Dissection of intercellular communication using the transcriptome-based framework ICELLNET

Latest Research

Expert-curated database of ligand–receptor interactions

In order to globally reconstruct cell communication networks, we curated a comprehensive database of ligand–receptor interactions from the literature3,11,12 and public databases10,13. An expert manual curation was performed based on a rigorous literature screening of original articles, applying the following criteria: 1) robustness of the findings, 2) consistency with international classifications and nomenclature, 3) experimental validation of the functionality of the ligand–receptor interaction (see Methods). We also used consensus reviews from leaders in the field, in particular for cytokines3,14,15,16. This helped solving controversies on how to classify some molecules, which have features of different molecular families. We did not include putative interactions based on protein–protein interaction predictions, as it is done in some other databases9. This led to the integration of 380 ligand–receptor interactions into the ICELLNET database (Supplementary Data 1). Whenever relevant, we took into account the multiple subunits of the ligands and the receptors (Fig. 1a). Interactions were classified into 6 major families of communication molecules, with a strong emphasis on inflammatory and immune processes: Growth factors, Cytokines, Chemokines, Immune Checkpoints, Notch signaling, and Antigen binding (Fig. 1b and Supplementary Data 1). Other families such as hormones or adhesion molecules were more scarcely represented. In order to simplify the subsequent graphical visualization, these were grouped as other in our current classification (Fig. 1b and Supplementary Data 1).

Fig. 1: Structure of the ligand–receptor database.
figure1

a Extract of the ligand–receptor database. The database is structured as the following scheme: the 4 first columns correspond to gene symbol of each subunit of interacting ligand and receptor, the next two columns state the classification into families and/or subfamilies of molecules, and the last column gives the source for manual curation (PubMed ID). b Histogram displaying the number of interactions classified in each family of communication molecules included in the database (antigen binding, checkpoint, chemokine, cytokine, growth factor, notch signaling). c Histogram displaying the number of interactions classified in the different defined subfamilies of cytokines: interleukin 1 family, interleukin 17 family, receptor tyrosine kinase (RTK) family, transforming growth factor beta (TGF) family, tumor necrosis factor family (TNF), type 1 cytokine family, and type 2 cytokine family, and unclassified cytokines.

Cytokine–receptor pairs were mapped in an exhaustive manner, by exploiting a series of reference articles and consensus classifications. They represent 50% of the total interactions included in the database (194 interactions), and were further classified into 7 sub-families according to structural protein motifs: type 1 cytokines, type 2 cytokines, IL-1 family, IL-17 family, TNF family, TGF-ß family, and RTK cytokines3,14,15,16 (Fig. 1c).

This database is integrating information on both multiple subunits of ligands and receptors, and a classification into molecular families/subfamilies.

Development of a computational pipeline to dissect intercellular communication

In the ICELLNET framework, we developed an automatized tool in R script to infer communication between multiple cell types by integrating: 1) prior knowledge on ligand–receptor interactions (Fig. 1); 2) computation of a communication score between pairs of cells based on their transcriptomic profiles, and; 3) several visualization modes to guide results interpretation. Quantification of intercellular communication was achieved by scoring the intensity of each ligand–receptor interaction between two cell types from their expression profiles (Fig. 2).

Fig. 2: ICELLNET pipeline to study intercellular communication from cell transcriptional profiles.
figure2

(top) Selection of transcriptomic profile of cell population of interest (central cell) and of other cells used to infer communication with among a public dataset (partner cells) (Supplementary Data 2). (middle) Genes corresponding to ligands/receptors included in the database are selected, scaled by maximum of gene expression for each gene, and used to compute a communication score between two cell types. (bottom) Graphical layers used to dissect intercellular communication scores between the central cell (CC) and partner cells (Macroph: macrophage, Fblast: fibroblast, Epith: epithelial cell, B cell, Treg, NK: natural killer cell): global communication network intensity assessment, contribution of each family or subfamily of molecules to the communication scores, statistical difference assessment of the communication scores from the same central cell to the different partner cell types, and plots of specific interactions most contributing to communication scores. Some schematic art pieces were used and modified from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License. http://smart.servier.com/.

From each transcriptomic profile, all genes or only differentially expressed genes could be used, and no filtering threshold was applied to gene expression. Taking advantage of the ICELLNET database, the genes coding for ligands/receptors were selected from all 380 interactions to compute the score, but it is also possible to restrict the database to specific families of molecules, depending on the biological question.

A unique feature and strength of ICELLNET is its ability to infer cell-to-cell communication even from an individual cell population-based transcriptome of interest (the “central cell”). The user should define this cell type of interest, and provide its transcriptional profile, which for example may be selected from different primary cell subsets, or from the same cell type cultured in different biological conditions (Fig. 2 top-left). ICELLNET separately considers other cell types with known transcriptomic profiles (hereafter called « partner cells ») that can connect to the central cell. These can be cell types coming from the same dataset as the central cell, or from any other transcriptomic dataset. The ICELLNET pipeline has integrated reference transcriptomic profiles by using the Human Primary Cell Atlas17,18. This public dataset includes transcriptomic profiles of 31 human cell types including immune cells, stromal cells, neural cells, and tissue-specific cell types, all generated with the same Affymetrix technology (Fig. 2 top-right). Human Primary Cell Atlas has been downloaded and added to ICELLNET framework, in order to be used as reference transcriptomic profiles of partner cell types (Supplementary Data 2). The user always has the possibility of providing original transcriptional profiles for central and partner cells, or to use the reference transcriptomic profiles included in ICELLNET R package to represent putative « partner » cells.

Establishment of a score to assess the communication between cells

From the transcriptomic profiles, we selected the genes coding for the ligands and the receptors in our database. We designed the tool to enable the user to focus on the ligands and/or receptors that are differentially expressed between conditions of study, or to use the entire ligand–receptor database to compute the communication score.

Since cell-to-cell communication is directional, we considered ligand expression from the central cell, and receptor expression from the partner cells in order to assess outward communication. Conversely, we then selected receptor expression from the central cell, and ligand expression from partner cells in order to assess inward communication (Fig. 2 middle). For each gene, expression levels were scaled by maximum of gene expression in the dataset, in order to avoid a communication score predominantly driven by highly expressed genes. Indeed, bioactivity of communication molecules varies a lot. Some cytokines, such as IL-12 and IL-4, are very bioactive at low concentrations, and often expressed at very low levels both in transcript and protein. Conversely, many chemokines are produced at much higher levels, without necessarily having a higher bioactivity. Not scaling the data before inferring a communication score would systematically favor a few highly expressed molecules, and would not allow detecting the contribution of important molecules expressed at much lower levels. In the ICELLNET framework, quantification of intercellular communication consists of scoring the intensity of each ligand–receptor interaction between two cell types with known expression profiles. Whenever relevant, we took into account multiple ligand subunits, or receptor chains, using logical rules to impose their co-expression in order to consider functionality. The score of an individual ligand–receptor interaction was computed as the product of their expression levels by the respective source (central) and target (partner) cell. Whenever a communication molecule (ligand or receptor or both) was not expressed by a cell, the score of this particular interaction was set to zero. Individual scores were then combined into a global metric assessing the overall exchange of information between the cell types of interest (Fig. 2 middle), defining a global communication score. ICELLNET provides a matrix summarizing all global communication scores as an output of these analytical steps.

ICELLNET offers different graphical representations allowing multiple layers of interpretation

ICELLNET generates a large quantity of data and scores, which are complex to interpret and analyze. In order to facilitate hypothesis generation, three graphical representations were implemented to help visualize and interpret the results (Fig. 2 bottom). The first representation allows the visualization of intercellular communication networks in directed connectivity maps. In these graphs, nodes represent cell types, the width of the edges connecting two cell types is proportional to their global communication score and the arrows indicate the direction of communication. The second visualization mode breaks down the global scores into the contribution of specific molecular families through a barplot representation. This allows the identification of patterns of co-expressed molecules from the same family, potentially contributing to coordinated biological functions. We implemented statistical analyses of the scores (see Methods) to evaluate the robustness of the differences between scores. The resulting p-values can be visualized as an additional heatmap. The third representation displays the highest contributing ligand–receptor pairs to the communication score within a given channel in a balloon plot. This enables the identification of specific interactions that may drive the global intercellular communication. Thus, the ICELLNET framework is a powerful tool to assess intercellular communication with different visualization modes that can be helpful to dissect underlying mechanisms and extend biological knowledge and understanding.

Application of ICELLNET to study human breast cancer-associated fibroblasts

Cancer-associated fibroblasts (CAFs) are stromal cells localized in the tumor microenvironment that are known to enhance tumor phenotypes, notably cancer cell proliferation, and inflammation. Recently, four subsets of CAFs have been identified and characterized in the context of previously untreated Luminal and Triple Negative Breast Cancer (TNBC)19. Notably, two subsets of CAFs named CAF-S1 and CAF-S4 specifically accumulated in TNBC microenvironment and CAF-S1 was associated with an immunosuppressive microenvironment. This study raised important questions about the regulatory mechanisms involved, in particular the role of cell-to-cell communication. Using the available transcriptional profiles of CAF-S1 and CAF-S4 in TNBC (Fig. 3a), we applied the ICELLNET pipeline to reconstruct the intercellular communication network with 14 other cell types potentially localized in the tumor microenvironment (TME) (Fig. 3b and Supplementary Data 3). The partner cells were selected from Human Primary Cell Atlas and included innate immune cells (monocytes, macrophages, pDC, DC1, DC2, NK cells, neutrophils), adaptive immune cells (CD4+ T cells, CD8+ T cells, Tregs, B cells), epithelial and stomal cells (fibroblasts and endothelial cells). In order to assess the global intercellular communication, we first used the network graphical visualization. This strongly suggested that CAF-S1 has a greater communication potential than CAF-S4 (Fig. 3b) to interact with other TME components. The rescaled communication scores were higher for CAF-S1 as compared to CAF-S4, and the differences were statistically significant for epithelial cells (score CAF S1 > Epith = 6, score CAF-S4 > Epith = 3, p value < 0.1), endothelial cells (score CAF-S1 > Endoth = 7, score CAF-S4 > Endoth = 4, p value < 0.1), plasmacytoid dendritic cells (score CAF-S1 > pDC = 6, score CAF-S4 > pDC = 4, p value < 0.1) and B cells (score CAF-S1 > B cells = 3, score CAF-S4 > B cells = 1, p value < 0.1) (Fig. 3b, c and Supplementary Data 3).

Fig. 3: Dissection of intercellular communication between Triple-Negative breast cancer infiltrating CAF subsets and the tumor microenvironment.
figure3

a Workflow of the analysis. b Connectivity maps describing outward communication from cancer associated fibroblasts CAF-S1 (n = 6 biologically independent samples) and CAF-S4 (n = 3 biologically independent samples) subsets to primary cells (Supplementary Data 2). The CAF subsets are considered as central cells and colored in gray. Primary cells are considered as partner cells (DC1 and DC2: conventional dendritic cell 1 and 2, pDC: plasmacytoid dendritic cell, Macroph: macrophage, Mono: monocyte, Endoth: endothelial cell, Fblast_B: breast fibroblast, Epith: epithelial cell, B cell, Treg, CD8 T cell, CD4 T cell, Neutrop: neutrophil, and NK: natural killer cell) and are colored depending on the cell compartment (green: stroma, orange: innate, blue: adaptive, pink: epithelium). The width of the edges corresponds to a global score combining the intensity of all the individual ligand/receptor interactions. A scale ranging from 1 to 10, corresponding to minimum and maximum communication scores, is shown in the legend. A selection of normalized scores is written directly on the network. c Barplot of communication score with contribution by families of communication molecules between CAF subsets (n = 6 biologically independent samples for CAF-S1, n = 3 biologically independent samples for CAF-S4) and a selection of partner cells. Significant differences are shown on the graph (two-sided wilcoxon test, and p values are adjusted with Benjamin–Hochberg method, *p-value ≤ 0.1). d Balloon plot of individual interaction scores between CAF subsets and Tregs. Only interactions with a score contribution above 10 to the score are displayed for clarity purpose. Two biologically interesting communication channels were highlighted by red boxes. e Barplot of communication score with contribution restricted to cytokines subfamilies between CAF subsets (n = 6 biologically independent samples for CAF-S1, n = 3 biologically independent samples for CAF-S4) and a selection of partner cells. Some schematic art pieces were used and modified from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License. http://smart.servier.com/ (a).

CAF-S4 uses specific communication channels to interact with the TME components

We focused on the biological composition of the score, to identify families of molecules highly involved in CAFs communication with the selected cells. We selected 4 partner cell types: epithelial cells, fibroblasts, Tregs and B cells. Using the barplot representation, we looked for differences between CAF-S1 and CAF-S4 in terms of genes coding for families of communication molecules. We found that genes coding for communication molecules inducing Notch signaling were specifically expressed by CAF-S4 to communicate with other cells (Fig. 3c). Looking at individual communication interactions between CAF subsets and Tregs demonstrated that gene coding for JAG1 protein was only expressed by CAF-S4 to interact with NOTCH receptors (NOTCH1 and NOTCH2 genes expressed), and thus potentially having a role in activating the Notch signaling pathway (Fig. 3d and Supplementary Data 3). For both CAF subsets, the barplot representation indicated that cytokines–receptors interactions were highly contributing to the global communication scores compared to other families of molecules (Fig. 3c). This observation led us to focus on cytokine-mediated communication using the ICELLNET pipeline (Fig. 3e). By considering only cytokine–receptor interactions, the CAFs appear to communicate more with other fibroblasts compared to other cell types with a significant p-value (Fig. 3e, Supplementary Fig. 1a). Also, this approach highlighted that RTK cytokines, and notably PDGFB coding for PDGF, were preferentially expressed by CAF-S4 compared to CAF-S1 (Fig. 3e, Supplementary Fig. 1b and Supplementary Data 3). We also applied ICELLNET pipeline to study inward communication between the partner cells and the CAF subsets, which revealed no difference between CAF-S1 and CAF-S4 in term of communication score intensities but also in terms of the families of molecules involved in communication (Supplementary Fig. 2). Thus, the ICELLNET framework allowed us to identify specific communication channels revealing potential interactions between CAF-S4 and TME components.

Lupus nephritis cell–cell communication network inferred from single-cell RNA-seq datasets using ICELLNET

Single-cell technologies are now largely employed in various biological fields to better characterize immune cell diversity and cell phenotypes. They also offer insightful datasets to reconstruct cell–cell interactions between different cell populations from the same sample or tissue. We applied ICELLNET to a published single-cell dataset of immune cells from lupus nephritis patients20. This dataset included several immune cell subpopulations of T and B lymphocytes, but also natural killer cells, macrophages, and dendritic cell populations20.We represented those cells into a Uniform Manifold Approximation and Projection (UMAP) (Fig. 4a). We decided to focus on the potential communication between the conventional dendritic cell (cDC) cluster (CM3) and two clusters of T cells, CT0a and CT3b, which respectively refer to effector memory CD4+ T cells and TFH-like cells according to the original study20 (Fig. 4b). Because of sparsity and drop-out that are inherent to single-cell data, we computed the average gene expression profile for each cluster. Communication scores were then computed with cluster’s mean expression profiles as input. The communication score between CM3 cluster and CT3b was higher than the score from CM3 to CT0a cluster (score CM3 > CT3b = 1527, score CM3 > CT0a = 1123) (Fig. 4b and Supplementary Data 4). In particular, it showed higher communication potential for checkpoints, chemokine, and growth factors (Fig. 4b). From this, we highlighted specific interactions that most differed between the two communication scores, such as CD86/CD28 (92 vs 40 for CM3 > CT3b and CM3 > CT0a, respectively), CD86/CTLA4 (92 vs 14, respectively), PDCD1LG2/PDCD1 (72 vs 19), CCL22/CCR4 (100 vs 39), or VEGFA/FLT1 (21 vs 0) (Fig. 4c and Supplementary Data 4).

Fig. 4: Evaluation of cell-to-cell communication potential between dendritic cells and T-cell subpopulations in lupus nephritis single cell data.
figure4

a Uniform Manifold Approximation and Projection (UMAP) visualization of the lupus nephritis dataset. 22 clusters were previously identified by the authors and their annotations are displayed on the right. Cell identity of each cluster can be found in the original article20. b ICELLNET framework applied on specific cluster to assess DC-T intercellular communication. Average expression profiles were computed from the single-cell data matrix counts for each cluster to then compute intercellular communication score. Barplots are displaying the contribution of the different communication molecules families to the communication scores. c Balloon plot representing specific individual interaction scores that differ from at least 10 between the two conditions (cutoff chosen for clarity purpose) for interaction belonging to either chemokine, checkpoint, or growth factor families of molecules. d Assessment of communication score variability when subsampling DC or T cluster. Cells were randomly selected for CT0a and CT3b cluster (left) or CM3 cluster (right) (number of cells selected according to their respective cluster size). Communication scores were computed with the other complete cluster (CM3 for left, T cell cluster for right). Standard deviations of the communication scores are displayed on the graphs (n = 20 random selection of cells). Some schematic art pieces were used and modified from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License. http://smart.servier.com/ (b).

We then assessed the robustness of the communication scores by subsampling the clusters of interest (either CM3 cluster or CT0a and CT3b clusters). We randomly selected cells (number defined by the size of the clusters) from the clusters and computed again communication scores to assess variability (Fig. 4d). We observed that the variability of the communication score (measured using standard deviation) was anti-correlated with the percentage of cells used from the cluster, ranging from 37 and 59 (value of SD, for CT0a and CT3b communication, respectively) for 90% of the cluster cell to 146 and 215 (value of SD, for CT0a and CT3b communication, respectively) for 10% for CM3 subsampling (Fig. 4d and Supplementary Data 4). Similar results were observed for CT0a and CT3b cluster subsampling. This demonstrated the relevance of using the average of cluster gene expression to assess communication, rather than unique cells.

We also used the Human Primary Cell Atlas transcriptomic profiles of cDC to compute the communication scores between dendritic cells and the same T-cell clusters from the lupus nephritis dataset (CT0a and CT3b) (Supplementary Fig. 3a). Even if the cDC transcriptomic profile came from a different technology, it led to similar communication scores when comparing with the ones computed exclusively from the original single cell dataset (Fig. 4b, c, and Supplementary Fig. 3). The 122 ligand–receptor pairs contributing to the communication score in the single-cell dataset were also found using Human Primary Cell Atlas dataset (Supplementary Fig. 3c). The slight differences can be explained by the fact that the cells were not in the same biological state but also some genes were not captured with the Affymetrix technology. Nevertheless, it demonstrated that the Human Primary Cell Atlas can be very valuable to infer communication potential with a cell type that is not represented in a studied dataset.

Application of ICELLNET to in vitro-generated immune cell states

We then focused on a controlled in vitro system to assess whether ICELLNET would be able to predict cell–cell communication. We addressed the communication and underlying mechanisms in resting and activated DC. We generated an original microarray dataset, and applied ICELLNET to assess intercellular communication. LPS-activated human monocyte-derived DC produce two important autocrine cytokines, TNF and IL-10, which play a major role in regulating inflammation. We asked whether this could be mediated by modulating DC communication with partner cells. To this end, LPS-activated DC were cultured in the presence and absence of blocking antibodies (Abs) to the TNF and IL-10 receptors (aTNFR and aIL-10R). No effect on cell viability was observed (Supplementary Fig. 4a). The most prominent effect of LPS on DC hallmark maturation markers was observed at the mRNA level in the time frame of 4 to 8 h following activation21. We performed large-scale microarray analysis after 4 and 8 h of DC culture with LPS, with and without blocking Abs to TNF and IL-10 receptors (Fig. 5a).

Fig. 5: IL-10R blocking activates a cell-to-cell communication module in LPS-stimulated DCs.
figure5

a Depicted are the four experimental conditions for which transcriptomics was generated (n = 6 biologically independent samples). b Connectivity maps describing outward communication from DCs to putative target cells in the conditions: Medium (M), LPS (L), LPS + aTNFR and LPS + aIL-10R. Twelve primary cell types are considered as partner cells (DC: conventional dendritic cell, Macroph: macrophage, Mono: monocyte, Ostblast: osteoblast, Endoth: endothelial cell, Fblast: fibroblast, Kerat: keratinocyte, Epith: epithelial cell, B cell, T cell, Neutrop: neutrophil and NK: natural killer cell) and are colored depending on the cell compartment (green: stroma, orange: innate, blue: adaptive, pink: epithelium). The width of the edges corresponds to a global score combining the intensity of all the individual ligand/receptor interactions, normalized to the medium condition. A scale ranging from 1 to 10, corresponding to minimum and maximum communication scores, is shown in the legend. c Gene corresponding to ligands (black) and receptors (white) counted in each loop signature (from n = 6 biologically independent samples for each conditions) and plotted according to regulation directionality: upregulated (Up) or downregulated (Down). Genes with separability score ≥4 were included in each condition’s signature. d Balloon plot representing a selection of individual interaction scores. e Protein levels of IL-6 (n = 10 biologically independent samples, except for Medium condition where n = 5), OSM (n = 4 biologically independent samples), IL-23 (n = 8 biologically independent samples) and IL-12p70 (n = 16 biologically independent samples, except for Medium condition where n = 8) demonstrating increased secretion in LPS + aIL-10R DC supernatant. Data are presented as mean values ± SEM. Some schematic art pieces were used and modified from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License. http://smart.servier.com/ (a).

Despite extensive studies of both TNF and IL-10 in the context of innate immunity, their different contribution to DC intercellular communication could not be predicted a priori at this systems level. We applied ICELLNET to reconstruct the intercellular networks between DCs and putative target cells. The network representation demonstrated an increase of the global communication score in all 12 channels, when comparing 8-hours LPS-activated DC to resting (medium) DC (Fig. 5b). Importantly, these maps revealed that blocking the IL-10 loop determined the largest amplification of DC communication with all 12 cellular targets, while blocking the TNF loop in LPS-activated DCs had a negligible effect on the global communication score (Fig. 5b, Supplementary Fig. 4b and Supplementary Data 5).

IL-10 controls an intercellular communication module in LPS-activated dendritic cells

We compared the transcriptomic profiles of each condition (aTNFR and aIL-10R) to the LPS-alone condition to extract the differentially expressed genes (DEG) (Supplementary Data 5). We then screened the IL-10 and TNF DEG to identify ligands and receptors included in the database. We were able to extract 27 ligands and 23 receptors which were differentially regulated from the aIL-10R condition, while there were only 12 ligands and 10 receptors differentially regulated from the aTNFR condition (Fig. 5c).

ICELLNET barplots suggested that cytokines were driving the increase in the communication score when blocking IL-10R. We looked at the subfamilies of cytokines to precisely identify the key communication channels (Supplementary Fig. 4b) Type 1 and TNF subfamilies were increased in aIL-10R condition compared to others. This was confirmed by individual channel communication scores (Fig. 5d). To confirm the hypothesis that IL-10 controls cytokine-mediated DC communication, we selected four important immunoregulatory molecules from the IL-6- and IL-12-families, and further validated expression at the protein level in 24 h culture DC supernatants using cytometric bead array (CBA) and ELISA (Fig. 5e).

Experimental validation of multiple IL-10-dependent communication channels

To assess communication efficiency, i.e., how increased connectivity translates into functional changes in target cells, we turned to experimental validation of predicted communication channels using immunological assays adapted to the output response of each cell type. Due to its physiopathological relevance, we first investigated the DC-T cell axis through co-culture experiments of T cells with DCs treated by LPS with or without TNFR and IL-10R blocking antibodies (Fig. 6). We found that naive CD4+ T cells, when co-cultured with LPS-DC in the absence of the IL-10 loop, globally increased and shifted their pattern of cytokine secretion, compared to LPS-DCs, while blocking the TNF loop had almost no effect (Fig. 6a). Similar results were obtained with memory T cells (Fig. 6b).

Fig. 6: IL-10 but not TNF loop dictates T helper polarization by LPS-DC.
figure6

a, b Supernatants of CD4+ naive (a) and memory (b) T cells, co-cultured with the indicated DCs, were analyzed for the presence of T helper cytokines by CBA: IL-2, IL-3, IL-4, IL-9, IL-10, IL-17A, IL-17F, and IFN-γ (a) and all the above in addition to IL-5, IL-13 TNF, and GM-CSF (b). Results are shown in a two-dimensional principal component analysis (PCA). Dots represent mean of 9 (a) or 6 (b) independent co-culture experiments. c Histogram representation of 4 cytokines present in the supernatant of naive (white bars, left axis) or memory (black bars, right axis) supernatant (n = 16 biologically independent samples for naïve CD4 T cells except Medium condition where n = 8 biologically independent samples, n = 6 for memory CD4 T cells). Data are presented as mean values ± SEM. d CD4+ naive T cells were analyzed for IL-17A, IL-9, and IFNg production using intracellular staining FACS. Percentage of positive producers is given. Shown is one representative out of 3 independent experiments. Some schematic art pieces were used and modified from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License. http://smart.servier.com/ (a).

Since the IL-10/IL-10R pathway may have a direct effect on T helper cells during the differentiation process, we verified that the observed T helper polarization was indeed due to the IL-10 loop blockade in the DCs, and not due to a direct effect on T cells (Supplementary Fig. 5a). It is possible that residual IL-10R blocking antibodies could have acted directly on T cells during DC-T co-culture. By adding IL-10R antibodies during DC-T co-culture (not only during DC activation) we demonstrated that any IL-10R antibodies in this setting would not have any direct effect on T-cell polarization.

Among the factors explaining the secretion profile of T cells determined by LPS + aIL-10R-DCs, we observed a remarkable emergence of Th17 cytokines (Fig. 6c), in line with murine studies22,23. Strikingly, IL-9 secretion was also increased (Fig. 6c), and produced by a T-cell population distinct from the Th17 cells producing IL-17A alone or co-expressed with IL-9 and IFN-g (Fig. 6d). This provides the demonstration that LPS-activated DCs, in the absence of an IL-10 loop, determine Th17 and Th9 polarization in humans, both of which participate in host defense and autoimmunity24,25.

In order to validate the model-based hypothesis that there is increased communication between DC and multiple cell types, we considered three additional types of target cells: keratinocytes, plasmacytoid DCs (pDC), and neutrophils. Similar to T cells, these cell types play key roles in the inflammatory microenvironment and had an increased global communication score. Target cells were cultured with DC-derived supernatants, and their activation assessed by qRT-PCR or FACS. LPS-DC supernatant induced marginal keratinocyte activation, as assessed by the expression of TNF, IL-1β and this was not affected by aTNFR (Fig. 7a). However, blocking the IL-10 loop dramatically increased both factors (Fig. 7a), validating a potent DC to keratinocyte communication controlled by IL-10. This extends DC-induced keratinocyte activation26,27 to the context of bacterial infection.

Fig. 7: IL-10 loop controls DC communication with keratinocytes, neutrophils, and pDCs.
figure7

a RT-PCR analysis of the expression of TNF and IL-1b mRNA in HaCat cells incubated with medium, LPS or with supernatant (diluted 1:10) of the indicated DCs for 4 h. Blocking antibodies for the cytokines IL-19, IL-36g, OSM and TNF were added to LPS + aIL-10R-DC supernatant for 1-h incubation before culturing with HaCat cells. Data are represented as mean values ± SEM, n = 4 or n = 8 biologically independent samples depending on the conditions, *p < 0.05 (two-sided paired t-test without correction). b, c Expression of maturation markers CD86, HLA-DR, and ICOSL (b) or CD11b and CD62L (c) analyzed by flow cytometry with surface staining on pDCs (n = 18 or n = 6 biologically independent samples depending on the conditions) cultured with supernatant (diluted 1:10) of the indicated DC for 24 h (b) and neutrophils (n = 9 or n = 6 biologically independent samples depending on the conditions) (c) cultured with supernatant (diluted 1:100) of the indicated DC for 1 h. Blocking antibodies for the cytokines GCSF, GM-CSF, TNF, and IL-12 (for pDC) or IL-6 (neutrophils) were added to LPS + aIL-10R-DC supernatant for 1 h incubation before culture. Each biological replicate comprised independent DC donor paired to independent pDCs/neutrophils donor. Data are represented as mean values ± SEM, *p < 0.05; **p < 0.01; ***p < 0.001 (two-sided paired t-test without correction). d For each target cell, we reduced the different activation markers to a single parameter normalized between 0 (Ø) and 1 (max) in the rectangles. The value 0 corresponds to the activation level induced by supernatants from untreated DC, while 1 corresponds to the maximum activation level from all the observed conditions. These experimentally validated activation scores were in qualitative agreement with the ICELLNET communication scores between DC and the target cells, represented by the width of the edges. Some schematic art pieces were used and modified from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License. http://smart.servier.com/ (a, b, c).

The DC-pDC communication channel was also controlled by IL-10, since LPS + aIL-10R-DC supernatants activated pDCs (as assessed by CD86, HLA-DR, and ICOSL surface expression), in comparison to LPS-DCs (Fig. 7b). DC-induced activation of pDC and keratinocytes was not due to the presence of residual aIL-10R (Supplementary Fig. 5b, c). DC-pDC crosstalk was suggested to be important in antiviral28, antibacterial29, and antitumor30 immune responses. Through our systems approach, we have shown that IL-10 controls DC-pDC connectivity.

Neutrophils contribute to DC migration to infection sites and to their subsequent activation31,32. Reciprocally, it was proposed that DCs can promote neutrophil survival33. LPS-DC supernatant induced only a mild activation of neutrophils (as evaluated by rapid upregulation of CD11b with concomitant downregulation of CD62L), while LPS + aIL-10R-DC supernatants led to a strong activation of neutrophils (Fig. 7c), establishing an IL-10 loop control of DC-neutrophils communication.

For all the abovementioned communication channels, we aimed at getting further mechanistic insight. First, we performed control experiments using exogenous LPS that formally excluded any direct effect of LPS at the concentrations found in the DC supernatants (Supplementary Fig. 5d). We then considered ligand–receptor interactions showing high intensity, and thus more likely to mediate cellular crosstalk as observed with the LPS + aIL-10R-DC supernatants (Supplementary Data 6). We blocked, in each DC communication channel, four of the ligands known as potential activators of the target cell type: GCSF, GM-CSF, IL-6, and TNF for neutrophils, IL-19, IL-36 gamma, OSM and TNF for keratinocytes, and G-CSF, GM-CSF, TNF and IL-12 for pDCs. Importantly, blocking TNF alone in the LPS + aIL-10R-DC supernatant was sufficient to inhibit keratinocyte, pDC and neutrophil activation (Fig. 7a–c). By comparing the predicted communication intensities with a global score describing the activation level of keratinocytes, pDC and neutrophils, we observed a qualitative agreement (Fig. 7d), demonstrating increased communication efficiency. In all cases, the target cells were most activated in LPS + aIL10R condition.

Products You May Like

Leave a Reply

Your email address will not be published.