Novel mouse models based on intersectional genetics to identify and characterize plasmacytoid dendritic cells

Valente, Michael; Collinet, Nils; Vu Manh, Thien-Phong; Popoff, Dimitri; Rahmani, Khalissa; Naciri, Karima; Bessou, Gilles; Rua, Rejane; Gil, Laurine; Mionnet, Cyrille; Milpied, Pierre; Tomasello, Elena; Dalod, Marc

doi:10.1038/s41590-023-01454-9

Download PDF

Technical Report
Open Access
Published: 16 March 2023

Novel mouse models based on intersectional genetics to identify and characterize plasmacytoid dendritic cells

Nature Immunology volume 24, pages 714–728 (2023)Cite this article

3892 Accesses
1 Citations
31 Altmetric
Metrics details

Subjects

Abstract

Plasmacytoid dendritic cells (pDCs) are the main source of type I interferon (IFN-I) during viral infections. Their other functions are debated, due to a lack of tools to identify and target them in vivo without affecting pDC-like cells and transitional DCs (tDCs), which harbor overlapping phenotypes and transcriptomes but a higher efficacy for T cell activation. In the present report, we present a reporter mouse, pDC-Tom, designed through intersectional genetics based on unique Siglech and Pacsin1 coexpression in pDCs. The pDC-Tom mice specifically tagged pDCs and, on breeding with Zbtb46^GFP mice, enabled transcriptomic profiling of all splenic DC types, unraveling diverging activation of pDC-like cells versus tDCs during a viral infection. The pDC-Tom mice also revealed initially similar but later divergent microanatomical relocation of splenic IFN⁺ versus IFN⁻ pDCs during infection. The mouse models and specific gene modules we report here will be useful to delineate the physiological functions of pDCs versus other DC types.

Main

Host survival from viral infections depends on IFN-I, exerting both antiviral and immunoregulatory functions¹. However, dysregulated IFN-I production fuels immunopathology in autoimmune diseases and certain viral infections^1,2. Hence, identifying the cellular sources of IFN-I and their molecular regulation is important to design treatments to boost or dampen IFN-I responses depending on the pathophysiological context.

The pDCs are specialized in rapid and high-level production of IFN-I in response to viruses^1,3. They engulf virus-derived material into endosomes equipped with toll-like receptor 9 (TLR9) for sensing unmethylated CpG DNA and TLR7 for single-stranded RNA. TLR7/9 activates an MyD88-to-IRF7 signaling cascade, leading to IFN-I production. Recently, new DC types sharing surface markers and gene expression with pDCs were identified, including pDC-like cells and transitional DCs (tDCs), which can contaminate pDC populations and confound their characterization^4,5,6.

Only a small fraction of pDCs produces IFN-I during viral infections⁷. How this process is regulated remains enigmatic. We do not know precisely when, where and how pDCs sense and sample virus-derived material, and how this shapes host antiviral defense³. Answering these questions has been hampered by the lack of mutant mouse models enabling specific and penetrant targeting of pDCs⁸. This bottleneck was caused by lack of a gene expressed in pDCs with high enough specificity to target them by classic knock-out or knock-in approaches⁹. In the Siglech-based deleter/reporter mice, other DC populations and macrophage subsets are targeted¹⁰. Moreover, activated pDCs downregulate SiglecH⁷. Hence, Siglech–green fluorescent protein (GFP) mice are not suitable for reliably identifying pDCs in vivo¹⁰, even if this was attempted by selecting cells with high GFP intensity and plasmacytoid morphology¹¹. Transgenic Siglech-Cre mice additionally suffer from a low pDC-targeting efficacy¹². In Itgax^Cre;Tcf4^flox/− mice, tDCs and pDC-like cell development is compromised⁶; macrophage and B cell subsets coexpressing Tcf4 and Itgax^6,13 might also be affected. BDCA2-DTR mice¹⁴ are the most trusted model for pDC depletion; however, they should be used with caution because their serial injection with diphtheria toxin causes artefactual chronic IFN-I production and severe immunopathology, with one dose sufficient to induce IFN-I (ref. ¹⁵). Thus, there is an unmet scientific need for mutant mouse models allowing specific and penetrant targeting of pDCs without technical artefacts⁸.

Results

The pDC-Tom mice allow specific pDC detection by flow cytometry

We generated mice knocked in for Cre expression from Siglech, which is highly expressed by mouse pDCs³. We crossed Siglech^iCre and Rosa26^{LoxP-STOP-LoxP(LSL)-RFP} mice¹⁶ to generate S-RFP mice for fate-mapping Siglech-targeted cells (Fig. 1a). Over 95% of splenic pDCs were red fluorescent protein positive (RFP⁺) (Fig. 1b and Extended Data Fig. 1a,b). Variable proportions of myeloid and lymphoid lineages expressed RFP (Fig. 1b and Extended Data Fig. 1a,b), consistent with Siglech expression^10,17. We reasoned that enhanced specificity could be achieved by harnessing intersectional genetics, driving expression of a reporter under the control of two genes coexpressed only in pDCs. We aimed for activation by Siglech-driven Cre of a conditional fluorescent reporter cassette knocked in a gene exclusively expressed by pDCs within Siglech fate-mapped cells. We selected Pacsin1, expressed exclusively in pDCs within hematopoietic cells¹⁸ and promoting their IFN-I production¹⁹. We generated Pacsin1^{LoxP-STOP-LoxP-tdTomato} (Pacsin1^LSL-tdT) mice, knocked in with a floxed cassette for tdTomato (tdT) conditional expression. We crossed them with Siglech^iCre mice, to generate pDC-Tom mice (Fig. 1c). In splenocytes from pDC-Tom mice, tdT was exclusively expressed in pDCs (Fig. 1d and Extended Data Fig. 1a,c). The tdT⁺ cells expressed neither lineage markers nor CD11b (Extended Data Fig. 1d). The CD45⁺ tdT⁺ cells isolated from different organs were CD11c^int and BST2^high (Fig. 1e), as expected for pDCs²⁰. CD45⁺tdT⁺ cells coexpressed Ly6D, B220, SiglecH and CCR9 (Fig. 1f), a combination specific to pDCs. Thus, tdT expression in pDC-Tom mice is sufficient to specifically and unambiguously identify most pDCs.

**Fig. 1: The pDC-Tom mice allow specific and unambiguous identification of pDCs in different organs.**

The pDC-Tom mice allow refining pDC gating strategies

Defining pDCs as coexpressing CD11c, BST2 and SiglecH can lead to contamination by conventional DCs (cDCs), pDC-like cells⁴ or tDCs^5,6,21. Moreover, on inflammation, such as during mouse cytomegalovirus (MCMV) infection, the expression of these markers is altered, causing a phenotypic convergence of pDCs and cDCs⁷. Hence, we harnessed pDC-Tom mice to define a gating strategy allowing unequivocal pDC identification both at steady state and during infection. We defined pDCs as tdT⁺ cells and used HyperFinder for unsupervised computational generation of a gating strategy to identify them, based on surface markers without using tdT. We included Ly6D, which is selectively expressed on pDCs and B cells, discriminating them from cDCs and tDCs^4,5. At steady state, splenic pDCs were identified as Bst2^highLy6D⁺B220⁺CD19⁻CCR9⁺SiglecH⁺ cells (Fig. 1g). During MCMV infection, they were identified as Ly6D⁺CX₃CR1^low/intCD19⁻CCR9^highB220^highBST2^high cells (Fig. 1h). Hence, current identification of pDCs as lin⁻CD11b⁻CD11c^low-to-intBST2^high cells²⁰ can be improved by addition of positivity for Ly6D or CCR9 and eventual exclusion of CX₃CR1^high cells. SiglecH is not a good marker postinfection (p.i.), because it is downregulated (Fig. 1i), especially on IFN-I-producing pDCs⁷. We thus propose identifying pDCs as Ly6D^highBST2^highCD19⁻B220⁺CD11b⁻CD11c⁺ cells.

In pDC-Tom mice, tdT soars in pDCs’ immediate precursors

We analyzed bone marrow cells to investigate tdT expression along the myeloid^{17,22,23,24,25} and lymphoid^4,5 ontogeny paths proposed for pDCs (Fig. 2a). Within differentiated cells, tdT was expressed exclusively in pDCs (Fig. 2b). The tdT was also detected, at a lower mean fluorescence intensity (MFI), in pDCs’ immediate precursors, the CD11c⁺ pre-pDCs (Fig. 2c and Extended Data Fig. 2a).

**Fig. 2: Expression of tdT is detectable in late bone marrow precursors selectively committed to the pDC lineage.**

Upstream along the myeloid path, very low tdT levels were detected in SiglecH⁺ pre-DCs, but none in SiglecH⁻ pre-DCs, irrespective of Ly6C expression (Fig. 2d and Extended Data Fig. 2a). This is consistent with SiglecH⁺Ly6C^+/− pre-DCs giving rise to cDCs and pDCs, whereas SiglecH⁻Ly6C⁺ and SiglecH⁻Ly6C⁻ pre-DCs generate mostly cDC2s or cDC1s, respectively¹⁷. The common DC progenitor (CDP), monocyte and DC progenitor (MDP) and common myeloid progenitor (CMP) were tdT⁻ (Fig. 2e and Extended Data Fig. 2a).

Upstream along the lymphoid path, low tdT levels were detected in CD127⁺SiglecH⁺Ly6D⁺ progenitors (Fig. 2g and Extended Data Fig. 2b), consistent with these cells giving rise to pDCs⁴. Very low tdT levels were detected in the Ly6D⁺SiglecH⁻ progenitor and none upstream (Fig. 2f–g and Extended Data Fig. 2b).

Thus, in pDC-Tom mice, tdT expression is exclusively induced in late bone marrow precursors committed to the pDC lineage, with a strong increase in CD11c⁺ pre-pDCs and maximal level in differentiated pDCs.

ZeST mice distinguish pDCs, pDC-like cells and tDCs

Refined, flow cytometry phenotypic keys can discriminate pDCs from pDC-like cells and tDCs at steady state⁶. This remains challenging in inflammation and in tissues by immunohistofluorescence. Within hematopoietic cells, Zbtb46 is specifically expressed in the cDC lineage including in pre-cDCs¹⁸, in pDC-like cells, as confirmed in Zbtb46^GFP mice^4,26, and in tDCs⁶. Therefore, to rigorously identify pDC-like cells and tDCs, we generated Zbtb46^GFP;Siglech^iCre;Pacsin1^LSL-tdT (ZeST) mice (Fig. 3a).

**Fig. 3: ZeST mice allow unambiguous discrimination of pDCs from tDCs and pDC-like cells.**

In ZeST mice, most of the lin⁻, SiglecH⁺ or CD11c^high cells expressed GFP or tdT in an almost mutually exclusive manner (Fig. 3b). Within tdT⁺ or GFP⁺ cells, we identified type 1 cDCs (cDC1s) as XCR1⁺ and type 2 cDCs (cDC2s) as CD11b⁺ (Fig. 3b). Within the XCR1⁻CD11b⁻BST2^highSiglecH⁺ gate, we identified pDCs as Ly6D⁺GFP⁻ and pDC-like cells as GFP⁺ according to their original definition⁴ but considering only the major Ly6D⁻ fraction. The small fraction coexpressing Ly6D and GFP (Zbtb46⁺Ly6D⁺ cells) was studied separately to determine whether they encompassed pDCs. The tDCs were originally characterized as Lin⁻CD11b⁻XCR1⁻SiglecH^−/highCD11c^low-to-highCX3CR1⁺ and split into two fractions: CD11c^lowLy6C⁺ cells, harboring higher levels of SiglecH and Tcf4, more similar to pDCs, versus CD11c^highLy6C⁻ cells, high for Zbtb46, more similar to cDC2s⁶. Comparing phenotypic markers and genes between CD11c^lowLy6C⁺ tDCs⁶ and pDC-like cells⁴ suggested that they were the same population. Thus, to avoid overlap between the gates for pDC-like cells and tDCs, we focused on the SiglecH^−/low tDCs, identified similarly to their original definition⁶ as CD11b⁻, XCR1⁻, SiglecH^−/low, BST2^−/lowCX3CR1⁺ cells, within singlet, live, nonautofluorescent, Lin⁻, SiglecH⁺ or CD11c^high, tdT⁺ or GFP⁺ cells (Fig. 3b). The SiglecH^−/low tDCs were all Ly6D⁻ (Fig. 3b–c) and expressed higher levels of CD11c than pDCs and pDC-like cells (Fig. 3c); hence we called them CD11c^high tDCs. Only very few of the Lin⁻, CD11c^high or SiglecH⁺ cells expressed neither GFP nor tdT, a fraction of which were Ly6D⁺SiglecH^high and thus probably tdT⁻ pDCs (Fig. 3b).

To validate the identity of the DC types gated manually, we characterized their phenotype further (Fig. 3c). As expected, beyond being Ly6D⁺SiglecH⁺, both tdT⁻ and tdT⁺ pDCs were CCR9^highCD11c^int. A fraction of Zbtb46⁺ Ly6D⁺ cells expressed lower levels of tdT and CCR9, and higher levels of CD11c, compared with tdT⁺ pDCs. CD11c^high tDCs were GFP^hightdT⁻. The pDC-like cells were GFP⁺tdT⁻CCR9^−/low. The surface phenotype of these populations was not modified during MCMV infection (Fig. 3c and Extended Data Fig. 3). Hence, tdT was expressed only in cells harboring a phenotype of bona fide pDCs, whereas GFP was mainly expressed in cDC1s, cDC2s, pDC-like cells and CD11c^high tDCs.

Next, we performed an unsupervised analysis of the phenotypic relationships for all the Lin⁻, CD11c^high or SiglecH⁺ cells. Onto a t-distributed stochastic neighbor embedding (t-SNE) representation of the data, we projected the populations identified through manual gating as defined in Fig. 3b. This analysis highlighted three main cell clusters, corresponding to cDC2s, cDC1s and pDCs (Fig. 3d). The other DC types formed a continuum between pDCs and cDC2s, with tdT⁻ pDCs and Zbtb46⁺Ly6D⁺ cells close to pDCs, versus pDC-like cells and CD11c^high tDCs close to cDC2s, consistent with previous observations⁶.

We sorted cDC2s, CD11c^high tDCs, pDC-like cells and pDCs from steady-state mouse spleens to examine their morphology (Fig. 3e). The cDC2s harbored many pseudopods or dendrites, translating into a low circularity index. Most pDC-like cells and pDCs harbored a round morphology, translating into a high circularity index. The CD11c^high tDC population was morphologically heterogeneous, with a bimodal distribution of circularity indices, half of the cells being dendritic, like cDC2s, and the other half round, like pDCs. Overall, quantitative and unbiased analysis of cellular morphology supported success in high-degree purification of the DC types.

Finally, to better discriminate the tdT signal from autofluorescence, and analyze more cell-surface markers in ZeST mice, we harnessed spectral flow cytometry (Extended Data Figs. 4 and 5). Unsupervised cell clustering based on all surface markers (Extended Data Fig. 4a,b), without considering tdT and GFP, showed that most GFP⁺ cells were cDC2s or cDC1s (Extended Data Fig. 4c). They also encompassed lymphoid cells and a cluster of myeloid cells, but with a low MFI (Extended Data Fig. 4d). The pDCs represented 81.5 ± 6% of tdT⁺ splenocytes (Extended Data Fig. 4e). The individual contribution of other phenotypic cell clusters to the tdT⁺ gate was very small and their tdT MFI below that of pDCs, barely above background (Extended Data Fig. 4f).

Complementary analysis by supervised identification of cell types via manual gating allowed study of both tDC populations: the Ly6C⁻ versus Ly6C⁺ fractions of Lin⁻CD11b⁻XCR1⁻, CD11c⁺ or BST2⁺, CX3CR1⁺CD26⁺ cells (Extended Data Fig. 5a–d). The populations phenotypically defined as pDC-like cells⁴ or Ly6C⁺ tDCs⁶ largely overlapped, as was the case for CD11c^high tDCs and Ly6C⁻ tDCs (Extended Data Fig. 5b–e). Most Lin⁻CD11b⁻XCR1⁻, CD11c⁺ or BST2⁺, Ly6D⁻CX3CR1⁻ cells were CD11c⁺CD26⁺GFP⁺CD64⁻ (Extended Data Fig. 5f), indicating DC lineage. They encompassed major histocompatibility complex II (MHC-II)^−/low and MHC-II^high cells putatively corresponding to pre-DCs versus differentiated DCs, respectively. A decrease in the absolute numbers of most of DC types was observed in the spleen at 48 h p.i. (Supplementary Table). This analysis confirmed the high specificity and penetrance of GFP expression in cDC1s and cDC2s, and both tDC populations, as well as the high specificity and penetrance of tdT expression in pDCs (Extended Data Fig. 5g).

ScRNA-seq confirms DC-type identification in ZeST mice

We harnessed ZeST mice to perform single-cell RNA sequencing (scRNA-seq) for the five DC types, on index sorting from the spleen of animals either infected or not infected with MCMV, using the gating strategy shown in Fig. 3b and the FB5P-seq (FACS-based 5′-end scRNA-seq) protocol^27,28.

We first analyzed 343 splenocytes isolated from noninfected (NI) mice. They were clustered and annotated for cell types (Supplementary Table) by Seurat, based on gene expression profiles (Extended Data Fig. 6a), and by Rphenograph, based on surface marker expression (Extended Data Fig. 6b,c). DC-type assignment to individual cells was rather consistent between these two strategies, but suggested heterogeneity of Seurat clusters II and III (Extended Data Fig. 6d). Therefore, to unambiguously and reliably assign a DC type to individual cells in an unbiased manner, we used a combinatorial strategy: to be selected, a cell had to belong to the expected intersection between the Rphenograph and Seurat clusters (colored cells in Extended Data Fig. 6d) and to be enriched for the corresponding DC-type-specific transcriptomic signatures^4,29 as tested with single-cell connectivity Map (sgCMap; Extended Data Fig. 6e). A final cell type was assigned to 205 cells: 103 pDCs, 26 pDC-like cells, 19 CD11c^high tDCs, 23 cDC2s and 34 cDC1s (Extended Data Fig. 6e,f and Supplementary Table).

Our next objective was to assign a DC type to the 951 cells from the whole dataset, from both NI and MCMV-infected mice (Extended Data Fig. 7a and Supplementary Table). We aimed at achieving a robust DC-type assignment, irrespective of cell states and infection conditions, by combining transcriptomic and phenotypic analyses. Based on the intersection of the clustering with Seurat (Extended Data Fig. 7b) versus Rphenograph (Extended Data Fig. 7c,d), a final DC type was assigned to 851 cells (colored cells in Extended Data Fig. 7e): 310 pDCs, 170 pDC-like cells, 146 CD11c^high tDCs, 103 cDC2s and 122 cDC1s, with 100 cells left unannotated (Fig. 4a). We confirmed DC-type assignment by a complementary method: we generated cell type-specific transcriptomic signatures from the dataset focused on cells from NI mice (Supplementary Table) and used them for sgCMap analysis of the whole dataset (Extended Data Fig. 7f). All the cells sorted as pDCs, and most of the tdT⁻ pDCs, were computationally assigned to pDCs (Extended Data Fig. 7e,f). The assignment to cDC1s was also consistent between cell sorting and deductive re-annotation. Cell clustering on sgCMap scores tended to confirm the distinction between cDC2s and CD11c^high tDCs, although many cells had a null score for the ‘tDC_vs_cDC2’ sgCMap signature, emphasizing the proximity between these two DC types (Extended Data Fig. 7f). As expected, Zbtb46⁺Ly6D⁺ cells were mostly assigned to pDC-like cells. Some cells sorted as pDC-like cells were in time assigned to CD11c^high tDCs (Extended Data Fig. 7e,f). Not only CD11c^high tDCs but also pDC-like cells were CX3CR1^+/high (Fig. 4b). Akin to cDCs, CD11c^high tDCs were GFP^high, whereas pDC-like cells expressed SiglecH, BST2, Ly6D and CCR9 to levels intermediate between those of pDCs (high) and cDCs (low). The pDCs and pDC-like cells shared high expression of Siglech (Fig. 4c,d), Tcf4 and Runx2 (Fig. 4c). CD11c^high tDCs and pDC-like cells shared high expression of Crip1, Lgals3 and Vim (Fig. 4c), previously reported to discriminate pDC-like cells from pDCs⁴. The pDC-like cells and CD11c^high tDCs shared with cDCs higher expression of Zbtb46, Spi1, Slamf7 and S100a11, compared with pDCs (Fig. 4c). Contrary to cDC2 cells, a fraction of pDC-like cells and CD11c^high tDCs expressed Cd8a (Fig. 4c,d), as reported for tDCs^6,21,29. The pDC-like cells, CD11c^high tDCs and cDC2s specifically expressed Ms4a4c (Fig. 4c,d) and Ms4a6c (Fig. 4c). CD11c^high tDCs and cDC2s selectively shared expression of Ms4a6b (Fig. 4d) and S100a4 (Fig. 4c). CD11c^high tDCs expressed higher levels of certain cDC genes than pDCs and pDC-like cells, including Batf3, Rogdi and Cyria (Fig. 4c). The pDC-like cells expressed very high levels of Ly6c2 (Fig. 4c,d). Hence, the pDC-like cells characterized in the present report were confirmed to align with both the originally described pDC-like cells⁴ and the CD11c^lowLy6C⁺ tDCs⁶.

**Fig. 4: ScRNA-seq confirmed proper identification of pDCs, pDC-like cells and tDCs in ZeST mice.**

All cDC1s expressed the XCR1 protein (Fig. 4b). However, a fraction was low/negative for Xcr1 and other genes specific of steady-state cDC1s (Gpr141b, Tlr3, Cadm1 and Naaa; Fig. 4c), consistent with DC-type activation decreasing the expression of many of the genes used to identify them at steady state^30,31,32, preventing use of individual genes for reliable DC-type identification in scRNA-seq datasets³³.

As expected, only pDCs expressed rearranged immunoglobulin (Ig) genes, Ccr9, Klk1 and Cox6a2 (Fig. 4c) as well as the tdT fluorescent reporter above autofluorescence levels (Fig. 4e). The tdT⁻ pDCs were largely overrepresented in our scRNA-seq dataset because we enriched them for characterization. As we did not use the tdT signal in our analysis, these results confirm the specificity of pDC-Tom mice for pDC identification; they also confirmed proper identification of pDC-like cells and CD11c^high tDCs in ZeST mice, and allowed refining of their characterization through side-by-side pangenomic transcriptomic profiling.

The pDC-Tom mice allow mapping of pDC microanatomical localization

We harnessed the pDC-Tom mice to determine the microanatomical localization of pDCs in various organs. In the spleen, a relatively high density of tdT⁺ cells was observed in the T cell zone (TCZ) within the white pulp (WP) (Fig. 5a). Scattered tdT⁺ cells were detected in the red pulp (RP), delimited by CD169⁺ metalophilic marginal zone (MZ) macrophages and densely populated by F4/80⁺ RP macrophages (Fig. 5b). Most TCZ or RP tdT⁺ splenocytes expressed BST2 (Fig. 5a,b). Conversely, whereas most TCZ BST2⁺ cells were tdT⁺ (Fig. 5a), this was not the case in the RP (Fig. 5b), consistent with BST2 expression on plasma cells, macrophage subsets, pDC-like cells and tDCs^4,6,34. We quantified the density of pDCs (tdT⁺ cells) in the whole spleen (Fig. 5c,d) and in its microanatomical compartments: the RP, the MZ, the TCZ and the B cell zone (BCZ) (Fig. 5c,e,f). The pDC density in the whole spleen was around 400 cells mm⁻² (Fig. 5d). Steady-state splenic pDCs were primarily located in the TCZ and RP (Fig. 5e). The pDC density was much higher in the TCZ (~1,500 cells mm⁻²) than in other splenic compartments (≤400 cells mm⁻²) (Fig. 5c,f). In lymph nodes, pDCs were frequent in paracortical T cell (CD3⁺) areas (Extended Data Fig. 8a). In the small intestine, pDCs had been observed in both Peyer’s patches³⁵ and the lamina propria (LP), primarily based on BST2 signal³⁶. However, interpretation may have been confounded by BST2 expression on macrophage and DC subsets in the gut^37,38. Hence, we re-examined pDC microanatomical location in the small intestine. The tdT⁺ cells were detectable in the LP, generally close to EpCAM⁺ epithelial cells (Extended Data Fig. 8b). Very few tdT⁺ cells were observed in Peyer’s patches (Extended Data Fig. 8c). The pDCs were not detectable in the LP of the large intestine (Extended Data Fig. 8d). Thus, at steady state, in the gut, pDCs are detectable quasi-exclusively in the small intestine LP.

**Fig. 5: The pDC-Tom mice allow studying pDC microanatomical location in the spleen.**

SCRIPT mice allow tracking IFN-I⁺ and IFN-I⁻ pDCs in situ

During MCMV infection, the Ifnb1^EYFP reporter mouse model³⁹ does not only allow detection, at 36 h p.i., of the pDCs actively producing IFN-β^7,27, but, in addition, at 48 h, of the pDCs that have already produced and secreted IFN-β but maintained enhanced YFP expression for >12 h²⁷. Hence, the Ifnb1^EYFP mice are reliable and well suited to fate-map the pDCs that have previously produced IFN-I, enabling determination of their activation trajectory in vivo during infection²⁷. The pDCs clustered in the MZ near to infected cells at 36 h p.i., at the time of peak IFN-I production, then migrated to the TCZ after termination of their IFN-I production between 40 h and 48 h p.i., while acquiring transcriptomic, phenotypic and functional features of mature cDCs²⁷. However, the lack of adequate markers prevented us from assessing the spatiotemporal repositioning of the pDCs that were not producing IFN-I. To answer this question, we generated Siglech^iCre;Ifnb1^EYFP;Pacsin1^LSL-tdT (SCRIPT) mice (Fig. 6a) for unambiguous identification of both the pDCs that are producing (or have produced) IFN-I (tdT⁺YFP⁺) and those that do not (tdT⁺YFP⁻) (Fig. 6a,b). By flow cytometry, both tdT⁺YFP⁻ and tdT⁺YFP⁺ cells were detectable at 36 h and 48 h p.i., expressing Ly6D and BST2 consistent with their pDC identity (Fig. 6b). We detected only very rare tdT⁻YFP⁺Ly6D⁻BST2^low cells, consistent with pDCs being the main source of IFN-I during infection^7,20,27,40. We examined the localization of both tdT⁺YFP⁻ and tdT⁺YFP⁺ pDCs in spleens from MCMV-infected mice (Fig. 6c–f). At 36 h p.i., both pDC populations formed clusters in the MZ (Fig. 6c,d), leading to an increase in the proportion of pDCs at this microanatomical site (Fig. 6e). Although pDCs were still detectable in the TCZ of infected animals, their local density was reduced compared with uninfected mice (Fig. 6c,e).

**Fig. 6: Diverging intrasplenic migration patterns and morphological changes between the pDCs producing and those not producing IFN-I during MCMV infection.**

Only IFN-I^fate_mapped pDCs entered the TCZ during infection

At 48 h p.i., we detected large aggregates of tdT⁺YFP⁻ cells in the MZ, whereas tdT⁺YFP⁺ cells were mostly detected in the TCZ (Fig. 6c,d). Hence, we observed an opposite trend in the spatiotemporal repositioning of the pDCs that either produced or did not produce IFN-I within the spleen during infection: whereas most of the pDCs fate-mapped for IFN-I production (YFP⁺) were located in the MZ at 36 h and then in the TCZ at 48 h, the reverse distribution was observed for the pDCs that did not produce IFN-I (YFP⁻) (Fig. 6f). This analysis showed that, at the peak of IFN production, the high recruitment of pDCs to the MZ occurs independent of their ability to produce IFN-I, whereas, at later time points, only the pDCs that have produced IFN-I are licensed to migrate to the TCZ; the other pDCs are retained in the MZ. Next, we sorted DC types from the spleens of infected mice and examined their morphology, also comparing YFP⁺ versus YFP⁻ pDCs at 60 h p.i. (Fig. 6g,h and Extended Data Fig. 9). A fraction of YFP⁺ pDCs acquired an irregular morphology with pseudopods or dendrites (Fig. 6g), harboring significantly lower circularity indices compared both with their YFP⁻ counterparts and with steady-state pDCs (Fig. 6h). Hence, the pDCs that had produced IFN-I during infection selectively acquired a dendritic morphology, consistent with their known transcriptomic and functional convergence toward cDCs²⁷.

The pDCs clustered around infected cells but few produced IFN-I

Due to the lack of specific markers to track pDCs in situ before their peak IFN-I production, determination of the early kinetics of their microanatomical redistribution in the spleen during infection has not been possible previously. The pDC-Tom mice allowed us to address this issue (Fig. 7). Recruitment of pDCs and their clustering in the MZ were detectable as early as 12 h p.i. (Fig. 7a,b), with a clear increase in the proportion of MZ pDCs approaching the plateau values observed between 18 h and 48 h (Figs. 6e and 7c). Cells replicating MCMV (expressing the viral immediate early gene 1, IE1) were already detectable at 12 h p.i., mainly in the MZ (Fig. 7d–f). The pDC clusters were already localized in close proximity to MCMV-infected cells at 12 h (Fig. 7d), consistent with the proximity between IFN-I-producing pDCs and infected cells observed at later time points²⁰. Thus, the recruitment of pDCs to the vicinity of infected cells in the MZ occurred early, already at 12 h p.i., 24 h before their peak IFN-I production.

**Fig. 7: Early during MCMV infection pDCs are recruited at the MZ where they contact infected cells.**

In the MZ of the spleen of 40-h infected mice, three-dimensional (3D) reconstructions from confocal microscopy images showed that IFN-I-producing pDCs established tight interactions with virus-infected cells (Supplementary Videos 1–3), consistent with in vivo establishment of interferogenic synapses as previously observed only in vitro^41,42.

The pDC-like cells and CD11c^high tDCs diverge on activation

We analyzed our FB5P-seq data to compare the responses of the five DC types to MCMV infection in vivo. Four Seurat clusters were identified for pDCs, corresponding to distinct activation states (Fig. 8a and Supplementary Table): quiescent pDCs, intermediate pDCs harboring induction of IFN-stimulated genes (ISGs) but lacking expression of cytokines genes, activated pDCs expressing moderate levels of cytokine genes and IFN-I-producing pDCs expressing high levels of Ifnl2, and all the genes encoding IFN-I. CD11c^high tDCs, cDC1s and cDC2s each split into three clusters, corresponding to quiescent, intermediate and activated states. The pDC-like cells split into two clusters only: quiescent and activated.

**Fig. 8: ScRNA-seq confirms the unique capacity of pDCs for high IFN-I/III expression during infection and shows divergent activation patterns for pDC-like cells and tDCs.**

The cell types and activation states identified by scRNA-seq were shared across individual mice (Supplementary Fig. 1a), showing similar proportions between mice at the same time p.i. (Supplementary Fig. 1b).

Expression of tdT remained highly selective in pDCs, irrespective of their activation states (Fig. 8b), despite SiglecH downregulation in IFN-I-producing pDCs⁷ (Extended Data Fig. 10a). Activation increased autofluorescence but did not induce tdT in other DC types (Fig. 8b). The pDCs remained Ly6D^highCCR9^high across activation states. BST2 expression increased on all activated DC types (Fig. 8b), as reported previously^7,43,44. Activation decreased CX3CR1 expression in CD11c^high tDCs (Extended Data Fig. 10). Together with their increased BST2 expression (Fig. 8b), this contributed to shifting a fraction of the CD11c^high tDC population from infected mice into the phenotypic gate used for sorting pDC-like cells, which was corrected by deductive cell-type reassignment on scRNA-seq computational analysis (Extended Data Fig. 7f and Supplementary Table). The cells assigned to CD11c^high tDCs expressed higher levels of CD11c and GFP than pDC-like cells (Extended Data Fig. 10a). Although the expression of XCR1 on cDC1s and CD11b on cDC2s decreased with activation, these surface markers remained clearly detectable, allowing phenotypic identification of these DC types (Extended Data Fig. 10a), contrasting with near extinction of Xcr1 expression in activated cDC1s (Extended Data Fig. 10b).

As observed in previous reports based on flow cytometry or bulk RNA-seq data⁶, quiescent CD11c^high tDCs and pDC-like cells were close together in the Uniform Manifold Approximation and Projection (UMAP) space, between pDCs and cDC2s (Fig. 8a). On activation, they moved in opposite directions, with activated pDC-like cells close to intermediate pDCs, but intermediate or activated CD11c^high tDCs and cDC2s close together (Fig. 8a). Hence, pDC-like cells and CD11c^high tDCs underwent divergent activation in vivo during infection. Indeed, whereas only 84 genes were differentially expressed between CD11c^high tDCs and pDC-like cells at the steady state, this number increased to 527 for activated cells (Supplementary Table). Whereas all DC types induced ISGs (Fig. 8c, cluster 1), only CD11c^high tDCs, cDC2s and cDC1s induced high levels of genes linked to DC maturation/migration and interactions with T cells, encompassing Fscn1, Ccr7, Il15, Il15ra, Cd80, Cd200 and Cd274 (Fig. 8c, cluster 3). Another set of ISGs and genes associated with DC interactions with T cells gradually increased in expression from intermediate to activated to IFN-I-producing pDCs (Fig. 8c, cluster 2). These genes were induced to similar levels in activated pDC-like cells, but to even higher levels in activated CD11c^high tDCs, cDC2s and cDC1s. Only a few genes were differentially expressed between CD11c^high tDCs and cDC2s: 16 at steady state, 1 at the intermediate activation state and 11 for activated cells (Supplementary Table). The number of genes differentially expressed between pDCs and pDC-like cells remained stable over activation: 113 at steady state versus 138 in activated cells (Supplementary Table). Hence, on activation, pDC-like cells behaved more like pDCs, although not producing IFN-I (Fig. 8c, cluster 5), whereas CD11c^high tDCs converged further toward cDC2s.

Activated pDC-like cells maintained a higher Ly6c2 expression (Extended Data Fig. 10c) and a lower CD11c MFI (Extended Data Fig. 10a) than CD11c^high tDCs, confirming the reliability of these markers even in inflammation. Genes encoding MHC-II molecules, such as H2-DMb2 and H2-Eb1, were higher in CD11c^high tDCs than in pDC-like cells, reaching the same levels as in cDC2s (Extended Data Fig. 10c). Tmem176a and Tmem176b, encoding cell-surface markers, were higher in CD11c^high tDCs than in pDC-like cells, pDCs, cDC1s and to a lesser extent cDC2s, across activation conditions (Extended Data Fig. 10c,d). Cd8a expression remained lower in cDC2s across activation conditions (Extended Data Fig. 10c). Ms4a4c was higher in activated pDC-like cells, intermediate and activated CD11c^high tDCs and cDC2s than in pDCs and cDC1s (Extended Data Fig. 10c). Apod was expressed to higher levels in pDC-like cells, although it was induced in all other DC types except cDC2s on activation (Extended Data Fig. 10c,d).

Finally, beyond confirming the unique ability of pDCs to produce IFN-I/III during MCMV infection^7,20,40, our analysis unraveled selective induction in cDC1s of genes encoding costimulation molecules or cytokines involved in the crosstalk with natural killer (NK) or T cells, encompassing Cd70, Pdcd1lg2, Pvr and Ccl22 (Fig. 8c, cluster 4), Il27 (Fig. 8c, cluster 3), Il18 and Il12b (Extended Data Fig. 10e). This is consistent with a division of labor between DC types, with cDC1s promoting NK and CD8 T cell antiviral responses^45,46.

Discussion

We generated the first reporter mouse model, to our knowledge, specifically and efficiently tagging pDCs, across organs and activation conditions: the pDC-Tom (Siglech^iCre;Pacsin1^LSL-tdT) mice.

We did not provide data to solve the current debate on pDC ontogeny^4,5,47, because tdT expression in our model starts only from the late CD11c⁺ pre-pDC differentiation state that is common to the ‘lymphoid’ and ‘myeloid’ paths.

Using pDC-Tom mice, we could study the choreography of the relocation of all pDCs in the spleen during MCMV infection. All pDCs were attracted to the microanatomical sites of viral replication early p.i. Thus, the failure of most pDCs to produce IFN-I is unlikely to result from restricted access to infected cells. Alternatively, pDC tight clustering raises the hypothesis that the first to produce IFN-I may have repressed this function in their neighbors, through a quorum-sensing mechanism, to prevent excessive inflammation and consequent immunopathology. Future studies using pDC-reporter mice and pDC-specific genetic manipulations should help to better understand this regulation.

We generated the first side-by-side transcriptomic comparison, to our knowledge, of pDCs, pDC-like cells and CD11c^high tDCs, through scRNA-seq. We uncovered a divergent activation between CD11c^high tDCs and pDC-like cells. Only CD11c^high tDCs underwent a maturation closely resembling that of cDCs.

CD11c^high tDCs and activated cDC2s selectively expressed high levels of H2-DMb2 and Tmem176a/b, new candidate markers for these DC types. Tmem176a/b colocalizes with human leukocyte antigen (HLA)-DM in the late endolysosomal system, promotes antigen presentation to naive T cells and contributes to tolerance/immunosuppression^48,49,50. This raises the question of the division of labor between CD11c^high tDCs and cDC2s, including for CD4 T cell tolerance. Our scRNA-seq dataset will be a valuable resource to mine the gene expression profiles of pDC-like cells and CD11c^high tDCs, compared with pDCs, cDC1s and cDC2s, to infer and experimentally test hypotheses on their functional specialization and molecular regulation.

Methods

Mice

All animal experiments were performed in accordance with national and international laws for laboratory animal welfare and experimentation (EEC Council Directive 2010/63/EU, September 2010). Protocols were approved by the Marseille Ethical Committee for Animal Experimentation (registered by the Comité National de Réflexion Ethique sur l’Expérimentation Animale under no. 14; APAFIS no. 1212-2015072117438525 v.5 and APAFIS no. 21626-2019072606014177 v.4). C57BL/6 mice were purchased from Janvier Labs. All other mouse strains were bred at the Centre d’ImmunoPhénomique (CIPHE) or the Centre d’Immunologie de Marseille-Luminy (CIML), under specific pathogen free-conditions and in accordance with animal care and use regulations. Mice were housed under a 12 h dark:12 h light cycle, with a temperature range of 20–22 °C and a humidity range of 40–70%. Siglech^iCre mice (B6-Siglech^{tm1(iCre)Ciphe})²⁰ and Pacsin1^{LoxP-STOP-LoxP-tdTomato(LSL-tdT)} (B6-Pacsin1^{tm1(tdT)Ciphe}) mice were generated by CIPHE. Siglech^iCre;Pacsin1^LSL-tdT mice (pDC-Tom) were generated by crossing Siglech^iCre mice with Pacsin1^LSL-tdT mice, and then maintained and used in a double homozygous state. SCRIPT mice were generated by crossing pDC-Tom mice with Ifnb1^Eyfp mice (B6.129-Ifnb1^tm1Lky)³⁹, and then maintained and used in a triple homozygous state. ZeST mice were generated by crossing Zbtb46^GFP mice (B6.129S6(C)-Zbtb46^tm1.1Kmm/J)²⁶ with pDC-Tom mice and were used in a heterozygous state. All animals used were sex and age matched (used between 8 and 16 weeks of age).

Virus and viral infection

Virus stocks were prepared from salivary gland extracts of 3-week-old, MCMV-infected BALB/c mice. All mice used in the experiments were infected intraperitoneally with 10⁵ plaque-forming units of Smith MCMV and sacrificed at the indicated time points.

Cell preparation for flow cytometry analysis or cell sorting

Spleens or lymph nodes were harvested and submitted to enzymatic digestion for 25 min at 37 °C with Collagenase IV (Worthington biochemical) and DNase I (Roche Diagnostics). Organs were then mechanically digested and passed over 100-µm cell strainers (Corning). Bone marrow cells were flushed from mouse femurs. Red blood cells (RBCs) were then lysed by using RBC lysis buffer (Life Technologies) for spleen and bone marrow cell preparation. Livers were harvested, minced and submitted to enzymatic digestion, as for the spleen. Liver pieces were then crushed and the cell suspension obtained was washed 2× with phosphate-buffered saline (PBS) 1×, before performing a 80:40 Percoll gradient. Small intestines were harvested, opened longitudinally and then cut into 1-cm pieces. Pieces were washed extensively with PBS 1×, then incubated 3× at 37 °C on shaking (200 r.p.m.) with PBS 1× containing 2% fetal calf serum (FCS) and 5 mM EDTA. At the end of each incubation, supernatants were collected and centrifuged. Pelleted cells, mainly intraepithelial lymphocytes, from the three incubations were pooled together and submitted to a 67:44 Percoll gradient. Cells isolated from the middle ring of Percoll gradients were washed once with PBS 1×, then used for flow cytometry.

Flow cytometry analysis

Extracellular staining was performed in PBS 1× supplemented with 2 mM EDTA (Sigma-Aldrich) and 1% FCS. All extracellular staining was performed for at least 30 min at 4 °C. Dead cell staining (LIVE/DEAD Fixable Aqua Dead Cell Stain or Blue Dead Cell Stain, Life Technologies) was performed in PBS 1× according to the manufacturer’s recommendations. Samples were acquired with a FACS Fortessa X20 (BD Biosciences) or sorted with FACS ARIA III (BD Biosciences) using BD Diva v.9.0. All data were analyzed with FlowJo v.10.8.1 software. For the unsupervised analysis, t-SNE plots were generated with the dedicated plugin in FlowJo software. Dimensional reduction was performed on singlets/nonautofluorescent/live dead⁻/lineage⁻/CD19⁻/CD11c⁺ and/or SiglecH⁺ cells. Dimension reduction was calculated with the following markers: B220, BST2, CD11b, CD11c, CCR9, CX3CR1, Ly6D, SiglecH, tdT, XCR1 and Zbtb46-GFP. We used the HyperFinder plugin of FlowJo for the generation of unsupervised gating strategy. The calculation was performed on singlets/nonautofluorescent/live dead⁻ cells. We defined the pDC population for the calculation as being lineage⁻/CD19⁻/CD11b⁻/XCR1⁻/CD11c^low/BST2^high/tdT⁺. The software calculated the best and fastest gating strategy with the following markers: B220, BST2, CD11b, CD11c, CD19, CCR9, CX3CR1, lineage (CD3/Ly6G/NK1.1), Ly6D, SiglecH and XCR1.

Spectral flow cytometry analysis

All antibodies were purchased from Becton Dickinson Biosciences, BioLegend or eBioscience. Dead cells were discriminated in all experiments using Live/dead (ZombieNIR) fixable dead stain (Life Technologies). All staining was carried out with Fc-block (2.4G2 and 9E9) in staining buffer (PBS, 2 mM EDTA, 0.5% bovine serum albumin (BSA) and 20% Brilliant Violet stain buffer (BD Biosciences)). Cell suspensions were stained first for 15 min at 37 °C by monoclonal antibodies targeting chemokine receptors and certain antigens (CCR9, CD26, CD64, CX3CR1, SiglecH and XCR1) and for 25 min more by monoclonal antibodies targeting other antigens (B220, Bst2, CD45, CD3, CD8a, CD11b, CD11c, CD19, CD88, F4/80, IgD, IgM, Ly6C, Ly6D, Ly6G, NK1.1 and MHC-II). Cells were washed 3× in FACS buffer (PBS, 2 mM EDTA and 0.5% BSA) and resuspended in PBS. In all flow cytometric plots, doublets, aggregates, dead cells and autofluorescence were excluded. Acquisition was performed on an Aurora five lasers (Cytek Biosciences) using SpectroFlo v.3.0.1 software (Cytek Biosciences); the quality control (QC) indicates a similarity and complexity index of 11.7. Data were analyzed using OMIQ (app.OMIQ.ai). First, the data were run through flowCut⁵¹ to check for aberrant signal patterns or events. Second, the data were cleaned by manual gating to remove doublets, debris and dead cells. Third, the autofluorescence of the tissue was subtracted. UMAP⁵² was run to reduce the dimensions to a two-dimensional space and thus group phenotypically similar events into ‘islands’ to illustrate differences both between and inside each population. PARC (phenotyping by accelerated refined community)⁵³ was subsequently used to cluster the events based on UMAP parameters.

ScRNA-seq data generation

For the generation of the scRNA-seq data, we followed the FB5P method previously published²⁸. Briefly, single cells were FACS sorted into ice-cold 96-well PCR plates (Thermo Fisher Scientific) containing 2 µl of lysis mix per well. Immediately after cell sorting, each plate was covered with an adhesive film (Thermo Fisher Scientific), briefly spun down in a benchtop plate centrifuge and frozen on dry ice. The reverse transcription (RT) reaction was performed with SuperScript II (Thermo Fisher Scientific) in the presence of RNaseOUT (Thermo Fisher Scientific), dithiothreitol (Thermo Fisher Scientific), betaine (Sigma-Aldrich), MgCl₂ (Sigma-Aldrich) and well-specific template-switching oligonucleotide. For complementary DNA amplification, KAPA HiFi HotStart ReadyMix (Roche Diagnostics) was used with adapted primers. For library preparation, amplified cDNA from each well of a 96-well plate was pooled, purified with two rounds of 0.6× solid-phase reversible immobilization beads (AmpureXP, Beckman or CleanNGS, Proteigene) and finally eluted in nuclease-free water. After tagmentation and neutralization, tagmented cDNA was amplified with Nextera PCR Mastermix containing Nextera i5 primer (Illumina) and customized i7 primer mix. Libraries generated from multiple 96-well plates of single cells and carrying distinct i7 barcodes were pooled for sequencing on an Illumina NextSeq2000 platform, with 100 cycles of P2 flow cells, targeting 5 × 10⁵ reads per cell in paired-end, single-index mode with the following cycles: Read1 (Read1_SP, 67 cycles), Read i7 (i7_SP, 8 cycles) and Read2 (Read2_SP, 16 cycles). Two to three individual mice were used as a source for the single cells for each time point, with three independent sorts performed with two or three animals each time (sorts for mice nos. 56 and 58 on 3 November 2020, for nos. 52, 53 and 61 on 17 December 2020 and for nos. 81, 84 and 86 on 11 February 2021); sorting plates were frozen until all samples had been collected and all libraries were generated and sequenced simultaneously to avoid eventual batch effects.

Bioinformatics analyses of scRNA-seq data

A mark-up file of how the scRNA-seq analysis was performed is provided (Supplementary Fig. 2). FB5P-seq data were aligned and mapped to a reference genome using STAR (v.2.5.3a) and HTSeqCount (v.0.9.1) and processed to generate a single-cell, unique molecular identifier (UMI) counts matrix as described²⁸. The counts matrix was loaded to R (v.4.0.3) and Seurat (v.3.2.0)⁵⁴ was used for downstream analyses as described²⁷. Gene expression is shown as log(normalized values) and protein expression as inverse hyperbolic arcsine (asinh) of fluorescence intensity-scaled values (asinh(fluorescence)/100). For dimensionality reduction, we performed UMAP, using the RunUMAP function. The differentially expressed genes (DEGs) were determined using the FindMarkers function.

A convergent transcriptional reprogramming occurs in all DC types during their maturation, which can lead to their clustering primarily according to their activation states rather than to their cell types³¹. In particular, many of the genes that are specific for a given DC type at steady state are strongly downregulated on activation^30,31,32, such that they cannot be used individually to identify DC types³³. These phenomena recurrently caused issues for identifying DC types in the analyses of certain scRNA-seq data, which led to an inadequate inflation of DC subset nomenclature, with new DC subset names coined for clusters that actually corresponded to activation states of already identified DC types. This problem is exemplified by the current use of the name ‘DC3’ for mature DC clusters in scRNA-seq studies, whereas ‘DC3’ is also being used for a human DC type that is ontogenetically and functionally distinct from human pDCs, cDC1s, cDC2s and monocyte-derived DCs, as discussed in a recent commentary⁵⁵. Similar to the authors of this commentary, we emphasize, in scRNA-seq data analysis, the necessity for clearly distinguishing distinct DC types from different activation states of the same DC type. To achieve this aim, proper strategies are required for robust assignment of a cell-type identity to each individual cell, irrespective of its activation state, in a manner enabling resolution of the heterogeneity of mixed-cell clusters. This can be achieved by integrating phenotypic and transcriptomic data (as enabled with index sorting or CITE-seq³³) or by performing a gene set enrichment analysis at the level of individual cells, by using the single-cell CMap algorithm⁵⁶ with specific composite up/down gene modules carefully defined from an independent dataset. We used both strategies in the present study as a technical guide to help readers in their future analyses of scRNA-seq data.

A first Seurat analysis was performed only on cells from uninfected mice (345 cells after QC, 2 of which were removed due to lack of index sorting data). Clusters were identified based on either gene expression using Seurat (k-nearest neighbor = 5; resolution = 0.2) or phenotypic marker expression using Rphenograph⁵⁷ (v.0.99.1) (number of nearest neighbors k = 20), taking into account the protein expression for GFP, B220, Ly6D, XCR1, CX3CR1, BST2, SiglecH, CCR9, CD11b and CD11c (but not for tdTomato). A CMap analysis⁵⁶ was performed on this dataset using cell-type-specific signatures (tDCs, cDC2s, cDC1s and pDCs) established on reanalysis with the BubbleGUM GeneSign module⁵⁸ of a published²⁹, independent, bulk RNA-seq dataset (Gene Expression Omnibus (GEO) accession no. GSE76132), and a relative pDC_vs_pDClike signature retrieved from a previously published study⁴. The GSE76132 dataset encompassed four cell types, tDCs, cDC2s, cDC1s and pDCs, analyzed through bulk RNA-seq. We retrieved and reanalyzed these data to generate signatures for each DC type compared with the three others. Hence, we ended up with four composite signatures, one for each DC type, with its ‘UP’ and ‘DN’ genes when compared with the three other DC types. The study by Rodrigues et al.⁴ encompassed only two cell types, pDCs and pDC-like cells, analyzed through scRNA-seq. Hence, from this dataset, we retrieved one composite signature encompassing the genes expressed higher in pDCs above pDC-like cells (‘UP’) and reciprocally the genes less expressed in pDCs than in pDC-like cells (‘DN’). The CMap algorithm has been made available on Github (https://github.com/SIgN-Bioinformatics/sgCMAP_R_Scripts), as well as an example and recommendations on how to use it (https://github.com/DalodLab/MDlab_cDC1_differentiation/blob/main/scRNAseq_pipeline.md#cell-annotation-using-cmap).

Integration of Seurat and Rphenograph cluster information together with the CMap scores allowed identification of, and focus on, bona fide steady-state pDCs, pDC-like cells, CD11c^high tDCs, cDC1s and cDC2s (205 total cells), from which we computed relative signatures based on all pairwise comparisons between cell types, using the FindMarkers function (default parameters), for consecutive single-cell CMap analyses.

A second Seurat analysis was then performed on cells from both uninfected and MCMV-infected mice (1,132 cells after QC). Clusters of contaminating cell types (macrophages, NK cells and a small cluster of proliferative cells of mixed types) were identified by their top markers using the FindMarkers function (test.use = ‘bimod’) and removed (181 cells). A third Seurat analysis was performed on the remaining cells (951 cells).

Clusters were identified based on gene expression using Seurat (k-nearest neighbor = 9; resolution = 0.7) or based on phenotypic marker expression using Rphenograph (number of nearest neighbors k = 50). Integration of Seurat and Rphenograph clusters allowed identification of unambiguous cell types and activation states (851 total cells), which was corroborated on performing a single-cell CMap analysis using the relative signatures identified in the previous step. DEGs between cell types at equivalent activation states or between activation states for a given cell type were then extracted, using the FindMarkers function (default parameters, threshold for adjusted P < 0.05) (see Supplementary Table).

The heatmaps were plotted using the Gene-E program.

Immunohistofluorescence, microscopy and image analysis

Organs were fixed with Antigen Fix (Diapath) for 2 h for the small intestine, colon and lymph nodes or 4 h for the spleen at 4 °C, and then washed several times in phosphate buffer (PB; 0.025 M NaH₂PO₄ and 0.1 M Na₂HPO₄). Organs were then immersed in a solution of 30% sucrose overnight at 4 °C. Organs were then embedded in OCT (optimal cutting temperature; Sakura), snap frozen and stored at −80 °C. Cryosections, 20 μm, were performed using a microtome (Leica 3050s Cryostat) at temperatures between −20 °C and −22 °C. For immunostaining, sections were blocked with PB, 0.1% Triton X-100 and 2% BSA for 30 min at room temperature and then stained overnight at 4 °C with primary antibodies (Supplementary Table) diluted in PB, 0.1% Triton X-100 and 2% BSA. After several washings with PB, sections were then stained in PB, 0.1% Triton X-100 and 2% BSA with secondary antibodies (Supplementary Table) for 2 h at 4 °C. To analyze YFP signal, after the incubation with secondary antibodies, sections were washed and incubated with PB, 0.1% Triton X-100, 2% BSA and 5% rabbit serum for 30 min at room temperature. Sections were then stained with anti-GFP antibodies directly labeled with Alexa 488 for 2 h at 4 °C. Finally, sections were washed with PB and mounted with a coverslip and Prolong Antifade Gold mounting medium (Life Technologies). Whole sections were acquired by spectral confocal microscope (Zeiss LSM 880) with ×20 or ×40 objectives. Pictures were then analyzed using ImageJ v.1.52p software, including through the development of specific macros (https://github.com/Imagimm-CIML/Micro-anatomical-location-of-splenic-IFNpos-vs-IFNneg-pDC-). The TCZ was defined as a CD3-rich region within the WP and the BCZ as a B220-rich area within the WP. The RP was defined as an F4/80-rich region and the MZ as the space between the CD169 staining and the F4/80 staining. For the calculation of pDC counts mm⁻², the function ‘analyze particles’ was used and a threshold for tdT intensity and the size (>8 μm²) was applied. For the quantification of the repartition of pDCs on infection, the intensity of tdT was calculated in the different zones after a threshold for the intensity had been applied. A ratio between the intensity in each zone and the total intensity of the whole section was then calculated.

Cell immunofluorescence and cell morphology analysis

The different cell subsets were FACS sorted in a cold 5-ml FACS tube containing 1 ml of RPMI (Roswell Park Memorial Institute) medium supplemented with 10% FCS, 1% l-glutamine (Gibco), 100 U ml⁻¹ of penicillin–streptomycin, 1% nonessential amino acids, 1% sodium pyruvate and 0.05 mM 2-mercaptoethanol. After washing in PBS, cells were resuspended in RPMI and 1% FCS. Then, 1,000–50,000 cells were allowed to adhere for 1 h at 37 °C to coverslips coated with 5 μg cm⁻² of poly(d-lysine) (Sigma-Aldrich). Cells were then washed in PBS, fixed in 4% paraformaldehyde for 10 min and permeabilized/blocked in PBS, 0.2% Triton X-100, 2% FCS, 2% rat serum, 2% goat serum and 2% donkey serum. F-actin staining was performed at 4 °C overnight in PBS, 0.1% Triton X-100 and 2% FCS with phalloidin AF405plus (Life Technologies). Samples were washed in PBS and mounted with Prolong Antifade Gold mounting medium. Images were acquired by spectral confocal microscope (Zeiss LSM 880) with a ×63 objective and analyzed with ImageJ v.1.52p software. For the cell morphology, a binary image was created for each individual cell based on the F-actin staining. The circularity index ((4π × area)/(perimeter × 2)) was then calculated with the adequate function of the software.

RT-qPCR

For each indicated cell type, 100 cells were FACS sorted into ice-cold, 96-well PCR plates (Thermo Fisher Scientific) containing 10 µl of TCL buffer (QIAGEN) supplemented with 1% 2-mercaptoethanol (Invitrogen). Cell lysates were sealed, vortexed and spun down at 300g and 4 °C for 1 min, immediately placed on dry ice and transferred for storage at −80 °C. The SS2 protocol was performed on cell lysate as previously described²⁷. Briefly, plates containing cell lysates were thawed on ice, followed by RNA purification and first-strand synthesis. Complementary DNA was amplified by PCR for 22 cycles. Serial dilutions of the cDNA obtained were used to perform quantitative (q)PCR with the ONeGreenFAST kit (Ozyme) and run on a 7500 Real Time PCR System apparatus (Applied Biosystems). Relative gene expression was calculated using the ΔΔ^Ct method with Actb as housekeeping gene for normalization. The primers used were as follows: Actb forward 5′-GGCTGTATTCCCCTCCATCG-3′; reverse 5′-CCAGTTGGTAACAATGCCATGT-3′; Xcr1 forward 5′-CCTACGTGAAACTCTAGCACTGG-3′; reverse 5′-AAGGCTGTAGAGGACTCCATCTG-3′; Siglech forward 5′-GGAGGCAAAACATGGAATTTATG-3′; reverse 5′-CACATCACATTGGTAGGACGAC-3′; Tmem176a forward 5′-GCCGGATGCTCATTGCTAAG-3′; reverse 5′-ATGGCCTATGTAGAGGGTTCC-3′; Tmem176b forward 5′-CAGTCCGCTCACATCAGCAT-3′; reverse 5′-GCTGCCCATAGTGGATTCTGG-3′; Apod forward 5′-TCACCACAGCCAAAGGACAAA-3′; reverse 5′-CGTTCTCCATCAGCGAGTAGT-3′.

Statistical analysis

No statistical methods were used to predetermine sample sizes but our sample sizes are similar to those reported in previous publications^20,27. Data distribution was assumed to be normal but this was not formally tested. All quantifications were performed with awareness of experimental groups, meaning not in a blinded fashion. Animals were matched in age and gender between experimental groups, without randomization. No animals or data were excluded. Statistical parameters including the definitions and exact value of n (number of biological replicates and total number of experiments) and the types of statistical tests are reported in the figures and corresponding legends. Statistical analyses were performed using Prism v.8.1.2 (GraphPad Software) or R v.4.0.3 statistical programming language. Statistical analysis was conducted on data with at least three biological replicates. Comparisons between groups were planned before statistical testing and target effect sizes were not predetermined. Error bars displayed on graphs represent the mean ± s.e.m. Statistical significance was defined as: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The scRNA-seq data have been deposited in the GEO repository under accession no. GSE196720. All other data generated or analyzed during the present study are included in this report (and its Supplementary Information files). Source data are provided with this paper.

References

Tomasello, E., Pollet, E., Vu Manh, T. P., Uze, G. & Dalod, M. Harnessing mechanistic knowledge on beneficial versus deleterious IFN-I effects to design innovative immunotherapies targeting cytokine activity to specific cell types. Front. Immunol. 5, 526 (2014).
Article PubMed PubMed Central Google Scholar
King, C. & Sprent, J. Dual nature of type I Interferons in SARS-CoV-2-induced inflammation. Trends Immunol. 42, 312–322 (2021).
Article CAS PubMed PubMed Central Google Scholar
Reizis, B. Plasmacytoid dendritic cells: development, regulation, and function. Immunity 50, 37–50 (2019).
Article CAS PubMed PubMed Central Google Scholar
Rodrigues, P. F. et al. Distinct progenitor lineages contribute to the heterogeneity of plasmacytoid dendritic cells. Nat. Immunol. 19, 711–722 (2018).
Article CAS PubMed PubMed Central Google Scholar
Dress, R. J. et al. Plasmacytoid dendritic cells develop from Ly6D⁺ lymphoid progenitors distinct from the myeloid lineage. Nat. Immunol. 20, 852–864 (2019).
Article CAS PubMed Google Scholar
Leylek, R. et al. Integrated cross-species analysis identifies a conserved transitional dendritic cell population. Cell Rep. 29, 3736–3750.e3738 (2019).
Article CAS PubMed PubMed Central Google Scholar
Zucchini, N. et al. Individual plasmacytoid dendritic cells are major contributors to the production of multiple innate cytokines in an organ-specific manner during viral infection. Int. Immunol. 20, 45–56 (2008).
Article CAS PubMed Google Scholar
Anderson, D. A. 3rd & Murphy, K. M. Models of dendritic cell development correlate ontogeny with function. Adv. Immunol. 143, 99–119 (2019).
Article CAS PubMed Google Scholar
Dalod, M. & Scheu, S. Dendritic cell functions in vivo: a user’s guide to current and next- generation mutant mouse models. Eur. J. Immunol. 52, 1712–1749 (2022).
Swiecki, M. et al. Cell depletion in mice that express diphtheria toxin receptor under the control of SiglecH encompasses more than plasmacytoid dendritic cells. J. Immunol. 192, 4409–4416 (2014).
Article CAS PubMed Google Scholar
Brewitz, A. et al. CD8⁺ T cells orchestrate pDC-XCR1⁺ dendritic cell spatial and functional cooperativity to optimize priming. Immunity 46, 205–219 (2017).
Article CAS PubMed PubMed Central Google Scholar
Puttur, F. et al. Absence of Siglec-H in MCMV infection elevates interferon alpha production but does not enhance viral clearance. PLoS Pathog. 9, e1003648 (2013).
Article CAS PubMed PubMed Central Google Scholar
Wohner, M. et al. Molecular functions of the transcription factors E2A and E2-2 in controlling germinal center B cell and plasma cell development. J. Exp. Med. 213, 1201–1221 (2016).
Article PubMed PubMed Central Google Scholar
Swiecki, M., Gilfillan, S., Vermi, W., Wang, Y. & Colonna, M. Plasmacytoid dendritic cell ablation impacts early interferon responses and antiviral NK and CD8⁺ T cell accrual. Immunity 33, 955–966 (2010).
Article CAS PubMed PubMed Central Google Scholar
Stutte, S. et al. Type I interferon mediated induction of somatostatin leads to suppression of ghrelin and appetite thereby promoting viral immunity in mice. Brain Behav. Immun. 95, 429–443 (2021).
Article CAS PubMed Google Scholar
Luche, H., Weber, O., Nageswara Rao, T., Blum, C. & Fehling, H. J. Faithful activation of an extra-bright red fluorescent protein in ‘knock-in’ Cre-reporter mice ideally suited for lineage tracing studies. Eur. J. Immunol. 37, 43–53 (2007).
Article CAS PubMed Google Scholar
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).
Article CAS PubMed Google Scholar
Robbins, S. H. et al. Novel insights into the relationships between dendritic cell subsets in human and mouse revealed by genome-wide expression profiling. Genome Biol. 9, R17 (2008).
Article PubMed PubMed Central Google Scholar
Esashi, E., Bao, M., Wang, Y. H., Cao, W. & Liu, Y. J. PACSIN1 regulates the TLR7/9-mediated type I interferon response in plasmacytoid dendritic cells. Eur. J. Immunol. 42, 573–579 (2012).
Article CAS PubMed PubMed Central Google Scholar
Tomasello, E. et al. Molecular dissection of plasmacytoid dendritic cell activation in vivo during a viral infection. EMBO J. 37, e98836 (2018).
Article PubMed PubMed Central Google Scholar
Bar-On, L. et al. CX3CR1+ CD8alpha+ dendritic cells are a steady-state population related to plasmacytoid dendritic cells. Proc. Natl Acad. Sci. USA 107, 14745–14750 (2010).
Article CAS PubMed PubMed Central Google Scholar
Naik, S. H. 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).
Article CAS PubMed Google Scholar
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).
Article CAS PubMed Google Scholar
Fogg, D. K. et al. A clonogenic bone marrow progenitor specific for macrophages and dendritic cells. Science 311, 83–87 (2006).
Article CAS PubMed Google Scholar
Auffray, C. et al. CX3CR1+ CD115+ CD135+ common macrophage/DC precursors and the role of CX3CR1 in their response to inflammation. J. Exp. Med. 206, 595–606 (2009).
Article CAS PubMed PubMed Central Google Scholar
Satpathy, A. T. et al. Zbtb46 expression distinguishes classical dendritic cells and their committed progenitors from other immune lineages. J. Exp. Med. 209, 1135–1152 (2012).
Article CAS PubMed PubMed Central Google Scholar
Abbas, A. et al. The activation trajectory of plasmacytoid dendritic cells in vivo during a viral infection. Nat. Immunol. 21, 983–997 (2020).
Article CAS PubMed PubMed Central Google Scholar
Attaf, N. et al. FB5P-seq: FACS-based 5-prime end single-cell RNA-seq for integrative analysis of transcriptome and antigen receptor repertoire in B and T cells. Front. Immunol. 11, 216 (2020).
Article CAS PubMed PubMed Central Google Scholar
Lau, C. M. et al. Leukemia-associated activating mutation of Flt3 expands dendritic cells and alters T cell responses. J. Exp. Med. 213, 415–431 (2016).
Article CAS PubMed PubMed Central Google Scholar
Crozat, K. et al. Cutting edge: expression of XCR1 defines mouse lymphoid-tissue resident and migratory dendritic cells of the CD8alpha+ type. J. Immunol. 187, 4411–4415 (2011).
Article CAS PubMed Google Scholar
Manh, T. P., 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).
Article CAS PubMed Google Scholar
Ardouin, L. et al. Broad and largely concordant molecular changes characterize tolerogenic and immunogenic dendritic cell maturation in thymus and periphery. Immunity 45, 305–318 (2016).
Article CAS PubMed Google Scholar
Maier, B. et al. A conserved dendritic-cell regulatory program limits antitumour immunity. Nature 580, 257–262 (2020).
Article CAS PubMed PubMed Central Google Scholar
Blasius, A. L., Cella, M., Maldonado, J., Takai, T. & Colonna, M. Siglec-H is an IPC-specific receptor that modulates type I IFN secretion through DAP12. Blood 107, 2474–2476 (2006).
Article CAS PubMed PubMed Central Google Scholar
Contractor, N., Louten, J., Kim, L., Biron, C. A. & Kelsall, B. L. Cutting edge: Peyer’s patch plasmacytoid dendritic cells (pDCs) produce low levels of type I interferons: possible role for IL-10, TGFbeta, and prostaglandin E2 in conditioning a unique mucosal pDC phenotype. J. Immunol. 179, 2690–2694 (2007).
Article CAS PubMed Google Scholar
Wendland, M. et al. CCR9 is a homing receptor for plasmacytoid dendritic cells to the small intestine. Proc. Natl Acad. Sci. USA 104, 6347–6352 (2007).
Article CAS PubMed PubMed Central Google Scholar
Bonnardel, J. et al. Innate and adaptive immune functions of peyer’s patch monocyte-derived cells. Cell Rep. 11, 770–784 (2015).
Article CAS PubMed Google Scholar
Da Silva, C., Wagner, C., Bonnardel, J., Gorvel, J. P. & Lelouard, H. The Peyer’s patch mononuclear phagocyte system at steady state and during infection. Front. Immunol. 8, 1254 (2017).
Article PubMed PubMed Central Google Scholar
Scheu, S., Dresing, P. & Locksley, R. M. Visualization of IFNbeta production by plasmacytoid versus conventional dendritic cells under specific stimulation conditions in vivo. Proc. Natl Acad. Sci. USA 105, 20416–20421 (2008).
Article CAS PubMed PubMed Central Google Scholar
Dalod, M. et al. Interferon alpha/beta and interleukin 12 responses to viral infections: pathways regulating dendritic cell cytokine expression in vivo. J. Exp. Med. 195, 517–528 (2002).
Article CAS PubMed PubMed Central Google Scholar
Assil, S. et al. Plasmacytoid dendritic cells and infected cells form an interferogenic synapse required for antiviral responses. Cell Host Microbe 25, 730–745.e736 (2019).
Article CAS PubMed Google Scholar
Megjugorac, N. J. et al. Image-based study of interferongenic interactions between plasmacytoid dendritic cells and HSV-infected monocyte-derived dendritic cells. Immunol. Invest. 36, 739–761 (2007).
Article CAS PubMed Google Scholar
Asselin-Paturel, C., Brizard, G., Pin, J. J., Briere, F. & Trinchieri, G. Mouse strain differences in plasmacytoid dendritic cell frequency and function revealed by a novel monoclonal antibody. J. Immunol. 171, 6466–6477 (2003).
Article CAS PubMed Google Scholar
Blasius, A. L. et al. Bone marrow stromal cell antigen 2 is a specific marker of type I IFN-producing cells in the naive mouse, but a promiscuous cell surface antigen following IFN stimulation. J. Immunol. 177, 3260–3265 (2006).
Article CAS PubMed Google Scholar
Alexandre, Y. O., Cocita, C. D., Ghilas, S. & Dalod, M. Deciphering the role of DC subsets in MCMV infection to better understand immune protection against viral infections. Front. Microbiol. 5, 378 (2014).
Article PubMed PubMed Central Google Scholar
Ghilas, S. et al. Natural killer cells and dendritic epidermal gammadelta T cells orchestrate type 1 conventional DC spatiotemporal repositioning toward CD8⁺ T cells. iScience 24, 103059 (2021).
Article CAS PubMed PubMed Central Google Scholar
Feng, J. et al. Clonal lineage tracing reveals shared origin of conventional and plasmacytoid dendritic cells. Immunity 55, 405–422 e411 (2022).
Article CAS PubMed PubMed Central Google Scholar
Lancien, M. et al. Dendritic cells require TMEM176A/B ion channels for optimal MHC class II antigen presentation to naive CD4⁺ T cells. J. Immunol. 207, 421–435 (2021).
Article CAS PubMed Google Scholar
Segovia, M. et al. Autologous dendritic cells prolong allograft survival through Tmem176b-dependent antigen cross-presentation. Am. J. Transpl. 14, 1021–1031 (2014).
Article CAS Google Scholar
Segovia, M. et al. Targeting TMEM176B enhances antitumor immunity and augments the efficacy of immune checkpoint blockers by unleashing inflammasome activation. Cancer Cell 35, 767–781.e766 (2019).
Article CAS PubMed PubMed Central Google Scholar
Meskas, J., Yokosawa, D., Wang, S., Segat, G.C. & Brinkman, R.R. flowCut: An R package for automated removal of outlier events and flagging of files based on time versus fluorescence analysis. Cytometry A 103, 71–81 (2023).
Article PubMed Google Scholar
Becht, E. et al. Dimensionality reduction for visualizing single-cell data using UMAP. Nat. Biotechnol. 37, 38–44 (2019).
Stassen, S. V. et al. PARC: ultrafast and accurate clustering of phenotypic data of millions of single cells. Bioinformatics 36, 2778–2786 (2020).
Article CAS PubMed PubMed Central Google Scholar
Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).
Article CAS PubMed PubMed Central Google Scholar
Ginhoux, F., Guilliams, M. & Merad, M. Expanding dendritic cell nomenclature in the single-cell era. Nat. Rev. Immunol. 22, 67–68 (2022).
Article CAS PubMed Google Scholar
See, P. et al. Mapping the human DC lineage through the integration of high-dimensional techniques. Science 356, eaag3009 (2017).
Article PubMed PubMed Central Google Scholar
Levine, J. H. et al. Data-driven phenotypic dissection of AML reveals progenitor-like cells that correlate with prognosis. Cell 162, 184–197 (2015).
Article CAS PubMed PubMed Central Google Scholar
Spinelli, L., Carpentier, S., Montanana Sanchis, F., Dalod, M. & Vu Manh, T. P. BubbleGUM: automatic extraction of phenotype molecular signatures and comprehensive visualization of multiple Gene Set Enrichment Analyses. BMC Genomics 16, 814 (2015).
Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank CIPHE for the generation of Siglech^iCre and Pacsin1^LSL-tdT mice and their assistance in the breeding of mice, as well as the staff of the CIML mouse houses, flow cytometry, histology, genomics bioinformatics and imaging (ImagImm) core facilities. We thank S. Henri (CIML) for the generous gift of the Zbtb46^GFP mice. We thank the two master trainees, A. Trinh and D. Anselmo, as well as M. Toxé, for technical assistance in some experiments. We thank C. Dong for help in the preprocessing of the FB5P-seq data. We thank the Centre de Calcul Intensif d’Aix-Marseille for granting access to its high-performance computing resources. We thank M. Bajénoff for discussions and advice. This research was funded by grants from the Fondation pour la Recherche Médicale (nos. DEQ20110421284 and DEQ20180339172, Equipe Labellisée, to M.D.) and from the French National Research Agency (ANR, no. ANR-21-CO12-0001-01, ‘RIPCOV’ to M.D. and no. ANR-21-CE15-0044-01, ‘DECITIP’ to E.T.). We also thank the DCBIOL Labex (ANR-11-LABEX-0043, grant no. ANR-10-IDEX-0001-02 PSL*), the A*MIDEX project (grant no. ANR-11-IDEX-0001-02), funded by the French government’s Investissements d’Avenir program managed by the ANR, and institutional support from CNRS (French National Centre for Scientific Research), INSERM (French National Institute of Health and Medical Research) and Aix-Marseille Université and Marseille Immunopole. This work was supported by the ANR through the Investments for the Future program (France-BioImaging, no. ANR-10-INBS-04).

Author information

Michael Valente
Present address: Veracyte, Luminy biotech entreprises, Marseille, France
These authors jointly supervised this work: Elena Tomasello, Marc Dalod.

Authors and Affiliations

Aix-Marseille University, CNRS, INSERM, CIML, Centre d’Immunologie de Marseille-Luminy, Turing Center for Living Systems, Marseille, France
Michael Valente, Nils Collinet, Thien-Phong Vu Manh, Dimitri Popoff, Khalissa Rahmani, Karima Naciri, Gilles Bessou, Rejane Rua, Laurine Gil, Cyrille Mionnet, Pierre Milpied, Elena Tomasello & Marc Dalod

Authors

Michael Valente
View author publications
You can also search for this author in PubMed Google Scholar
Nils Collinet
View author publications
You can also search for this author in PubMed Google Scholar
Thien-Phong Vu Manh
View author publications
You can also search for this author in PubMed Google Scholar
Dimitri Popoff
View author publications
You can also search for this author in PubMed Google Scholar
Khalissa Rahmani
View author publications
You can also search for this author in PubMed Google Scholar
Karima Naciri
View author publications
You can also search for this author in PubMed Google Scholar
Gilles Bessou
View author publications
You can also search for this author in PubMed Google Scholar
Rejane Rua
View author publications
You can also search for this author in PubMed Google Scholar
Laurine Gil
View author publications
You can also search for this author in PubMed Google Scholar
Cyrille Mionnet
View author publications
You can also search for this author in PubMed Google Scholar
Pierre Milpied
View author publications
You can also search for this author in PubMed Google Scholar
Elena Tomasello
View author publications
You can also search for this author in PubMed Google Scholar
Marc Dalod
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

M.V. designed, performed and analyzed most of the experiments and wrote the manuscript. N.C. performed part of the microscopy experiments, genotyping of mutant mice, and generation and stabilization of SCRIPT mice in the triple homozygous state. T.-P.V.M. performed the bioinformatics analysis and contributed to writing the manuscript. D.P. and C.M. contributed to design, and performed and analyzed the spectral flow cytometry experiments. K.R. contributed to performing and analyzing the micro-bulk RT-qPCR experiments. R.R. gave advice on the spectral flow cytometry experiments. K.N. performed the crosses and genotyping for the generation and stabilization of the S-RFP and pDC-Tom mouse strains. G.B. contributed to the genotyping of mutant mice and performed the crosses and genotyping for the generation and stabilization of the ZeST mice. L.G. trained M.V. for the generation of the FB5P-seq data. P.M. coordinated the generation and first-level analysis of the FB5P-seq data. E.T. directed and contributed to funding the study, designed experiments, performed and analyzed some experiments and wrote the manuscript. M.D. directed and funded the study, designed experiments, contributed to the bioinformatics analyses and wrote the manuscript. All authors contributed to the revision of the manuscript.

Corresponding authors

Correspondence to Michael Valente, Elena Tomasello or Marc Dalod.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Immunology thanks Roxane Tussiwand, Meredith O’Keeffe and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. L. A. Dempsey was the primary editor on this article and managed its editorial process and peer review in collaboration with the rest of the editorial team. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Expression of RFP and tdT in different immune lineages in S-RFP and pDC-Tom mice.

a, Gating strategy used for the identification of the different immune cell types studied. b, c, Splenocytes were isolated from S-RFP (b) or pDC-Tom (c) animals, stained with fluorescently labeled antibodies and analyzed by flow cytometry. The expression of RFP (b, red histograms) or tdT (c, orange histograms) was evaluated in different immune cell types as indicated. d, The expression of Lineage cocktail antibodies and CD11b was assessed on tdT⁺ cells and compared to Live non-autofluorescent cells (Red) (Lineage, left panel) or cDC2 (Blue) (CD11b, right panel). C57BL/6 mice were used as negative controls (black histograms). The data shown are from one mouse representative of 8 animals for (b) and 6 animals for (c, d) from 2 independent experiments.

Extended Data Fig. 2 Gating strategy for bone marrow precursor analysis of pDC-Tom mice.

Bone marrow cells were isolated from pDC-Tom animals, stained with fluorescently labeled antibodies and analyzed by flow cytometry. The gating strategy used to identify myeloid progenitors (a) or lymphoid progenitors (b) is depicted. The data shown are from one mouse representative of 5 pDC-Tom animals from 2 independent experiments.

Extended Data Fig. 3 The phenotype and the proportions of Zbtb46⁺ Ly6D⁺ cells, CD11c^high tDC and pDC-like cells are relatively stable during MCMV infection.

Splenocytes from ZeST mice were stained with fluorescently labeled antibodies and analyzed by flow cytometry. Representative contour plots from uninfected animals (a-c), MCMV-infected mice at 36 h p.i (d-f) and from MCMV-infected mice at 48 p.i (g-i). The data shown are from one ZeST mouse representative of at least 10 uninfected animals (a-c), and for 7 MCMV-infected animals at 36 h p.i (d-f) or 8 MCMV-infected animals at 48 h p.i. (g-i).

Extended Data Fig. 4 Spectral flow cytometry-based unsupervised characterization of the expression pattern of tdT and GFP in splenocytes of ZeST mice.

a, Unsupervised dimensional reduction, and cell clustering, for the analysis by spectral flow cytometry of the expression of surface markers on CD45⁺ splenocytes from ZeST mice and one control C57BL/6 mouse, without considering tdT and GFP signals. The UMAP was calculated for all samples together, with downsampling to 200,000 CD45⁺ cells/sample, and is shown for each infection time point with individual mice pooled together (ZeST not infected, n = 2; 36 h, n = 4; 48 h, n = 3; control C57BL/6 mouse, not infected, n = 1). The cluster color code represents related cell types in similar colors (lower right, see also panel b). b, Cell cluster annotation for cell type identities, based on cell surface marker expression. The mean fluorescence intensity (MFI) of each cluster for each marker was calculated by averaging cluster MFI values across the 10 mice analyzed, and used for hierarchical clustering (heatmap). The corresponding expression patterns were then used to assign cell cluster to indicated cell types. c, Percent of each cluster within GFP⁺ cells, shown first as mean percent of each cluster within GFP⁺ cells across all ZeST mice, and then for each mouse. The color code of the clusters is the same as in (a, b) and their ordering in each pie chart shown in the color key (lower right). d, GFP MFI (mean on scaled data, Y-axis) for each ZeST mouse (one dot per mouse) and each cluster (X-axis, with same color code and ordering as in (a, c, d). For each cluster, the mean ± SEM of the MFI across all ZeST mice is shown as the black lines. For comparison, the autofluorescent signal in the GFP channel in each cell cluster in the C57BL/6 mouse is shown as a black square. e, Percent of each cluster within tdT⁺ cells, designed as in (c). f, tdT MFI for each cluster and each ZeST mouse, designed as in (d). The data shown are from one experiment representative of two independent ones.

Source data

Extended Data Fig. 5 Spectral flow cytometry-based supervised analysis of the expression of tdTomato and GFP in splenocytes of ZeST mice.

a, Gating strategy followed to identify the indicated lymphoid vs myeloid cell populations. Splenocytes were isolated from uninfected or MCMV-infected ZeST mice. Cells were previously gated as CD45⁺ after exclusion of dead cells and doublets. b, c, The expression of CD11c (b), GFP (c) and of CD26 vs CD64 (d) was analyzed on Ly6C⁺ tDC cells (orange), Ly6C^neg tDC (purple) and CD26^neg Ly6C^neg cells gated as shown in (a). e, CD11c^high tDC (purple) and pDC-like cells (green) were gated from the spectral flow cytometry dataset according to the strategy used with the initial regular flow cytometry dataset (as shown in Fig. 3b), and then analyzed for CD26 vs Ly6C expression, for comparison with the gating strategy used to identify Ly6C^neg tDC and Ly6C⁺ tDC (panel a). f, Analysis of the expression of GFP vs the indicated cell markers upon gating on Ly6D^neg CX₃CR1^neg cells. g, Percentages (mean ± SD) of GFP⁺ cells (top) or tdT⁺ cells (bottom) in indicated cell populations gated as shown in (a). For (a) and (f), the analysis shown is representative one ZeST mouse out of 9 tested. For (b-e) n = 2 for NI, n = 4 for MCMV 36 h and n = 3 for MCMV 48 h.

Source data

Extended Data Fig. 6 Analysis of the FB5P-seq data for cells from uninfected mice to generate DC type-specific signatures for helping annotation of the total FB5P-seq dataset.

a, UMAP dimensionality reduction of the gene expression data for DC types isolated from the spleen of 3 uninfected ZeST mice. Cells were index sorted into the 5 DC types studied (see Fig. 3b), and used for single cell RNA sequencing. After quality controls, 343 cells were kept for the analysis. The color code indicates belonging of the cells to the 7 Seurat clusters obtained. b, Projection onto the UMAP space of the phenotype of cells based on their belonging to the Rphenograph clusters (color code) obtained upon re-analysis of the fluorescent signals for 10 of the phenotypic markers acquired during index sorting as listed in panel c. c, Annotation of the Rphenograph clusters for DC type identity based on mean fluorescent intensities per marker and cluster as shown on the heatmap. d, Number of cells belonging to the intersections between Seurat (rows) and Rphenograph (column) clusters. Seurat clusters were annotated for cell types based on analysis of their marker genes (see Supplementary Table). e, Heatmap showing sgCMAP scores of individual cells (rows) for DC type-specific signatures (columns) generated from published independent RNA-seq datasets. Hierarchical clustering was performed, using the Pearson’s minus one metric for signatures (columns), and the Euclidian distance for individual cells (rows), with annotation of individual cells (rows) for i) belonging to Seurat clusters, ii) belonging to Rphenograph clusters, iii) sorting phenotype, and iv) final cell type assignment. 205 cells were assigned a cell type identity based on consistency between their belonging to Seurat and Rphenograph clusters (panel d) and their sgCMAP scores. 138 cells were left non-annotated (NA). f, Projection onto the UMAP space of the final cell type assignment.

Extended Data Fig. 7 Analysis of the FB5P-seq data for all cells from uninfected and infected mice.

a, UMAP dimensionality reduction of the gene expression data for DC types isolated from the spleens of eight ZeST mice (3 uninfected, 3 MCMV-infected for 36 h and 2 infected for 48 h, see graphical legend). Cells were index sorted into the 5 DC types studied (see Fig. 3b), and used for single cell RNA sequencing. After quality controls, 951 cells were kept for the analysis. b, Projection onto the UMAP space of the 13 Seurat clusters obtained (see graphical legend). c, Projection onto the UMAP space of the phenotype of cells based on their belonging to the Rphenograph clusters (color code) obtained upon re-analysis of the fluorescent signals for 10 of the phenotypic markers acquired during index sorting as listed in panel d. d, Annotation of the Rphenograph clusters for DC type identity based on mean fluorescent intensities per marker and cluster as shown on the heatmap. e, Number of cells belonging to the intersections between Seurat (rows) and Rphenograp (column) clusters. Seurat clusters were annotated for cell types based on analysis of their marker genes (see Supplementary Table 2). f, Heatmap showing sgCMAP scores of individual cells (rows) for the DC type-specific sgCMAP signatures (columns) generated from the analysis of the data focused on cells from uninfected mice (see Extended Data Fig. 6). Hierarchical clustering was performed, using the Pearson’s minus one metric for signatures (columns), and the Euclidian distance for individual cells (rows), with annotation of individual cells (rows) for i) belonging to Seurat clusters, ii) belonging to Rphenograph clusters, iii) sorting phenotype, and iv) final cell type assignment. 851 cells were assigned a cell type identity based on consistency between their belonging to Seurat and Rphenograph clusters (panel e), which was well corroborated by the sgCMAP scores. 100 cells were left non-annotated (NA).

Extended Data Fig. 8 pDC-Tom mice allow to determine pDC micro-anatomical location in the different organs.

20μm organ cryosections from uninfected pDC-Tom mice were stained with anti-tdT (magenta), anti-CD169 (white), anti-CD3 (green), anti-CD11c (red) and anti-BST2 (cyan) antibodies (a) or with anti-tdT (magenta), anti-Epcam (white), anti-CD3 (green) and anti-B220 (cyan) antibodies (b-d). Representative images are shown for LN of 3 mice (a), small intestine of 4 mice (b, c) and colon of 3 mice (d).

Extended Data Fig. 9 Circularity indices for cDC1 and cDC2 upon MCMV infection.

a, b, Quantitative and unbiased assessment of the cellular morphology of cDC1 and cDC2 isolated from 60 h MCMV-infected SCRIPT mice, as compared to cDC1 and cDC2 from uninfected mice. a, One representative confocal microscopy image of each DC type is shown. b, The distribution of the circularity indices for individual cells across DC types is shown as dots, with the overlaid color bars showing the mean circularity indices of each DC type. The data shown are from 2 independent experiments, each performed with one mouse, with 35 to 46 individual cells analyzed for each DC type as indicated below the graph. The data are shown as mean ± SEM.

Source data

Extended Data Fig. 10 Expression of selected phenotypic markers and genes across DC types and activation states.

a, Violin plots showing the expression of phenotypic markers across DC types and activation states. b, Violin plot showing expression of the Xcr1 gene across DC types and activation states. c, Violin plots showing mRNA expression levels of selected genes across DC types and activation states. d, Mini-bulk qRT-PCR analysis of the expression level of control (Siglech and Xcr1) and new candidate marker genes (Tmem176a, Tmem176b and Apod) across DC types and activation states. e,Violin plots showing mRNA expression levels of Il18 and Il12b across DC types and activation states.

Source data

Supplementary information

Supplementary Information

Reporting Summary

Peer Review File

Interactions between IFN-I-producing pDCs and MCMV-infected cells or pDCs that do not produce IFN-I. Representative videos of IFN-I-producing pDCs in direct contact with MCMV-infected cells (IE1⁺) at 40 h after MCMV infection in the MZ of the spleen of SCRIPT mice (pDCs: magenta; Ifnb1-YFP reporter: yellow; IE1: cyan). The 3D reconstructions were generated using the IMARIS software v.9.9.1 (n = 4 mice, thickness: 20–30 µm, distance between slices: 1 µm).

Same as for Supplementary Video 1.

Supplementary Table

The first spreadsheet gives the absolute numbers of DC types in the spleen early during MCMV infection. These data are derived from the spectral flow cytometry experiments. For each spleen, the percentage of each DC type within the live cell spectral flow cytometry gate was retrieved from the analysis illustrated in Extended Data Fig. 5 and multiplied by the absolute number of viable cells. Statistical analyses were performed using the unpaired Student’s t-test. The subsequent spreadsheets give the metadata for the scRNA-seq experiment. Specifically, the second sheet lists the names of each of the following sheets and describes their content. The third sheet indicates, for each cell, the experimental condition (‘time point’), the identity of the mouse from which it was isolated (‘mouse_ID’), its initial sorting phenotype, the results of the QCs, its belonging to Seurat or Rphenograph clusters for the different analyses performed, its final assignment to a DC type and activation state, and its fluorescence intensity for the cell-surface markers used for sorting. Gene markers for Seurat clusters, and the lists of DEGs between Seurat clusters or subclusters, according to the different bioinformatics analyses performed, are given in the following spreadsheets, as are the signatures used for the sgCMap analyses, as listed and described in the second sheet.

Source data

Source Data Fig. 1

Statistical source data.

Source Data Fig. 2

Statistical source data.

Source Data Fig. 3

Statistical source data.

Source Data Fig. 5

Statistical source data.

Source Data Fig. 6

Statistical source data.

Source Data Fig. 7

Statistical source data.

Source Data Extended Data Fig. 4

Statistical source data.

Source Data Extended Data Fig. 5

Statistical source data.

Source Data Extended Data Fig. 9

Statistical source data.

Source Data Extended Data Fig. 10

Statistical source data.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Cite this article

Valente, M., Collinet, N., Vu Manh, TP. et al. Novel mouse models based on intersectional genetics to identify and characterize plasmacytoid dendritic cells. Nat Immunol 24, 714–728 (2023). https://doi.org/10.1038/s41590-023-01454-9

Download citation

Received: 07 February 2022
Accepted: 03 February 2023
Published: 16 March 2023
Issue Date: April 2023
DOI: https://doi.org/10.1038/s41590-023-01454-9

This article is cited by

Plasmacytoid dendritic cells turn red!
- Roxane Tussiwand
Nature Immunology (2023)

Subjects

Abstract

Main

Results

The pDC-Tom mice allow specific pDC detection by flow cytometry

The pDC-Tom mice allow refining pDC gating strategies

In pDC-Tom mice, tdT soars in pDCs’ immediate precursors

ZeST mice distinguish pDCs, pDC-like cells and tDCs

ScRNA-seq confirms DC-type identification in ZeST mice

The pDC-Tom mice allow mapping of pDC microanatomical localization

SCRIPT mice allow tracking IFN-I+ and IFN-I− pDCs in situ

Only IFN-Ifate_mapped pDCs entered the TCZ during infection

The pDCs clustered around infected cells but few produced IFN-I

The pDC-like cells and CD11chigh tDCs diverge on activation

Discussion

Methods

Mice

Virus and viral infection

Cell preparation for flow cytometry analysis or cell sorting

Flow cytometry analysis

Spectral flow cytometry analysis

ScRNA-seq data generation

Bioinformatics analyses of scRNA-seq data

Immunohistofluorescence, microscopy and image analysis

Cell immunofluorescence and cell morphology analysis

RT-qPCR

Statistical analysis

Reporting summary

Data availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Extended data

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

This article is cited by

Search

Quick links

SCRIPT mice allow tracking IFN-I⁺ and IFN-I⁻ pDCs in situ

Only IFN-I^fate_mapped pDCs entered the TCZ during infection

The pDC-like cells and CD11c^high tDCs diverge on activation