Fibroblastic reticular cells form reactive myeloid cell niches in human lymph nodes

Abstract

Lymph nodes play a key role in maintaining fluid balance in homeostatic and inflamed tissues and provide fibroblastic niche environments for optimal immune cell positioning and interaction. Here, we used single-cell and spatial transcriptomic analyses in combination with high-resolution imaging to molecularly define and functionally characterize niche-forming cells that control inflammation-driven remodeling in human lymph nodes. Fibroblastic reticular cells responded to inflammatory perturbation with activation and expansion of poised niche environments. Inflammation-induced adaptation of lymph node infrastructure and topography included the expansion of peptidase inhibitor 16 (PI16)-expressing reticular cell (PI16⁺ RC) networks that enwrap the perivenular conduit system. Interactome analyses indicated that macrophage-derived oncostatin M directs PI16⁺ RC activation in inflamed lymph nodes and thereby promotes immune cell aggregation in the perivenular space. In conclusion, these data demonstrate that the inflammatory remodeling of human lymph nodes results in the formation of reactive myeloid cell niches by PI16⁺ RCs.

Fig. 1.. Characterization of the immune cell landscape in quiescent human lymph nodes.

(A-B) Immunofluorescence images depicting quiescent human lymph nodes from two different patients. B cell follicles (arrows) and T cell zones (asterisks) are highlighted. (C-H) Flow cytometric analysis of hematopoietic cells in quiescent human lymph nodes. (C,E) Phenograph-based clustering projected on the UMAP showing CD45⁺ cells (C) or CD3⁺ T cells (E) from one representative lymph node together with relative marker expression used to identify different immune cell populations. Manual gating of CD19⁺ B cell populations projected on the UMAP (G). Relative abundance of immune cell populations gated either on all CD45⁺ cells (D), CD3⁺ T cells (E) or CD19⁺ B cells (H). Relative abundance was calculated according to the gating strategy shown in fig. S1G, data are shown as the mean and SEM. In D,F data is representative of n=5 patients from 3 independent experiments and in H of n=4 patients from 3 independent experiments. q values were calculated using the Kruskal-Wallis test with Benjamini, Krieger and Yekutieli correction. (I) Immunofluorescence analysis of B cell follicles with or without germinal centers (GCs) shown with BCL6 staining. BCL6⁺ B cell follicles were quantified based on n=6 quiescent lymph nodes. P value was calculated using the two-sided Mann-Whitney test.

Fig. 2.. Stromal cell topography and composition in immunologically quiescent human lymph nodes.

(A-B) Immunofluorescence images depicting representative quiescent human lymph nodes from n=2 patients. PDPN high B cell zone reticular cell (BRC) networks (arrows), interfollicular zones (arrowhead), and PDPN low T cell zones (asterisks) are highlighted. (C-D) Immunofluorescence images showing B cell zone reticular cells (BRC, arrow), interfollicular reticular cells (IFRC, arrowhead), T cell zone reticular cells (TRC, asterisk, C) and medullary reticular cells (MedRC, D). (E) Immunofluorescence images showing COL6A1⁺ conduit networks in different lymph node regions, the subcapsular sinus (SCS, arrow), the T cell zone (arrowhead), around high endothelial venules (HEV, asterisk) and the medulla (double arrow). (F-H) Flow cytometric analysis of non-hematopoietic cells in quiescent human lymph nodes. (F) Phenograph-based clustering projected on the UMAP showing all CD45⁻ CD235⁻ cells from one representative lymph node. (G) Relative expression of marker proteins projected onto the UMAP of one representative LN. (H) Relative abundance of different stromal cell populations in resting human LNs of n=4 patients from 2 independent experiments. Relative abundances were calculated according to the gating strategy shown in Fig. S2C, data are shown as mean and SEM. q values were calculated using the Kruskal-Wallis test with Benjamini, Krieger and Yekutieli correction for multiple comparisons.

Fig. 3.. Single cell transcriptomics-based analysis of FRC subsets in quiescent human lymph nodes.

(A-C) ScRNA-seq data of VSMCs and FRCs from quiescent human lymph nodes. (A) UMAP representation colored by subset identity. (B) Dotplot indicating the average expression of signature genes across FRC subsets. (C) Projection of FRC, VSMC and perivascular niche signatures onto the UMAP. Signatures are shown as module scores calculated from the average expression of indicated genes. (D) Immunostaining of one representative quiescent human lymph node processed for spatial transcriptomics with the 10X Genomics Visium platform. Immunostaining with CD20, CD4, CD31 and ACTA2 was used to localize B cell follicles, T cell zone and vasculature, respectively. (E-F) Cell type decomposition maps showing the inferred relative frequency of immune cell types (E) and FRC subsets (F). Lymph node scRNA-seq data is representative of n=5 patients and represent 18,322 cells from 5 independent experiments.

Fig. 4.. Characterization of perivascular FRC subsets along the vascular tree.

(A) Cell type decomposition maps showing the inferred relative frequency of endothelial cell subsets in one representative quiescent human lymph node. (B) Spatial proximity enrichment between pairs of FRCs and endothelial cell subsets based on the inferred cell type decomposition maps. (C-F) Representative confocal microscopy images of different lymph node regions stained for the indicated markers to confirm the predicted localization of FRC subsets and VSMCs in perivascular niches. Co-expression of ACTA2 and MYH11 was used to localize VSMCs (E). Antibodies against ACTA2, FABP4 and CCL19 were used to stain for ACTA2⁺ PRCs (F). (G) Schematic representation of the localization of different FRC subsets in perivascular niches of quiescent human lymph nodes. ArtBEC, arterial BEC; BF, B cell follicle; CapBEC, capillary BEC; M, Medulla; MedSinusLEC, medullary sinus LEC; SCScLEC, subcapsular sinus ceiling LEC; SCSfLEC, subcapsular sinus floor LEC; TZ, T cell zone; VenBEC, venous BEC.

Fig. 5.. Phenotypic characterization of FRCs in inflamed human lymph nodes.

(A) Immunofluorescence images depicting one representative inflamed human lymph node. (B-C) Immunofluorescence images showing B cell follicles in the subcapsular sinus (SCS, B) and paracortical/medullary areas (C) in inflamed human lymph nodes. (D-E) Imaging-based quantification of B cell follicles (D) and BCL6⁺ B cell follicles (E) in sections of quiescent and inflamed human lymph nodes. Data is representative for n=5 patients per group. Statistical analysis was performed using the Mann-Whitney test. (F-H) Flow cytometric analysis of hematopoietic cells in quiescent and inflamed human lymph nodes. Relative abundances of immune cell populations gated either on all CD45⁺ cells (F), CD4⁺ T cells (G) or CD3⁺ T cells (H). Data are shown as the mean and SEM. In (F-H) data is representative of n=5 patients from 3 independent experiments. P values were calculated using the two-sided Mann-Whitney test (D-H). (I-J), ScRNA-seq data of VSMCs and FRCs from quiescent and inflamed human lymph nodes. (I) UMAP representation split by lymph node condition and colored by subset identity. (J) Boxplots showing the average log fold change of the top 100 upregulated genes for each FRC subset in inflamed versus quiescent lymph nodes. ScRNA-seq data is representative of 18,322 FRCs/VSMCs and 98,713 immune cells (quiescent) from n=5 patients and 25,449 FRCs/VSMCs and 69,585 immune cells (inflamed) from n=5 patients. Data from each patient was processed as independent experiment.

Fig. 6.. PI16⁺ RC activation and interaction with immune cells in inflamed human lymph nodes.

(A) ScRNA-seq data of PI16⁺ RCs from inflamed human lymph nodes. Heatmap plot of significantly enriched GO terms in PI16⁺ RCs from inflamed versus quiescent lymph nodes determined by enrichment tests on the top 100 up-regulated genes. (B,C) Immunoflourescence images of PI16⁺ RC networks (arrows) in perivenular regions in quiescent (B) and inflamed human lymph nodes. (D) Histological quantification of perivascular (upper panel) and perivenular (lower panel) PI16⁺ networks in sections of quiescent and inflamed human lymph nodes. The number of PI16⁺ networks was quantified based on n=4 quiescent and n=5 inflamed lymph nodes. (E) Zoomed-in immunofluorescence images showing localization of PI16⁺ RCs and COL6A1⁺ conduit networks in paracortical/medullary regions. Connection of COL6A1⁺ conduits to endothelial basement membrane of blood vessels is highlighted (arrows). (F) Cellchat analysis of PI16⁺ RCs and immune cells from inflamed and quiescent human lymph nodes. Dotplot indicates the inferred communication probability of receptor-ligand pairs from enriched interaction pathways between the indicated immune cell type and PI16⁺ RCs. (G) Immunofluorescence images depicting membrane contact of PI16⁺ RCs and CD11c⁺ myeloid cells. The surface contact area is visualized in turquoise. (H-K) In-vitro cultured FRCs from n=7 human lymph nodes and PI16⁺ RCs from n=5 human lymph nodes were stimulated with recombinant proteins for 48 h. PI16 (H), ICAM1 (I), IL6 (J) and SEMA3C (K) mRNA fold change was measured by quantitative PCR. P values were calculated with the two-sided Wilcoxon test (H) and two-way ANOVA test (I-K).

Fig. 7:. Characterization of OSM-expressing macrophages and interaction with PI16⁺ RCs in human lymph nodes and tonsils.

(A) Immunofluorescence images of Oncostatin-M (OSM) producing myeloid cells in close proximity to PI16⁺ RCs in perivenular regions of inflamed human lymph nodes. (B) UMAP representation of re-embedded myeloid cell populations from quiescent and inflamed human lymph nodes colored by assigned cell types. (C) Dotplot representation of the average expression of marker genes across myeloid cell subsets from quiescent and inflamed human lymph nodes. (D) Dotplot representation of the average expression of stimulatory receptors and ligands in myeloid cell subsets derived from the interactome between myeloid cells and PI16⁺ RCs. (E) Cellchat analysis of PI16⁺ RCs and myeloid cell subsets from quiescent and inflamed tonsils. Dotplot indicating the inferred communication probability of receptor-ligand pairs from enriched interaction pathways between the indicated myeloid cell type and PI16⁺ RCs.