Sex-biased human thymic architecture guides T cell development through spatially defined niches..stimulates in vivo production of thymulin by human Sex hormones and the thymus in relation to thymocyte proliferation and maturation.

At the Cluny Museum, medieval culture showcases its ancestral knowledge. It took five centuries to discover that the thymus and the genitals are connected, as seen in this statue of the first man to experience desire, through a dream about a mythical serpent.....

Within the thymus, regulation of the cellular crosstalk directing T cell development depends on spatial interactions within specialized niches. To create a spatially defined map of tissue niches guiding human postnatal T cell development, we employed the multidimensional imaging platform co-detection by indexing (CODEX) as well as cellular indexing of transcriptomes and epitopes sequencing (CITE-seq) and assay for transposase accessible chromatin sequencing (ATAC-seq). We generated age-matched 4- to 5-month-old human postnatal thymus datasets for male and female donors, identifying significant sex differences in both T cell and thymus biology. We demonstrate a possible role for JAG ligands in directing thymic-like dendritic cell development, identify important functions of a population of extracellular matrix (ECM)− fibroblasts, and characterize the medullary niches surrounding Hassall’s corpuscles. Together, these data represent an age-matched spatial multiomic resource to investigate how sex-based differences in thymus regulation and T cell development arise, providing an essential resource to understand the mechanisms underlying immune function and dysfunction in males and females.

 

The thymus is the primary organ responsible for the generation and selection of mature, functional, and self-tolerant T cells.1 Effective T cell development is a critical component of our immune system’s ability to accurately and exclusively identify and kill foreign entities such as pathogens. During early postnatal T cell development—the period in life when T cell development is most active2—thymic seeding progenitors migrate to the thymus and mature into thymocytes. Thymic architecture is highly organized to provide spatially defined, stage-specific signaling cues to migrating thymocytes that guide development toward functional mature T cells.3,4,5,6

Recent single-cell sequencing resources demonstrating the diversity of human thymus tissue are incongruous with our current framework of thymus structure and organization,7,8,9,10,11,12,13,14,15,16,17,18,19 which describe a general migratory path thymocytes take through the cortex and medulla during conventional αβT cell development. Spatial transcriptomic sequencing of human thymus has demonstrated a deeper granularity of thymic niches and their evolution during fetal development to support different waves of non-conventional T cells.19,20 However, our understanding of how human postnatal thymus niches support conventional and non-conventional T cell development, T-lineage branching, and alternative lineage development remains limited.3,4,6 T cells generated at this stage of postnatal human development will become the foundation of our immune system, patrolling the body for decades.21 Thus, insights into early postnatal thymus niche biology are crucial to understand how our adaptive immune system is built and how perturbations in postnatal T cell development may emerge as immune dysfunction later in life.

To create a spatially defined map of tissue niches guiding human postnatal T and alternative lineage cell development, we employed multi-dimensional spatial proteomic imaging using co-detection by indexing (CODEX),22,23 single-cell transcriptomic-proteomic profiling using cellular indexing of transcriptomes and epitopes sequencing (CITE-seq),24 and single-cell assay for transposase accessible chromatin sequencing (ATAC-seq).25 Given the emerging recognition of sex differences in thymus gene expression and function,26,27,28,29,30,31 we collected and analyzed samples from male and female donors. Our analysis identifies significant sex differences during early postnatal development that affect T cell and thymus biology through common and cell type-specific mechanisms. Additionally, we highlight key cell types contributing to thymic involution that exhibit sex-based differences in thymic growth and early transition toward adipogenesis. These data suggest that kinetic differences in thymic involution are present between sexes and, importantly, that mechanisms driving thymic involution begin early in life. Altogether, these data represent a powerful age-matched spatial multiomic resource to investigate how sex-based differences in thymus biology and T cell development arise, and how they contribute to sex differences in diseases caused by immune dysfunction.

Results

Spatial multiomic profiling of human postnatal thymus identifies sex-based differences in T cells and thymus biology

We performed single-cell CITE-seq, ATAC-seq, and CODEX imaging on 4–33 months human postnatal thymuses, including 6 (3 female and 3 male) 4- to 5-month-old age-matched samples (Table S1). Each donor sample was processed simultaneously for CODEX imaging and sequencing (Figure 1A). We included a comprehensive 137 antibody panel (Data S1), allowing us to compare epigenomic, transcriptomic, and proteomic expression kinetics across developing thymocytes and enabling direct comparison of cells identified via phenotypic expression in CODEX with cells captured via CITE-seq. Prior to sequencing, we enriched CD45− non-hematopoietic cells and CD25+CD8− regulatory T (Treg) cells to ensure coverage of low-abundance cell types. After quality control and computational merging of individually sequenced patient datasets, we obtained a total of 74,334 cells with CITE-seq, including 19,434 non-T-lineage cells, and captured 25,717 nuclei with ATAC-seq. Importantly, cell proximity in CODEX tissue niches was used to screen predicted receptor-ligand interactions.

 

Figure 1 Spatial multiomic analysis identifies sex-biased characteristics of thymic niches

Show full captionFigure viewer

CITE-seq cells were clustered based on transcriptional expression and annotated based on marker gene and surface protein expression (Figure S1A; Table S2).7,8 ATAC-seq clusters were computationally labeled using CITE-seq reference cluster labels, which identified 34 ATAC-seq cluster transfer labels for dataset integration (Figures 1B and S1B). We captured 54,900 thymocytes spanning development from early thymic progenitors (ETPs) to mature single positive (SP) T cells, immature innate cells, innate-like cells, and Tregs. We identified three Treg populations expressing canonical lineage markers, namely Treg progenitors (Pro-Tregs), thymic Tregs (tTregs), and recirculating/resident Tregs (rrTregs).32 We also identified antigen-presenting cells, including B cells, mast cells, monocytes, and six populations of dendritic cells (DCs).33 In addition to the activated DCs (aDCs), plasmacytoid DCs (pDCs), DC1, and DC2/3 populations described by Park et al.,7 we found proliferating populations of pDCs and DC1. We also captured 7,093 epithelial cells, including cortical epithelial cells (cTECs), medullary epithelial cells (mTECs), activated mTECs, and mimetic TECs.

Importantly, we captured 7,721 mesenchymal cells, which contribute to negative selection and thymic involution.9,19,34,35,36 Subclustering identifies important mesenchymal cell types, including two populations of endothelial cells (ECs) defined by differential expression of Notch ligands (ECs, ECs (Notch)). Additionally, we identified lymphatic ECs (LECs), pericytes, vascular smooth muscle cells (VSMCs), and five distinct fibroblast cell types, including DPP4+ capsular fibroblasts (DPP4+ capFibs), capsule fibroblasts (capFibs), medullary fibroblasts (mFibs), KRT+ fibroblasts (KRT+ Fibs), and proliferating fibroblasts (Fibs (P)).

We imaged each tissue sample with a custom 48 antibody CODEX panel to study the architecture and function of niches guiding thymocyte development, aiming to define the niche characteristics guiding T-lineage branch points. Stage-specific thymocyte phenotyping markers (CD62L, CCR7, CD1A, CD5, CD7, CD4, CD8, CD3, CD45RO, CD45RA, FOXP3, and SATB1) identified CD3+ double positive cells (DPs) undergoing T-lineage commitment toward CD4 or CD8 T cells. Phenotyping markers for non-T-lineage hematopoietic cells (CD19, CD11c, CD11b, and CD68), epithelial cells (EPCAM and KRT5/8), mural cells (MCAM and SMA), ECs (CD31), and fibroblasts (PDGFRA) identified the remaining major cell types defining thymic niche architecture. Finally, we included functional markers to define patterns of antigen presentation (CD86), human leukocyte antigen (HLA) class I and II expression (HLA-ABC and HLA-DR,DP,DQ), adhesion ligands (ICAM and VCAM), Notch ligands (DLL1, DLL4, JAG1, and JAG2), T cell activation (PD-1), self-tolerance (PD-L1), proliferation (Ki67), and enzymatic regulation (15-PDGH). In sum, our CODEX panel enabled investigation of spatially regulated mechanisms directing human T cell development.

Using neural-network-driven cell segmentation and Leiden-based clustering,23 we identified individual cells within thymic tissue for each sample (Figure S1C). We annotated cell types based on tissue location and phenotypic expression compared with CITE-seq clusters (Figure 1C), performed proximity-based neighborhood clustering to identify niches,23 and annotated niches based on location and cell type composition (Figure 1D; Figure S1D). This analysis quantified proximity-based cell-cell interactions (Figure S1E) and served as a platform to interrogate spatially defined thymic niche biology via integrated sequencing-imaging analysis.

Because of known sex differences in thymus and T cell gene expression,31 we compared our age-matched male and female samples separately. In line with prior reports of sex-biased gene expression on autosomes,37,38,39,40 only 2% of male differentially expressed genes (DEGs) were found on the Y chromosome and 0.3% of female DEGs were found on the X chromosome (Tables S3 and S4). Gene set enrichment analysis (GSEA) on male vs. female cells for each cell type identified pathways commonly upregulated in either sex (Figure 1E; Data S1). Pathways differentially regulated across hematopoietic, epithelial, and stromal cells represent cell-intrinsic sex-based differences. Female cells have higher gene expression of transcription, energy regulation, and antigen presentation. Male cells, by contrast, have increased gene expression of proinflammatory signaling, amino acid metabolism, and G protein-coupled receptors (GPCR) signaling. The top differentially expressed energy regulation and metabolism pathways were similarly sex-biased in human kidney,41 suggesting multiple organs show consistent sex-biased enrichment of pathways linked to metabolism and energy production. Our data align with sex-biased trends identified in human induced pluripotent stem cell (iPSC) lines42 and other human organs,43 indicating these pathways often differ between male and female cells across various cell types.

By contrast, some pathways showed cell type-specific sex-biased enrichment. Female T and hematopoietic cells showed enrichment of interferon signaling, and female fibroblast and perivascular cells were enriched in extracellular matrix (ECM)-centric pathways (Figure 1E). Our dataset also identified differential sex-specific pathway enrichment between cell types. Gene expression indicated higher cytokine signaling in T cells and hematopoietic cells in females and in epithelial and mesenchymal cells in males (Figure 1E). These data show significant gene expression differences in male and female thymic cells. To demonstrate sex differences at the proteomic level, we identified genes with a log fold change greater than 1 that contributed to increased chemokine signaling in male T cells. CXCR4, an important chemokine receptor in thymocyte migration and development, had increased expression in male progenitor T (pro-T) cells, which we confirmed via flow cytometry (3 male, 3 female; p = 0.03; Figure S1F). As higher levels of cytokine and interferon signaling have been previously shown to influence thymus and T cell biology,44,45 our data suggest male and female T cells develop in different signaling environments and may respond differently to cytokine stimuli.

Next, we quantified cell type abundance within male and female tissues, demonstrating differences in cortical and medullary cell distributions between sexes. When normalized to the total number of cells per lobe, female thymus lobes contained significantly more DPs (p = 0.011) and cTECs (p = 0.0023). In males, we found significantly more SPs (p = 4.2 × 10−4), CD3+ DPs (p = 9.9 × 10−4), activated mTECs (p = 0.0014), and VSMCs (p = 2.4 × 10−6) (Figure 1F). Given that thymus lobules with more DPs and cTECs would have a greater proportion of cells undergoing positive selection and lobules with more medullary cells would have more cells undergoing negative selection, these data suggest that sex differences in cell type abundance may influence the resources directed toward specific stages of thymocyte selection. Alternatively, these results may suggest that male and female thymuses are developmentally asynchronous, with males exhibiting faster growth and involution kinetics, resulting in decreased cortical-to-medullary ratios even in early neonatal stages. We focused further analyses on sequential developmental niches, including analysis of sex differences in cell types and niches at each stage.

JAG1 skews ETP development toward thymic DCs

We first analyzed the cortico-medullary junction (CMJ) where cells home to the thymus (Figure 2A). This region recruits and supports ETPs10 and is composed of ECs, VSMCs, and pericytes expressing the Notch ligand JAG1 (Figures 2B and 2C). CITE-seq demonstrated that the cell adhesion molecule used by ETPs to enter the thymus, CD62L, is quickly downregulated upon CMJ entrance through the vasculature (Figure S2A). However, recently immigrated CD62L+ double negative cells are frequently located in the subcapsular zone (Figure S2B), suggesting that ETPs enter the thymus and rapidly migrate to a subcapsular niche where DLL4, a more potent Notch ligand, is highly expressed on fibroblasts and subcapsular epithelial cells (Figures 2D and S2C). However, the concentrated presence of JAG1 at the entry point indicates that ETPs are first exposed to this Notch ligand.

 

Figure 2 Thymic progenitors entering via the corticomedullary junction are exposed to a gradient of Notch ligands, which influence lineage specification

Show full captionFigure viewer

CellChat46 pathway analysis showed that JAG1-NOTCH1 interactions between endothelial and perivascular cells are enriched with ETPs (Figure 2E), while JAG1-NOTCH2 and JAG1-NOTCH3 interactions are enriched with DC1, DC1 (P), DC2/3, and aDCs (Figures 2E–2G). These data suggest that JAG1 could induce commitment toward other hematopoietic lineages, such as pDCs, conventional DCs (cDCs), or macrophages, which are known to develop within the thymus.10 As JAG ligands induce weaker Notch induction,47,48,49,50 we hypothesized that early contact with ETPs could maintain T-lineage potential while cells migrate toward DLL4 in the subcapsular niche.

We first analyzed the ability of the four thymic Notch ligands to induce T-lineage commitment or alternative lineage development from cord-blood-derived CD34+ hematopoietic stem and progenitor cells (HSPCs) in a defined, feeder-free culture system44 (Figure 2H). We included titrated concentrations of granulocyte-macrophage colony-stimulating factor (GM-CSF), which is produced by mast cells at the CMJ, to support DC development.51 We found that only DLL1 and DLL4 ligands induce T-lineage commitment, whereas JAG ligands or no ligand controls supported myeloid cell development and did not induce T-lineage commitment (Figure S2D). Specifically, JAG ligands with GM-CSF skewed CD68+ DC development toward CD14− DC1 cells, while no ligand controls skewed CD68+ DC development toward CD14+ DC2/3 cells (Figures 2I and S2E).

Next, to test our hypothesis that Notch signals via JAG1 ligands could act as a bridge toward later DLL4 interactions, we analyzed cells grown on JAG1 for 3, 5, or 7 days prior to DLL4 transfer (Figure 2J). We found that cells cultured on JAG ligands or no ligands for 3 days maintained reduced T-lineage commitment compared with DLL1 or DLL4 cells (pJAG1 = 0.033; pJAG2 = 0.017), whereas cells cultured on JAG ligands for longer than 3 days lost T-lineage potential (Figure 2K), indicating that JAG ligands could not support T-lineage potential.

We next analyzed the contribution of different Notch ligands to the development of male and female ETPs (Figures S2F and S2G). Our data suggest that JAG ligand interactions are more abundant and diverse in females, with JAG1-NOTCH1 interactions enriched in female ETPs and DLL4 interactions enriched in male ETPs.

Together, these data suggest that timely migration from the CMJ to DLL4 ligands at the subcapsular zone is critical for T-lineage commitment, and exposure to JAG ligands at the CMJ can guide alternative lineage development toward thymic-derived DCs. Our data further demonstrate previously unrecognized sex-biased regulation by Notch ligands.

Analysis of the subcapsular zone identifies sex-based differences in fibroblast regulation of DP development and thymus growth

From the CMJ, ETPs migrate to the subcapsular zone via a CCL25-CCR9 chemokine gradient established by cTECs and directed to pro-T, DP (P), and DP2 (Q), but not DP1 (Q) cells (Figure 3A; Figure S3A). The subcapsular niche consists of JAG1+ VCAM1+ DCs, cTECs, capsular fibroblasts, DPP4+ capsular fibroblasts, and proliferating fibroblasts, which secrete and maintain spatially regulated ECM ligands to support sequential thymocyte development (Figures 3B and 3C; Figure S3B and S3C).

 

Figure 3 Fibroblasts in the subcapsular zone contribute to regulation of thymus biology and T cell progenitor development

Show full captionFigure viewer

GSEA showed that DPP4+ capsule fibroblasts were enriched in HSP90 chaperone cycle for steroid hormone receptors (padjusted = 0.0065; 18/52 pathway genes significantly upregulated) (Data S1), suggesting an enhanced response to steroid hormones and supporting their role in sex hormone-based thymic involution.9 By contrast, capFibs were enriched for genes related to cytokine (interleukin [IL]-33, padjusted = 1.50 × 10−6; IL-34, padjusted = 3.56 × 10−7) and chemokine signaling (CCL2, padjusted = 5.10 × 10−40; CXCL3, padjusted = 0.020; CXCL12, padjusted = 1.78 × 10−8; CXCL14, padjusted = 3.63 × 10−15), functions previously attributed to TECs. Furthermore, CellChat identified cortical fibroblasts as major contributors to insulin growth factor (IGF) signaling through predicted signaling to cTECs, which are found in close proximity in the cortex (Figure S3D), via IGF2-IGF1R and IGF1-IGF1R axes, and to ETPs and β-selection cells, which were found under the capsule (Figure S2B), via an IGF2-IGF2R axis (Figures 3D–3F).

We next explored the role of proliferating fibroblasts. GSEA comparisons between capFibs and Fibs (P) showed marked differences in signal transduction pathways. CapFibs resembled traditional fibroblasts, which upregulate tyrosine kinase, angiogenesis, and ECM regulation and deposition pathways, whereas Fib (P) upregulates WNT signaling and cell sensing pathways, including genes involved in transient receptor potential (TRP) channels in the stimuli sensing channels pathway and taste receptors (TASRs) (Figure 3G; Data S1). Interestingly, CODEX images identified ECM− PDGFRa+ fibroblasts lacking extra domain A fibronectin (EDA-FN) expression, indicating that Fibs (P) are not involved in fibrotic matrix deposition unlike capFibs (Figure 3H; Figure S3B). Fibs (P) form a network of PDGFRa+ cells throughout the cortex that does not overlap with the cTEC network, yet maintain cell-cell contact in specific niches and often localize near cortical capillaries (Figure S3D).

We found sex-specific differences in vascular endothelial growth factor A (VEGFA) signaling within ECM− fibroblasts (Fib (P)) and other mesenchymal cells. Although all thymic fibroblasts produce the angiogenesis growth factor VEGFA, male fibroblasts express more than female cells (Fibs (P): padjusted = 0.0306; DPP4+ capFibs: padjusted = 0.0318; mFibs: padjusted = 1.85 × 10−6) (Figure 3I). Given that postnatal male thymuses are larger than female thymuses in humans and primates26 (Figure S3E), male fibroblasts may provide increased VEGFA to support angiogenesis and rapid thymic growth observed during postnatal development.52 Additionally, male mFibs have higher expression of FGF7 (padjusted = 0.0154), which regulates thymus size.53 CellChat predicts that male Fibs (P) are enriched in FGF10 compared with females, which supports cTEC proliferation and vascular growth,53,54 and only male VSMCs express FGF18 (Figures S3F–S3H). These sex biases in fibroblast growth factor (FGF) gene expression may contribute to the larger size of early postnatal male thymuses by stimulating epithelial and EC growth and proliferation.

Comparison of DEGs between male and female mesenchymal cells found increased expression of adipogenesis, cytokine, and GPCR signaling pathways in DPP4+ capFibs (Figure 3J). We also found increased expression of APOD, a gene associated with androgen, estrogen, progesterone, and glucocorticoid signaling,55,56 across male fibroblast populations (Fibs (P): padjusted = 2.18 × 10−26, mFibs: padjusted = 8.45 × 10−32) (Figure S3I). Given the association of hormone signaling with thymic involution,29,52,57 these findings suggest early initiation of thymic involution in postnatal males.

In sum, we identified three roles for fibroblasts within the subcapsular niche: maintaining tissue structure and organization via ECM and chemokine signaling, directly regulating cTEC maintenance and expansion, and potentially coordinating T cell development directly through growth factors and cell-cell interactions.

Human postnatal thymocytes may self-select in the cortex to support positive selection of conventional αβT cells

Upon exiting the subcapsular zone, DPs migrate into the inner cortex toward the medulla, where they receive positive selection signals that guide T-lineage branching toward CD4 or CD8 SP cells (Figure 4A). For DPs to transition toward the CD4 lineage, cells must receive T cell receptor (TCR) stimulation through HLA class II interactions, yet previous mouse studies have shown transcriptional downregulation of HLA class I and II in DPs.58,59 Low transcriptional expression is hypothesized to prevent thymocyte-thymocyte self-selection during positive selection, necessitating DP interactions with cTECs to receive positive selection signals.

 

Figure 4 HLA class I and II interactions may support thymocyte positive selection in the inner cortical zone

Show full captionFigure viewer

Analogous to mouse literature, quiescent human DPs do not express HLA class II transcripts and have closed CIITA promoters (Figures 4B and 4C). Despite the lack of class II mRNA, thymocytes express low levels of HLA class II protein throughout development (Figure 4B). Additionally, in contrast to mouse data, we observe constitutive class I mRNA expression, which increased as cells transitioned toward SPs (Figure 4D). This is consistent with ATAC-seq data demonstrating that the B2M promoter is open throughout thymocyte development (Figure 4E). We confirmed HLA expression via flow cytometry and found that approximately 25% of DPs express both class I and II, and over 65% of DPs are class I+ (Figure S4A). Thus, thymocyte self-selection within the cortex could support positive selection. In support of this notion, CODEX enabled us to identify locations within the cortex devoid of epithelial, fibroblast, endothelial, or DCs but packed with DPs expressing class II+ molecules concentrated at cell junctions (Figure 4F). We confirmed the absence of spindle-like cTEC projections in this niche via confocal imaging (Figure 4G). Additionally, we quantified cell-cell interactions and identified a niche (positive selection niche 1) consisting of class II+ DPs and CD3+ DPs and a niche (self-selection niche) containing mainly class II+ DPs (Figure 1D). Finally, we sorted thymocytes to isolate immature DPs (CD4+CD8+CD3−TCR−) and mature DPs (CD4+CD8+CD3+TCR+) from three donors and cultured them for 7 days in a feeder-free assay. In the absence of epithelial cells, both immature and mature DPs upregulate HLA class II proteins (Figure 4H), and immature DPs continue to mature along their developmental pathway, as indicated by increased percentage of CD27+ DPs in culture after 7 days (Figure 4I).

Next, we identified a niche that directs T-lineage commitment toward CD4 or CD8SPs. We performed differential gene expression analysis on clusters representing this lineage branch point to identify markers for our CODEX panel (Figure S4B). We found SATB1 expression increased as DPs transitioned toward SPs (Figure S4C), and compared with CD8SP transition cells, CD4SP transition cells had higher expression of this master transcription factor60 (Figures S4D and S4E). Imaging analysis confirmed increased SATB1 expression coincides with CD3 upregulation, consistent with a role in late DP development and lineage branching (Figure 4J).7 Neighborhood analysis identified a niche enriched for mature CD3+ DPs in the inner cortex, suggesting that there either exists a niche specifically for late DP development and CD4 lineage transition or that cells are pre-disposed to CD4 lineage development through their TCR and migrate as clonal populations after proliferation at the outer cortex.

We compared cortical niche organization between sexes and found differences in niche organization supporting conventional T cell development, self-selection, and cross presentation. Females showed increased neighborhood interactions between the cortical DC niche containing JAG1+ VCAM+ DCs and the mature DP niche containing CD3+ DPs, the positive selection niche 1 containing class II+ DP cells and CD3+ DP cells, and the positive selection niche 3 containing DCs and DPs (Figure S4F) as well as increased cell-cell interactions between cTECs and class II+ DPs (Figures S4G and S4H). Conversely, males had increased cell-cell interactions between cTECs and CD3+ DPs (Figures S4G and S4H). These data suggest that the proportionally larger female cortex could increase cross presentation from DCs and cTECs to class II+ DPs, possibly facilitating greater use of self-selection as an alternative mechanism for positive selection.

Taken together, spatial multiomic analysis of the inner cortex identified cortical niches supporting specific stages of DP development, including three positive selection niches, a specialized niche for self-selection, and a mature DP niche thymocytes migrate through prior to entering the medulla.

Spatial multiomics identifies key mechanisms regulating negative selection niches in the medulla

Mature DPs enter the medulla, an environment specialized for negative selection, and transition toward CD4 or CD8 lineages (Figure 5A). Within the medulla, cells specialized for negative selection localize around keratinized structures called Hassall’s corpuscles (HCs).61 HCs appear during late prenatal development and are abundant in human postnatal thymuses but rare in mice.62 Here, we demonstrate that HCs can be divided into three major components: an external epithelial border of highly keratinized cells, an inner border of cells expressing prostaglandin-degrading enzyme 15-PGDH (HPGD), and a central PDGFRa+ mass (Figure 5B). HCs produce thymic stromal lymphopoietin (TSLP),61 an analog of IL-7, which activates DCs to increase expression of class II and co-stimulatory molecules CD80 and CD86. Importantly, subclustering stromal populations identified a population of KRT+ fibroblasts resembling cells undergoing epithelial-to-mesenchymal transition (EMT)63 (Figures S5A and S5B). CITE-seq identified TSLP and 15-PGDH mRNA expression in KRT+ Fibs, mFibs, mTECs, activated mTECs, and aDCs (Figure 5C), implicating these cell types as potential contributors to the function of HCs. Finally, given the inner layer of 15-PGDH+ cells, we explored the role of prostaglandin signaling regulation within the medulla. We found that DC1 cells express high levels of PGE2, whereas DC2/3 cells and monocytes express the PTGER2 and PTGER4 receptors, and aDCs express the PTGER3 receptor (Figure 5C), suggesting prostaglandin signaling is a major regulator of DC activity near HCs.

 

Figure 5 HCs represent scalable organizing centers for negative selection in the neonatal thymic medulla

Show full captionFigure viewer

CODEX imaging suggests HCs act as sub-medullary organizational centers to segregate the inner medulla into specialized niches for negative selection. CD86+ APCs, a subset of which express the co-stimulatory ligand CD40, localize near HCs and in direct contact with CD45RA+ mature SPs (Figure 5D; Figure S5C). In addition, approximately 30% of medullary area is composed of CD19+ B cells,64 which cluster into niches surrounding HCs (Figure S5D). These B cells are found in close contact with—and are often enveloped within—mTECs, potentially facilitating cross presentation with epithelial cells (Figure 5E). These results suggest thymic B cells may comprise an important source of antigen presentation for negative selection.64,65 We quantified medullary neighborhoods and identified six niches, including an mTEC maturation niche, a cross-presentation niche, and four niches specialized for negative selection, which vary in relative location to HCs or the CMJ, as well as their composition of APCs, epithelial, and T cells (Figure 1D; Figure S1D).

Negative selection niches surrounding HCs play a key role in conventional T cell and tTreg development.61 We enriched CD25+ cells for sequencing and found a population of CD25hi pro-Tregs expressing canonical Treg markers CTLA-4, TNFRSF1B (TNFR2), and TNFRSF4 (OX40); positive/negative selection markers (ITM2A, RANBP1, NCL, NME1, MIF, and ATP5G1); Treg developmental long non-coding RNA (MIR155HG)66,67,68,69; and other markers described in mice (Figure S5E). Whereas pro-Tregs expressed high levels of pro-apoptotic gene BCL2L11, mature tTreg subsets expressed the anti-apoptotic gene BCL2. Gene network reconstruction via SCENIC70 identified transcription factor networks activated during pro-Treg to tTreg transition (Figure 5F).

The thymus also contains mature, highly activated Tregs, labeled as rrTregs, believed to have recirculated from the periphery.71,72 rrTregs lack expression of CCR7 or thymic egress markers (KLF2 and S1PR1) but express IL1R2 (Figure S5F), which sequesters the inflammatory cytokine IL-1β to reduce local concentrations.73 CODEX imaging identified tTregs and rrTregs dispersed throughout the medulla, with rrTregs primarily adjacent to CD68+ DCs (Figure 5G). CellChat supported the potential of rrTregs to sequester inflammatory cytokines through interactions with DC2/3 via an IL-1β-IL-1R2 axis (Figure S5G). rrTregs also exhibited a tissue resident Treg phenotype (BATFhigh CCR8+) associated with wound healing and tissue regeneration function,74 and expressed remodeling and tissue repair-related genes such as matrix metalloproteinase enzymes (MMP25 and ADAM19) (Figure S5H). Overall, these findings illustrate Treg diversity in the thymus with their developmental trajectories and functions yet to be elucidated.

Comparisons of male and female rrTregs showed that male rrTregs had higher expression of IL-4 and IL-13, heat shock factor protein 1 (HSF1), and IL-1 signaling pathways (Figure 5H), suggesting rrTreg-mediated regulation of IL-1R2-mediated anti-inflammatory feedback checkpoints is a more prominent mechanism in male tTreg development in early postnatal thymus. Notably, male-activated mTECs have higher expression of CD40 and tumor necrosis factor (TNF) inflammatory pathways than females, possibly resulting in higher rrTreg activity (Figure S5I).

Finally, as Tregs have been shown in mouse to contribute to thymic involution through JAG1,75 we explored sex-based differences in tTreg gene expression. GSEA showed male rrTregs and tTregs have higher expression of adipogenesis pathways (Figures 5H and 5I). Given the presence of cells undergoing EMT, our data underlie the aggressive timeline of thymic involution and suggest that sex-based differences in thymus functional decline begin early in life.

Our detailed examination of the medulla identifies several niches specialized for negative selection, cross presentation, and mTEC maturation around HCs and demonstrates sex biases in inflammatory pathways and thymic involution kinetics within these niches.

Discussion

We performed spatial multiomics to construct a tissue atlas of niches guiding T cell development in early human postnatal thymus. These datasets characterize how key developmental niches drive lineage branch decisions, identify a possible mechanism for conventional αβT cell development through self-selection, and suggest additional functions for mesenchymal cell types governing thymus biology. Furthermore, we discovered several sex-specific differences in thymus cell and niche biology. As T cell development is a dynamic migratory process, knowledge of cell position in combination with proteomic, transcriptomic, and epigenomic sequencing data provides an invaluable resource to predict niche-specific signaling cues directing T cell development, and mechanisms responsible for maintaining tissue structure and directing thymic involution.

We describe an approach to sequencing analysis using multidimensional imaging to establish benchmarks for the location, ligand expression, and composition of key niches in T cell development. This enables us to analyze cell-cell interactions guided by niche composition, identifying physiologically relevant ligand-receptor interactions based on cell proximity within the tissue. Ultimately, this approach maps epigenomic, transcriptomic, and proteomic data to distinct tissue niches at single-cell resolution. Furthermore, we included equal numbers of male and female age-matched thymus samples, enabling comparison between sexes across platform modalities. Our analysis of sex-matched human early postnatal thymus demonstrates the highly plastic nature of thymus lobule organization and resource dedication. Each niche responds to sex-biased developmental kinetics, supporting robust T cell development to ultimately produce functional immune systems in different manners (Figure 6). The findings herein describe only a subset of the data, and we encourage the community to capitalize on this resource to provide further insight into sex differences and targeted niche-specific inquiries.

 

Figure 6 The human early postnatal thymus lobule is spatially organized into sex-biased niches to support stage-specific T cell development

Show full captionFigure viewer

In our analysis of Notch ligands, we complemented our in silico approach with in vitro analysis. Our analysis suggests that JAG1 at the CMJ cannot support T-lineage commitment as cells migrate toward the subcapsular zone but instead skew alternative lineage development toward a CD14− DC1 subset (Figure 6). CD14 expression on DCs is linked with increased inflammatory cytokine production,76 suggesting that JAG ligands promote non-inflammatory DC phenotypes. These results highlight the importance of precise Notch signaling strength and timing in the thymus and emphasize the need for strict spatial control of different Notch ligands within thymic niches. Our observation of high JAG1 expression in the medulla and decreased DLL4 expression on cTECs outside the subcapsular zone aligns with previous studies on human postnatal thymus.77

In the subcapsular zone, we characterize the important roles of specialized fibroblasts. DPP4+ capFibs, described in mouse as cells with progenitor and anti-fibrotic potential,78,79,80,81,82 are observed as a fibroblast subset responsive to changes in systemic hormone levels. Since thymic function and involution are regulated by sex hormone levels,57,83,84,85 DPP4+ capFibs likely control these processes and are potential targets for addressing age-related thymic involution.86 Previously, only medullary fibroblasts were linked to thymocyte development and selection in the medulla.82 We demonstrate that capFibs may directly support thymocyte development in the cortex by producing growth factors like IGF2 (Figure 6). Blocking IGF2 signaling arrests thymocytes at the double negative stage,87 and our data identify capFibs as the IGF2 source, suggesting capFibs as an additional cell source of cytokines and growth factors for in vitro developmental systems. Finally, we demonstrate that ECM profiles of thymic fibroblasts are tightly regulated based on spatial localization. Future work should characterize how tissue stiffness changes as thymocytes migrate through developmental thymic niches to improve biomaterial strategies for in vitro T cell development.88

Furthermore, we identify a population of ECM− cortical fibroblasts that are enriched in cell sensing pathways, such as TASRs and TRP channels. Interestingly, TASRs regulate cell responses to local soluble substances, such as glucose, modulating release of hormones and other signaling molecules.89 Similarly, TRP channels play roles in cell sensing, such as pheromone signaling, nociception, temperature sensation, and osmoregulation.90 Given the proximity of these cells to vasculature in the cortex, Fibs (P) may play a critical role as regulatory cells by sensing environmental changes and modulating thymus size (Figure 6). Their lack of ECM production and network-like structure resemble fibroblast reticular cells (FRCs) in the lymph node, which rapidly proliferate and remodel the cortex during infection.91 Our data are generated from early postnatal thymus samples, an age with active T cell development, suggesting these fibroblasts expand the thymic cortex similarly to FRCs during infection, signaling through FGF and IGF to stromal and epithelial cells to orchestrate remodeling.

While the dogma in thymocyte positive selection suggests that DPs downregulate class II RNA to prevent self-selection and force interactions with cTECs,58,59 several studies suggest that T-lineage cells can select off each other to support CD4 T cell development.20,92,93,94 Here, we describe an inner cortical niche where class II+ DPs reside that may support positive selection via DP-DP self-selection (Figure 6). We show that immature DPs cultured without epithelial cells upregulate HLA class II and continue to mature and receive positive selection signals. Additionally, upregulated SATB1 expression identifies mature DPs in an inner cortical niche and the CD4 branch of their progeny, suggesting it may determine early lineage specificity. Future work should investigate critical features of this niche and SATB1’s role in thymocyte development.

Within the medulla, we identified a niche adjacent to HCs specialized for negative selection and highlighted the role of rrTregs in modulating the medullary inflammatory environment (Figure 6). The abundance of HCs in human but not mouse, and their proximity to negative selection niches, suggests these structures evolved to provide niche-level organization within the larger human medulla or to regulate negative selection more stringently in longer-lived species.

Comparing male and female tissue showed sex differences in both T cell and thymus biology. Studies on post-pubertal males and females show that sex hormones differentially regulate thymic involution between sexes,26,27,28,29,30,52,57,84,86 and that androgen blockers increase FOXN1 expression, thymic involution, and increased rejuvenation.29,30,52,84,86 Additionally, older males produce fewer recent thymic emigrants and have smaller thymuses compared with females.26,28 Some studies describe decreased numbers of AIRE+ mTECs with age and in females,95 potentially predisposing females who maintain greater thymic function later in life to autoimmune disease.29 These studies also observe less interlobular fat in young female thymus,26 suggesting differences in thymic involution kinetics begin pre-puberty. However, current literature has not addressed transcript-level sex differences underlying functional differences in thymic and immune function. Our analysis uncovers that female thymic cells upregulate energy regulation, transcription, and antigen-presentation pathways, whereas male cells increase proinflammatory signaling, amino acid metabolism, and GPCR signaling. These cell metabolic differences align with transcript-level sex differences in other organs41,42,43 and highlight the need for sex-based cell culture optimization in in vitro T cell culture systems.

In addition to changes common to other organs,40,41 we identify thymus-specific differences affecting key processes in thymocyte development and training. Females have a larger proportion of cortical cells per lobule, aligning with lower thymic involution rates and a larger cortex/medulla ratio.26,27,52 ETPs have enriched interactions with JAG1 as they migrate away from the CMJ, suggesting increased JAG1 interactions could skew ETP lineage commitment toward less inflammatory DC phenotypes (Figure 6). In the female cortex, we observe increased cTEC and class II+ DP interactions and increased interactions between cortical DC and positive selection niches, suggesting thymocyte self-selection may play a larger role during positive selection (Figure 6). Conversely, the female medulla shows decreased inflammatory pathway activation and fewer medullary cells. These data suggest females prioritize generating a larger repertoire of DPs over deleting autoreactive cells through negative selection, potentially contributing to sex differences in autoimmune disease prevalence in females.96

In males, we observe enriched DLL4 interactions with ETPs, which aligns with previous data demonstrating that androgen levels positively correlate with DLL4 on cTECs.29 The male cortex shows increased interactions with mature CD3+ DPs and cTECs, suggesting male thymocytes may have lower proliferation rates post β-selection, allowing sufficient space for positive selection. In the medulla, male-activated mTECs exhibit increased inflammatory pathway markers, and male Tregs exhibit higher inflammatory modulation and activate thymic involution pathways.75 Upregulation of inflammatory modulation by male rrTregs may regulate the higher proinflammatory signaling in male cells (Figure 6). Interestingly, post-pubertal males have more Tregs and fewer CD4 T cells than females, possibly due to a more inflammatory medullary environment skewing CD4 development toward the Treg lineage.31

We further explore sex differences in thymus size control mechanisms. Among fibroblast populations, we find significant differences in expression of growth and angiogenesis factors, such as VEGFA and FGFs, potentially contributing to the size difference in male and female thymuses at this age (Figure 6). These data align with and extend known sex differences in growth factor expression, including sex-biased expression of growth hormone and IGF-1 in regulating size of different tissues.97,98 Importantly, these results indicate sex-specific differences in early thymus structure maintenance and growth, which could skew T cell development. We also establish an early transition toward an adipogenic environment in males. These observations align with findings in model organisms, where young male rats exhibit higher rates of thymic involution52 and early postnatal male primates have a larger interlobular fat area.26 Together, these factors define two possible mechanisms contributing to a male-female difference in thymus size and involution kinetics.

Future studies should test how sex differences at the transcript, niche, and organ level impact differential T cell production and quality as well as explore how sex differences in other organs contribute to known differences in immune responses. Defined in vitro and organoid culture systems replicating the thymic microenvironment present powerful platforms to test if the cell type-specific and sex-specific differences identified here lead to increased autoimmune disease incidence among females and increased infection susceptibility in males. Furthermore, given the surprising sex-based differences at this early postnatal stage, future work should examine aged thymus to investigate how cellular level differences in thymic involution kinetics may translate to larger impacts on our immune system later in life.

Limitations of the study

Our analysis of intra-sex variation is limited by access to patient samples as well as the inability to conduct mechanistic experiments in the context of a whole organism. There is an opportunity for future work to further validate and expand on predicted ligand-receptor interactions.

 

The thymic epithelium is responsible for the secretion of thymic peptides, which intervene in some steps of intra- and extrathymic T cell differentiation. Recent data suggest that thymic hormone secretion is modulated by the neuroendocrine network, comprising thyroid, adrenals, and gonads. However, the role of the pituitary gland in this regulation is still poorly understood. In the present paper we studied the in vivo and in vitro influences of PRL on the secretion of thymulin, one of the chemically defined thymic hormones, by thymic epithelial cells (TEC). When injected daily (20-100 micrograms/20 g) in young or old C57BL/6 mice, PRL induced a specific increase in thymulin synthesis and secretion, respectively, measured by the number of thymulin-producing cells in the thymus and the peripheral levels of the hormone. This stimulation was dose dependent and reversible after the end of treatment. Similar findings have been made in animals with pituitary dwarfism, known to have low levels of circulating thymulin. This stimulatory effect was also observed in primary cultures of human and mouse TEC when PRL (10(-7) to 10(-8) M) was applied to culture supernatants, thus suggesting that PRL could act directly on TEC. In addition, we induced in vivo experimental hypoprolactinemia, treating mice with bromocriptine, a dopamine receptor agonist that inhibits pituitary PRL secretion. Bromocriptine treatment (100-200 micrograms/20 g) yielded a significant decrease in thymulin secretion that could be reversed by coincident treatment with PRL. In the light of previous observations that bovine GH can also increase thymulin production in aged dogs, we performed a series of experiments in vitro to evaluate whether GH has a direct effect on TEC. We observed that only human GH preparations that are known to have a PRL-like effect were efficient in stimulating thymulin biosynthesis and release into the culture supernatants. The effects of PRL on TEC were not restricted to thymic hormone production. We observed that TEC proliferation, as well as the numbers of a TEC subset defined by the expression of cytokeratins 3 and 10, could also be increased by PRL treatment. All these findings show that the pituitary gland directly affects TEC in terms of cytoskeletal and secretory protein expression as well as cell cycle.. This paper reviews the mechanism of sex hormone actions on the thymus, presenting mainly our data obtained at the cellular and molecular levels. First, data supporting the "genomic" action via the nuclear sex hormone receptor complexes are as follows: 1) sex hormone receptors and the thymic factor (thymulin) are co-localized in thymic epithelial cells, but not in T cells; 2) production/expression of thymic factors (thymulin, thymosin alpha 1) are remarkably inhibited by sex hormone treatment; 3) sex hormones cause changes in T cell subpopulations in the thymus; and 4) sex hormones strongly influence the development of thymus tumors in spontaneous thymoma BUF/Mna rats through their receptor within the tumor cells. Secondly, data indicating the "non-genomic" action of sex hormones via a membrane signal-generating mechanism are as follows: 1) the proliferation/maturation of thymic epithelial cells is mediated through protein kinase C activity introduced by sex hormones; 2) sex hormones directly influence DNA synthesis and cdc2 kinase (cell cycle-promoting factor) activity..

pubmed.ncbi.nlm.nih.gov/2737149/

 

www.cell.com/developmental-cell/fulltext/S1534-5807(24)00539-2

Done