+ 
+ 
+
+ 
+ 
+ 
+ 
Abstract
+High-dimensional approaches revealed emerging heterogeneity within +dendritic cells (DC), including a population of transitional DC (tDC) present in +mouse and human. However, tDC origin and relationship to other DC subsets are +not fully understood. Here, we show that tDC are distinct from other +well-characterized DC and conventional DC precursors (pre-cDC). We demonstrate +that tDC originate from bone marrow progenitors shared with plasmacytoid DC +(pDC). In the periphery, tDC contribute to the pool of ESAM+ type 2 +DC (DC2), and these DC2 harbor pDC-related developmental features. Different +from pre-cDC, tDC have lower turnover, capture antigen, respond to stimuli, and +activate antigen-specific naïve T cells, all characteristics of +differentiated DC. Different from pDC, viral sensing by tDC results in +IL-1β secretion and fatal immune pathology in a murine coronavirus model. +Our findings suggest that tDC are a distinct pDC-related subset with a DC2 +differentiation potential and unique pro-inflammatory function during viral +infections.
+Keywords: Dendritic cells, transitional DC, tDC, ASDC, pre-cDC, ESAM+ DC2, IL-1β, murine coronavirus, murine hepatitis virus, pDC-like cells
INTRODUCTION
+Dendritic cells (DC) are key immune sentinels comprised of subsets that +initiate a broad range of immune responses1. Traditional nomenclature groups DC into conventional DC +(cDC), which includes DC1 and DC2, and plasmacytoid DC (pDC). DC1 and DC2 are +specialized in T cell activation1, +whereas pDC secrete interferon type I (IFN-I) following exposure to nucleic +acids2. This classification +does not include emerging DC, such as transitional DC (tDC), also reported as +AXL+ DC, ASDC or pre-cDC3-6. This is because the +origin and in vivo function of tDC remains unclear.
+All DC differentiate from a CX3CR1-expressing hematopoietic progenitor +localized in the bone marrow (BM) through a FLT3L-dependent pathway7-9. Despite this common origin, immediate precursors for each +subset have been reported. cDC arise from CD115-expressing conventional DC +progenitors (pro-cDC)8,9, with DC1 and DC2 originating from +SiglecH−Ly6C− and +SiglecH−Ly6C+ immediate precursors, +respectively10,11. pDC originate from a +CX3CR1+Ly6D+ pro-pDC progenitor7. The molecular mechanisms dictating each DC +subset lineage specification have been described12,13. DC1 require +expression of several transcription factors (TFs) including IRF812,13, +whereas pDC are dependent on TCF42. +DC2 are defined by the co-expression of the non-specific markers +CD11chiCD11b+MHCII+CD172a+; however, +several molecular mechanisms converge into cells expressing these markers12,14. One of these mechanisms depends on the TF IRF4 and Notch2 +signaling, and results in ESAM+ DC215-18. Another +mechanism requires the TF KLF4, and results in CX3CR1+ DC216,17,19. Mouse +ESAM+ and CX3CR1+ DC2 subpopulations seem to parallel +human CD5+ DC2 and CD14+CD163+ DC3, +respectively16,20-22, +although further work is required to align these populations between species.
+tDC were identified by single-cell high-dimensional approaches in human +blood3,5,6, and +shown to be conserved between species4. Since human tDC express AXL and SIGLEC6, they were initially +called AXL/SIGLEC6 DC (ASDC) or AXL+ DC3,6. +However, these markers are not conserved in mice. Therefore, the term +“transitional DC” was suggested to highlight their spectrum of +transcriptomic, phenotypic and functional features that spans cells similar to pDC +and DC23,4,6. This spectrum of +features allows to sub-divide tDC into CD11clo tDC (hereafter called +tDClo) more similar to pDC, and CD11chi tDC (hereafter +called tDChi) similar to DC2. Until now, most studies have inadvertently +analyzed one of these tDC sub-populations, which acquired different names, e.g., +non-canonical DC (mainly tDChi cells)4,23 and pDC-like cells +(mainly tDClo cells)24,25. Consequently, it is still +uncertain if tDClo and tDChi are related, and if their pDC and +DC2 features can be attributed to different origins. Moreover, tDC identity has been +further challenged as their capacity to convert into cDC is becoming clearer in +mouse and human5,6,26. +However, whether tDC correspond to previously identified cDC precursors (pre-cDC) or +are a different cell population is unknown. Importantly, given the lack of models to +deplete tDC, the function of these cells has not yet been assessed in +vivo.
+Murine coronavirus (M-CoV), also known as murine hepatitis virus, can be used +as a model to assess DC subset function in vivo. Early M-CoV +containment is dependent on pDC capacity to rapidly produce IFN-I27,28. pDC elimination results in mild pathology characterized by +higher liver viral titers, without increasing mortality27,28. +Although these studies are informative, previous pDC depletion models, i.e., +anti-BST2 antibody (Ab) and Tcf4 conditional knockout mice +(TCF4cKO), also eliminate tDC, confounding interpretations. +Consequently, it is imperative to evaluate pDC and tDC function during M-CoV +infection using depletion models that allow their distinction in +vivo.
+Here, we address tDC origin, heterogeneity, and function in +vivo. We found that tDC possess a core transcriptome that is unique to +them. We resolve the developmental relationship of tDClo and +tDChi, and demonstrate that both populations are related to pDC. We +show that tDC convert into DC2, as recently reported26. However, tDC differentiate into +ESAM+ DC2 that retain pDC developmental features. Importantly, tDC +are not pre-DC2, indicating they represent a distinct source of DC2, and are not +conventional precursors as suggested. Finally, tDC specifically produce +IL-1β, which promotes fatal immunopathology during M-CoV infection. +Altogether, our data indicate that the pDC lineage should be expanded to include +tDC, a DC population that has a distinct role during viral infection and contributes +to the developmental heterogeneity of DC2 in mice and humans.
RESULTS
+tDC are a distinct DC population related to pDC
+We queried if tDC have a unique core transcriptomic signature different +from other DC subsets by performing bulk-RNAseq analysis of purified cells +(Extended Data Fig.1a). Despite +expressing pDC- and DC2- associated genes (Extended Data Fig.1b), tDC segregate from other DC subsets by +Principal Component Analysis (PCA), implying distinct gene expression (Fig.1a and Extended Data Fig.1c). Differentially expressed gene (DEG) analysis +revealed a tDC-specific signature including the expression of Vim, Ngfr, +Rab3il1, Cd209e, Chil5, L1cam, Ccr2 and Cd200r1 (Fig.1b; Supplementary Table1). tDC also +differentially expressed Klf12 and Bcl11b, TFs +shown to be active by ATACseq analysis29.
+FIGURE 1. tDC are a distinct DC population developmentally related to pDC.
+
a. Spleen DC subsets were sorted from +hCD2Cre+/− +Rosa26EYFP+/− (hCD2EYFP) mice as +shown in Fig.1f and Extended Data Fig.1a (our rationale for using +hCD2EYFP mice is explained in Fig.4), and prepared for bulk RNAseq analysis. PCA plot of the top +500 most variable DEG in each DC subset (differences between EYFP+ +and EYFP− DC2 are analyzed in Fig.4).
+b. Heatmap of selected genes from (a). Shown is Z-score of normalized +expression.
+c. Splenocytes of WT mice were enriched in CD135+ cells and +stained for cell sorting. Populations #1 and #3 were purified, mixed 1:1 and +processed for droplet-based genomics. Excluded lineage contains +CD3+/CD19+/NK1.1+/Ly6G+ +cells.
+d. KNetL plot of scRNAseq data. After filtering and removing a small +fraction of monocytes, 2,075 cells were analyzed using the KNetL pipeline. Shown +are KNetL clusters (top) and annotations based on subset-specific scores using +CIBERSORTx (bottom)(Extended Data +Fig.1f).
+e. Heatmap of selected genes differentially expressed in (d) clusters. +Gene expression after normalization and imputation is shown.
+f. Gating strategy used for the analysis of spleen DC subsets. +CD135-enriched splenocytes were stained and analyzed by flow cytometry. Excluded +lineage contains +CD3+/CD19+/NK1.1+/Ly6G+ +cells.
+g. Left shows schematic of BMC generated by transplanting 50% WT +(CD45.1) and 50% TCF4cKO (CD45.2) BM into lethally irradiated CD45.1 +WT mice. Right shows bar graphs (mean + SD) of the contribution of donor cells +to splenic DC populations (gated as in Fig.1f) in BMC, normalized to NK cells (n=6 mice in 2 experiments). +Statistics were determined by one-way ANOVA with Dunnett’s multiple +comparison test.
+h. Bar graph (mean + SD) showing the percentage of EYFP+ +splenic DC (gated as in Fig.1f) in +hCD2EYFP mice (n=5 mice in 4 experiments).
+i. Bar graph (mean + SD) showing the percentage of TdTomato+ +splenic DC (gated as in Fig.1f) in +CD300cTdT mice. Percentage relative to pDC is shown (n=4 mice in +3 experiments; see also Extended Data +Fig.2c).
We applied scRNAseq to uncover cell-to-cell heterogeneity within tDC. We +increased tDC representation by enriching CD135+ splenocytes and +sorting two populations that were equally mixed for sequencing (Fig.1c). A total of 2,110 single cell transcriptomes +were generated and 9 clusters identified by Seurat. Cell identity was assigned +unbiasedly using CIBERSORTx and the bulk-RNAseq data, which resulted in one tDC +cluster (Extended Data Fig.1d-e). To resolve the expected transcriptional +heterogeneity within tDC, we instead applied K-nearest-neighbor-based Network +graph drawing Layout (KNetL), a dimensionality reduction that improves +resolution. KNetL returned 8 clusters (Fig.1d), whose identity was again assigned unbiasedly (Extended Data Fig.1f; Supplementary Table2). tDC were +represented in clusters 3 and 4, expressing the core genes identified by +bulk-RNAseq, e.g., Vim, Cd200r1, Chil5, Lgals3, Ms4a6c (Fig.1e; Supplementary Table3). Cluster 3 +expressed pDC-associated genes, e.g., Bst2, Cd300c, Ly6c2 and +Tcf4. Cluster 4 expressed DC2-associated genes, e.g., +Irf4 and Zbtb46. The expression of these pDC- and DC2- +associated genes was also observed by pair-wise comparisons of bulk-RNAseq +(Extended Data Fig.1b; Supplementary Table4). We concluded +that cluster 3 and 4 corresponded to tDClo and tDChi, +respectively. In sum, despite transcriptional heterogeneity within the +population, tDC express a core transcriptional program that defines their +identity.
+tDClo and tDChi share developmental features with +pDC, including expression of pre-T cell receptor alpha and re-arranged IgH +segments, and are decreased in Itgax-Cre+/− +Tcf4fl/fl mice (TCF4cKO)4. To rule out a cell extrinsic +role for TCF4 in tDC development, we generated WT/TCF4cKO competitive +bone marrow chimeras (BMC), and analyzed each DC population (Fig.1f-g)4. Similar to +pDC, tDClo and tDChi were mostly derived from +TCF4-proficient BM, indicating that TCF4 is intrinsically required for both tDC +sub-populations (Fig.1g, Extended Data Fig.2a-b). To further evaluate the developmental relationship of +tDClo, tDChi and pDC, we used pDC-specific lineage +tracing models. Mice expressing Cre under the human CD2 promoter +(hCD2-iCre+/−) were bred to Cre-inducible +EYFP reporter mice (Rosa26LSL-EYFP+/−) to +generate hCD2EYFP mice that label pDC30,31. EYFP labeling was observed in 100% of pDC, and ~85% +of tDClo and tDChi (Fig.1h). We also generated a new fate mapping model based on the +specific expression of CD300c by pDC32. When mice with Cre knocked into the +Cd300c locus +(Cd300cCre+/+) were bred +with Rosa26LSL-TdTomato+/− mice +(CD300cTdT), tDClo and tDChi showed +~75% of TdTomato labeling relative to pDC (Fig.1i, Extended Data Fig.2c). +Notably, in both lineage tracing models there was virtually no labeling +(<10%) of DC1, but detectable (~25%) labeling of DC2, a result +that will be discussed later. Altogether, these experiments cement the +developmental relationship between tDClo, tDChi and pDC, +suggesting a common origin.
tDC originate from BM progenitors at steady state
+Mouse and human activated pDC can give rise to tDC29,33. We queried if pDC could give rise to tDC at steady state. +Spleen and BM pDC were adoptively transferred to non-irradiated congenic mice +and their outcome was followed over time (Fig.2a and Extended Data +Fig.2d). All recovered donor-derived cells were pDC, indicating a +non-pDC origin for tDC at steady state. Accordingly, tDClo and +tDChi numbers were not affected by the temporal depletion of pDC +in BDCA2-DTR+/− mice (Fig.2b).
+FIGURE 2. tDC originate from BM progenitors.
+
a. Percentage (mean + SD) of each DC subset recovered after the +adoptive transfer of 30,000 BM or splenic CX3CR1EGFP CD45.1 pDC into +CD45.2 WT non-irradiated congenic mice. Transferred cells were analyzed in the +spleen of recipient mice after 2-8 days. BM: n=4 at 2- and 4-days, and n=3 at +8-days. Spleen: n=3 at 2- and 4-days, and n=2 at 8-days. N represents mice and +experiments.
+b. Bar graphs (mean + SD) of the total number of each splenic DC subset +(gated a in Fig.1f) in littermate controls +(C; n=11 BDCA2-DTR−/− mice) or +BDCA2-DTR+/− mice inoculated with DT and +analyzed 1 day later (n=8 mice), or inoculated with DT every second day and +analyzed at day 7 (n=8 mice). Data pooled from 5 exp. Statistical differences +were determined by one-way ANOVA with Tukey’s multiple comparison +test.
+c. Left shows schematic of BM cells cultured with FLT3L +/− +OP9-DL1 cells. Right shows the gating strategy to detect DC subsets at 6-days of +BM culture, analyzed by flow cytometry (excluded lineage contains +CD3+/CD19+/F4/80+ cells). One +representative of 5 experiments.
+d. Histograms of hCD2EYFP (left) or CD300cTdT +(right) expression in BM-derived DC as defined in (c), analyzed at 6-days. One +representative of 3 experiment.
+e. pDC and tDC percentage (mean + SD) in BM cultures from (c), gated on +total live and single cells. N=5 samples/time point, except n=6 samples for +FLT3L+DL1 at 4-days. Data pooled from ≥ 5 experiments. Statistical +differences were determined by Two-way ANOVA with Sidak multiple comparisons +test.
+f. Histograms of protein expression of DC markers in BM-derived DC as +defined in (c) or splenic DC as defined in Fig.1f. Number indicates gMFI x 102. One representative +of 3 experiments.
We next asked if tDC could originate from BM progenitors by culturing +unfractionated BM cells with FLT3L for DC differentiation34. +B220+BST2+SiglecH+ cells, previously +characterized as pDC, could be further divided into +Ly6D+CX3CR1− pDC and +Ly6D−CX3CR1+ tDC (Fig.2c). Both of these populations labeled the highest when using +hCD2EYFP and CD300cTdT lineage tracing models (Fig.2d). BM-derived tDC peaked at 6-days and +their development was abolished when OP9 stromal cells expressing Delta-like 1 +(DL1) were added to the cultures (Fig.2e), +as described for pDC35. +Similar to their splenic counterparts, BM-derived tDC expressed TCF4 and +intermediate levels of pDC and cDC markers (Fig.2f). BM-derived tDC did not segregate into tDClo and +tDChi, suggesting that in vivo signals are +missing in the culture.
+We concluded that tDC originate from BM progenitors at steady state. +Interestingly, we detected a small number of tDC in blood but almost none in the +BM, suggesting that tDC quickly exit the BM after development (Extended Data Fig.2e-f).
tDC originate from BM pro-pDC at steady state
+Our data so far suggest that tDC originate from BM progenitors shared +with pDC. We tested this unbiasedly by inferring the trajectory of tDC +development in CD135-enriched BM cells analyzed by scRNAseq (Fig.3a). Application of the splenic tDC signature +score to the clusters identified by Seurat labeled most pDC- and cDC- primed +cells, which was attributed to tDC transcriptional similarities with pDC and +DC2. Thus, the splenic pDC signature score was applied instead, which colored +clusters expressing Siglech and Ly6d, but not +Csf1r encoding CD115 (Fig.3b-c). Analysis of cells +with a pDC score >0.15 resulted in 6 clusters (Fig.3d). Cluster 0 represented pDC expressing +Iglc3 and Klk1, whereas cluster 5 represented tDC +expressing Lgals3, Ms4a6c and Cx3cr1 (Extended Data Fig.3a). Both cluster 0 and +cluster 5 expressed H2 genes and had gene enrichment for immune +pathways, whereas clusters 2-4 were enriched for cell cycle pathways, a +characteristic of progenitors (Extended Data +Fig.3b). We then applied Slingshot and Velocity analyses to identify +putative developmental relationships between these clusters (Fig.3e-f). Both +algorithms revealed cluster 2 branching to pDC (cluster 0) and tDC (cluster 5). +Gene expression dynamics demonstrated upregulation of Ccr9 and +Klk1 in pDC, and Ms4a6c, Cx3cr1 and Lgals3 in tDC +(Fig.3g). Besides expressing +Siglech and Ly6d (Fig.3c), cluster 2 also expressed +Itgax encoding for CD11c (Extended Data Fig.3a), markers of pDC progenitors +(pro-pDC)7.
+FIGURE 3. tDC originate from pro-pDC.
+
a. CD135-enriched BM cells were sorted as +live/singlets/CD45+/Lineage− +(CD3−/CD19−/CD335−/Ly6G−) +cells and prepared for scRNAseq using droplet-based genomics. After filtering +for low quality cells, 2831 cells were clustered using Seurat and represented in +a UMAP.
+b. Splenic pDC-signature in BM progenitors. A signature matrix was +generated using RNAseq data of splenic DC and CIBERSORTx. Labeled are cells with +a positive gene enrichment score.
+c. UMAP showing the expression of the indicated genes.
+d. UMAP of “pDChigh” cells in the BM. Cells +from (b) with a pDC-specific gene enrichment score of >0.15 were selected +and re-clustered using Seurat.
+e. Inference of differentiation trajectory of the clusters defined in +(d) using Slingshot.
+f. Inference of differentiation trajectory of the clusters defined in +(d) using RNA velocity. Velocity vector is indicated as streamlines.
+g. Velocity illustrating the expression of selected pDC- and tDC- +specific genes.
+h. Unbiased X-shift clustering of BM progenitors analyzed by CyTOF and +represented by UMAP. CD135-enriched BM cells were gated as +Lineage− +(CD3−/CD19−/CD335−/Ly6G−) +B220−/MHCII−/CD117int-lo/CD135+, +and concatenated with 10,000 B220+ pDC (left). UMAP was plotted based +on the expression of all markers excluding lineage, and colored based on the +expression of the indicated proteins (right).
+i. Heatmap depicting marker enrichment modeling (MEM) scores in +clusters identified in (h). imm pDC: immature pDC; early pro: early +progenitors.
+j. CD135-enriched BM cells were gated as Lineage− +(CD3−/CD19−/ +CD335−/Ly6G−) +B220−/MHCII−/ +CD117int-lo/CD135+ as in (h). Graphs show bi-axial +gating strategy to identify immature pDC (cluster 4; +CD11c+CX3CR1−Ly6D+SiglecH+), +pro-pDC (cluster 3; +CD11c+CX3CR1+Ly6D+SiglecH+) and +pro-cDC (clusters 2 and 7; +CD11c−CX3XR1+CD115+ +Ly6C+/−).
+k. Bar graph (mean + SD) showing the percentage of BM pDC, imm pDC, +pro-pDC and pro-cDC labeled with EYFP in hCD2EYFP mice. N=4 mice and +4 independent experiments.
+l. Clusters gated as in (j) were overlaid on the CyTOF UMAP of (h).
+m. 15,000-30,000 pro-cDC, imm pDC or pro-pDC sorted from BM of +CX3CR1EGFP CD45.1 mice using the strategy described in (j) were +adoptively transferred into non-irradiated WT CD45.2 congenic mice. The spleen +of recipient mice was analyzed 2-4-days later to evaluate the outcome of +transferred cells. Shown is the percentage (mean + SD) of each DC subset +recovered in the spleen of recipient mice. Pro-cDC: n=3 at 2-days and n=5 at +4-days. Imm pDC: n=3 at 2- and 4-days. Pro-pDC: n=3 at 2-days and n=5 at 4-days. +N represents number of mice and independent experiments.
+n. As in (m), but splenic tDC recovered following the adoptive transfer +of pro-pDC were further divided into tDClo and tDChi. A +representative gating strategy (left) and percentage (mean + SD) of each tDC +population (right) are shown. N=3 at 2-days and n=5 at 4-days. N represents +number of mice and independent experiments.
+o. As in (m), but progenitors were sorted from hCD2EYFP mice +and adoptively transferred into congenic non-irradiated recipients. The spleen +of recipient mice was analyzed for the presence of transferred cells at 4-days, +which were gated as described in Fig.1f and +further analyzed for EYFP expression. Shown are tDC recovered after the adoptive +transfer of pro-pDC (left), DC2 recovered after the adoptive transfer of pro-pDC +(middle), and DC2 recovered after the adoptive transfer of pro-cDC (right). One +representative of 3 exp.
+p. Clusters gated as in Extended Data +Fig.4a were overlaid on the CyTOF UMAP of (h).
We aimed to unbiasedly establish a gating strategy to isolate pro-pDC +and evaluate their capacity to give rise to tDC. We performed CyTOF of +CD135-enriched BM cells, and identified 7 clusters by UMAP. Marker enrichment +analyses (MEM) pinpointed cluster-specific identifiers (Fig.3h-i). +Cluster 6 was recognized as pDC and probably some tDC that were not resolved by +the clustering algorithm. Cluster 4 corresponded to immature pDC expressing low +levels of pDC markers. Cluster 3 had SiglecH, Ly6D, CX3CR1 and CD11c, an +expression pattern consistent with pro-pDC. Cluster 2 and 7 both expressed CD115 +and CX3CR1 but not SiglecH, corresponding to conventional DC progenitors +(pro-cDC). Also, the expression pattern of cluster 2 and 7 correlated with +pro-DC2 (Ly6Chi) and pro-DC1 (Ly6Clo), respectively. Based +on the MEM analysis, a gating strategy to purify pro-pDC (cluster 3), immature +pDC (cluster 4) and pro-cDC (clusters 2 and 7) was designed (Fig.3j), which was corroborated by overlaying onto the +UMAP, and evaluating EYFP expression when using hCD2EYFP mice (Fig.3k-l). Only pDC, immature pDC and pro-pDC, but not pro-cDC, had >80% +EYFP expression. Similar expression patterns were observed in +CD300cTdT mice (Extended Data +Fig.3c).
+We assessed the potential of these progenitors to give rise to tDC +following their adoptive transfer (Fig.3m +and Extended Data Fig.3d-e). Immature pDC gave rise to pDC, whereas pro-cDC +produced almost exclusively cDC, confirming the identity of the clusters. +Pro-pDC gave rise to tDC and pDC. Analysis of recovered tDC revealed the +presence of mainly tDClo at 2-days but the appearance of +tDChi at 4-days post-transfer, suggesting a trajectory of +conversion from tDClo to tDChi (Fig.3n). Notably, recovered tDClo and +tDChi labeled 100% with EYFP when transferred from +hCD2EYFP mice, corroborating their common origin (Fig.3o). Surprisingly, pro-pDC also gave rise to DC2 +(defined as in Fig.1f), as well as a small +number of DC2 lacking CD11b, a population that is analyzed below (Fig.5). Interestingly, DC2 derived from pro-pDC were +labeled 100% with EYFP when using hCD2EYFP mice (Fig.3o). In contrast, DC2 derived from pro-cDC were +EYFP−, indicating that hCD2EYFP can trace the +heterogenous origins of DC2.
+Finally, we reconciled our new gating strategy with previously described +approaches. BM CD115+, CD127+, and double negative (DN) +progenitors were purified and adoptively transferred (Extended Data Fig.4). tDC arose from DN +SiglecH+Ly6D+ progenitors. When these DN progenitors +were overlaid onto our UMAP of CyTOF data, we observed they fell within clusters +3 and 4, suggesting this prior gating strategy does not separate pro-pDC from +immature pDC (Fig.3p). Even though the +majority of the CD115+Ly6C+/− progenitors fell +within pro-cDC in the UMAP, a small fraction overlapped with pro-pDC (Fig.3p). +CD127+SiglecH+Ly6D+ cells, suggested to be +the immediate pDC progenitor25, partially overlapped with cluster 4 but not cluster 3, +indicating that this gate captures a few immature pDC but not pro-pDC. +Altogether, our gating strategy provides improved resolution of BM +progenitors.
+We concluded that tDC originate from a CD115− BM +progenitor that expresses SiglecH, Ly6D, CD11c and CX3CR1, and corresponds to +the recently described pro-pDC progenitor7.
Splenic tDC differentiate into ESAM+ DC2
+In hCD2EYFP and CD300cTdT lineage tracing models, +~25% of DC2 were labeled (Fig.1h-i). Thus, we queried if +tDC could convert into these labeled DC2 (Fig.4a and Extended Data +Fig.5a). tDC acquired DC2 markers, specifically +CD11chiMHCII+CD172a+ expression, following +adoptive transfer into non-irradiated congenic recipients. At 2-days +post-transfer, some DC2 lacked CD11b expression (CD11b− DC2), +but most were CD11b+ by 4-8-days suggesting a transition from a +CD11b− to a CD11b+ DC2 (evaluated in Fig.5). When tDC were purified from +hCD2EYFP mice, ~85% of recovered DC2 were EYFP+ +independently of their expression of CD11b (Fig.4b). This data suggests that tDC can convert into DC2.
+FIGURE 4. tDC convert into mouse ESAM+ DC2 and human CD5+ +DC2.
+
a. Schematic of adoptive transfer (left). 30,000 splenic tDC (All) +sorted from CX3CR1EGFP CD45.1 mice using Fig.1f gating strategy, and adoptively transferred +into WT CD45.2 non-irradiated congenic mice. The percentage (mean + SD) of each +DC subset recovered in the spleen of recipient mice is shown 2-8 days after +transfer. N=5 at 2-days, n=6 at 4-days, and n=4 at 8-days. N represents number +of mice and independent experiments.
+b. As in (a), but tDC were sorted from hCD2EYFP mice. +Histograms of hCD2EYFP expression in sorted tDC pre-transfer (left), +and CD11b− and CD11b+ DC2 recovered in the spleen +of recipient mice at 8-days. One representative of 2 experiments.
+c. Bar graphs (mean + SD) of the total number of splenic DC subsets +(gated a Fig.1f) in hCD2EYFP +CX3CR1DTR mice inoculated (n=4 mice) or not (n=3 mice) with DT +every second day and analyzed at day 7. Statistics were determined by unpaired +two-tailed t-test. Data pooled from 2 experiments.
+d. As in (c), but DC2 gated as in Fig.1f were further divided in EYFP+ and +EYFP− cells, and their frequencies were plotted.
+e. Splenic EYFP+ and EYFP− DC2 were sorted +from hCD2EYFP mice and prepared for bulk-RNAseq analysis (Extended Data Fig.1a). MA plot comparing +differences between EYFP+ and EYFP− DC2. Genes with +2-fold Log2 change are colored in red (up) or blue (down).
+f. Splenic DC2 from hCD2EYFP mice were gated as in Fig.1f and further analyzed to delineate +ESAM+ and CX3CR1+ subpopulations (left). EYFP +expression in ESAM+ and CX3CR1+ DC2 (right). One +representative of 8 experiments.
+g. DC2 from the spleen of hCD2EYFP mice were gated on +CX3CR1+, ESAM+ EYFP+ and +EYFP− as shown in (f)(left). Each DC2 subpopulation was +analyzed for their relative expression of surface markers. Heatmap shows the +Z-scored gMFI (right). Z-score values was determined from n ≥ 2 mice. +Data pooled from 8 independent experiments.
+h. Dot plot of ESAM+ DC2 gated as in (f) was colored base on +EYFP expression (top). Bar graph (mean + SD) showing the percentage of +EYFP+ in ESAM+ CD24− and +ESAM+ CD24+ (bottom)(n=5 mice in 4 experiments). +Statistics were determined by unpaired two-tailed t-test.
+i. Freshly isolated tDC were sorted from human blood and stimulated +with CD40L for 2-days. Cells were analyzed by CyTOF and Scaffold at time 0 and +2-days post-stimulation (n=3 samples in 3 independent experiments).
+j. As in (i), but bar graph shows the frequency (mean + SD) of tDC +mapped to each scaffold landmark node at day 0 and 2 post-stimulation. N=2 at 0 +days and n=3 at 2-days. N represents number of samples and independent +experiments.
+k. Heatmap of the expression of the indicated markers in freshly +isolated (day 0) and stimulated (day 2) human tDC analyzed by CyTOF and flow +cytometry. The frequency of positive cells for each marker is shown (≥ 2 +donors/per marker in ≥ 2 experiments).
We then aimed at evaluating if tDC conversion into DC2 occurs in +vivo at homeostasis. Given that tDC express CX3CR1 and are labeled +in hCD2EYFP mice, we hypothesized that they would be specifically +depleted in hCD2-iCre+/− x +Cx3cr1LSL-DTR+/− mice. One DT dose +resulted in tDC depletion without affecting other DCs (Extended Data Fig.5b). Continuous DT administration +for 7-days also depleted pDC due to elimination of pro-pDC, as +described7(Extended Data Fig.5b-c); however, total DC2 numbers were unaffected. To +reconcile the adoptive transfer and tDC depletion experiments, we crossed +hCD2EYFP with Cx3cr1LSL-DTR+/− +mice (hCD2EYFP CX3CR1DTR), which allowed us to monitor +EYFP+ DC2 following DT inoculation. tDC and pDC were decreased +7-days post-DT administration, but total DC2 numbers were unaffected (Fig.4c). However, the fraction of +EYFP+ DC2 decreased ~2-fold, with a corresponding increase +in EYFP− DC2 (Fig.4d). +Thus, tDC are a source of a fraction of DC2 in vivo, which can +be traced in hCD2EYFP mice. To confirm that EYFP-labeling +distinguishes tDC-derived DC2, we evaluated the expression of IgH +rearrangements, a feature of tDC and pDC that is passed to progeny +cells4. IgH +rearrangements were detected in EYFP+ but not EYFP− +DC2 (Extended Data Fig.5d). Notably, +CD11b− DC2 were also eliminated post-DT administration, +supporting a relationship with tDC (Fig.4c).
+To gain a comprehensive understanding of tDC-derived DC2, we sorted +EYFP+ and EYFP− DC2 from hCD2EYFP +mice and performed bulk-RNAseq (Fig.4e). +DEG analysis revealed that EYFP− DC2 uniquely expressed +Fcgr2b (encoding CD32), Cx3cr1, +Clec12a, Clec7a, resembling +CX3CR1+ DC216,17(Supplementary Table5). Despite a +small contamination with EYFP+ lymphoid cells, tDC-derived +EYFP+ DC2 expressed higher levels of Gpr4 and +the Notch target Dtx1, genes expressed by ESAM+ +DC216,17. Flow cytometry confirmed that +EYFP+ were contained within ESAM+ DC2 and not +CX3CR1+ DC2 (Fig.4f). When +comparing relative expression of DC2 markers between the different DC2 +subpopulations, EYFP+ ESAM+ DC2 lacked expression of CD32, +CLEC7A, CLEC12A, all markers of CX3CR1+ DC216,17(Fig.4g). +EYFP+ ESAM+ DC2 expressed CD4 and CD5, and differed +from EYFP− ESAM+ DC2 in their higher and lower +levels of CD24 and CD11b, respectively (Fig.4g-h). These results were +confirmed using CD300cTdT mice (Extended Data Fig.5e). Thus, tDC can convert into a subpopulation of +DC2 expressing ESAM.
+Human tDC may give rise to DC1 and DC25. However, our mouse data indicates that +tDC only convert into ESAM+ DC2, which is presumably the equivalent +to human CD5+ DC216. We re-evaluated human tDC conversion into cDC using CyTOF and +unbiased Scaffold analysis. Freshly isolated blood tDC localized in the tDC +landmark node (day 0). By 2-days in culture, ~60% tDC mapped to DC2, but +not DC1 (Fig.4i-j). We observed a few tDC localized in the pDC +landmark node (~5%), which was attributed to a loss of AXL. Human tDC +upregulated CD33, CD172a, CD32, CD1c, but not the DC1-specific marker CLEC9A +(Fig.4k). tDC also upregulated CD5, but +failed to upregulate CD14 and CD163, suggesting conversion into DC2, but not +DC322. Altogether, our +data indicate that tDC differentiate into mouse ESAM+ DC2 and human +CD5+ DC2.
tDC are distinct from cDC-committed precursors
+We sought to clarify tDC relationship with previously described +DC2-committed precursors. We focused on the spleen because of the high tDC +numbers vs BM or blood. Splenic precursors have differential lineage commitment +based on the expression of SiglecH and Ly6C11. SiglecH+Ly6C+ (pop#5; Fig.5a) are recognized as uncommitted +precursors that give rise to DC1 and DC2, whereas +SiglecH−Ly6C− (pop#3; Fig.5a) and SiglecH−Ly6C+ +(pop#4; Fig.5a) are DC1- and DC2-committed +precursors, respectively. We observed that EYFP labeling was similar between +pop#5 and tDC when using hCD2EYFP mice (Fig.5b). Contrarily, only ~25% of pop#4 and ~10% of +pop#3 were EYFP-labeled. We then mapped these 3 splenic precursor populations +onto a UMAP of all splenic DC. Pop#5 mapped within tDC; however, only fell +within tDClo and represented ~25% of all the tDC cluster +(Fig.5c-d). Pop#3 mapped within pre-DC1, although some mapped within DC1 +suggesting that the reported gating strategy includes contaminating mature +XCR1+ DC1. Most of pop#4 (~85%) mapped outside the tDC +gate, with only ~4% contaminating all tDC. Thus, tDC are distinct from +pop#3 and pop#4. Notably, our lineage tracing data implies that pop#5 are tDC +and not uncommitted precursors that give rise to pop#3 and pop#4, as +suggested11.
+FIGURE 5. tDC are distinct from pre-cDC.
+
a. CD135-enriched splenic cells were stained for flow cytometry to +identify pre-cDC. Lineage includes +CD3+/CD19+/NK1.1+/Ly6G+ cells. +One representative of 3 experiments.
+b. Populations gated as in (a) were analyzed for their percentage of +EYFP+ (mean + SD) in hCD2EYFP mice (n=3 mice in 3 +experiments). tDC were gated as in Fig.1f.
+c. Frequency of pop#4 and pop#5 within the tDC gate is shown as the +mean of n=3 mice in 3 experiments.
+d. CD135-enriched splenocytes from hCD2EYFP mice were +labeled for flow cytometry, gated as +live/singlets/CD3−/CD19−/NK1.1−/Ly6G− +cells and analyzed by UMAP. DC subset assignment (left) was done by gating cells +as in Fig.1f. UMAP was labeled for EYFP +(right). One representative of 3 experiments.
+e. CD135-enriched splenocytes gated as in Fig.1f were further analyzed for the identification of +CD11b− DC2, pre-DC2 and pre-DC1. One representative of 5 +experiments.
+f. Splenocytes from hCD2EYFP mice gated as in Fig.1f and Fig.5e +were analyzed for the frequency (mean + SD) of EYFP+ cells. N=5 mice +in 5 experiments.
+g. CD135-enriched splenocytes were stained and analyzed by CyTOF. +CD3−/CD19−/NK1.1−/Ly6G− +cells were gated as in Fig.1f and Fig.5e, and overlaid into the UMAP of all +cells (left)(n=3 mice in 1 exp.). Heatmap of the Z-scored expression of each +marker analyzed by CyTOF or flow cytometry (right)(n=3 mice/surface marker).
+h. Percentage (mean + SD) of each DC subset recovered after the +adoptive transfer of 10,000-30,000 CX3CR1EGFP CD45.1 cell populations +(gated as described in Fig.1f and Fig.5e). Transferred cells were analyzed in +the spleen of recipient mice after 2 or 4 days. tDClo and +tDChi: n=3 at 2-days and n=5 at 4-days. +CD11b−DC2: n=3 at 2-days and n=4 at 4-days. Pre-DC2: n=3 at 2- +and 4-days. Pre-DC1, n=2 at 2-days and n=4 at 4-days. N represents number of +mice and independent experiments.
+i. As in (h), but cells were sorted from hCD2EYFP mice and +adoptively transferred into congenic non-irradiated recipients. The spleen of +recipient mice was analyzed for the expression of ESAM and CX3CR1 (top panels) +or EYFP (bottom panels) at 4-days post-transfer. One representative of 3 +experiment.
+j. As in (h), but splenic cells recovered following the adoptive +transfer of tDClo were analyzed as tDClo and +tDChi. One representative of 3 experiments.
+k. WT mice were pulsed with BrDU for 14 days. 6-8-days post-BrDU +removal, DC subsets gated as in Fig.1f and +Fig.5e were analyzed for their BrDU +content by flow cytometry. The frequency of BrDU+ cells (mean + SD) +is shown. N=4 mice in 2 experiments. Statistical differences were determined by +one-way ANOVA with Tukey’s multiple comparison test.
+l. Bar graphs (mean + SD) showing the total number of splenic DC (gated +a Fig.1f and Fig.5e) in hCD2EYFP CX3CR1DTR mice inoculated +or not with DT every second day, and analyzed at day 10 (n=3 mice in two +experiments). See also Extended Data +Fig.6d-e. Statistical +differences were determined by unpaired two-tailed t-test.
We designed a gating strategy to analyze pop#3 and pop#4 precursors +without contaminating DC1 or tDC (called pre-DC1 and pre-DC2, hereafter) (Fig.5e). Our new gating strategy yielded +pre-DC2 and pre-DC1 with ~10% EYFP labeling in hCD2EYFP mice +(Fig.5f). It also allowed separation of +CD11b− DC2, that label similar to tDC in +hCD2EYFP mice. As expected, EYFP labeling was only evident in +ESAM+ DC2, but not in CX3CR1+ DC2. We validated our +new gating strategy by mapping all populations onto our UMAP of CyTOF data +(Fig.5g). tDClo, +tDChi, CD11b− DC2 and pre-DC2 mapped in +different locations of the UMAP, indicating distinct phenotype. tDClo +and pre-DC2 shared the expression of CX3CR1, but tDClo expressed +higher levels of Ly6C, IRF8, TCF4 and SiglecH (Fig.5g). tDChi expressed CD11c, CD172a, MHCII and some +CD8, all markers absent in pre-DC2. We observed higher expression of CCR2 in +pre-DC2, a chemokine receptor known to be expressed in cDC precursors36. Notably, +CD11b− DC2 and DC2 shared most markers; however, +CD11b− DC2 expressed CD135 and low levels of CD24. +Interestingly, pDC, tDC and CD11b− DC2 expressed higher levels +of CD45RB and CD45RA, suggesting that these markers may be useful to identify +these cells in models other than hCD2EYFP mice. Thus, our +multi-dimensional analyses unmasked key phenotypic differences between tDC and +pre-DC2.
+We compared tDC and pre-DC2 capacity to give rise to DC2 after their +adoptive transfer (Fig.5h). Pre-DC1 were +used as controls, and only gave rise to DC1. Transferred pre-DC2 only gave rise +to DC2, whereas tDClo, tDChi, CD11b− DC2 +also gave rise to some tDC and CD11b− DC2 (Fig.5h). Pre-DC2 gave rise to EYFP− +cells, whereas tDClo, tDChi and CD11b− +DC2 gave rise to EYFP+ cells when purified from hCD2EYFP +mice (Fig.5i). Importantly, all tDC-related +populations gave rise to CX3CR1− but not CX3CR1+ +DC2, and most of these cells (~70%) expressed ESAM (Fig.5i). Transferred tDClo cells gave rise +to ~60% tDChi by 4-days indicating a transition from +tDClo to tDChi (Fig.5j). Similarly, few tDC were recovered when tDChi +were transferred, and none when CD11b− DC2 were transferred. +Together, this transfer data implies a transition from tDClo → +tDChi → CD11b− DC2 → +ESAM+ DC2.
+To further contrast tDC and pre-DC2, we compared their turnover rate, +which is high in DC precursors37, by performing a BrdU “pulse-chase” +experiment (Extended Data Fig.6a). Pre-DC2 +had the highest turnover rate, clearing ~90% of BrdU by 8-days (Fig.5k), whereas tDClo and +tDChi only cleared ~50% and ~35% of BrdU. +Accordingly, Ki-67 proliferation marker was highest in pre-DC2 (Extended Data Fig.6b). Thus, tDC have a low turnover +rate in comparison to pre-DC2, which could explain why we did not observe a DC2 +decrease after tDC depletion for 7-days. Consistently, analysis of +hCD2EYFP CX3CR1DTR mice following 10-days of DT +treatment showed ~30% DC2 decrease (Fig.5l). Only ESAM+ EYFP+ DC2, but not +CX3CR1+ DC2 were decreased (Fig.5l, Extended Data Fig.6d). +Importantly, pre-DC2 were not affected, which may explain the maintenance of +~50% ESAM+ DC2 following tDC depletion. Similarly, monocytes, +neutrophils and lymphocytes were not affected (Extended Data Figure 6e)
+Altogether, our data show that tDC are distinct from pre-DC2 in their +phenotype, turnover and proliferative rate. It also implies that +ESAM+ DC2 have two different origins, i.e., BM +CD115− pro-pDC and CD115+ pro-cDC.
tDC require IRF4 to convert into DC2
+The TFs IRF4 and KLF4 are distinctly required for ESAM+ and +CX3CR1+ DC2 development, respectively12,17,19. We queried +if IRF4 is necessary for tDC conversion into ESAM+ DC2. We correlated +Irf4 expression with EGFP levels following the +Itgax-Cre-mediated recombination of +Irf4fl/fl +(IRF4cKO mice), and found highest EGFP expression in +tDChi (Fig.6a). +IRF4cKO mice had slightly increased tDC and decreased DC2 +numbers, specifically ESAM+ DC2 (Fig.6b). Interestingly, IRF4cKO mice had almost no +CD11b− DC2, but normal numbers of pre-DC2, suggesting that +only tDChi → CD11b− DC2 → +ESAM+ DC2, but not pre-DC2 ESAM+ DC2 conversion was +affected. We next generated competitive BMC to evaluate the intrinsic role of +IRF4 in each population. tDC did not require IRF4; however, +CD11b− DC2 were IRF4-dependent as these cells derived +almost exclusively from IRF4-proficient BM (Fig.6c). Contrarily, IRF4 was not required for pre-DC2. As expected, +ESAM+ but not CX3CR1+ DC2 required IRF4.
+FIGURE 6. IRF4 is required for tDC transition to CD11b− DC2.
+
a. Bar graphs (mean + SD) showing EGFP expression in splenic DC from +IRF4cKO, gated as described in Fig.1f and Fig.5e. N=11 +mice/subset, except n=8 mice for pre-DC2, in ≥ 4 experiments. Statistical +differences were determined by one-way ANOVA followed by Tukey multiple +comparison test.
+b. Bar graphs (mean + SD) show the frequency of DC in +IRF4control mice and IRF4cKO mice analyzed for the +proportion of each splenic DC population, gated as in Fig.1f and Fig.5e. pDC, DC2 and DC1: n=9 (IRF4control) and n=11 +(IRF4cKO). tDC (All), tDClo and tDChi: n=9 +(IRF4control) and n=11 (IRF4cKO). +CD11b−DC2: n=5 (IRF4control) and n=5 +(IRF4cKO). Pre-DC2: n=5 (IRF4control) and n=5 +(IRF4cKO). ESAM+DC2 and CX3CR1+DC2: n=7 +(IRF4control) and n=8 (IRF4cKO). N represents mice in +≥4 independent experiments. Statistical differences were determined by +two-way ANOVA with Sidak multiple comparison test, and t-test for +CD11b−DC2 and pre-DC2.
+c. Left shows schematic of BM chimera generated by transplanting 50% WT +(CD45.1) and 50% IRF4cKO (CD45.2) BM. Right shows bar graphs (mean + +SD) of the contribution (ratio) of donor cells to splenic DC populations (top) +or DC2 subpopulations (bottom) in mixed BMC, normalized to NK cells (n=7 mice in +2 experiments). Statistical differences were determined by one-way ANOVA with +Dunnett’s multiple comparison test.
+d. Schematic of adoptive transfer (left). 30,000 tDC (All) were sorted +from CD45.2 IRF4control and IRF4cKO, and transferred to +CD45.1 WT. Right graphs show the total # of recovered cells (mean + SD) in the +spleen 2-8-days post-transfer (n=3 mice in 3 experiments/time point). +Statistical differences were determined by two-way ANOVA with Sidak multiple +comparison test. Colored p-values represent statistical difference between DC2 +and tDCs (cyan), or DC2 and CD11b− DC2 (gray).
We validated our results by performing adoptive transfer experiments +(Fig.6d). IRF4control tDC +yielded DC2, whereas IRF4cKO tDC yielded no cells by 8-days (Fig.6d). IRF4cKO tDC still yielded +some tDC at 4-days post-transfer, suggesting arrested conversion to DC2. Similar +results were observed when tDC from FLT3L BM cultures were purified and +re-cultured in vitro (Extended +Data Fig.6c). Altogether, our results indicate that tDChi +→ CD11b− DC2 → ESAM+ DC2 conversion +is dependent on IRF4.
tDC respond to TLR agonists and promote T cell proliferation
+Mouse and human tDC are able to respond to the TLR9 agonist CpG, +upregulate MHCII/costimulatory molecules, and promote the proliferation of +allogenic T cells4, suggesting +that these cells are not just precursors, but harbor functions of differentiated +DC. To look further at the functional capabilities of tDC vs cDC, we first +evaluated TLR expression. tDC expressed TLR7 and TLR9, but low levels of TLR4 +and TLR3 (Extended Data Fig.7a). +tDClo and tDChi have lower basal expression of MHCII +and CD864, but were able to +quickly upregulate these markers post-TLR7 or TLR9 stimulation (Extended Data Fig.7b-c; see4 for TLR9). +Also, as shown for TLR94, tDC +produced IL-12p40 in response to TLR7. Thus, tDC quickly respond to TLR7 and +TLR9 agonists.
+We assessed unbiasedly the transcriptional changes that tDC undergo +following TLR stimulation. Splenocytes from hCD2EYFP mice were +stimulated with a TLR cocktail to activate all DC simultaneously (TLR3, 4, 7, +and 9) and bulk-RNAseq was performed 3hr later in purified cells. +tDClo-hi were analyzed as one population, and DC2 were divided +into EYFP+ and EYFP− DC2. PCA showed that +stimulated DC clustered separately from unstimulated cells, accounting for +~50% of variance (Fig.7a). +Stimulated tDC clustered independently from other DC subsets, implying that +their gene expression program was shaped by their activation status and core +signature genes, as previously shown for pDC and cDC38. Activated tDC upregulated pathways +related with DC maturation, including hallmark inflammatory, TNF, interferon +responses, and genes of the IL-1-family (Il1b, +Il1r1, and Il1rap)(Fig.7b-c). Thus, +tDC upregulate key programs for the induction of adaptive immune responses, a +feature shared with pDC and cDC.
+FIGURE 7. tDC respond to TLR stimulation and activate antigen-specific naïve T +cells.
+
a. CD135-enriched splenic DC were activated or not with a cocktail of +TLR agonists consistent on LPS/PolyIC/Resiquimod/CpGA (stimulated) for 3 hrs +before sorting (as described in Extended Data +Fig.1a), and prepared for bulk RNAseq (n=2 mice in 2 exp.). PCA plot +of the top 500 most variable DEG in each unstimulated or stimulated DC +subset.
+b. GSEA of selected pathways using the 500 most variable DEG. Each DC +subset was compared to their respective unstimulated control.
+c. Heatmap of selected genes from the GSEA analysis in (b).
+d. Sorted CTV-labeled naïve OT-II CD4+ T cells were +co-cultured with sorted splenic DC subsets (purified using the gating strategy +in Fig.1f and Fig.4f) at a ratio of 5:1 (T cell:DC), in the presence of OVA +protein. Four days after co-culture, cells were analyzed by flow cytometry. +Left: representative flow cytometry analysis. Right: bar graphs (mean + SD) of T +cell proliferation. N=5 for CD11b− DC2, n=7 for +tDChi, and n=8 for all the other groups. N represents +biologically independent samples in ≥ 2 experiments. Statistical +differences were determined by one-way ANOVA with Tukey’s multiple +comparison test.
We next evaluated if tDC share other functions with cDC, e.g., antigen +capture, processing and presentation to T cells. FITC-beads or PKH-labeled sheep +red blood cells (SRBC) were administered i.v. for their in vivo +capture by splenic DC subsets. FITC-beads and PKH-labeled SRBC were taken up +comparably between tDC and cDC (Extended Data +Fig.7d-e). Notably, pre-DC2 or +pre-DC1 were unable to capture antigen in vivo. tDC also +captured, processed, and presented ovalbumin (OVA) protein to naïve +antigen-specific CD4+ T cells, which underwent several rounds of +division (Fig.7d). Although +tDChi were more efficient than tDClo at activating T +cell proliferation, both populations were significantly better than pDC. +tDChi and CD11b− DC2 were slightly better than +ESAM+ DC2, which may be consequence of differential survival of +these cells in culture, an observation that remains to be confirmed.
+In summary, tDC respond to TLR7/9 stimulation, capture, process and +present antigen to T cells, indicating that they are not just DC2 precursors but +a functional DC population.
tDC promote immunopathology during coronavirus infection
+Finally, we sought to understand tDC function in vivo. +We hypothesized that tDC would play a role during viral infection, given their +developmental relationship with pDC and expression of TLR7/9. We chose M-CoV +since pDC are known to be essential for its early containment27,28. Similar to influenza infection4, tDC accumulated in the primary infected +tissue, which is the liver for M-CoV (Extended +Data Fig.8a-c). We reasoned +that models for the specific depletion of pDC and tDC (hCD2EYFP +CX3CR1DTR + DT mice; pDCΔtDCΔ +mice hereafter) vs pDC only (BDCA2-DTR + DT mice; +pDCΔ mice hereafter) would allow us to dissect tDC +function during infection. pDCΔtDCΔ mice +depleted liver pDC and tDC, whereas pDCΔ mice depleted only +pDC (Fig.8a). None of these models affected +other liver immune cells (Extended Data +Fig.8a-b, d).
+FIGURE 8. Secretion of IL-1β by tDC results in immunopathology in pDC-depleted +mice.
+
a. Liver pDC and tDC numbers (mean + SD) at 2- and 5-days following +M-CoV infection of pDCΔ and +pDCΔtDCΔ (n=3/group at 2-days and +n=5/group at 5-days) vs control mice (n=11 at 2-days and n=9 at 5-days). +pDCΔ mice are +BDCA2-DTR+/− mice inoculated with DT one +day before M-CoV infection. pDCΔtDCΔ are +hCD2EYFP CX3CR1DTR mice inoculated with DT every other +day for 5-days before M-CoV infection. Control mice are a combination of +BDCA2-DTR+/− or hCD2EYFP +CX3CR1DTR mice inoculated with PBS, and +DTR−/− mice inoculated with DT (no difference in +control mice was observed). N represents mice. Data pooled from ≥ 3 +experiments.
+b. Percent weight loss (left, mean + SD) and survival curve (right) of +control (n=12), IFNAR1KO (n=8), pDCΔ (n=7) and +pDCΔtDCΔ (n=8) mice. Health status was +monitored twice daily, and moribund mice were euthanized. N represents mice. +Data pooled from 4 experiments.
+c. Serum ALT levels (mean + SD) determined 5-days post-M-CoV infection +(n=4 mice for control; n=6 mice for pDCΔ; n=7 mice for +pDCΔtDCΔ). Data pooled from 3 +independent experiments.
+d. The number of liver monocytes and neutrophils (mean + SD) were +evaluated by flow cytometry 2- and 5-days post-M-CoV infection as in (a). Day 2: +n=14 for control, n=4 for pDCΔ and n=6 mice for +pDCΔtDCΔ. Day 5: n=10 for control, n=5 +mice for pDCΔ and n= 6 mice for +pDCΔtDCΔ. N represents mice. Data pooled +from 3 or more exp.
+e. Liver viral titers (log10 pfu / gr. tissue) were +determined 5-days post-M-CoV infection (n=10 mice for control, n=7 mice/group +for pDCΔ and pDCΔtDCΔ). +Violin plots show data pooled from 4 experiments.
+f. Serum levels of IL-1β were determined 2-days post-M-CoV +infection by Luminex (n=7 mice for control, n=8 mice/group for +pDCΔ and pDCΔtDCΔ). +Data pooled from 4 experiments.
+g. pDCΔ mice were inoculated with anti-IL-1β +Ab 2-days post-M-CoV-infection, and the percent weight loss (mean + SD) was +measured over time. N=5 mice for pDCΔ + control Ab, and n=6 +mice for pDCΔ + α-IL1-β Ab. Data pooled from 3 +experiments.
+h. As in (g), but serum ALT levels (mean + SD) were evaluated at 5-days +post-infection. N=4 mice for control, n=7 mice for pDCΔ + +control Ab, and n=6 mice for pDCΔ + α-IL1-β Ab. +Data pooled from 3 experiments.
+i. As in (g), but numbers of liver monocytes and neutrophils (mean + +SD) were quantified by flow cytometry at 5-days post-M-CoV infection. N=4 mice +for control, n=5 mice for pDCΔ + control Ab, and n=6 mice for +pDCΔ + α-IL1-β Ab. Data pooled from 3 +independent experiments.
+j. As in (g), but liver viral titers (log10 pfu / gr. +tissue) were evaluated 5-days post-M-CoV infection. N=4 mice for control, n=7 +mice for pDCΔ + control Ab, and n=6 mice for +pDCΔ + α-IL1-β Ab. Violin plots show data +pooled from 3 experiments.
+k. DC subsets gated as in Fig.1f +and Fig.4g were sorted and incubated with +M-CoV or mock (supernatant from non-infected L929 cells) using an MOI of 1. +Culture supernatant was analyzed by CBA 14-16 hrs later (n=5 samples in 3 +experiments). Minimum and maximum values are indicated via whiskers, while the +interquartile range and median are marked with the box.
+l. Human blood enriched DCs were left unstimulated (control) or +stimulated with CpG-A for 6-9 hrs, and then analyzed for intracellular +IL-1β by flow cytometry (circles) or CyTOF (square). N=5 donors for tDC +and pDC and n=3 donors for DC1 and DC2. Data pooled from 4 experiments. Minimum +and maximum values are indicated via whiskers, while the interquartile range and +median are marked with the box.
+m. As in (l), but cells were stimulated with CpG-A or influenza virus, +and analyzed by CyTOF in one experiment.
+n. As in (l), but pDC and tDC were sorted and incubated with MRC5 +fibroblast cell line alone or MRC5 infected with CMV (CMV-MRC5) for 6 hrs, and +analyzed for intracellular IL-1β by flow cytometry in one experiment.
+Statistical differences were determined by Two-way ANOVA with +Tukey’s multiple comparisons test (a-c,g), by One-way ANOVA with +Tukey’s multiple comparisons test (d-f,h-l) or Mantel-Cox test for +survival curves (b).
Weight change in infected mice was used as a measure of infection +severity. Control mice developed asymptomatic infection with almost no weight +loss, whereas IFN-I receptor knockout mice (IFNR1KO) displayed +critical pathology and died by 2-days post-infection (p.i.), as +described27(Fig.8b). By 3-days p.i., +pDCΔ and pDCΔtDCΔ mice +started to lose weight. By 5-days p.i., pDCΔ mice were +moribund experiencing severe pathology and >20% weight loss. In contrast, +pDCΔtDCΔ mice displayed milder pathology +and complete survival. Severe pathology in pDCΔ mice +correlated with exacerbated liver damage measured by increased serum alanine +transaminase (ALT), pro-inflammatory monocyte and neutrophil infiltration (Fig.8c-d). pDCΔ mice also had a late spike in viral titers +(5-days p.i.), likely consequence of excessive liver damage (Fig.8e and Extended +Data Fig.8f). On the other hand, milder pathology in +pDCΔtDCΔ mice correlated with less liver +damage and pro-inflammatory cells. To confirm that eliminating both pDC and tDC +resulted in milder pathology to M-CoV, we also infected TCF4cKO mice, +which harbor significantly less numbers of liver pDC/tDC without affecting cDC +(Extended Data Fig.9a). Similar to +pDCΔtDCΔ mice, TCF4cKO mice +had mild pathology, limited liver damage and modest increase in viral titers, as +previous observed28(Extended Data Fig.9b-d). These results suggest that severe M-CoV pathology +is not mediated by the sole elimination of pDC, which occurs in both mouse +models, but rather by tDC presence in pDC-depleted mice.
+To investigate the underlying tDC-mediated mechanism driving severe +pathology, we analyzed cytokines/chemokines in serum 2-days p.i. (Extended Data Fig.9e). Severe pathology in +pDCΔ mice correlated with a significant increase in +pro-inflammatory cytokines and chemokines, including IL-1β, IL-18, TNF, +IL-6, which was not observed in pDCΔtDCΔ +mice. Acutely increased serum levels of IL-12p70 and IFNγ in +pDCΔ mice suggest augmented innate lymphocyte numbers or +function. However, natural killer (NK) and innate lymphocyte 1 (ILC1) numbers +were not affected, and their depletion did not ameliorate M-CoV infection +severity in pDCΔ mice, indicating that these cells do not +mediate pathology (Extended Data Fig.8d +and 9f-g). Among the other pro-inflammatory cytokines upregulated in +pDCΔ mice, IL-1β could mediate higher +chemokine/cytokine levels resulting in increased liver damage39,40. Indeed, IL-1β increased 2-5-fold in the serum +of pDCΔ vs control and +pDCΔtDCΔ mice, as early as 2-days p.i. +(Fig.8f). A single dose of +anti-IL-1β blocking antibody in pDCΔ mice significantly +reduced pathology, i.e., lower weight loss, ALT levels, liver-infiltrating +leukocytes and viral titers (Fig.8g-j and Extended +Data Fig.9h).
+So far, our data suggest that tDC mediate severe pathology in +pDCΔ mice through a dysregulated IL-1β production, +which is consistent with an increase in Il1b and related genes +following TLR stimulation (Fig.7c). We +tested tDC capacity to secrete IL-1β in response to M-CoV. Purified tDC +secreted IL-1β at higher levels than either pDC, DC1, EYFP+ +and EYFP− DC2 (Fig.8k), +whereas pDC secreted TNF, IL-6 and some IL-12p70 (Extended Data Fig.9i). We extended these observations to human tDC +stimulated with CpG (Fig.8l-n). Again, human tDC rapidly produced IL-1β, +which was more prominent than either pDC or cDC. Moreover, human tDC also +produced IL-1β in response to influenza and cytomegalovirus +(CMV)-infected lung fibroblasts (MRC5), whereas pDC produced IFN-α (Fig.8m-n, and41). Thus, tDC +have a distinct capacity to rapidly produce IL-1β in response to viruses +and their mimics, a feature that is conserved in mouse and human.
+In summary, our data supports a model in which dysregulated secretion of +IL-1β by tDC is associated with severe coronavirus pathology.
DISCUSSION
+In this study, we unravel tDC identity, development and function. Our data +show that tDC are not pre-DC2 as suggested, but rather a DC population that develops +from BM pro-pDC. tDC have functions of cDC, and a distinct capacity to rapidly +produce IL-1β in response to viruses, which promotes fulminant liver +pathology during M-CoV infection. Finally, tDC distinctly convert into mouse +ESAM+ DC2 or human CD5+ DC2. Our data indicate that +current definitions of the DC compartment should be expanded to encompass tDC, a +pDC-related DC subset that has a distinct role during viral infections and +contributes to the developmental heterogeneity of DC2 in mice and humans.
+It has been suggested that tDClo and tDChi may have +different developmental origins, i.e., lymphoid vs myeloid, respectively. Here, we +show ample evidence that this is not the case. Both tDClo and +tDChi share a core transcriptome that differentiate them from other +subsets, as also observed by the group of M. Dalod24. These subpopulations exhibit developmental +features shared with pDC, such as IgH rearrangements and PTCRA expression4, and the need of TCF4 for their +differentiation. Both tDClo and tDChi labeled similarly to pDC +when using two different tracing mouse models (hCD2EYFP and +CD300cTdT), and emerged from pro-pDC at steady state. Further +experiments are needed to evaluate if pro-pDC are bi-potent progenitors giving rise +to tDC and pDC. Finally, tDClo gives rise to tDChi after +adoptive transfer experiments. Thus, tDClo and tDChi are +different states within the same population of cells. As tDClo transition +to tDChi, they upregulate several DC2 TFs including ZBTB464. Thus, ZBTB46 expression is one of +the many transitional features of this population and should not be interpreted as +an origin marker from a pro-cDC progenitor25. In fact, ZBTB46 regulates cDC activation rather than +development42.
+Adoptive transfer experiments revealed that tDC convert exclusively into +ESAM+ DC2, passing through a CD11b− DC2 stage. Two +fate tracing models suggest that 30-40% of ESAM+ DC2 are tDC-derived. +This was confirmed by the presence of IgH rearrangements. Finally, tDC-derived +ESAM+ DC2, but not CX3CR1+ DC2, are depleted by the +administration of DT into hCD2EYFP CX3CR1DTR. During the +preparation of this manuscript, a report by Rodriguez et +al.26 implied +that pDC-like cells, which encompass a portion of tDClo, convert into +CX3CR1+ DC2 following adoptive transfer into irradiated recipient +mice. The discrepancies between our observations and Rodriguez et +al. may arise from different gating strategies to define DC2, as well +as the cell transfer conditions. Rodriguez et al. used only two +markers to define DC2, MHCII+ and CD172a+. However, this +strategy likely results in contamination of DC2 with +ESAM−CX3CR1+ tDChi. Further, different +from Rodriguez et al., our cell transfer experiments were performed +in non-irradiated congenic recipients to avoid unphysiological proliferation of DC +following radiation. Finally, our data unraveling tDC conversion into mouse +ESAM+ DC2 parallels human tDC conversion into CD5+ DC2, +but not CD14+ DC3. Notably, human DC3 are known to develop via a +GM-CSF-dependent monocyte/dendritic cell progenitor (MDP)20,21. +Although the origin of mouse DC3 requires further investigation, we show here that +tDC arise from the FLT3L-dependent pro-pDC pathway, and not MDP.
+We show that tDC are distinct from pre-DC2. tDC trace back to pro-pDC and, +consequently, share developmental features with pDC. These features are only +observed in tDC and tDC-derived DC2, but not pre-DC2 derived DC2. Second, tDC have +phenotypic differences with pre-DC2 . Despite that tDC and pre-DC2 give rise to +ESAM+ DC2, tDC do so passing through a CD11b− DC2 +stage. Different from pre-DC2, tDC have lower turnover and proliferation rate. +Unlike pre-DC2, very few tDC reside in the BM suggesting that these cells exit the +organ quickly post-development. Finally, mouse tDC exhibit functions of cDC +including widespread transcriptional changes following TLR stimulation, antigen +capture, and activation of antigen-specific T cells; functions conserved between +species4-6. Importantly, tDC require different TF than +pre-DC2. TCF4 is required for tDC development from pro-pDC, whereas IRF4 is required +for tDChi → CD11b− DC2 → ESAM+ +DC2 conversion; however, IRF4 does not seem to be necessary for pre-DC2 → +ESAM+ DC2 conversion. Notably, IRF4cKO mice have reduced +CD24+ DC2 in several tissues13,15, consistent with +our observation that CD24 expression is slightly higher in splenic tDC-derived +ESAM+ DC2. Further studies are needed to evaluate the contribution of +tDC and pre-DC2 to the DC2 compartment in different tissues. Nevertheless, the +unique characteristics of tDC warrant their distinction from pre-DC2.
+Interestingly, in situations where tDC development is impaired at the level +of pro-pDC (e.g., TCF4cKO mice), DC2 are present at normal numbers +suggesting that they are maintained by pre-DC2. Similarly, in situations where +pre-DC2 development is impaired (e.g., Clec9aCre +Rosa26DTA mice), DC2 are present at normal numbers but +harbor IgH rearrangements43, +suggesting they may be derived from tDC. Thus, the ESAM+ DC2 compartment +seems to be maintained by two distinct developmental pathways that converge into +cells that are phenotypically similar (Extended Data +Fig.10). The contribution of these two distinct developmental pathways to +the DC2 pool may be signaled by environmental cues (e.g., age and IFN-I +availability)44,45. To dissect the in vivo +functions of these two developmental pathways, it would be helpful if DC2 origin is +reflected in their name, e.g., tDC2 for tDC-derived DC2 and cDC2 for pre-DC2-derived +DC2. Importantly, it is imperative to identify surface markers to distinguish these +two developmental origins of DC2, and CD45RA and CD45RB described here may assist +this goal.
+Mouse and human tDC distinctively produce high levels of IL-1β in +response to viruses, which likely plays a protective role during immune responses. +However, the dysregulated IL-1β secretion by tDC culminated in excessive +immunopathology during M-CoV infection. This excessive immunopathology recapitulates +several features of SARS-CoV-2 including higher viral titers, neutrophil and +monocyte recruitment, and excessive tissue damage46,47. +Notably, similar to our mouse depletion model, pDC from severe SARS-CoV-2 patients +are dysfunctional and unable to produce IFN-I48,49, suggesting that +the dysregulated tDC-mediated IL-1β pathway described here may be active also +in humans. Accordingly, anti-IL-1β Abs and IL-1β/α inhibitors +have shown benefit in SARS-CoV-2 patients50. Altogether, our work unveils an innate immune balance +between efficient antiviral vs pro-inflammatory response that is controlled by +developmentally related cells, pDC and tDC, suggesting that these two DC populations +form a distinct lineage of cells with key complementary functions during viral +infections.
METHODS
+Human specimens and PBMCs
+Donors provided informed consent in accordance with IRB protocols +approved by the Stanford University Administrative Panel on Human Subjects in +Medical Research. De-identified blood and buffy coats were obtained from healthy +adult human donors (20-45 years old) following the guidelines of the Research +and Laboratory Environmental Health and Safety program of Stanford University or +from the New York Blood Center (NYBC). Peripheral blood mononuclear cells +(PBMCs) were isolated by density gradient centrifugation using Ficoll-Paque PLUS +(GE Healthcare).
Mice
+Mice (reporting summary) were purchased from The Jackson Laboratory or +Charles River. Cx3cr1EGFP+/− were crossed +with CD45.1 mice in-house. +Itgax-Cre+/− mice +were crossed to Irf4fl/fl in-house and screened +routinely for the appearance of Irf4 germline +deletions51. Side by +side comparison showed that Itgax-Cre+/− x +Irf4fl/fl and +Itgax-Cre+/− x +Irf4fl/deletion mice have similar phenotypes, so +mice were pooled for experiments and labeled as IRF4cKO. +IRF4control group includes +Itgax-Cre−/− x +Irf4fl/fl and +Itgax-Cre−/− x +Irf4fl/deletion mice. The +Cd300cCre strain was generated +by knocking Cre recombinase ORF and BGH polyA into the first coding exon of +Cd300c. This strain description will be provided in a +separate manuscript (Perez et al., in preparation) and is available upon request +to B.R. For this study, Cd300cCre+/+ x +Rosa26TdTomato+/− (CD300cTdT) +mice were used. Mixed bone marrow chimeras (BMC) were generated by lethally +irradiating (two doses of 6Gy) 8-week-old females and transplanting with +3x106 BM cell suspensions (50% WT CD45.1 and 50% cKO CD45.2). BM +cells were engrafted for ≥8 weeks. For DT inoculation, 50ng/g body weight +was inoculated i.v. at day 0, followed by 25ng/g every other day. For BrDU +experiments, 1 mg of BrDU (SIGMA) was inoculated i.p. at day 0, followed by +continuous feeding with 0.8 mg/mL of BrDU in drinking water (prepared fresh +daily). After 14 days, BrDU was removed and mice euthanized at different time +points. For in vivo antigen uptake experiments, +1x108 sheep red blood cells (SRBCs) stained with 2μM PKH26 +(Sigma-Aldrich) or 1x109 Yellow-Green Polystyrene Beads +(Polysciences, Inc.) were injected i.v. for 3 hrs. For viral infection, 50 PFU +M-CoV was injected i.p. In some experiments, M-CoV-infected mice were inoculated +with 100 μg of anti-IL-1β or control Ab (BioXCell, clone B122 and +Cat#BE0091, respectively) on day 2, or 200 μg of anti-NK1.1 or control Ab +(BioXCell, clone PK136 and 2A3, respectively) on day −1, 0 and 2 +post-infection. Animals were maintained in ventilated cages under specific +pathogen-free conditions and a 12 hr dark:12 hr light cycle at 20-22°C +and a humidity range of 30-70%. Animals were fed with Teklad Irradiated Rodent +Diet (Inotiv #2918). Animals were used at 6-12 weeks of age in accordance with +Stanford and New York University Administrative Panel on Laboratory Animal Care +and overseen by the Institutional Animal Care and Use Committee.
Cell suspension preparation
+Blood was obtained from anesthetized mice via retro-orbital puncture +into a vial containing 1mM EDTA (Corning), and mononuclear cells were enriched +using Lympholyte M (Cedarlane). Blood serum was used for measuring alanine +2-oxoglutarate-aminotransferase (ALT)(Sigma-Aldrich), or cytokines by Luminex +(ThermoFisher; HIMC Stanford University). Spleens, peripheral LN (inguinal, +brachial, and axillary) and perfused livers were digested with 400 U/mL +Collagenase D (Millipore-Sigma) and 50 μg/mL DNase I (Millipore-Sigma) +for 30 min at 37°C. 10 mM EDTA was added for the last 5 min of culture. +For liver, leukocytes were enriched by density gradient centrifugation using 42% +Percoll (Cytiva). Perfused lungs were digested with 0.13 U/mL Liberase TM +(Millipore-Sigma) and 50 μg/mL DNase I for 25 minutes at 37°C. For +BM, femurs and tibias were flushed using un-supplemented RPMI (Corning). All +cell suspensions were filtered through a 70 μm strainer. Red blood cells +from spleen and BM were lysed using ACK Lysis buffer (Lonza).
Enrichment of DC and precursors
+Mouse cell suspensions were incubated with anti-CD135-biotin (Biolegend) +for 30 min at 4°C, washed and incubated with Ultra-Pure anti-Biotin +microbeads (Miltenyi) for 30 min at 4°C. CD135+ cells were +selected using LS columns (Miltenyi). Human myeloid cells were negatively +enriched from PBMCs using anti-CD3 (OKT3), anti-CD14 (HCD14), anti-CD19 (HIB19) +and anti-CD335 (9E2) Abs followed by anti-mouse Dynabeads (Invitrogen) at a +concentration of 4 beads per target cell52. For IL-1β experiments, human DC were negatively +enriched using EasySep Human pan-DC Pre Enrichment Kit (StemCell +Technologies).
Staining for flow cytometry and cell sorting
+Abs (reporting summary) were purchased from Biolegend, R&D, MBL +International and ThermoFisher. Anti-TCF4 (Abcam) was labeled in-house using the +Alexa 647 Labeling Kit (ThermoFisher). Mouse cell suspensions were incubated +with anti-CD16/CD32 (clone 2.4G2, produced in-house) to block non-specific +binding for 15 min at 4°C. Cells were incubated with Ab cocktails in +mouse FACS buffer (2 mM EDTA, 2% FBS in PBS) for 20 min at 4°C. DAPI +(Sigma) was added to exclude dying cells before sample acquisition. For +transcription factor staining, cells were stained with surface Abs and LIVE/DEAD +Fixable Blue (ThermoFisher) in PBS for 20 min at 4°C, then fixed with +FoxP3 Transcription Factor Fix/Perm Buffer (ThermoFisher) for 2-12 hrs and +stained intracellularly for 30 min in 1X Permeabilization Buffer (ThermoFisher). +For cytokines and EYFP co-staining, cells were surface stained and fixed with 2% +paraformaldehyde (Electron Microscopy Sciences) for 40 min, followed by +Cytofix/Cytoperm (BD Biosciences) for 12 minutes. Intracellular cytokine +staining was performed in 1X Permwash (BD Biosciences) for 30 min. For BrDU +staining, cells were surface stained, washed and fixed with Cytofix/Cytoperm for +30 min at 4°C, followed by incubation with Cytoperm Permeabilization +Buffer Plus (BD Biosciences) for 20 min at 4°C and a second fixation with +Cytofix/Cytoperm for 10 minutes at 4°C. Cells were then incubated with +160 U/mL DNase I for 1 hr at 37°C, followed by incubation with anti-BrDU +PE (Biolegend) for 20 min at RT. For mouse cell sorting, lineage included Abs +against CD3, CD19, NK1.1 and Ly6G for splenocytes plus MHCII for BM (all in +Alexa700 or Efluor450), and streptavidin-Alexa647 to detect CD135 expression +from CD135-enriched cells. Human PBMCs were incubated with human gamma-globulin +(ThermoFisher) to block non-specific binding for 15 min at 4°C, followed +by Abs in human FACS buffer (2 mM EDTA, 2% Donor equine serum in PBS) for 20 min +at RT. For transcription factor staining, cells were stained with surface Abs +and LIVE/DEAD Fixable Blue in PBS for 20 min at RT, fixed with FoxP3 +Transcription Factor Fix/Perm Buffer for 1 hr at 4°C, and stained +intracellularly for 20 min in 1X Permeabilization Buffer. Human tDC were sorted +as lineage negative +(CD3−CD19−CD20−CD335−CD66b−), +monocyte/DC3 negative (CD14−CD16−) and +HLA-DR+CD123+BDCA1−AXL+ +(using CD123 clone 6H6 to allow for IL-3 binding in culture). For IL-1β +human experiments, pDC and tDC were sorted as +HLA-DR+CD123+AXL−SIGLEC6− +and HLA-DR+CD123+AXL+SIGLEC6+, +respectively. In all cases, analysis was performed in a 5-laser LSRFortessa X-20 +(BD Biosciences), and cell sorting in a FACSAria Fusion using BD Diva software +(BD Biosciences), and SY3200 or SH800Z sorter and software (SONY Biotechnology). +Data was analyzed using FlowJo (v.10.8.1; Tree Star, Inc). Unstained cells and +single-fluorochrome-stained cells were used for accurate compensation. Control +samples included fluorescence minus one.
Staining for CyTOF
+Metal-labeled Abs (reporting summary) were obtained from Fluidigm or +labeled using the MaxPar X8 labeling kit (Fluidigm). CD135-enriched mouse cell +suspensions were stained with 0.25 μM cisplatin (Fluidigm) for 5 min at +RT to exclude dead cells, washed with CyFACS buffer (2 mM EDTA, 1% BSA, 1% in +PBS) and stained with a heavy-metal-labeled Abs for 30 min at 4°C. Cells +were washed twice with CyFACS then fixed with Foxp3 Transcription Factor +Fix/Perm Buffer for 2 hrs followed by intracellular staining in 1X Permwash for +30 min. After staining, samples were washed and incubated with 2% PFA in PBS +containing 125 nM Iridium intercalator (Fluidigm) overnight. Cells were washed +with water, filtered, and acquired. Human PBMC and sorted tDC were cisplatin and +surface stained, followed by fixation as described above for mouse. The +intracellular cocktail included Abs affected by fixation52. Sorted tDC were pooled with mouse +splenocytes to avoid cell loss during washes52. In all cases, cells were acquired in a CyTOF2 or a +Helios mass cytometer (Fluidigm) at the Shared FACS Facility at Stanford or Mt. +Sinai, using CyTOF software.
Adoptive transfer experiments
+10,000-30,000 cells sorted from 1-2 mice were adoptively transferred +into WT congenic non-irradiated mice through tail vein injection. Analysis was +performed 2-8 days later in the spleen. During analysis, splenocytes of +recipient mice were enriched by density gradient centrifugation using 30% Bovine +Serum Albumin (BSA, Millipore-Sigma)4.
+In vitro DC stimulation
+8x106 mouse splenocytes were resuspended in 0.5mL of complete +RPMI (complete medium contains: 10% FBS, GIBCO; 2 mM L-glutamine, Corning; 100 +IU Penicillin, Corning; 100 mg/mL Streptomycin, Corning; 55 mM +2-Mercaptoethanol, GIBCO), and cultured for 6h at 37°C. Splenocytes were +left untreated or stimulated with 200ng/mL LPS, 200ng/mL MPLA, 25μg/mL +Poly(I:C), or 2.5μg/mL Resiquimod. For cytokine detection, 5 μg/mL +Brefeldin A (Millipore-Sigma) was added to the culture. Costimulatory marker and +cytokines were detected by surface and intracellular staining, respectively. +2x105/mL purified splenic DC were incubated with M-CoV (MHV A59; +MOI of 1) or mock control (supernatant from non-infected L929 cells) and +cultured for 14-16 hrs at 37°C. IL-1β, TNF, IL-6 and IL-12p70 were +measured by Cytometric Bead Array (CBA) Mouse Enhanced Sensitivity Flex Set and +Mouse Enhanced Sensitivity Master Buffer Kit (BD Biosciences). Concentration of +each cytokine was normalized to the mock control. Sorted human tDC were cultured +in 96 well U-bottom plates at 37°C at a concentration of 5,000 cells in +200 μL complete RPMI. Media was supplemented with 10ng/mL IL-3 (R&D +Systems) and 200ng/mL CD40L (R&D Systems) for 48 hrs. tDC phenotype was +evaluated by CyTOF and flow cytometry. For IL-1β detection, enriched or +sorted human DC were stimulated with 2.5 μM CpG-A (ODN 2216), MRC5, +CMV-MRC5 and H1N1 influenza virus A/PR/8/34 (PR8) at 10 MOI for 6 to 9 +hrs41. 10 μg/ml +of Brefeldin A (BFA) was added for the last 3-6 hours of incubation prior to +flow cytometry and CyTOF experiments41.
Bone marrow-derived DC
+2x106 BM cells were plated in a 24-well plate with 2mL +complete DMEM supplemented with 1.8mM sodium pyruvate (SIGMA), 1X MEM-NEAA +(GIBCO #11140050), and 10% supernatant from B16-FLT3L cell line. In some cases, +1 mL of cell suspension was transferred at 3-days to a 24-well plate containing +a mono-layer of mitomycin-treated OP9-DL1 cells35. At 4-7 days, cell suspensions were +analyzed by flow cytometry. Lineage (CD3+ and CD19+) and +macrophages (SSC-Ahigh F4/80high cells) were gated +out.
IgH D-J Rearrangement
+Genomic DNA was isolated from sorted DC using the Nucleospin Tissue XS +kit (Takara Bio). The amount of DNA per reaction was equalized by qPCR using +primers against Actb4. The primers used were as follows: Actb +forward 5’-GGTGTCATGGTAGGTATGGGT-3’, reverse +5’-CGCACAATCTCACGTTCAG-3’; Mu0-JH1 (germline) forward +5’-CCGCATGCCAAGGCTAGCCTGAAAGATTACC-3’, reverse +5’-GGTCCCTGCGCCCCAGACA-3’; DHL-JH3 forward +5’-TGGCAGGGATTTTTGTCAAGGGATCTACTACTGTG-3’, reverse +5’-CTAATTCTCACAAGAGTCCGATAGACCCTGG-3’. Products were run on a 2% +agarose gel and imaged with a ChemiDoc MP (Bio-Rad).
+In vitro T cell activation
+Spleen, skin draining LN, mesenteric LN, and iliac LN of OT-II mice were +collected and mechanically dissociated. Cells were lysed using ACK Lysis buffer +(Lonza), filtered through a 70 μm strainer, and incubated with Abs +against B220, F4/80, MHCII, NK1.1 and CD8 (produced in-house) for 30 min at +4°C followed by sheep anti-Rat IgG Dynabeads (Invitrogen). Negatively +enriched CD4+ T cells were labeled with 5 μM CellTrace Violet +(CTV, ThermoFisher), and naïve T cells sorted as +CD3+CD4+CD44loCD45RB+. +Naïve T cells were cultured with sorted splenic DC (1:5 ratio) and 20 +μg/mL ovalbumin (OVA) for 4 days. Controls included T cells incubated +with OVA in the absence of DC.
Viral stocks and titers
+Influenza virus H1N1 (PR8), CMV stocks and infected cell lines were +previously generated41. Thawed +MRC5 and CMV-MRC5 were plated at the density of 6x104 +cells/cm2 for > 8 hrs before the addition of +5x103 - 4x104 human DC. M-CoV (also known as murine +hepatitis virus, MHV A59; BEI Resources) was propagated on L929 cells (ATCC). +M-CoV viral titers were determined after homogenization by standard plaque assay +using L929 cells.
Transcriptomic data generation
+For bulk-RNAseq, splenic DC from 2 mice were CD135-enriched and split in +two samples: 30% was stained to sorting, and 70% was stimulated with the LPRC +adjuvant cocktail [100ng/mL LPS, 25μg/mL Poly(I:C), 2.5μg/mL +Resiquimod and 6μg/mL CpG-A (ODN 2216)] at a concentration of +107/mL for 3 hrs before staining and sorting. RNA was extracted +using the NucleoSpin RNA XS kit (Takara Bio). For scRNAseq, CD135-enriched cells +were stained and sorted. Splenocytes were mixed 1:1 from 2 sorted gates to +enrich for tDC: +CD3−CD19−NK1.1−Ly6G− +and +CD3−CD19−NK1.1−Ly6G−XCR1−CD11blo +(Fig.1c). BM cells were sorted as +CD3−CD19−NK1.1−Ly6G−. +600 sorted cells per μL were processed for 10X Genomics. For NanoString, +sorted DC were resuspended in 1/3 RNeasy Lysis Buffer RLT (Qiagen) at a +concentration of 1,000-5,000 cells/μL and analyzed on the NanoString +nCounter® Mouse Myeloid Innate Immunity V2 Standard +Platforms, following the manufacturer's instructions. Samples were +processed on the NanoString Digital Analyzer to yield a reporter code count +(RCC) dataset, which was analyzed via +ROSALIND® (https://rosalind.bio).
CyTOF data analysis
+Files in FCS format were normalized with the R package Premessa. Mouse +splenocytes were gated as +live/singlets/CD3−/CD19−/CD335−/Ly6G−, +exported, and analyzed by UMAP and X-shift in FlowJo. Analyzed files were +imported into MATLAB (v.9) for visualization. MEM was performed following R +scripts53. The +Scaffold R package was downloaded from GitHub (https://github.com/nolanlab/scaffold)54. Human DC were gated from PBMC in FlowJo +and imported into Scaffold to create a reference map. Events from stimulated tDC +were imported into the reference map for clustering29.
Bulk-RNAseq analysis
+The count matrix was generated using the “Seq-N-Slide” +pipeline (https://github.com/igordot/sns). Data was aligned using STAR +(v2.7.3a) to the mm10 reference genome. featureCounts was used to generate the +gene-count matrix. The count matrix was analyzed using DESeq2 (v3.14)55. Differential gene expression +analysis was performed using DESeq2’s DESeq function. Log fold change +shrinkage was performed by the “ashr” method56 and used for volcano plots and MA plots +(DESeq2)55. Variance +stabilizing transformation was performed using DESeq2 for heatmap and PCA +visualization.
scRNAseq analysis
+Cell Ranger (10x Genomics) was used for demultiplexing and generation of +barcodes and count matrices. Spleen and BM data were analyzed using Seurat +(v3)57 in R (4.1.2). +Cells were selected for number of genes over 300, number of total counts less +than 50000, and % of mitochondrial genes less than 20%. Data was normalized +using the “LogNormalize” method (scale factor 10000) of +Seurat’s NormalizeData function. PCA was performed in Seurat and the top +25 dimensions were used for UMAP. Differentiating markers for each cluster were +identified using the Wilcoxon rank sum test implemented in Seurat’s +FindAllMarkers function (min.pct=0.25, logfc.threshold =0.25). KNetL graphs were +generated using iCellR (v1.6.5)58. Cells were filtered for <20% mitochondrial genes, +unique genes expressed between 300-8000 genes, and UMIs less than 50,000. +Monocytes were filtered using cell IDs of cells in Cluster 9 from the spleen +UMAP analysis in Seurat. Data was normalized using the default +“ranked.glsf” method in iCellR. PCA was calculated using the top +2000 genes. KNetL analysis was performed using 1:20 dimensions with a zoom of +175 and UMAP as the dimensionality reduction method. Clusters were calculated +using the KNetL results with a sensitivity of 150. Differential gene expression +was calculated for each cluster using the findMarkers function of iCellR with +default parameters except with adjusted p-value <0.05 set as the +threshold of significance.
GSEA analysis
+For bulk-RNAseq, a gene list was created by ranking all genes post log +fold change shrinkage by the log2FC value. The mouse hallmark gene +set was downloaded (MSigDB website)59,60 and used as +the input for gene sets database. GSEAPreranked analysis was run on GSEA +(v4.2.2) with 1000 permutations. For scRNAseq, a ranked gene list was created by +computing markers for each cluster using Seurat’s FindMarkers with +min.pct and logfc.threshold parameters set to 0 to extract all genes. Genes were +then ranked on the logFC value and analyzed using the same parameters as +above.
Cell type signature generation
+CPM counts from the sorted populations of the bulk-RNAseq were used to +generate a signature matrix using CIBERSORTx with default parameters61. The signature matrix was then +used to calculate signature scores for each cell of the spleen scRNA sequencing +using the imputing cell fraction function of CIBERSORTx (permutations = 500, +B-mode batch correction). Raw counts for tDC, pDC, DC1, and DC2 from the spleen +scRNAseq were used to generate a signature matrix using CIBERSORTx with default +parameters. The signature matrix was then used to calculate signature scores for +each cell of the BM scRNAseq by imputing cell fraction (CIBERSORTx) +(permutations = 500). pDC-signature “high” cells were defined as +signature score >0.15.
Trajectory analysis
+BM cells with pDC signature >0.15 were selected and re-clustered +using Seurat. UMAP was generated using the top 25 statistically significant PCA +components. Trajectory analysis was performed on UMAP embeddings from Seurat +clustering using Slingshot (v2.2.0)62. For RNA velocity analysis, Velocyto +(0.17.17)63 was used +to generate the loom file containing spliced and un-spliced transcripts. The +mask file, mm10_rmsk, was downloaded from the UCSC genome browser. The genome +annotation file, mm10_2020_A, was downloaded from 10x Genomics. The data was +then loaded into scVelo (v.0.2.4)63,64 for further +analysis using Python (v3.10). scVelo was run with default settings using the +“filter_and_normalize” and “moments” functions to +filter and compute PCA/nearest neighbors, respectively. Results were overlayed +onto Seurat-clustered UMAP of the BM.
Statistical analysis
+For Fig.1-7 experiments, no statistical methods were used to +predetermine sample sizes. For Fig. 8, the +number of mice was estimated using power analysis (G*Power 3). Data distribution +was assumed to be normal, but this was not formally tested. Individual +experiments used mice of the same sex and age, without randomization. Data +collection and analysis were not performed blinded. No data points were +excluded. For scRNAseq, pre-established criteria for single-cell exclusion, +i.e., low number of unique genes, abnormally high read count, and high +mitochondrial gene content, was used. Statistical details of experiments, +including statistical tests used, n value, number of experiments, and the type +of statistical tests can be found in figure legends. Statistical analysis was +performed with at least three biological replicates using GraphPad Prism 9 +(GraphPad Software) or R (v.4.1.2) statistical programming language. All graphs +show mean +/− SD. gMFI indicates geometric mean fluorescence intensity. +Graphics created with BioRender.com.
Extended Data
+Extended Data Fig. 1. tDC are distinct transcriptionally.
+
a. CD135-enriched splenocytes from hCD2EYFP mice were +stained and sorted. Briefly, lineage containing +CD3+/CD19+/NK1.1+/Ly6G+ +cells were removed, and cells were gated using the strategy described in +Fig.1f. DC2 were further separated +as EYFP+ and EYFP− cells. Shown are cells +before (upper panels) and after sort (bottom panels).
+b. MA plots of bulk-RNAseq data comparing tDC vs DC2 and tDC vs pDC. +Genes with 2-fold Log2 change are shown in red (up) and blue +(down) (see also Supplementary Table S4).
+c. MA plot comparing tDC vs all other DC (pDC, DC2 and DC1). Genes +with 2-fold Log2 change are shown in red (up) and blue (down) +(see also Supplementary +Table S1).
+d. scRNAseq of splenic DC subsets. Splenic DCs were sorted as shown +in Fig.1c and sequenced using +droplet-based genomics. After filtering, 2,110 cells were analyzed. UMAP of +clusters detected using Seurat’s pipeline (right). Heatmap of top DEG +between clusters (left).
+e. DC signature scores generated using CIBERSORTx in the bulk-RNAseq +data from (a) was overlayed on UMAP from (d).
+f. As in (e), but DC signature scores were overlayed on the KNetL +Plot from Fig.1d.
Extended Data Fig. 2. tDC originate from BM progenitors.
+
a. Gating strategy used for the analysis of skin-draining lymph node +(LN) and lung DC subsets excluded. Lineage contains +CD3+/CD19+/NK1.1+/Ly6G+ +cells.
+b. Bar graphs (mean + SD) showing the contribution of donor cells to +LN and lung DC populations (gated as in (a)) in mixed BMC generated by +transplanting 50% CD45.1 WT and 50% CD45.2 TCF4cKO BM, normalized +to NK cells (n=6 mice in 2 experiments). Statistics determined by two-way +ANOVA with Dunnett’s multiple comparison test.
+c. Bar graph (mean + SD) showing the percentage of splenic DC (gated +as in Fig.1f) labeled with the lineage +tracing CD300cTdT (n=4 mice in 3 experiments).
+d. Adoptive cell transfer of 30,000 BM (left) or splenic (right) +CX3CR1EGFP CD45.1 pDC into CD45.2 WT non-irradiated congenic +mice. Recipient mice were euthanized at different time points (2-8-days) and +the number of recovered cells (gated on CD45.1+ and +CX3CR1+) calculated per 1000 cells transferred (mean + SD). +N=3/time point except n=2 for splenic pDC at 8-days. N represents number of +mice and experiments.
+e. CyTOF analysis of blood and spleen DC, analyzed by UMAP. Cells +were gated as +CD3−CD19−NK1.1−Ly6G−. +Populations were delineated manually (top) and colored by marker expression +(bottom)(n=2 mice in 2 experiments).
+f. Spleen, blood and BM of hCD2EYFP mice were analyzed +for the presence of EYFP+ cells, that were further separated in +Ly6D+ pDC and CX3CR1+ tDC. Lineage includes +CD3+/CD19+/NK1.1+/Ly6G+ +cells. One representative of 3 experiments.
Extended Data Fig. 3. tDC originate from pro-pDC.
+
a. CD135+ BM cells were enriched, sorted as +CD45+/CD3−/CD19−/Ly6G−, +and prepared for scRNA-seq. Cells with a pDC-specific gene enrichment score +of >0.15 were selected and re-clustered using Seurat as shown in +Fig.3d. Violin plots show the +expression of the indicated genes in the clusters defined in Fig.3d.
+b. As in (a) but GSEA of selected pathways from clusters defined in +Fig.3d. Bubble size indicates the +normalized enriched score (NES), and color scale depicts False Discovery +Rate (FDR).
+c. TdTomato expression in bone marrow progenitor cells from +CD300cTdT mice, gated as in Fig.3j. One representative experiment.
+d. Imm pDC, pro-pDC and pro-cDC from the BM of CX3CR1EGFP +or hCD2EYFP CD45.1 mice were purified as described in Fig.3j and transferred into CD45.2 WT +mice. 2- and 4-days post-transfer, recipient mice were analyzed for +transferred cells in the spleen. Shown is the number of cells recovered in +the spleen of recipient mice (mean + SD). Imm pDC: n=3 at 2- and 4-days. +Pro-pDC: n=3 at 2-days, and n=5 at 4-days. Pro-cDC: n=3 at 2-days, and n=5 +at 4-days. N represents number of mice and independent experiments.
+e. BM cells from CX3CR1EGFP or hCD2EYFP CD45.1 +mice were CD135-enriched and purified by cell sorting using the gating +strategy shown in Fig.3j. Purified +progenitors were adoptively transferred into congenic non-irradiated CD45.2 +WT mice and analyzed 4-days later in the spleen of recipient mice. Left +panels show post-sort purity of progenitors. Right panels show transferred +cells recovered in the spleen of recipient mice, analyzed using the gating +strategy described in Fig.1f. In both +cases, lineage includes +CD3+/CD19+/CD335+/Ly6G+ +cells. One representative of ≥ 3 experiments.
Extended Data Fig. 4. tDC originate from DN progenitors.
+
a. Gating strategy of CD135-enriched BM progenitors from +hCD2EYFP mice. Cells were stained for flow cytometry. Lineage +contain CD3+/CD19+/NK1.1+/Ly6G+ +cells. EYFP labeling is shown in the lower panels. One representative of 3 +experiments.
+b. As in (a), but CD115 progenitors were further gated based on the +expression Ly6C, whereas CD127 and DN progenitors were gated based on the +expression of Ly6D and SiglecH (top). Bottom histograms show EYFP expression +in each population. One representative of 3 experiments.
+c. The schematic shows adoptive transfer of 15,000-30,000 from +CX3CR1EGFP CD45.1 BM progenitors into WT CD45.2 +non-irradiated congenic mice. Transferred cells were identified in the +spleen of recipient mice as CD45.1+ and CX3CR1+ cells. +Bar graphs (mean + SD) show the percentage of each population recovered in +the spleen (gated as shown in Fig.1f). +N ≥ 2/mice and experiments.
+d. As in (c), but bar graphs (mean + SD) number of cells recovered +in the spleen of recipient mice. N represents number of mice and independent +experiments.
Extended Data Fig.5. tDC convert into DC2.
+
a. tDC were purified using the gating strategy described in Fig.1f. Top graph shows cell purity after +sort. Bottom graphs show the outcome of transferred cells in the spleen of +recipient mice analyzed at 4-days. One representative of 6 experiments.
+b. Bar graphs (mean + SD) of the total number of splenic DC in +hCD2-iCre+/− +Cx3cr1DTR+/− mice inoculated with DT and +analyzed 1 day after (n=5 mice in 4 experiments), or inoculated with DT +every second day and analyzed 7 days later (n=5 mice in 4 experiments). +“C” represents control mice (n=8 mice in 4 experiments), which +are a combination of hCD2-iCre−/− +Cx3cr1DTR+/− mice inoculated with DT or +hCD2-iCre+/− +Cx3cr1DTR+/− mice inoculated with PBS (no +differences were observed between these control mice). Statistics determined +by one-way ANOVA with Tukey’s multiple comparison test.
+c. As in (b), but BM progenitors from hCD2EYFP +CX3CR1DTR mice were analyzed after 7 days of DT inoculation +(n=4 mice). Control mice are hCD2EYFP CX3CR1DTR mice +inoculated with PBS (n= 3 mice). Bar graphs (mean + SD) of the frequency of +total progenitor populations from Fig.3j. Data pooled from 2 experiments. Statistical differences +were determined by unpaired two-tailed t-test.
+d. Mouse splenic DC subsets were sorted using the gating strategy +show in Fig.1f and Fig.4g, and analyzed by PCR assay for IgH D-J +rearrangements. Actin and IgH germline (GL) are also shown. One +representative of 3 experiments.
+e. DC2 from CD300cTdT mice were gated as indicated in +Fig.4f, and further separated in +TdTomato+ and TdTomato−. Heatmap indicating +the relative expression (Z-score of gMFI) of surface markers. Z-score was +calculated base of n=2 mice in 2 experiments.
Extended Data Fig. 6. tDC have lower turnover rate than pre-DC2 and required IRF4.
+
a. Mice were inoculated i.p. with 1 mg of BrdU on day 0 and then fed +continuously for 14 days with 0.8 mg/mL of BrdU in drinking water. After 14 +days, BrDU was removed. Mice were euthanized at different time points +post-BrdU removal, spleen cell suspension prepared and stained for BrdU and +the identification of DC subsets as described in Fig.1f and Fig.5e. N ≥ 2/mice per time point in 2 independent +experiments.
+b. Expression of Ki-67 in each splenic DC population gated as +described in Fig.1f and Fig.5e, analyzed by flow cytometry. One +representative of 3 experiments.
+c. BM cells from IRF4cKO or IRF4control mice +were cultured with FLT3L as described in Fig.2c. At day 6, tDC were sorted and re-cultured in complete +media and 1% FLT3L for 1 and 2 days. Bar graphs (mean + SD) show the +percentage of each DC recovered. N=3 biologically independent samples in 3 +experiment.
+d. Bar graphs (mean + SD) of the total number of splenic DC (gated a +Fig.1f and Fig.4f) in hCD2EYFP +CX3CR1DTR mice inoculated with DT every second day (n=3 mice) +or left untreated (n=3 mice), and analyzed at day 10. Statistics determined +by unpaired two-tailed t-test.
+e. As in (d), but shown is the number of other immune cells.
Extended Data Fig. 7. tDC respond to microbial stimulation and uptake antigen in +vivo.
+
a. Splenic DC subsets purified as described in Fig.1f and Fig.4f were analyzed by Nanostring. Heatmap shows the expression +of TLRs by each DC subset. N=2 mice in 2 experiments.
+b. Whole spleen cell suspensions were activated with LPS, MPLA, +Resiquimod or PolyIC for 6 hrs. Heatmap showing the upregulation of MHCII +and CD86 in pDC, tDC, ESAM+ EYFP+ and +EYFP− DC2, and DC1. Min and max correspond to the +lowest and highest gMFI per row, respectively. N=6 mice per group. Data +pooled from 4 experiments.
+c. As in (b), but bar graphs (mean + SD) of cytokine production. +Whole spleen cell suspensions were activated with LPS (n=6), MPLA (n=4), +Resiquimod (n=6) or PolyIC (n=6) for 6 hrs. TNF and IL-12p40 were detected +by intracellular cytokine staining. N represents mice and data was pooled +from 4 experiments. Statistical differences were determined by One-way ANOVA +with Tukey’s multiple comparisons test.
+d. Mice were inoculated with yellow-green polystyrene (YG-PS) beads +i.v. Three hrs later, spleens were harvested and the phagocytosis of the +particle was assessed by flow cytometry in each DC population, gated as in +Fig.1f (n=3 mice in 3 experiments). +Bars represent mean + SD. Statistical differences were determined by One-way +ANOVA with Tukey’s multiple comparisons test.
+e. As in (d), but mice were inoculated with PKH26-labeled SRBC (n=6 +mice in 4 experiments).
Extended Data Fig. 8. Number of immune cells, viral titer and ALT in pDC- and pDC/tDC- depleted +animals infected with M-CoV.
+
a. Liver cell suspensions were enriched in immune cells using a +Percoll gradient, and stained with a cocktail of Ab for the analysis of +neutrophils and lymphocytes. One representative of 3 experiments.
+b. As in (b), but cell suspensions were stained for the analysis of +DC and other myeloid cells. Lineage contains +CD3+/CD19+/NK1.1+/Ly6G+ +cells. One representative of 3 experiments.
+c. Liver of WT mice were analyzed for the number of DC subsets by +flow cytometry 2- and 5-days post-M-CoV infection. Fold change of the number +of each subset to day 0 is shown. N=3 at 0 days, n=14 at 2-days, and n=11 at +5-days. Data pooled from ≥ 4 experiments. p-values represent +statistical differences between tDC and DC2 vs DC1.
+d. Liver immune cell numbers 2-days post-M-CoV infection of +pDCΔ (n=3); +pDCΔtDCΔ (n=6) and control (n=5) +mice. pDCΔ mice are +BDCA2-DTR+/− mice inoculated with DT +one day before M-CoV infection. pDCΔtDCΔ +are hCD2EYFP CX3CR1DTR mice inoculated with DT every +other day for 5-days before M-CoV infection. Control mice are a combination +of BDCA2-DTR+/− or hCD2EYFP +CX3CR1DTR mice inoculated with PBS, and +DTR−/− mice inoculated with DT (no difference +in control mice was observed). Shown is mean + SD. N represents independent +mice in 2 experiments.
+e. As in (d), but serum ALT values (mean + SD) were determined 2-, +5- and 7-days post-M-CoV infection. N ≥ 4 mice/group/time point in 4 +experiments.
+f. As in (e), but viral titers (log10 pfu/gr. Tissue) in +liver (top) and spleen (bottom) were determined 2-, 5- and 7-days post-M-CoV +infection. N ≥ 4 mice/group/time point. Violin plots show +distribution of data pooled from 4 exp.
+Statistical differences determined by Two-way ANOVA with +Tukey’s multiple comparisons test (c,e,f), or One-way ANOVA with +Tukey’s multiple comparisons test per subset (d).
Extended Data Fig. 9. tDC promote immune pathology following M-CoV infection.
+
a. TCF4cKO vs TCF4control mice were generated +by transplanting Itgax-Cre+/− +Tcf4fl/fl or +Itgax-Cre−/− +Tcf4fl/fl BM into lethally irradiated CD45.1 WT mice, +respectively. 2 months post-reconstitution, mice were infected with M-CoV, +and the frequency of liver DC (mean + SD) analyzed 2-days later (n=6 +mice/group in 2 experiments).
+b. Percentage weight loss (left, mean + SD) and survival (right) of +TCF4cKO (n=7) vs TCF4control (n=10) mice in 3 +experiments.
+c. As in (b), but viral titers (log10 pfu/gr. tissue) +were determined 5-days post-M-CoV infection. N=7 mice for +TCF4control and n=6 mice for TCF4cKO. Violin plots +show distribution from 3 experiments.
+d. As in (b), but shown is serum ALT (mean + SD; n=6 mice/group in +3 experiments).
+e. Serum cytokines were determined by Luminex in control (n=7 +mice), pDCΔ (n=8) and +pDCΔtDCΔ (n=8) 2-days p.i. in 4 +experiments. Shown are cytokines significantly different between +pDCΔ vs pDCΔtDCΔ +mice.
+f. Percentage weight loss (left, mean + SD) and survival (right) of +pDCΔ mice inoculated with anti-NK1.1 Ab or isotype +control on day -1, 0 and 2 post-infection with M-CoV. N=3 mice/group except +n=5 mice for pDCΔ + isotype, pooled from 2 +experiments.
+g. As in (f), but viral titers were evaluated at 5-days p.i. (n=3 +mice/group). Violin plots show distribution from 2 experiments.
+h. As in Fig.8j, but splenic +viral titers (log10 pfu/gr. tissue) were determined. N=5 +mice/group except n=8 for pDC + control Ab. Violin plots show distribution +from 2 experiments.
+i. As in Fig.8k but heatmaps +show secretion of TNF, IL-6 and IL-12p70 by DC subsets (N=4 mice/subset in 4 +experiments).
+Statistical differences were determined by unpaired two-tailed +t-test (a,c-d), One-way ANOVA (e,h) or Two-way ANOVA (b,f) with +Tukey’s multiple comparisons test, and Mantel-Cox test for survival +(b,f).
Extended Data Fig. 10. Summary of DC development at steady state.
+
Supplementary Material
+Supplementary Table 1. Differentially expressed genes +between tDC and all other DC subsets. Significance threshold set to +padj<0.05.
+Supplementary Table 2. Differentially expressed genes +between clusters of KNetL graph. Calculated using findMarkers function +(padjval<0.05 and fold.change>2).
+Supplementary Table 3. List of genes shown in Figure 1e were grouped by KNetL clusters +and annotated if significant in pairwise comparisons of tDC vs pDC and tDC +vs DC2 of bulk-RNAseq data (threshold set to padj < 0.05).
+Supplementary Table 4. Differentially expressed genes +between tDC vs pDC, and tDC vs DC2. Significance threshold set to padj +< 0.05.
+Supplementary Table 5. Differentially expressed genes +between EYFP+ DC2 and EYFP− DC2. Significance +threshold set to padj < 0.05.
+ACKNOWLEDGMENTS
+This work was supported by NIH grants to J.I. (AI158808 and AI163775), B.R. +(AI072571, AI147501 and AI128949), and O.A.P. (CA232666). F.B.S. and R.A.H are +recipient of the Dean’s fellowship (Stanford), R.M-A. is supported by a +fellowship from SECTEI (CDMX, Mexico), H.M.R-M. by NSF GRFP fellowship +(#2019276237), N.M.A. by the Damon Runyon postdoctoral fellowship (DRG 2408-20), and +A.G-B. by a KL2 Mentored Career Development Award (KL2TR003143). CyTOF was performed +on the Stanford FACS Facility instrument funded by S10OD016318-01. We thank the +Idoyaga and Reizis Lab members for technical support and discussions.
Footnotes
+COMPETING INTERESTS
+J.I. serves on the scientific advisory board of Immunitas Therapeutics. +All the other authors declare no competing interests.
+DATA AVAILABILITY
+Mass cytometry data generated in this study are deposited in FlowRepository +(FR-FCM-Z6CW, FR-FCM-Z6CN and FR-FCM-Z6CK). RNAseq data generated in this study are +deposited in NCBI GEO (GSE198553 and GSE198720). NanoString data is deposited in +NCBI GEO (GSE233143). All other data generated in this study are included in the +source data files.
CODE AVAILABILITY
+This study did not generate unique codes.
REFERENCES
+-
+
- +1.Guilliams M. et al. +Dendritic cells, monocytes and macrophages: a unified +nomenclature based on ontogeny. Nat Rev Immunol +14, 571–578 +(2014). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +2.Reizis B. Plasmacytoid Dendritic Cells: Development, Regulation, +and Function. Immunity +50, 37–50 +(2019). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +3.Alcántara-Hernández M. et al. +High-Dimensional Phenotypic Mapping of Human Dendritic Cells +Reveals Interindividual Variation and Tissue Specialization. +Immunity +47, 1037–1050.e6 +(2017). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +4.Leylek R. et al. +Integrated Cross-Species Analysis Identifies a Conserved +Transitional Dendritic Cell Population. Cell Reports +29, 3736–3750.e8 +(2019). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +5.See P. et al. +Mapping the human DC lineage through the integration of +high-dimensional techniques. Science +356, eaag3009 +(2017). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +6.Villani A-C +et al. +Single-cell RNA-seq reveals new types of human blood dendritic +cells, monocytes, and progenitors. Science +356, eaah4573 +(2017). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +7.Feng J. et al. +Clonal lineage tracing reveals shared origin of conventional and +plasmacytoid dendritic cells. Immunity +55, 405–422.e11 +(2022). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +8.Naik SH +et al. +Development of plasmacytoid and conventional dendritic cell +subtypes from single precursor cells derived in vitro and in +vivo. Nat Immunol +8, 1217–1226 +(2007). [DOI] [PubMed] [Google Scholar] +
- +9.Onai N. et al. +Identification of clonogenic common Flt3+M-CSFR+ plasmacytoid and +conventional dendritic cell progenitors in mouse bone +marrow. Nat Immunol +8, 1207–1216 +(2007). [DOI] [PubMed] [Google Scholar] +
- +10.Grajales-Reyes GE +et al. +Batf3 maintains autoactivation of Irf8 for commitment of a +CD8α+ conventional DC clonogenic progenitor. +Nat Immunol +16, 708–717 +(2015). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +11.Schlitzer A. et al. +Identification of cDC1- and cDC2-committed DC progenitors reveals +early lineage priming at the common DC progenitor stage in the bone +marrow. Nat Immunol +16, 718–728 +(2015). [DOI] [PubMed] [Google Scholar] +
- +12.Nutt SL & Chopin M +Transcriptional Networks Driving Dendritic Cell Differentiation +and Function. Immunity +52, 942–956 +(2020). [DOI] [PubMed] [Google Scholar] +
- +13.Bosteels C & Scott CL +Transcriptional regulation of DC fate +specification. Mol Immunol +121, 38–46 +(2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +14.Anderson DA, Dutertre C-A, Ginhoux F & Murphy KM +Genetic models of human and mouse dendritic cell development and +function. Nat Rev Immunol +21, 101–115 +(2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +15.Schlitzer A. et al. +IRF4 transcription factor-dependent CD11b+ dendritic cells in +human and mouse control mucosal IL-17 cytokine responses. +Immunity +38, 970–983 +(2013). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +16.Brown CC +et al. +Transcriptional Basis of Mouse and Human Dendritic Cell +Heterogeneity. Cell +179, 846–863.e24 +(2019). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +17.Lewis KL +et al. +Notch2 Receptor Signaling Controls Functional Differentiation of +Dendritic Cells in the Spleen and Intestine. +Immunity +35, 780–791 +(2011). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +18.Satpathy AT +et al. +Notch2-dependent classical dendritic cells orchestrate intestinal +immunity to attaching-and-effacing bacterial pathogens. +Nat Immunol +14, 937–948 +(2013). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +19.Tussiwand R. et al. +Klf4 expression in conventional dendritic cells is required for T +helper 2 cell responses. Immunity +42, 916–928 +(2015). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +20.Bourdely P. et al. +Transcriptional and Functional Analysis of CD1c+ Human Dendritic +Cells Identifies a CD163+ Subset Priming CD8+CD103+ T Cells. +Immunity +53, 335–352.e8 +(2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +21.Cytlak U. et al. +Differential IRF8 Transcription Factor Requirement Defines Two +Pathways of Dendritic Cell Development in Humans. +Immunity +53, 353–370.e8 +(2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +22.Dutertre C-A +et al. +Single-Cell Analysis of Human Mononuclear Phagocytes Reveals +Subset-Defining Markers and Identifies Circulating Inflammatory Dendritic +Cells. Immunity +51, 573–589.e8 +(2019). [DOI] [PubMed] [Google Scholar] +
- +23.Bar-On L. et al. +CX3CR1+ CD8α+ dendritic cells are a steady-state +population related to plasmacytoid dendritic cells. +Proc National Acad Sci +107, 14745–14750 +(2010). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +24.Valente M. et al. +Novel mouse models based on intersectional genetics to identify +and characterize plasmacytoid dendritic cells. Nat Immunol +24, 714–728 +(2023). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +25.Rodrigues PF +et al. +Distinct progenitor lineages contribute to the heterogeneity of +plasmacytoid dendritic cells. Nat Immunol +19, 711–722 +(2018). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +26.Rodrigues PF +et al. +pDC-like cells are pre-DC2 and require KLF4 to control +homeostatic CD4 T cells. Sci Immunol +8, eadd4132 +(2023). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +27.Cervantes-Barragán L. et al. +Control of coronavirus infection through plasmacytoid +dendritic-cell-derived type I interferon. +Blood +109, 1131–1137 +(2007). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +28.Cervantes-Barragan L. et al. +Plasmacytoid dendritic cells control T-cell response to chronic +viral infection. Proc National Acad Sci +109, 3012–3017 +(2012). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +29.Leylek R. et al. +Chromatin Landscape Underpinning Human Dendritic Cell +Heterogeneity. Cell Reports +32, 108180 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +30.Dress RJ +et al. +Plasmacytoid dendritic cells develop from Ly6D+ lymphoid +progenitors distinct from the myeloid lineage. Nat Immunol +20, 852–864 +(2019). [DOI] [PubMed] [Google Scholar] +
- +31.Siegemund S, Shepherd J, Xiao C & Sauer K +hCD2-iCre and Vav-iCre mediated gene recombination patterns in +murine hematopoietic cells. PloS one +10, e0124661 +(2015). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +32.Kaitani A. et al. +Leukocyte mono-immunoglobulin-like receptor 8 (LMIR8)/CLM-6 is an +FcRγ-coupled receptor selectively expressed in mouse tissue +plasmacytoid dendritic cells. Scientific Reports +8, 8259–13 +(2018). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +33.Abbas A. et al. +The activation trajectory of plasmacytoid dendritic cells in vivo +during a viral infection. Nat Immunol +21, 983–997 +(2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +34.Naik SH +et al. +Cutting Edge: Generation of Splenic CD8+ and CD8− +Dendritic Cell Equivalents in Fms-Like Tyrosine Kinase 3 Ligand Bone Marrow +Cultures. J Immunol +174, 6592–6597 +(2005). [DOI] [PubMed] [Google Scholar] +
- +35.Kirkling ME +et al. +Notch Signaling Facilitates In Vitro Generation of +Cross-Presenting Classical Dendritic Cells. Cell Reports +23, 3658–3672.e6 +(2018). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +36.Cabeza-Cabrerizo M. et al. +Recruitment of dendritic cell progenitors to foci of influenza A +virus infection sustains immunity. Sci Immunol +6, eabi9331–eabi9331 +(2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +37.Liu K. et al. +In vivo analysis of dendritic cell development and +homeostasis. Science +324, 392–397 +(2009). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +38.Manh TV, Alexandre Y, Baranek T, Crozat K & Dalod M +Plasmacytoid, conventional, and monocyte-derived dendritic cells +undergo a profound and convergent genetic reprogramming during their +maturation. Eur. J. Immunol +43, 1706–1715 +(2013). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +39.Bawazeer AO +et al. +Interleukin-1β exacerbates disease and is a potential +therapeutic target to reduce pulmonary inflammation during severe influenza +A virus infection. Immunol Cell Biol +99, 737–748 +(2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +40.Grabherr S. et al. +An Innate Checkpoint Determines Immune Dysregulation and +Immunopathology during Pulmonary Murine Coronavirus +Infection. J Immunol +210, 774–785 +(2023). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +41.Yun TJ +et al. +Human plasmacytoid dendritic cells mount a distinct antiviral +response to virus-infected cells. Sci Immunol +6, (2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +42.Meredith MM +et al. +Zinc finger transcription factor zDC is a negative regulator +required to prevent activation of classical dendritic cells in the steady +state. J Exp Med +209, 1583–1593 +(2012). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +43.Salvermoser J. et al. +Clec9a-Mediated Ablation of Conventional Dendritic Cells Suggests +a Lymphoid Path to Generating Dendritic Cells In Vivo. +Front Immunol +9, 699 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +44.Lutz K. et al. +Ly6D+Siglec-H+ precursors contribute to conventional dendritic +cells via a Zbtb46+Ly6D+ intermediary stage. Nat Commun +13, 3456 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +45.Papaioannou NE +et al. +Environmental signals rather than layered ontogeny imprint the +function of type 2 conventional dendritic cells in young and adult +mice. Nat Commun +12, 464 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +46.Schulte-Schrepping J. et al. +Severe COVID-19 Is Marked by a Dysregulated Myeloid Cell +Compartment. Cell +182, 1419–1440.e23 +(2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +47.Mann ER +et al. +Longitudinal immune profiling reveals key myeloid signatures +associated with COVID-19. Sci Immunol +5, eabd6197 +(2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +48.Severa M. et al. +Differential plasmacytoid dendritic cell phenotype and type I +Interferon response in asymptomatic and severe COVID-19 +infection. Plos Pathog +17, e1009878 +(2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +49.Hadjadj J. et al. +Impaired type I interferon activity and inflammatory responses in +severe COVID-19 patients. Science +369, 718–724 +(2020). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +50.Geng J, Wang F, Huang Z, Chen X & Wang Y +Perspectives on anti-IL-1 inhibitors as potential therapeutic +interventions for severe COVID-19. Cytokine +143, 155544 +(2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
METHODS-ONLY REFERENCES
+-
+
- +51.Persson EK +et al. +IRF4 Transcription-Factor-Dependent CD103+CD11b+ Dendritic Cells +Drive Mucosal T Helper 17 Cell Differentiation. +Immunity +38, 958–969 +(2013). [DOI] [PubMed] [Google Scholar] +
- +52.Alcántara-Hernández M & Idoyaga J +Mass cytometry profiling of human dendritic cells in blood and +tissues. Nat Protoc +16, 4855–4877 +(2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +53.Diggins KE, Greenplate AR, Leelatian N, Wogsland CE & Irish JM +Characterizing cell subsets using marker enrichment +modeling. Nat Methods +14, 275–278 +(2017). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +54.Spitzer MH +et al. +An interactive reference framework for modeling a dynamic immune +system. Science +349, 1259425 +(2015). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +55.Love MI, Huber W & Anders S +Moderated estimation of fold change and dispersion for RNA-seq +data with DESeq2. Genome Biol +15, 550 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +56.Stephens M. False discovery rates: a new deal. +Biostatistics +18, kxw041 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +57.Stuart T. et al. +Comprehensive Integration of Single-Cell Data. +Cell +177, 1888–1902.e21 +(2019). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +58.Tang KH +et al. +Combined Inhibition of SHP2 and CXCR1/2 Promotes Antitumor T-cell +Response in NSCLC. Cancer Discov +12, 47–61 +(2021). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +59.Subramanian A. et al. +Gene set enrichment analysis: A knowledge-based approach for +interpreting genome-wide expression profiles. Proc National Acad Sci +102, 15545–15550 +(2005). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +60.Mootha VK +et al. +PGC-1alpha-responsive genes involved in oxidative phosphorylation +are coordinately downregulated in human diabetes. +Nature genetics +34, 267–273 +(2003). [DOI] [PubMed] [Google Scholar] +
- +61.Newman AM +et al. +Determining cell type abundance and expression from bulk tissues +with digital cytometry. Nat Biotechnol +37, 773–782 +(2019). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +62.Street K. et al. +Slingshot: cell lineage and pseudotime inference for single-cell +transcriptomics. BMC genomics +19, 477–16 +(2018). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +63.Manno GL +et al. +RNA velocity of single cells. +Nature +560, 494–498 +(2018). [DOI] [PMC free article] [PubMed] [Google Scholar] +
- +64.Bergen V, Lange M, Peidli S, Wolf FA & Theis FJ +Generalizing RNA velocity to transient cell states through +dynamical modeling. Nat Biotechnol +38, 1408–1414 +(2020). [DOI] [PubMed] [Google Scholar] +
Associated Data
+This section collects any data citations, data availability statements, or supplementary materials included in this article.
+Supplementary Materials
+Supplementary Table 1. Differentially expressed genes +between tDC and all other DC subsets. Significance threshold set to +padj<0.05.
+Supplementary Table 2. Differentially expressed genes +between clusters of KNetL graph. Calculated using findMarkers function +(padjval<0.05 and fold.change>2).
+Supplementary Table 3. List of genes shown in Figure 1e were grouped by KNetL clusters +and annotated if significant in pairwise comparisons of tDC vs pDC and tDC +vs DC2 of bulk-RNAseq data (threshold set to padj < 0.05).
+Supplementary Table 4. Differentially expressed genes +between tDC vs pDC, and tDC vs DC2. Significance threshold set to padj +< 0.05.
+Supplementary Table 5. Differentially expressed genes +between EYFP+ DC2 and EYFP− DC2. Significance +threshold set to padj < 0.05.
+Data Availability Statement
+Mass cytometry data generated in this study are deposited in FlowRepository +(FR-FCM-Z6CW, FR-FCM-Z6CN and FR-FCM-Z6CK). RNAseq data generated in this study are +deposited in NCBI GEO (GSE198553 and GSE198720). NanoString data is deposited in +NCBI GEO (GSE233143). All other data generated in this study are included in the +source data files.
+This study did not generate unique codes.
