CITE-seq object. The surface protein markers were determined by DE analysis for each cluster versus all other clusters. This was performed by running the Seurat function FindAllMarkers, which utilizes the Wilcoxon rank-sum test. The markers are ordered according to the appended score column (score = “pct.1/(pct.2 + 0.01) × avg_logFC”).

Table S3. GOEA and DS results pIC and pIC-LNP conditions CITE-seq

Significant pseudobulk DS analysis RNA results between the 2-h pIC and pIC-LNP condition and the SS condition at the EM cDC1 stage and also between the 8-h pIC and pIC-LNP condition and the SS condition at the LM cDC1 stage, as determined by muscat utilizing the DESeq2 method. Benjamini-Hochberg correction was used to adjust the p values for multiple testing. This is followed by Gene Ontology enrichment analysis (GOEA) results performed on the up- and downregulated differentially expressed genes (DEGs), respectively, between the aforementioned conditions. All biological pathway, cellular compartment, and molecular function (MF) GO terms are included and ordered according to adjusted p value.

Table S4. Triwise coordinates and DS results from analysis, including comparison to SS immature cDC1s

This table contains, in the first tab, the triwise coordinates for all the differentially expressed genes (DEGs) in the pairwise comparisons between SS immature cDC1s, 8-h eLNP LM cDC1s, and 8-h pIC-LNP LM cDC1s as determined by muscat utilizing the DESeq2 method. The second tab contains the triwise coordinates for the non-DEGs in the aforementioned comparisons. In the following tabs, the DS results of the DEGs are shown. The DEGs are split up into seven different DEG sets: set1 = DEGs only present in pIC_LNP_LM vs. SS_immature; set2 = DEGs only present in eLNP_LM vs. SS_immature; set3 = DEGs only present in pIC_LNP_LM vs. eLNP_LM; set4 = DEGs present in all three comparisons; set5 = DEGs present in pIC_LNP_LM vs. SS_immature and eLNP_LM_vs._SS_immature; set6 = DEGs present in eLNP_LM vs. SS_immature and pIC_LNP_LM vs. eLNP_LM; set7 = DEGs present in pIC_LNP_LM vs. SS_immature and pIC_LNP_LM vs. eLNP_LM.

Table S5. Triwise coordinates and DS results of 8-h LM cDC1s CITE-seq

This table contains, in the first tab, the triwise coordinates for all the differentially expressed genes (DEGs in the LM cDC1 pairwise comparisons between 8-h samples for the eLNP, pIC-LNP, and CpG-LNP conditions, as determined by muscat utilizing the DESeq2 method. The second tab contains the triwise coordinates for the non-DEGs in the aforementioned comparisons. In the following tabs, the DS results of the DEGs are shown. The DEGs are split up into seven different DEG sets: set1 = DEGs only present in CpG-LNP_8h_LM vs. eLNP_8h_LM; set2 = DEGs only present in eLNP_8h_LM vs. pIC_LNP_8h_LM; set3 = DEGs only present in CpG_LNP_8h_LM vs. pIC_LNP_8h_LM; set4 = DEGs present in all three comparisons; set5 = DEGs present in CpG_LNP_8h_LM vs. eLNP_8h_LM and eLNP_8h_LM vs. pIC_LNP_8h_LM; set6 = DEGs present in eLNP_8h_LM vs. pIC_LNP_8h_LM and CpG_LNP_8h_LM vs. pIC_LNP_8h_LM; set7 = DEGs present in CpG_LNP_8h_LM vs. eLNP_8h_LM and CpG_LNP_8h_LM vs. pIC_LNP_8h_LM.

Table S6. Triwise coordinates and DS results of 8-h LM cDC1s CITE-seq v2

This table contains, in the first tab, the triwise coordinates for all the DEGs in the pairwise comparisons between SS LM cDC1s, 8-h eLNP LM cDC1s, and 8-h pIC-LNP LM cDC1s, as determined by muscat utilizing the DESeq2 method. The second tab contains the triwise coordinates for the non-DEGs in the aforementioned comparisons. In the following tabs, the DS results of the DEGs are shown. The DEGs are split up into seven different DEG sets: set1 = DEGs only present in pIC_LNP_LM vs. SS_LM; set2 = DEGs only present in eLNP_LM vs. pIC_LNP_LM; set3 = DEGs only present in eLNP_LM vs. SS_LM; set4 = DEGs present in all three comparisons; set5 = DEGs present in pIC_LNP_LM vs. SS_LM and eLNP_LM_vs._ pIC_LNP_LM; set6 = DEGs present in eLNP_LM vs. pIC_LNP_LM and eLNP_LM vs. SS_LM; set7 = DEGs present in pIC_LNP_LM vs. SS_LM and eLNP_LM vs. SS_LM.

Table S7. Meta-analysis overview of the gene lists

Overview of the various gene lists used in the meta-analysis regarding DC maturation signatures. In the first tab, some basic info is given for each gene list as well as the number assigned. In the second tab, the names of the variables that are used in the R script to analyze all the gene lists are given. In the subsequent tabs, all the genes of each numbered gene list are listed. These are the significant upregulated DEGs found when comparing a mature DC population with an immature DC population in a specific biological setting. See the materials and methods for extra information regarding the different comparisons.

Table S8. CITE-seq antibody panel

Mouse cell surface protein antibody panel containing 155 oligo-conjugated antibodies (TotalSeq-A, BioLegend) and 9 TotalSeq-A isotype controls, used in the CITE-seq experiment.

Table S9. DS surface proteins between EM cDC1s CITE-seq

Significant pseudobulk DS analysis ADT results for the EM cDC1 pairwise comparison between 2-h samples for the eLNP, pIC-LNP, and CpG-LNP conditions, as determined by muscat, utilizing the DESeq2 method. Benjamini-Hochberg correction was used to adjust the p values for multiple testing.

Table S10. ADT Triwise coordinates and DS results of 8-h LM cDC1s CITE-seq

This table contains, in the first tab, the triwise coordinates for all the differentially expressed surface proteins in the LM cDC1 pairwise comparisons between 8-h samples for the eLNP, pIC-LNP, and CpG-LNP conditions, as determined by muscat, utilizing the DESeq2 method. The second tab contains the triwise coordinates for the non-DE surface proteins in the aforementioned comparisons. In the following tabs, the DS results of the DE proteins are shown. The DE proteins are split up into seven different DE sets: set1 = DE proteins only present in CpG-LNP_8h_LM vs. eLNP_8h_LM; set2 = DE proteins only present in eLNP_8h_LM vs. pIC_LNP_8h_LM; set3 = DE proteins only present in CpG_LNP_8h_LM vs. pIC_LNP_8h_LM; set4 = DE proteins present in all three comparisons; set5 = DE proteins present in CpG_LNP_8h_LM vs. eLNP_8h_LM and eLNP_8h_LM vs. pIC_LNP_8h_LM; set6 = DE proteins present in eLNP_8h_LM vs. pIC_LNP_8h_LM and CpG_LNP_8h_LM vs. pIC_LNP_8h_LM; set7 = DE proteins present in CpG_LNP_8h_LM vs. eLNP_8h_LM and CpG_LNP_8h_LM vs. pIC_LNP_8h_LM