Introduction

The thymus is the primary lymphoid organ dedicated to the generation of a diverse and self-tolerant repertoire of T lymphocytes, a process fundamental to adaptive immunity (1). This intricate T cell developmental program unfolds within a highly organized microenvironment composed of thymic epithelial cells (TECs), endothelial cells, and various hematopoietic populations, which collectively provide the essential signals for thymocyte differentiation, proliferation, and selection (2–4). The developmental journey of a T cell begins when bone marrow-derived thymic seeding progenitors (TSPs) enter the thymus at the cortico-medullary junction (CMJ) and commit to the T cell lineage, initiating a multi-stage differentiation cascade (5–7).

This journey is punctuated by a series of stringent checkpoints that assess the integrity and functionality of the nascent T cell receptor (TCR) repertoire (8,9). The earliest stages occur in the thymic cortex, where progenitors, lacking expression of CD4 and CD8, termed double-negative (DN) thymocytes (6), start to differentiate, which is characterized by the expression of CD44 and CD25 (3,4). Progression from the DN1 (CD44⁺CD25^-) to the DN2 (CD44⁺CD25⁺) and DN3 (CD44^-CD25⁺) stages is marked by the initiation of V(D)J recombination at the Tcrb, Tcrg, and Tcrd gene loci (1). The DN3 stage represents the first critical checkpoint in αβ T cell development, known as β-selection (12,13). Here, thymocytes that have successfully rearranged a functional TCRβ chain pair it with the invariant pre-TCRα chain (pTα) to form the pre-TCR complex. Successful signaling through this complex provides vital survival, proliferation, and differentiation cues, allowing the cell to progress to the DN4 stage and subsequently to the CD4⁺CD8⁺ double-positive (DP) stage (14,15). Thymocytes failing to generate a functional pre-TCR are eliminated via apoptosis, ensuring that only cells with a viable TCRβ chain continue their developmental progression (14).

Following β-selection, DP thymocytes undergo further TCRα chain rearrangement and are subjected to two additional selection events. Positive selection ensures that the fully assembled αβ TCR can recognize self-peptides presented by major histocompatibility complex (MHC) molecules on cortical (c)TECs, a process essential for TCR MHC-restriction (16–18). Subsequently, negative selection in the medulla eliminates thymocytes whose TCRs bind with high affinity to self-peptides presented by medullary (m)TECs and dendritic cells (DCs), thereby establishing central tolerance (19,20). These selection processes are notably stringent, resulting in the apoptotic death of over 95% of all developing thymocytes (21).

Integral to this journey and processes are thymic macrophages (TMs), which have long been recognized for their critical homeostatic function. Residing in both the cortex and medulla, TMs are exceptionally efficient phagocytes, responsible for the swift and silent clearance of the millions of apoptotic thymocytes generated daily (22–25). This process, known as efferocytosis, is vital for preventing the release of potentially immunogenic and inflammatory intracellular contents and maintaining the integrity of the thymic architecture (26). However, this view of TMs as mere "housekeepers" is rapidly evolving. Recent advances, particularly single-cell RNA sequencing (scRNA-seq), have revealed significant heterogeneity within the TM compartment, suggesting functional specialization (27). Two principal subsets have been identified in the adult mouse thymus: cortical TIMD4+ macrophages, which have potent efferocytosis function, and CX3CR1+ macrophages, which are enriched at the CMJ and express genes associated with antigen presentation, suggesting at a potential role in negative selection (27).

Despite these advances, our understanding of the full spectrum of TM function remains incomplete. While a role in clearing the cellular debris from positive and negative selection is well-established, and a contribution to negative selection itself is hypothesized, the involvement of TMs in the earliest stages of T cell development has been largely overlooked. The β-selection checkpoint, a nexus of proliferation, survival, and lineage commitment, occurs in a microenvironment rich in TMs. Yet, whether these macrophages are passive bystanders or active participants in this critical decision-making process is unknown.

The objective of this study is to investigate the role of thymic macrophages in early T cell development, with a specific focus on the β-selection checkpoint. By integrating high-resolution transcriptomic analysis, fate-mapping models, and functional assays, we sought to dissect the interactions between distinct TM subsets and developing DN thymocytes. Our findings reveal a previously unappreciated function for thymic macrophages in actively shaping the outcome of β-selection, thereby influencing the survival and differentiation of early T cell precursors. This work redefines the role of thymic macrophages, expanding their function beyond efferocytosis to include direct participation in the foundational checkpoints of T cell repertoire formation.

Materials and methods

Animals

Macrophage Fas-Induced Apoptosis (MaFIA), MafB-mCherry-Cre, Spic^−/−, and C57BL/6J mice were purchased from the Jackson Laboratory. The Flt3-Cre × Rosa-mTmG mice and Flt3-Cre × Rosa-mTmG mice in the Ccr2^-/- background were kindly provided by Dr. Slava Epelman (University Health Network, University of Toronto) (28). All animals were bred and maintained in specific pathogen-free conditions at the Comparative Research Facility of the Sunnybrook Research Institute (SRI). All animal procedures were approved by the Sunnybrook Health Sciences Centre Animal Care Committee (Toronto, Ontario, Canada).

Timed pregnancies and embryonic thymus analysis

Timed pregnancies were set up by housing male mice (8–24 weeks of age) in separate cages and mating them with female mice (8–24 weeks of age) overnight. The following morning, females were separated from the males, and successful matings were identified by the presence of a vaginal plug. The gestational age of the embryos was designated as embryonic day 0.5 (E0.5) on the day the plug was observed. Thymuses were harvested from embryos at E15 and E17, as well as from postnatal mice at 1 day, 1 week, and 3 weeks of age, in addition to adult (6-8 weeks) mice. Following collection, thymuses were subjected to enzymatic digestion to dissociate the tissue into single-cell suspensions. The resulting cells were stained for flow cytometric analysis.

Intravenous CD45 staining

A CD45-Phycoerythrin (PE) antibody (30-F11; BioLegend) was diluted in phosphate buffered saline (PBS) to a final concentration of 1.5 µg/150 µL. The antibody solution was administered via tail vein injection using a 1 mL insulin syringe. Three minutes after the injection, mice were euthanized, and thymuses were harvested. The collected tissue was then processed and stained for flow cytometric analysis.

Cell isolation from thymus

Thymocytes were harvested and subjected to enzymatic digestion using the Spleen Dissociation Kit (Miltenyi Biotec) and the gentleMACS™ Dissociator (Miltenyi Biotec) according to the manufacturer’s instructions. The resulting cell suspensions were filtered through a 70 µm cell strainer to remove debris, resuspended in PBS, and kept on ice until further processing.

Flow cytometry

Single-cell suspensions (1 × 10⁷ cells) were prepared from the thymus and stained in a 96-well conical bottom plate (ThermoFisher Scientific). The staining cocktail included the fixable viability dye eFluor 450 (eBioscience), diluted 1:800 in PBS, anti-CD16/32 to block Fc receptors, and the desired fluorochrome-conjugated antibodies targeting macrophages and T cells (see Supplemental Tables 1 and 2 for details). Cells were initially stained with eFluor 450 in PBS for 15 minutes on ice. After washing with fluorescence-activated cell sorting (FACS) buffer, anti-CD64, anti-CCR2, and anti-VCAM1 antibodies were added to the wells, and cells were stained for an additional 15 minutes on ice. To block non-specific binding, Fc-block antibody was added to the cells 15 minutes prior to the addition of the remaining antibody cocktail, which was incubated with the cells for 30 minutes on ice.

Following staining, cells were washed and resuspended in FACS buffer for flow cytometric acquisition using the FACSymphony™ A5 Cell Analyzer (BD Biosciences). Data were analyzed with FlowJo software (TreeStar). The complete list of antibodies and their corresponding clones is provided in Supplemental Tables 1 and 2.

Cell sorting

Thymuses were harvested from 6-week-old C57BL/6 mice and subjected to enzymatic digestion using the Spleen Dissociation Kit (Miltenyi Biotec) and the gentleMACS™ Dissociator (Miltenyi Biotec) to generate single-cell suspensions. After cell isolation, cells were counted and stained with allophycocyanin (APC)-conjugated anti-mouse CD64 antibody in FACS buffer for 20 minutes on ice (see Supplemental Table 1 for antibody details). Following a wash with FACS buffer, cells were stained with anti-APC MicroBeads (Miltenyi Biotec) for 20 minutes on ice. The cells were then washed again with FACS buffer and loaded onto LS columns (Miltenyi Biotec) for positive selection of CD64⁺ cells. After selection, the cells were washed and further stained with antibodies against F4/80, TIMD4, Ly6C, and CD45 for 30 minutes on ice (see Supplemental Table 1 for antibody details). The stained cells were sorted using the FACSAria™ Fusion Flow Cytometer (BD Biosciences). CD64⁺ sorted cells were collected in 1.5 mL Eppendorf tubes containing 300 µL of FBS, counted, and resuspended in PBS supplemented with 0.04% BSA. Cells were then encapsulated, and library preparation was performed using the Chromium Single Cell 3’ Reagent Kit (10X Genomics). Gene expression libraries were sequenced on a NovaSeq 6000 platform (Illumina) at Azenta Life Sciences. In total, 3,214 and 2,815 cells were sequenced from two independent experiments.

Fetal thymus organ culture (FTOC) and TM depletion in MaFIA mice

Timed pregnancies were set up for MaFIA mice as described above. Fetal thymuses were harvested at E15.5 for FTOC. Briefly, culture wells were set up in a 12-well-plate 24 hour before FTOC. In each well, Whatman® Nuclepore™ Track-Etched membrane (WHA110409; Sigma) was placed on top of SURGIFORM® absorbable gelatin sponge (1974; Ethicon) in 1.5 mL of complete Dulbecco’s Modified Eagle Medium (DMEM) (supplemented with 10% Fetal Bovine Serum, 2 mM L-glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin) (29,30). Three fetal thymuses were placed on top of the membrane for each well. B/B homodimerizer stock was diluted to 2.5 μM in complete DMEM and added to the thymus lobes at the time of culture setup and again on the following day. On day 2, the membrane with the thymus lobes was transferred to a fresh culture plate containing new gelatin sponge and complete DMEM. Cultures were maintained in 37 °C incubator until day 6, at which point the thymic lobes were processed for flow cytometry analysis as described above.

Transcriptomic analysis

CD64⁺ cells were isolated from the thymus of C57BL/6J adult mice and processed for scRNA-seq as described below. Data from each scRNA-seq experiment underwent pre-processing as detailed below. Quality control filters were applied to remove cells with more than 12.5% mitochondrial gene expression, fewer than 500 reads, or more than 6,000 reads, ensuring the retention of high-quality cells for analysis. Macrophages and monocytes were identified using a module scoring approach, and their respective cell barcodes were recorded. These barcodes were used for subsetting before merging the datasets from the two scRNA-seq experiments with published TM data from Dzhagalov’s group. The merging was performed using Harmony to align the datasets and correct for batch effects. Following integration, cells were clustered, and module scores were assigned to all identified clusters. Non-relevant cell types, including B cells, T cells, and DCs, were excluded from subsequent analyses to focus on macrophages and monocytes.

Analysis of scRNA-Seq and data integration

All analyses were performed using R software version 4.4.1. Cells expressing fewer than 500 genes or more than 6,000 genes were excluded from the dataset. Additionally, cells with greater than 12.5% mitochondrial gene content were removed to ensure data quality. The dimensionality of the dataset was assessed using an Elbow plot to determine the appropriate number of principal components. Gene ontology (GO) enrichment analysis of differentially expressed genes (DEGs) was conducted to identify overrepresented biological processes using the FindMarkers function from the Seurat package (31). The data were then integrated with previously published scRNA-seq datasets on TMs (GEO: GSE185460), as well as datasets from heart, liver, and lung macrophages (GEO: GSE188647), from the laboratories of Drs. Dzhagalov and Epelman, respectively. Batch correction across datasets was performed using the Harmony package (27,32,33). Following integration, the merged dataset was normalized using the SCTransform method to account for technical variation. For data visualization, Uniform Manifold Approximation and Projection (UMAP) was applied to reduce dimensionality and enable visualization of cellular clustering patterns.

Trajectory analysis using Monocle3

Trajectory analysis was performed using Monocle3 to infer the developmental trajectory of TMs (34). Prior to this, TM datasets were normalized and integrated using the SCTransform method implemented in Seurat. From the resulting Seurat object, the expression matrix, cell metadata, and gene metadata were extracted to construct a CellDataSet (CDS) object, which serves as the foundational data structure in Monocle3. Pre-processing steps were applied to the CDS to ensure high-quality trajectory inference. Principal component analysis (PCA) was first conducted to reduce dimensionality while preserving variance across the dataset. This was followed by non-linear dimensionality reduction using Uniform Manifold Approximation and Projection (UMAP) to visualize the global structure of the dataset in a low-dimensional space. The cells were then re-clustered within Monocle3 to refine group assignments, ensuring that clusters corresponded accurately to biological populations. Trajectory inference was performed with the root node manually assigned to monocytes (defined by Ly6c2 and Ccr2 expression), reflecting their position as the presumptive progenitor population in the trajectory.

Statistical analysis

Data are presented as the mean ± standard error of the mean (SEM). Univariate comparisons between two groups were performed using the Mann-Whitney U test. For analyses involving more than two groups, the normality of the data was assessed using the D’Agostino-Pearson test and/or inspection of Q-Q plots, while homoscedasticity (equal variance) was evaluated using the Brown-Forsythe test or by examining residual plots. Non-parametric comparisons across multiple groups were conducted using the Kruskal-Wallis test, followed by post-hoc pairwise comparisons of the mean ranks of each group with the mean rank of the control group. Multiple comparisons were adjusted using Dunn’s test to control for type I error. For pairwise comparisons involving more than two groups, the Holm-Šídák method was applied to correct for multiple comparisons, reducing the likelihood of type I error. All statistical analyses were performed using GraphPad Prism version 10.

Results

Identification of Two Distinct Tissue-Resident Macrophage Subsets in the Adult Thymus

To investigate the heterogeneity of phagocytic cells within the thymus, we performed multi-parameter flow cytometry on hematopoietic cells (CD45⁺) from the thymus of adult mice. We first identified the total TM population based on the co-expression of the canonical markers CD64 and F4/80 (Figure 1A-B). The majority of these CD64⁺ F4/80⁺ cells also expressed the phagocytic receptor MerTK, consistent with their identity as macrophages (35).

Figure 1.

To resolve heterogeneity within this population, we further analyzed for the expression of TIMD4 and VCAM1. This revealed three distinct subsets within the CD64⁺ F4/80⁺ gate: a major TIMD4⁺ VCAM1⁺ population, a smaller TIMD4^- VCAM1⁺ population, and a minor TIMD4^-VCAM1^- population (Figure 1C-E). To clarify the identity of these subsets, we utilized MafB-mCherry and Csf1r-EGFP reporter mice. Both the TIMD4⁺ VCAM1⁺ and TIMD4^- VCAM1⁺ subsets expressed high levels of MafB (mCherry⁺), a transcription factor crucial for mature tissue-resident macrophages (TRMs), while the TIMD4^- VCAM1^- subset was MafB-negative and expressed high levels of CSF1R (GFP⁺), a marker associated with monocytes (Figure 1F).

To definitively distinguish these thymus resident populations from circulating cells and further define their identity, we performed intravenous (IV) labeling with an anti-CD45 antibody. Both the TIMD4⁺ VCAM1⁺ and TIMD4^- VCAM1⁺ macrophage populations were largely protected from IV labeling (CD45^-), confirming their parenchymal, tissue-resident localization (Figure 2A, B, C). Furthermore, these two subsets lacked expression of classical monocyte-associated markers, including Ly6C, CCR2, and CX3CR1 (Figure 2D, E). In contrast, the TIMD4^- VCAM1^- subset expressed high levels of these monocyte markers, confirming its identity as a population of thymic monocytes rather than mature macrophages.

Figure 2.

Taken together, our analysis robustly identifies two distinct subsets of bona fide tissue-resident thymic macrophages, a dominant TIMD4⁺ VCAM1⁺ population and a TIMD4^- VCAM1⁺ population, alongside a separate population of thymic monocytes. This phenotypic framework allows for the specific investigation of the functional roles of each macrophage subset in thymic biology.

Transcriptomic Profiling by scRNA-seq Confirms Two Functionally Distinct Macrophage Subsets

To gain deeper insight into the transcriptional identities of these macrophage populations, we performed scRNA-seq on sorted thymic CD45⁺ CD64⁺ cells. Unbiased clustering of the resulting 3,153 cells revealed five distinct myeloid clusters, including monocytes, proliferating macrophages, and two major quiescent macrophage populations that corresponded directly to our flow cytometry findings (Figure 3A).

Figure 3.

A major macrophage cluster was defined by high expression of Timd4, Vcam1, Mertk, and the transcription factor Maf, confirming its identity as the TIMD4⁺ VCAM1⁺ population (Figure 3B). In contrast, the other major macrophage cluster was defined by high expression of Vcam1 and Cx3cr1, aligning it with the TIMD4^- VCAM1⁺ subset (Table 1), as identified above. Intriguingly, the TIMD4⁺ VCAM1⁺ CX3CR1^- cluster also showed uniquely high expression of Spic, the lineage-defining transcription factor for splenic red pulp macrophages (RPMs) (36). To test for a shared developmental dependency, we analyzed SpiC knockout (Spic^−/−) mice. While Spic^−/− mice exhibited the expected loss of RPMs, the numbers of both thymic macrophage subsets were unaltered, suggesting that TMs utilize a Spic-independent developmental pathway despite their transcriptional similarities (Figure 3C).

Table 1.

To formally assess the functional specialization suggested by these transcriptional profiles, we performed Gene Ontology (GO) pathway analysis. This revealed a stark functional dichotomy between the two subsets. The TIMD4^- VCAM1⁺ CX3CR1⁺ TMs were significantly enriched for pathways related to "antigen processing and presentation via MHC class I" and "interferon signaling" consistent with their high expression of H2-K1, H2-D1, and Runx3 and suggesting a role in T cell education (Figure 4A). Conversely, the TIMD4⁺ VCAM1⁺ CX3CR1^- TM cluster was dominated by pathways associated with "phagocytosis" and "endocytosis," supporting a primary role in efferocytosis for apoptotic cell clearance (Figure 4B).

Figure 4.

This functional role was confirmed by the identification of a sub-cluster within the TIMD4⁺ VCAM1⁺ CX3CR1^- population that was actively engaged in efferocytosis in vivo. These cells co-expressed canonical T-cell transcripts such as Cd3g and Trac alongside their macrophage-specific genes (Figure 3B). Thus, their T-cell transcriptional signature provides direct evidence of ongoing efferocytosis, capturing macrophages in the act of engulfing apoptotic thymocytes and confirming the potent phagocytic activity of TIMD4⁺ VCAM1⁺ CX3CR1^- TMs.

The Thymic Macrophage Compartment Undergoes Dynamic Remodeling During Development

To understand how the TM compartment evolves in parallel with the thymus itself, we analyzed the number and composition of TM subsets during ontogeny, from embryonic development through adulthood. Consistent with known thymic biology, the absolute number of total thymocytes and TMs peaked in young, 3-week-old mice, before declining with age, indicating that the size of the TM pool is coupled to the overall cellularity of the organ (Supplemental Figure 1A-B).

Further analysis of TM subset composition revealed a profound and dynamic remodeling process across the lifespan. At embryonic day 15 (E15), the nascent thymus was populated almost exclusively by macrophages with a progenitor-like phenotype (TIMD4⁻ VCAM1⁻), which co-expressed the monocyte-associated markers CD11b and CCR2 (Figure 5A). This suggests that the embryonic thymus is initially seeded by monocyte-derived precursors. Following birth, this landscape was dramatically reshaped. We observed a rapid postnatal expansion of the TIMD4⁺ VCAM1⁺ CX3CR1^- subset, which became the major TM population in young mice, coinciding with the period of peak thymic size (Figure 5B). This indicates that the establishment of the mature, homeostatic TM pool is a key feature of postnatal thymic development. Interestingly, as mice aged and the thymus began to involute (6 weeks and older), the proportion of TIMD4⁺ VCAM1⁺ CX3CR1^- macrophages steadily declined. This decline was accompanied by a progressive accumulation of the TIMD4^- VCAM1⁺ CX3CR1⁺ TM subset, which became a more prominent feature of the aging thymus.

Figure 5.

Together, these data reveal that the thymic macrophage compartment is not a static entity but is actively remodeled throughout life. Distinct TM subsets dominate at specific developmental windows, with an initial wave of progenitor-like cells being replaced by a homeostatic TIMD4⁺ VCAM1⁺ CX3CR1^- population, which in turn gives way to a TIMD4^- VCAM1⁺ CX3CR1⁺ population during the start of the thymic involution process.

Thymic Macrophage Subsets Arise from Distinct Developmental Origins

Having established the dynamic nature of the TM compartment, we next sought to define the developmental origins of the two major macrophage subsets. We first utilized Flt3-Cre × Rosa-mTmG fate-mapping mice, in which cells passing through a FLT3-expressing hematopoietic stem cell (HSC)-derived progenitor stage are permanently labeled with GFP, while cells from an FLT3-independent pathway (i.e., yolk sac-derived) remain Tomato⁺ (Figure 6A) (28). This fate-mapping analysis revealed a clear divergence in origin. The majority (>50%) of the homeostatic TIMD4⁺ VCAM1⁺ CX3CR1^- macrophages were Tomato⁺, indicating they arise from an FLT3-independent, likely embryonic/fetal, precursor and persist into adulthood (Figure 6B-C). In stark contrast, the vast majority of TIMD4^- VCAM1⁺ CX3CR1⁺ TMs and thymic monocytes were GFP⁺, demonstrating their origin from adult HSC-derived, FLT3-dependent progenitors.

Figure 6.

Based on these findings, we hypothesized that the maintenance of the TIMD4^- VCAM1⁺ CX3CR1⁺TM subset, but not the TIMD4⁺ VCAM1⁺ CX3CR1^- subset, would depend on the continuous recruitment of CCR2-dependent monocytes from the bone marrow. To test this, we analyzed TM populations in Ccr2 knockout (Ccr2^−/−) mice. While overall thymocyte numbers were unaffected, the composition of the TM compartment was significantly altered (Figure 7A). While there was no change in TIMD4⁺ VCAM1⁺ CX3CR1^- macrophages number, the TIMD4^- VCAM1⁺ CX3CR1⁺macrophage population and thymic monocytes was significantly decreased in Ccr2^−/− mice (Figure 7B-D).

Figure 7.

Together, these data establish a dual-origin model for the thymic macrophage network: a stable, self-maintaining population of TIMD4⁺ VCAM1⁺ CX3CR1^- macrophages established from fetal, FLT3-independent precursors, which coexists with a dynamic, CCR2-dependent population of TIMD4^- VCAM1⁺ CX3CR1⁺ macrophages that is continuously replenished by circulating monocytes.

Depletion of Csf1r-Expressing Myeloid Cells in Thymic Organ Culture Impairs T Cell Development at the β-Selection Checkpoint

To directly test the functional requirement for TMs during T cell development, we employed the MaFIA (Csf1r-EGFP/Fas-FKBP) mouse model, which permits the inducible depletion of Csf1r-expressing cells (Supplemental Figure 2A-B) (37). Because both TM subsets and thymic monocytes express the Csf1r-EGFP reporter, this system enables the targeted depletion of the thymic myeloid compartment (Supplemental Figure 2C-E). To circumvent the systemic effects of in vivo depletion, we utilized an ex vivo fetal thymic organ culture (FTOC) system (38). Thymic lobes from E15.5 MaFIA embryos were cultured with the dimerizer drug AP20187 to induce apoptosis in Csf1r⁺ cells (Figure 8A).

Figure 8.

To validate the specificity of our approach, we first examined the non-myeloid stromal compartment. Flow cytometry analysis confirmed that the number of TECs, identified as EpCAM⁺ CD45⁻, appeared unaffected by the treatment (Figure 8B). In contrast, depletion was highly effective against the intended myeloid targets. We observed a significant reduction in the total number of thymocytes and thymic DCs (Figure 8C-D). Crucially, the treatment also resulted in the profound loss of total TMs (Figure 8E). This effect was specific to the MaFIA system, as treatment of control C57BL/6 mice with the dimerizer did not cause TM depletion (Figure 8F). Analysis of the TM populations revealed that both the TIMD4⁺ VCAM1⁺ CX3CR1^- and the TIMD4^- VCAM1⁺ CX3CR1⁺ subsets were significantly depleted (Figure 8G-H).

Depletion of Csf1r⁺ cells led to an impairment in αβ T cell development (Figure 9A). AP20187-treated thymic lobes exhibited a significant reduction in the number of CD4⁺ CD8⁺ DP thymocytes (Figure 9B-C). While the development of γδ T cells was unaffected (Figure 9D). To identify the origin of this defect, we analyzed the upstream DN progenitor stages (Figure 10A). This revealed a developmental block, as treated FTOCs showed a significant accumulation of DN3 (CD44⁻ CD25⁺) thymocytes and a corresponding decrease in the subsequent DN4 (CD44⁻ CD25⁻) population (Figure 10B).

Figure 9.

Figure 10.

To pinpoint the impairment more precisely, we examined the expression of CD27, a surface marker upregulated upon successful pre-TCR signaling. Within the arrested DN3 population of treated lobes, the proportion of CD27⁺ cells was significantly reduced (Figure 10C-D). This points to a failure of thymocytes to receive or transduce the signals required to pass the β-selection checkpoint. Collectively, these data reveal that the Csf1r-expressing myeloid compartment, which includes thymic macrophages, appears to be important for the successful transition of thymocytes through β-selection, a critical early checkpoint in T cell development.

Discussion

The thymus provides a unique and specialized niche for T cell development, yet the contributions of its resident myeloid populations have been historically underappreciated. This study redefines the role of TMs, moving beyond their established function as mere "housekeepers" to reveal them as a heterogeneous and dynamic population that also is integral for the earliest checkpoints in T cell development. By combining high-resolution phenotyping, fate-mapping, and functional depletion, we demonstrate that the thymus contains two distinct macrophage subsets with separate origins and functions, and we uncover a critical role for the myeloid compartment in supporting thymocyte progression through β-selection.

Our work addresses the complexity of the TM landscape by integrating flow cytometry and single-cell transcriptomics. We identified a dominant TIMD4⁺ VCAM1⁺ CX3CR1^- population transcriptionally programmed for efferocytosis, expressing high levels of Mertk and sharing a gene signature (Spic, MafB) with splenic red pulp macrophages (36). However, using Spic knockout mice, we showed that TMs, unlike their splenic counterparts, follow a Spic-independent developmental pathway, highlighting their unique adaptation to the thymic niche. In contrast, the TIMD4^- VCAM1⁺ CX3CR1⁺ subset was enriched for genes involved in antigen presentation and interferon signaling, positioning it as a candidate for shaping the T cell repertoire, a role previously attributed almost exclusively to DCs (20).

Furthermore, we establish a dual-origin model for this network. The homeostatic TIMD4⁺ VCAM1⁺ CX3CR1^- macrophages are likely long-lived, self-maintaining cells derived from embryonic, FLT3-independent precursors. Conversely, the TIMD4^- VCAM1⁺ CX3CR1⁺ subset is continuously replenished from HSC-derived, Flt3-dependent monocytes, representing a more dynamic population likely involved in immune surveillance (39,40). This clarifies conflicting reports in the literature and aligns TM biology with the broader principles of tissue-resident macrophage ontogeny observed in other organs (23,27).

An impactful finding of this study is the appreciation of novel role for thymic myeloid cells in early T cell development. Using an ex vivo FTOC system to avoid the confounding stress of in vivo depletion, we demonstrated that the loss of Csf1r-expressing cells leads to a profound developmental block at the DN3 to DN4 transition. The accumulation of DN3 cells, coupled with a specific reduction in CD27 expression, pinpoints a failure at the β-selection checkpoint (41,42). While traditionally viewed as a cell-autonomous event, our data strongly suggest that pre-TCR signaling is not independent of the microenvironment (12). This support from myeloid cells could be multifaceted, involving the presentation of peptide-MHC to the pre-TCR (43–45) or other receptor-ligand interactions, the provision of essential survival factors (46–48), or the maintenance of a non-inflammatory niche through the clearance of apoptotic cells (49).

Although our depletion strategy also affected DCs, several lines of evidence implicate TMs as the primary mediators of this effect. The β-selection checkpoint occurs in the thymic cortex, a region densely populated by TMs, whereas DCs are largely restricted to the medulla and cortico-medullary junction (50). Nevertheless, future studies using more targeted depletion strategies will be crucial to definitively dissect the relative contributions of TMs and DCs. Additionally, the absence of TMs might lead to an accumulation of apoptotic cells, creating a potentially hostile environment for early thymocyte development. Macrophages, through their role in clearing apoptotic cells, produce anti-inflammatory cytokines like IL-10 and TGF-β, while suppressing pro-inflammatory cytokines such as IL-12 (51–53). Without these regulatory influences, TM depletion may result in localized thymic inflammation, further compromising thymocyte survival and development (54). Together, these findings highlight a supportive role of TMs in β-selection, not only by facilitating essential pre-TCR interactions but also by preserving a supportive microenvironment for thymocyte maturation.

In conclusion, this study elevates thymic macrophages from simple scavengers to critical architects of the initial TCRβ repertoire. By defining their heterogeneity, origins, and functional specialization, we reveal their integral role in thymic homeostasis and uncover their essential contribution to the successful navigation of the earliest checkpoints in T cell development. These findings not only advance our fundamental understanding of thymic biology but also open new avenues for exploring how modulation of this myeloid niche could influence immune tolerance and T cell-based therapies.

Data availability

Single cell RNAseq datasets have been deposited to the Gene Expression Omnibus (GEO) repository, accession number GSE326594.

Supplemental Figures

Supplemental Figure 1.

Supplemental Figure 2.

Additional information

Funding

Canadian Institutes of Health Research (CIHR) (FDN154332)

• Juan Carlos Zúñiga-Pflücker

Canadian Institutes of Health Research (CIHR) (PJT192050)

• Juan Carlos Zúñiga-Pflücker

Natural Sciences and Engineering Research Council of Canada (NSERC) (RGPIN-2024-05661)

• Juan Carlos Zúñiga-Pflücker

Canadian Institutes of Health Research (CIHR) (Postdoctoral Fellowship)

• Vinothkumar Rajan