Formalin fixed paraffin embedded (FFPE) tissue blocks used for this study were retrieved from the tissue bank of the Department of Pathology (ASST Spedali Civili di Brescia, Brescia, Italy). Human tissues included reactive lymph nodes (n=104; Supplementary Tables S1 , S2 ), bone marrow, spleen, thymus and tonsils (n= 60; Supplementary Table S3 ), non-lymphoid normal tissues (n=53; Supplementary Table S4 ), and fetal lymph nodes (n=41; Supplementary Table S5 ) from seven autopsies. For cell suspension used for flow cytometry (FC) analysis, a 1–10-mm³ fragment of the lymph node biopsy (n=5) was finely minced with a scalpel on a 100-mm cell culture dish. Tissue dissociation into single-cell suspension was accomplished using gentleMACS™ Dissociator (Miltenyi Biotec). The suspension was filtered onto a 40-µm cell strainer and centrifuged at 1200 RPM for 5 min. Red blood cell lysis was performed using 1× RBC Lysis Buffer (BioLegend, #420301) following the manufacturer’s instructions. Finally, cell suspensions were suspended in fetal bovine serum (FBS) with 10% dimethyl sulfoxide (DMSO) and stored in liquid nitrogen.

Four-micron-thick tissue sections were used for immunohistochemical staining. RORγt (clone 6F3,1, diluted 1:1500 from Merck) was revealed after 40 min in ethylenediaminetetraacetic acid (EDTA) buffer pH 8.0 as antigen retrieval using Novolink Polymer (Leica Microsystems, #RE7280-CE) followed by 3,3’-diaminobenzidine (DAB) as chromogen. For double immunohistochemistry, RORγt antibody was coupled with a panel of markers detailed in Supplementary Table S6 . Briefly after completing the first immune reaction, the second was visualized using Mach 4 MR-AP (Biocare Medical, #M4U536), followed by Ferangi Blue (Biocare Medical, #FB813) as chromogen. For triple immunohistochemistry, the third antigen was revealed using Mach 4 MR-AP and StayRED/AP (Abcam, #ab103741) as chromogen. For quadruple immunohistochemistry, the third antigen was revealed using Novolink Polymer followed by 3-amino-9-ethylcarbazole (AEC, Vector Laboratories #30013) and the fourth antigen using Mach 4 MR-AP followed by StayRED/AP (Abcam, #ab103741) as chromogen.

Isotype control (Mouse IgG2a) staining was performed on reactive lymph node. Sequential stainings were also performed to recognize the transcription factors and RORγt as previously described (28).

Immunofluorescence was performed with a tyramide signal amplification system on 2 μm FFPE. The antigen retrieval was performed using microwave in EDTA pH 8.00 or TRIS-EDTA pH 9.00 buffers. Aspecific binding was prevented through incubation with BSA (5% in TBS/Tween). Primary antibodies were incubated 1 h RT ( Supplementary Table S6 ) and revealed using Novolink Polymer (Leica Biosystems, #RE7280-CE) followed by tyramide fluorophore AF488 (Thermo Fisher Scientific, #B40953), CF555 (Biotium #96021), and CF647 (Biotium #96022). After completing the first immune reaction, the second and the third were visualized combining the different tyramide fluorophores.

Sections were digitized using Axioscan7 (Zeiss) at 20× magnification using Colibri 7 lamp as LED light source with different illumination wavelength (450–488; 540–570; 615–648) depending on the different modules of the LED (385 nm, 475 nm, 567 nm, and 630 nm).

RORγt⁺ nuclei diameters were obtained using the ruler tool of ImageScope (Leica Microsystems). The nuclear areas of RORγt⁺CD3⁻ (n=100 cells) and RORγt⁺CD3⁺ (n=100 cells) were measured on digitalized slides from reactive lymph nodes (n=10) double stained for RORγt and CD3, using the Positive Pixel count algorithm (ImageScope). Proliferation rate of PIGR⁺ cells was defined as a fraction of PIGR⁺Ki-67⁺ cells on total PIGR⁺ cells in reactive lymph nodes (n=5). Consecutive stains on the same section were performed to characterize the morphology of RORγt⁺CD3− cells; briefly, the section was stained for H&E, scanned, and stained for RORγt; reacquired, virtual slides were then realigned.

Digitalized immunofluorescence slides were analyzed using QuPath-0.4.3; briefly, at first RORγt signals were used as a nuclear marker to detect the cells, and then a single measurement classifier was trained to detect CD3, CD127, and RORγt positivity. At last, a composite classifier was used to identify the different cell populations; because CD127 stain overlaps with CD3, the RORγt⁺CD127⁺ population was defined by only CD3⁻ cells. The RORγt⁺ cell populations were graphed for density/mm² (the lymph node areas were obtained by digital image selection), percentage of all nucleated cells (H&E digitalized slides were processed using IHC nuclear algorithm to count all nodal cells as shown in Supplementary Figure S2E ), or percentage of the single population of total counted RORγt⁺ cells as well as absolute number in the lymph node.

Absolute count/mm² of a cohort of cases (n=25, five areas each randomly chosen) was performed preliminarily to establish quantification categories (+= 0.01–5 cells/mm², ++= 6–25 cells/mm², +++= >26 cells/mm²). The perimeter was drawn on digitalized slides through ImageScope to achieve the total nodal area in mm².

RORγt-DC lymph nodes (n=4, score +++) triple and quadruple stained on serial sections for CLA/Heca 452 and dendritic cell markers were digitalized and synchronized. Drawn squares of determined area of 0.025 mm² were cut and pasted around high endothelial venules (HEV; n=15) with RORγt-DC enrichment to count the major dendritic cell populations nearby.

Cell lysates from HEK293T were obtained from cells seeded in a 24-well plate in complete DMEM medium. Cell lysates from thymocytes and lymph node cell suspensions were obtained from 2×10⁶ cells. Cells were lysed in RIPA lysis buffer (Thermo Fisher Scientific, #89900) supplemented with Protease Inhibitor Cocktail (Merck, #78440) and incubated in ice for 20 min. Cell lysates were centrifuged at 13,000 RPM for 15 min, and the suspension was collected and stored at −20°C. Protein concentration was determined by Bradford assay, and 15 µg of total proteins was loaded on 4%–12% NuPAGE^® Bis–Tris Mini Gels (Invitrogen, #NP0335) under reducing condition and transferred onto a PVDF membrane (Invitrogen, #LC2007). Membranes were blocked with 5% milk (Biotium, #22012) in TBS-T (TBS with 0.05% Tween 20; Invitrogen, #28360) for 1 h at room temperature. Primary antibodies to RORγt (clone 6F3.1, diluted 1:1000 from Merck) and Histone H3 (polyclonal, diluted 1:5000, Merck, #06-755) were incubated o/n at 4°C in TBS-T with 5% BSA (Merck, #A3059). The anti-rabbit (Thermo Fisher Scientific, #31460) or the anti-mouse (Cell Signaling Technology, #7076) secondary antibodies conjugated with horseradish peroxidase were incubated for 1 h at room temperature. Detection was performed using the SuperSignal™ West Pico Chemiluminescent Substrate (Thermo Fisher Scientific, #34577) and visualized by autoradiography.

Analysis of RORγt expression was performed on immune cells starting from three healthy donors’ peripheral blood (PB) samples and from five frozen lymph nodes (LN) cell suspensions. Specifically, 250 µL of whole blood were lysed with 2 mL of FACS Lysing Solution (BD Biosciences, #349202), and at least 5×10⁵ cells/tube of LN suspension were stained with Far Red Live/Dead Fixable Stain kit (Thermo Fisher Scientific, #L10120) following the manufacturer’s instructions. Then, cells were incubated with appropriate mixes of fluorochrome-conjugated antibodies to identify target cell subsets ( Supplementary Table S6 ). For all panels, intracellular staining was performed to identify RORγt-expressing cells using the Transcription Factor (TF) Buffer Set (BD Biosciences, #562574), according to the manufacturer’s instructions. Specifically, the unconjugated mouse IgG2a anti-human ROR gammaT monoclonal antibody (Merck, #MABF81) was used as the primary antibody, followed by a PE-conjugated goat anti-mouse IgG2a secondary antibody (Southern Biotech, #1080-09). A minimum of 2×10⁵ CD45⁺ gated leukocytes were acquired and identified based on the forward light scatter (FSC) versus side light scatter (SSC) profile and singlets selection. Lymphocytes, monocytes, granulocytes, innate lymphoid cells (ILCs), and dendritic cell subsets were identified using gating strategies, as follows, while the fluorescence minus one (FMO) controls were used for all the immune cell subsets tested to discriminate between background fluorescence and RORγt2 positive cells.

Lymphocytes: CD4⁺ or CD8⁺ T-cell subsets were gated on CD45⁺CD3⁺ cells. Th17 cells were identified as CCR6⁺CD161⁺ double positive cells gated on CD4⁺CD45RA⁻ cells. Similarly, CD45RA⁻CCR6⁺CD161⁺ cells were selected within CD8⁺ T cells. Analysis of RORγt expression was performed also on CD45⁺CD3⁻ lymphocytes, which comprised CD19⁺ B cells, CD56⁺ CD16^+/−NK cells. Monocytes: CD14/CD16 cells were gated as CD4⁺HLA-DR⁺ cells on the lineage negative (Lin⁻; CD19⁻CD3⁻CD15⁻) fraction and comprised classic CD14⁺CD16⁻, non-classic CD16⁺CD14⁻, and intermediate CD16⁺CD14⁺ monocytes. Granulocytes: CD16⁺ population was gated on SSC⁺⁺CD15⁺ PMN cells. Dendritic cells: DC subsets were defined as CD4⁺HLA-DR⁺ mononuclear cells on Lin⁻ (CD3⁻CD14⁻CD16⁻CD19⁻) gate. Three different DC subsets were then gated according to expression of cell surface specific markers: CD141⁺ for cDC1, CD1C⁺ for cDC2, and CD303⁺ for pDC cells, respectively. Samples were processed on a FACSCanto II flow cytometer (BD Biosciences). Datasets were analyzed using the FlowJo X software (Tree Star Inc.) and FACSDIVA software (BD Biosciences).

Isolation of total RNA was carried out on cell suspensions from lymph nodes and tonsils with an RNeasy mini kit (Qiagen, #74104) according to the manufacturer’s protocol. The cDNA was synthesized from 500 ng total RNA using the qScriberTM cDNA Synthesis Kit 5× and 20× (TwinHelix, #RTK0104). RORγ1/RORC1 and RORγt/RORC2 expression analyses were performed by RT-qPCR technology using the Gene Expression Assay RORC RefSeqNM_001001523/NM_005060 (Hs01076112_m1, Thermo Fisher Scientific, #4453320) and RORC RefseqNM_001001523 (Hs02892670_m1, Thermo Fisher Scientific, #4448892) in combination with TaqMan^® Universal Master Mix II (Thermo Fisher Scientific, #4364340). The glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Thermo Fisher Scientific, #4331182Hs02758991-g1) transcript was used as normalization control. The results were obtained using the Vii-A-7 Real-Time PCR System and analyzed using QuantStudio™ Real-Time PCR software (Thermo Fisher Scientific). The threshold cycle (Ct) was determined for each sample, and quantification was performed using the comparative Ct method. ΔCt was derived as Ct Target − Ct Housekeeping.

To localize RORγt/RORC2 and PRMD16 transcripts by RNAscope assay (Advanced Cell Diagnostics), we used RNAscope 2.5 HD Assay-RED, Hs-RORGT probe (#311061) recognizing the nt 152 to 744 of the RORγt reference sequence NM_001001523 and Hs-PRMD16 (#581961) recognizing the nt 1633–3244 of the PRMD16 reference sequence NM_022114.4. Sections from fixed human tissue blocks were treated following the manufacturer’s instructions, as previously reported (25). Briefly, freshly cut 3-μm sections were deparaffinized in xylene and treated with the peroxidase block solution for 10 min at room temperature followed by the retrieval solution for 15 min at 98°C and by protease plus at 40°C for 30 min. The hybridization was performed for 2 h at 40°C. The signal was revealed using RNAscope 2.5 HD Detection Reagent and FAST RED. Control probes included Hs-POLR2a-C2 (#310451) and dapB-C2 (#310043-C2). For RNAscope and immunohistochemistry, dual stain detections by RNAscope were followed by immunoreaction visualized using Novolink Polymer (Leica Microsystems, #RE7280-CE) followed by DAB or using Mach 4 MR-AP (Biocare Medical, #M4U536) followed by Ferangi Blue (Biocare Medical, #FB813). For RNAscope and immunohistochemistry triple stain, we used the Co-detection Ancillary kit (Advanced Cell Diagnostics, #323180) following the manufacturer’s instructions. Briefly, the antibody was diluted using the Co-detection antibody diluent and incubated overnight. RNAscope proceeded as usual after a step of fixation (using buffered formalin), and after its revelation, a Co-detection blocker was used followed by Novolink polymer and DAB. The third stain was obtained by incubating the last antibody overnight and revealing it through Mach 4 MR-AP followed by Ferangi Blue.

To quantify the usage of each isoform, we used the intermediate output files from vast-tools (eej2 files) for all sample available in VastDB for hg38 (29, 30). These files contain the number of read counts mapping to each exon–exon junction. Thus, to obtain the percent of each isoform, we used the following exon–exon junction read counts: RORγ1 (exons 2γ-3) and RORγt isoform (exons 1γt-3). Only samples with a minimum of 10 read counts in total were considered.

The RNAseq datasets used in this study were downloaded from the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under the following accession numbers:

GSE107011 for Th17 and MAIT cells (31) and GSE136107 for tonsil slan⁺ cells, type 2 conventional CD1C⁺ DCs (cDC2) and CD11B⁺CD14⁺ macrophages (MΦ), as well as peripheral cDC2, classical (CL), intermediate (INT), NC, and slan⁺ NC monocytes (32). In brief, contaminant adapters in the fastq files were detected using FastQC v0.12.1. Subsequently, the removal of adapters and base quality trimming were performed using Trim Galore (v 0.6.19) with the length parameter set to 50. The quantification of transcript isoforms was performed by Salmon (v1.2.0) using a decoy-aware transcriptome (human transcriptome Ensembl version 110) in pseudo-alignment mode and the following settings: library type (−l A), correction of sequence-specific biases (−-seqBias) and position-specific biases (−-posBias). Additionally, Salmon provided the values of transcripts per million (TPM) which were then imported into R using the tximport package with the “dtuScaledTPM” scaling method and visualized using the ggplot2 R package.

Data analyses were performed using the Seurat R package (v4.0.4). Cells with less than 1000 and more than 5000 unique detected features were excluded. Prior to analysis, unwanted cell types were removed from the dataset using cell type specific markers. Normalization was achieved using log transformation. Following normalization, highly variable genes were identified and selected for Principal Component Analysis (PCA) reduction of high-dimensional data and for graph-based clustering analysis. The resolution in the FindClusters function was set at values between 0.2 and 0.5, and the clustering results were shown in a computed Uniform Manifold Approximation and Projection (UMAP) plot (dims = 1:7). Different cell types were grouped and annotated based on cell type defining markers. Selected signature genes were constructed as follows: cDC1: XCR1, BTLA, BATF3, CLEC9A; cDC2: CD1C, NOTCH2, SIRPA, CD14; RORγt-DC: RORC, PIGR, PRDM16, CLEC4A. Differential expression testing was done using non-parametric Wilcoxon rank sum test. Dot plots were generated using the DotPlot function in Seurat. Violin plots were generated using the VlnPlot function in Seurat. Volcano plots were generated using the EnhancedVolcano function in Seurat.

Standard descriptive statistic was used for data summarizing, expressing means, medians, interquartile range, and ranges. Kruskal–Wallis test was used for group comparisons in quantitative variables, and Fisher’s exact test for qualitative ones, as appropriate. A heatmap was built for graphical evaluation of the results. In all the analyses, a significance level of 5% was used. R (version 4.4.1) and R studio were used for statistical analysis.

Human samples were obtained following informed written consent. All experimental protocols were approved by the local IRB (WV, institutional review board code NP906; FF, institutional review board code NP 3719). The methods were carried out in accordance with the approved guidelines. All analyses of human samples, including retrospective analysis of archival material, were conducted in compliance with the Declaration of Helsinki and with policies approved by the local IRB.

Mouse studies have shown that the RORγt isoform encoded by the Rorc locus is restricted to distinct immune cell types, while the RORγ isoform is expressed in many tissues, such as liver, adipose tissue, skeletal muscle, and kidney (33) ( Supplementary Figure S1A ). The tissue distribution of human variants encoded by the RORC locus is not as defined. We noted that the human RORC gene encodes a set of transcription variants, with three of them encoding proteins (from ensemble genome browser https://www.ensembl.org/index.html) (34). The full-length RORγ1/RORC1 (variant 1, isoform 1 or a) generates a 518 aa protein; the RORγt/RORC2 (variant 2, isoform b or 2) encodes a shorter isoform of 497 aa with a different N-terminus region; a third transcript is generated by skipping of exon 3 and encodes for a putative protein of 423 aa, the function of which has never been investigated ( Supplementary Figure S1B ). To examine the tissue distribution of these isoforms, we studied the VastDB RNAseq data (https://vastdb.crg.eu/wiki/Main_Page). We found that the proximal promoter giving rise to the RORγt/RORC2 variant (blue in Supplementary Figure S1B ) is predominantly active in lymphoid tissues ( Figure 1A ), while the distal promoter giving rise to RORγ1/RORC1 is preferred in most non-lymphoid tissues ( Supplementary Figure S1B ; Figure 1A ). This is in line with the observation that RORγt/RORC2 is restricted to immune cells (34). To validate and extend these findings to human lymphoid organs, we tested predesigned TaqMan probes specific for the RORγt/RORC2 transcript. This analysis confirmed the expression of RORγt/RORC2 transcript in RNA extracted from reactive lymph nodes and tonsils. Moreover, analysis of RNA from fractions of cell suspensions of human tonsils revealed RORγt expression in sorted T cells as well as in the lymphoid-depleted fraction (likely containing ILC3 and non-ILC3 innate cells), whereas the B-cell fraction rendered negative results ( Figure 1B ).

In a recent study, we mapped the distribution of RORγt⁺ cells in small intestine and tonsils using the antibody clone 6F3.1 suitable for formalin-fixed tissues (25). Here we extended this analysis to various human tissues. Clone 6F3.1 displayed a strong reactivity in the large majority of human cortical thymocytes, whereas thymus epithelial and stromal cells as well as most medullary lymphoid cells rendered negative results ( Figure 1C ). In the human thymus, RORγt expression has been recently shown in double positive (DP) thymocytes using single-cell genomic approaches (35). Accordingly, in multiplex immunofluorescence on fixed human thymus (n=3), the anti-RORγt 6F3.1 confirmed the nuclear RORγt protein positivity in DP thymocytes ( Figure 1C ). Western blot analysis with the same antibody confirmed high abundance of the RORγt protein in a primary thymocyte suspension, whereas lower amounts were detected in reactive lymph nodes ( Supplementary Figure S1C ).

We further studied RORγt expression at the single-cell level in human lymph node suspensions and sections ( Figures 1D–F ). RORγt was regularly present in lymph nodes in a fraction of CD3⁺ T cells including a subset of γδ T cells, whereas the remaining lymphoid and stromal cells rendered negative results ( Figure 1D ). Lymph node RORγt⁺CD3⁺ T cells were particularly abundant within granulomas in sarcoidosis and mycobacterial infection ( Figure 1E ), two conditions recognized as IL-17-related disorders because of the infiltration by Th17 and Tc17 (36). Flow cytometry (FC) of single-cell suspension of toxoplasma lymphadenitis by clone 6F3.1 also identified numerous CD45⁺RORγt⁺CD3⁺ CD4⁺ T cells and CD45⁺RORγt⁺CD3⁺CD4^– T cells ( Figure 1F ), which corresponded to Th17 and Tc17 based on their expression of CCR6 and CD161. Altogether, these findings indicate that nodal T-cell populations expressing the RORγt are specifically traced in human archival tissues.

In human lymph nodes, a bi-modal spectrum of distribution of the RORγt⁺ cells was found, both in terms of signal intensity and nuclear morphometry (size and morphology). Specifically, in addition to positive small round cells showing weak expression of RORγt and corresponding to T cells, we could also detect a stronger nuclear reactivity for RORγt in cells showing a more heterogeneous morphology also including DC-like cells with large nuclei, with variable shape ranging from kidney-like to polylobate, folded or indented ( Figures 2A, B ). These cells were localized in the inter-follicular area ( Supplementary Figure S2A ). They were found nearby the high endothelial venules ( Figures 2C, D ; Supplementary Figure S2A ), as confirmed by double stain for RORγt and CLA/Heca 452 ( Figure 2C ), and in the marginal sinuses, as shown by co-stain with PDPN ( Figure 2E ). These cells were detectable in a large set of lymph nodes (n=104), throughout all ages in reactive or pathological patterns ( Supplementary Figure S2B and Supplementary Tables S1 , S2 ); among relevant association with the variables of the cohort, we found that this population was particularly enriched in larger lymph nodes showing reactive follicular hyperplasia, whereas their frequency was reduced in granulomatous lymphadenitis ( Supplementary Figure S2B , Supplementary Tables S1 , S2 ).

By multiplex immunostaining, a fraction of RORγt⁺CD3⁻ cells were positive for CD45⁺ ( Supplementary Figure S2C ): these cells encompassed a variable combination of ILC3 and non-ILC3 cells, based on the expression of CD127, CD117, and CD56 ( Supplementary Figure S2D ). Digital image quantification on triple immunofluorescence staining (n=5; Figure 2F ; Supplementary Figure S2E ) showed that RORγt⁺CD3⁻CD127⁻ non-ILC3 cells represented a very small fraction of nodal cells ( Supplementary Figure S2F ) and a minor percentage of RORγt⁺ cells ( Figure 2F ). To further support this observation, we tested anti-RORγt antibody on single-cell suspensions from reactive lymph nodes (n=4) by multiparametric flow cytometry. Within CD45⁺ cells, flow cytometry identified lin⁻CD127⁺CD117⁺HLA-DR⁻RORγt⁺ ILC3s (mean ± SD: 34.32% ± 21.16%), including a CD56⁺ subset consistent with NCR⁺ ILC3. RORγt⁺ ILC3s displayed an evident lymphoid-like morphology according to forward (FSC) and side scatter (SSC) ( Figure 2G ). A second population of lin⁻CD127⁻CD117⁻HLA-DR⁺RORγt⁺ cells were also recognized (mean ± SD: 62.30% ± 21.74%), showing monocytoid features, according to their size and granularity by FSC and SSC ( Figure 2G ). By expanding their phenotypic characterization (n=4) on tissue sections, we could confirm that most nodal RORγt⁺CD3⁻ cells lack lineage markers γδ and αβ TCR chains, CD20, CD56, CD16, and CD14 ( Supplementarys Figure S3A-F ), as well as of markers of follicular dendritic cells (CXCL13 and PDPN; Supplementary Figures S3G, H ), fibroblastic reticulum cells (SMA and CXCL12; Supplementary Figures S3I, J ), and endothelial cells (CD31 and CD34; Supplementary Figures S3K, L ). In conclusion, flow cytometry and immunohistochemistry identify a small fraction of HLA-DR⁺RORγt⁺ non-ILC3 cells in the lymph nodes of a large cohort of cases, which likely correspond to RORγt-DCs.

In our previous study, the characterization of RORγt-DCs found in gut lamina propria was mainly based on transcriptomic analysis (25). Among specific transcript expressed by RORγt-DCs previously identified, we could validate high levels of RORγt and PRDM16 expression on lymph node sections by RNAscope ( Figures 3A–C ). We further confirmed expression of HLA-DR, CLEC4A, AIF1/IBA1, SPI1, LYZ, IL22RA/IL22BP, CD74, and the polymeric immunoglobulin receptor (PIGR) ( Figures 3D–K ). PIGR is a glycoprotein located at the basolateral side of intestinal epithelial cells that mediates the transcytosis of IgA, produced by plasma cells in the gut lamina propria. Remarkably, PIGR was highly restricted to RORγt-DCs among immune cells ( Figure 3K ). We could also detect recurrent expression of S100 protein (using an antibody that recognizes S100B, S100A1, and S100A6 isoforms), CD103, CIITA ( Figures 3L–N ), and CD11C ( Figure 3O ). RORγt-DCs did not express other additional markers expressed in macrophages and DC subsets. Specifically, they rendered negative results for many conventional DCs and Langerhans cells markers (i.e., CD1A, CD207, CD1C, CD303, CD123, CD141, CLEC9A; n=3, Figures 4A–G ), classical macrophage markers (i.e. APOE, CD68 [PG-M1], CD163, CD163L, CD169, CD206, F13A, FOLR2, TIM4, and TREM2; Figures 4H–P ), and transcription factors (TFs) typically found in MΦs (ZEB2 and MAFB; Figures 4Q, R ) and DCs (IRF4 and TCF4; Figures 4S, T ). No reactivity was found for markers of myeloid cells CD66B and S100A9 ( Figures 4U, V ) as well as the cytotoxic molecules GZMB and PRF1 ( Figures 4W, X ). We also explore the abundance of RORgt-DCs and other DCs in the interfollicular area of lymph nodes by analyzing cases (n=4) containing an enriched RORgt-DCs population ( Supplementary Table S1 ). In two of them, the RORγt-DCs component was dominant and outnumbered PDCs, cDC1, and cDC2 DCs in selected areas around HEV ( Supplementary Figure S4 ). All these findings confirm that RORγt-DCs found in draining lymph nodes represent a distinct DC population, as also recently proposed (37).

We further expand and validate this finding by confirming their occurrence, similarity with other nodal DC populations, and antigen presenting features by analyzing human scRNAseq dataset curated in the Cross-tissue Immune Cell Atlas (CTICA) (38). This dataset includes genes of 45 well-annotated immune cell types from 16 different tissues of 12 adult organ donors, providing defined signatures of known immune cell subtypes and valuable information of their tissue distribution. To increase the resolution of our analysis, we first bioinformatically removed known cell types, such as T cells, B cells, monocytes, plasma cells, and conventional ILCs, based on cell type specific markers. We then extracted all remaining cells expressing the XCR1, CD1C, and RORC transcripts from the CTICA and recluster them on UMAP, obtaining a total of eight clusters ( Figure 5A ). Cluster 6 corresponded to RORγt-DCs, given the defining gene signature of RORC, PIGR, PRDM16, and CLEC4A ( Figures 5B, C ). Cluster 3 represented cDC1 population with the expression of XCR1, BTLA, BATF3, and CLEC9A ( Figures 5B, C ). Based on the high expression level of CD1C, NOTCH2, SIRPA, and CD14, we defined clusters 0, 1, 2, 4, 5, 7 as cDC2. Analysis of the tissue abundance of each cell type revealed that RORγt-DCs were prevalently found in spleen, lung-draining lymph nodes (LLN), and mesenteric lymph node (MLN) ( Figure 5D ). We then asked whether nodal RORγt-DCs display a specific transcriptomic profile. To this end, we selected LLN for downstream analysis, given the relative abundance of RORγt-DCs over that of MLN. Many differentially expressed genes (DEGs) were observed in RORγt-DCs compared to nodal myeloid DCs ( Figures 5E, F ); however, the comparison between RORγt-DCs and cDC1 did not yield a robust DEG profile ( Figure 5E ), likely due to the scarcity of these cells. We further examined the expression of the antigen presentation machinery on RORγt-DCs and myeloid DCs ( Figure 5G ). We observed that HLA-DR and HLA-DQ were indeed expressed in RORγt-DCs in LLN ( Figure 5H ). Together, these data identified a small subset of RORγt-DCs which are distinct from conventional nodal myeloid DCs but also express genes related to the antigen presentation machinery in human lymph nodes.

We subsequently analyzed in detail the presence of similar cells in peripheral lymphoid organs, such as thymus, spleen, tonsils, and bone marrow ( Figure 6 ; Supplementary Figure S5 and Supplementary Table S3 ), and non-lymphoid tissues ( Supplementary Figure S6 and Supplementary Table S4 ). RORγt-DCs were largely absent in peripheral non-lymphoid tissues ( Supplementary Figure S6 and Supplementary Table S4 ), except for scatter cells in the gut lamina propria as already described (25), confirmed in this analysis by PIGR co-expression. RORγt-DCs are regularly found in the thymic medulla (n=4; Figure 6A ), in the T-cell area of the spleen (n=10; Figure 6B ; Supplementary Figure S5A ), and in the intraepithelial and interfollicular areas on tonsils (n=24; Figure 6C ; Supplementary Figures S5B-E ); no RORγt-DCs were observed in the bone marrow (n=22; Figure 6D ). We also explored fetal lymph nodes and could confirm a tiny fraction of RORγt-DCs among RORγt⁺ cells ( Supplementary Figure S7 ). Notably, the large majority of thymic and nodal RORγt-DCs were negative for AIRE ( Figures 6E, F ), distinguishing them from mouse TCI and III and JC. Conversely, comparison of the phenotypes of RORγt-DCs with those of TCs reveals more similarity of human RORγt-DCs with mouse TCII ( Figure 6G ). Overall, these data indicate that RORγt-DCs populate human lymphoid organs.

To shed light on the origin of tissue RORγt-DCs, we tested the peripheral blood compartment. By exploring the scRNAseq dataset of peripheral blood mononuclear cells, we could not detect any cluster corresponding to RORγt-DCs (25). We further explored this finding by using flow cytometry on whole blood obtained from healthy donors (n=3). Among CD3⁺ T cells, the CD4⁺Th17 and CD8⁺ Tc17 subsets, identified by their expression of CCR6 and CD161, were positive for RORγt, whereas no reactivity was detected on CD3⁻ mononuclear cells and CD16⁺ granulocytes ( Figure 7A ). Among mononuclear cells, RORγt protein expression was negative on monocytes subsets (based on CD14 and CD16 expression) as well as on BDCA2⁺ plasmacytoid DCs, CD141⁺ cDC1, and CD1C⁺ cDC2 ( Figure 7A ). In keeping with the immunostain results, RORγt/RORC2 variant transcript could not be identified in sorted purified blood monocytes and blood DC populations ( Figure 7B ). Emerging findings indicate that also the DC and MΦ compartments are maintained by the local proliferation of their progenitor or differentiated counterparts (39–41). Double staining for Ki-67 and PIGR and Ki-67 and PRDM16 ( Figure 7C ) as well as stains for Ki-67 and RORγt on sequential lymph node sections ( Figure 7D ) suggested that a significant fraction (mean 31.6%; n=5) of nodal RORγt-DCs in the interfollicular area enter the cell cycle ( Figures 7C, D ). All these findings indicate that RORγt-DCs are distinct from other well-recognized human DC populations and are likely maintained by local renewal.