Single-cell mapping of thymus-dependent naïve t-cell regeneration reveals CD61-associated states after hematopoietic stem cell transplantation

Abstract

Thymus-dependent de novo T-cell generation is essential for maintaining T-cell diversity and immune competence. Although thymic involution after puberty reduces thymic output, residual thymic function remains important for immune surveillance and immune reconstitution following hematopoietic stem cell transplantation (HSCT). However, the cellular features that distinguish thymus-derived naïve T cells from those maintained through peripheral homeostatic expansion remain poorly understood. Here, we employed a congenic GFP-labeled mouse transplantation model combined with longitudinal single-cell transcriptomic and T-cell receptor repertoire analyses to define the origin and dynamics of naïve T-cell regeneration after HSCT. By distinguishing thymus-derived de novo generated T cells from peripherally expanded T cells, we found that recent thymic emigrants and both CD4⁺ and CD8⁺ naïve T cells preferentially reconstitute through a thymus-dependent pathway. We further identified distinct naïve T-cell states enriched within the thymus-derived compartment, including CD61-associated and Ly6C-associated naïve T-cell states. These populations displayed dynamic changes during immune reconstitution and were associated with thymic functional status in transplantation and physiological thymic involution mouse models. In human datasets and primary peripheral blood samples, CD61-associated features correlated with age-related thymic activity, sjTREC levels, and preliminary measures of thymic recovery after allogeneic HSCT. Collectively, our study defines cellular features of thymus-dependent naïve T-cell regeneration after HSCT and identifies CD61-associated naïve T-cell states as candidate peripheral features linked to thymic recovery. These findings provide new insights into thymus-dependent immune reconstitution and establish a framework for evaluating peripheral signatures of thymic regeneration after HSCT.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40164-026-00824-x.

Introduction

The thymus is the central organ for T-cell development and TCR repertoire formation. From fetal life through adolescence, thymic TCR rearrangement and positive and negative selection generate a diverse T-cell pool with broad antigen-recognition capacity and self-tolerance, thus supporting immune surveillance against infections and malignancies [1–3]. However, after puberty, the thymus undergoes involution, with reduced thymic output and decreased generation of newly formed naïve T cells [4, 5]. Consequently, the adult naïve T-cell pool is generally thought to rely primarily on peripheral homeostatic expansion, rather than sustained thymic export [6]. Nevertheless, the adult thymus is not dispensable. In adult cardiothoracic surgery cohorts, thymectomy was associated with reduced production of newly formed CD4⁺ and CD8⁺ T cells, increased levels of inflammatory cytokines, higher cancer incidence, and increased all-cause mortality [7, 8]. Subsequent analyses suggested that excess mortality after thymectomy was mainly due to cancer-related deaths rather than cardiovascular deaths [9]. These findings suggest that the adult thymus may continue to support long-term immune homeostasis through new T-cell output, TCR repertoire diversity, and immune surveillance.

Hematopoietic stem cell transplantation (HSCT) is not only a standard treatment for hematologic malignancies but also a potent therapeutic approach for rebuilding immune homeostasis [10–12]. Our previous research showed that patients undergoing haploidentical transplantation exhibited enhanced thymic output [13], suggesting that thymic function can be restored even in adulthood and is essential for re-establishing immune homeostasis. The thymus is highly susceptible to immune damage, and transplant-related complications, such as graft-versus-host disease (GVHD), can further impair thymic function [14, 15]. Restoration of thymic function is a key marker of immune homeostasis recovery and favorable post-transplant prognosis [8]. However, the cellular and molecular features that distinguish thymus-derived naïve T cells from those maintained through peripheral homeostatic expansion remain incompletely understood. Furthermore, although several markers of recent thymic emigrants (RTEs) have been described, reliable approaches for tracking thymus-dependent naïve T-cell regeneration, particularly in murine models and regenerative settings such as HSCT, remain limited.

Here, we focus on thymus-dependent naïve T-cell regeneration after transplantation, and use a physiological setting of thymic involution as an independent context in which to test the markers identified. We established a congenic GFP-labeled mouse transplantation model that enabled direct discrimination between peripherally expanded T cells and thymus-derived de novo generated T cells during immune reconstitution. By combining this system with longitudinal single-cell transcriptomic and TCR repertoire analyses, we characterized the cellular states and dynamic trajectories of naïve T-cell regeneration following HSCT. We found that RTEs and naïve T cells preferentially reconstituted through a thymus-dependent pathway. In addition, we identified distinct naïve T-cell states enriched within the thymus-derived compartment, including Itgb3 (CD61)-high and Ly6c-high populations. These populations exhibited transcriptional and repertoire features associated with thymus-dependent regeneration and were further evaluated in transplantation and aging models. We also examined the translational relevance of CD61/ITGB3 in human datasets and found that it was associated with established measures of thymic activity. Together, our study defines cellular features of thymus-dependent naïve T-cell regeneration after HSCT and provides a framework for investigating peripheral signatures associated with thymic recovery and immune homeostasis restoration.

Materials and methods

Clinical samples

Peripheral blood samples were collected from healthy individuals at Peking University People’s Hospital, including a young group (9–20 years old, n = 23) and an old group (55-73 years old, n = 18). All participants had no known underlying diseases, active infections, autoimmune disorders, or history of immunomodulatory treatment. Peripheral blood mononuclear cells (PBMCs) were isolated by density-gradient centrifugation and subsequently used for flow cytometric analysis and sjTREC quantification by qRT-PCR.

In addition, peripheral blood samples were obtained from allogeneic HSCT recipients at Peking University People’s Hospital, collected cross-sectionally at 6 months (n = 23) and 12 months (n = 31) post-transplantation (total n = 54; 3 patients contributed samples at both time points). Recipients underwent HSCT for hematologic malignancies (AML/MDS/CMML/MPAL) using haploidentical or HLA-matched donors. PBMCs were isolated and used for flow cytometric analysis. Detailed baseline characteristics and clinical outcomes, including acute and chronic graft-versus-host disease, and relapse are summarized in Supplementary Table S4.

Analyses of public scRNA-seq and microarray datasets are described in the subsequent bioinformatic analysis section.

This study was approved by the Ethics Committee of Peking University People’s Hospital (Approval No. 2023PHB098-001), and written informed consent was obtained from all participants in accordance with the Declaration of Helsinki.

Mice

Female C57BL/6 (CD45.2, H-2^b, 8–12 weeks old), CB6F1 (CD45.2, H–2^b/d, 8–12 weeks old) mice, and B6. SJL (CD45.1, H-2^b, 8–12 weeks old) mice were purchased from Beijing Vital Laboratory Animal Technology Company, Ltd. (Beijing, China). Rosa26-CAG-EGFP mice (C57BL/6 background) were purchased from Shanghai Model Organisms Center, Inc. (Shanghai, China). Young (9-week-old) and middle-aged (48-week-old) C57BL/6 mice were used to model graded thymic activity under physiological conditions. 48 weeks was selected to represent an intermediate stage of thymic involution. Mice were purchased from Beijing Vital Laboratory Animal Technology Company, Ltd. (Beijing, China). All mice were bred and maintained in the specific pathogen-free (SPF) animal facility of Peking University People’s Hospital. This study was approved by the Ethics Committee of Peking University People’s Hospital (Approval No. 2023PHE027). All animal experiments were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Mouse model

For the GFP-labeled transplantation model, recipient B6. SJL (CD45.1, H-2^b) mice received 8 Gy total body irradiation (TBI), administered as two fractions of 4 Gy separated by 4 h. In this model, mature donor T cells were co-infused with T-cell-depleted bone marrow (TCD-BM). Because donor T cells facilitate engraftment, 8 Gy TBI was sufficient to achieve adequate myeloablation and stable donor chimerism (Figure S1A). TCD-BM cells were isolated from C57BL/6 (CD45.2, H-2) donor mice using anti-CD90.2 microbeads (Miltenyi Biotec, 130-121-278). GFP⁺ splenic T cells were isolated from Rosa26-CAG-EGFP donor mice using a Pan T Cell Isolation Kit II (Miltenyi Biotec, 130-095-130) by negative selection, with a purity greater than 95%. Recipient mice were intravenously infused with 5 × 10⁶ TCD-BM cells together with 5 × 10⁶ GFP⁺ splenic T cells on the day of irradiation. Spleens were collected at 4, 8, and 24 weeks after transplantation for FACS sorting and single-cell transcriptomic analysis.cell

For congenic and allogeneic transplantation models, recipient B6.SJL(CD45.1, H-2^b) mice received 13 Gy TBI, administered as two fractions of 6.5 Gy separated by 4 h. A higher conditioning dose was required in this model because only TCD-BM was infused, without mature donor T cells. In preliminary dose-finding experiments, conditioning with 8–10 Gy did not result in donor engraftment under these conditions, whereas 13 Gy achieved consistent engraftment, consistent with the more intensive conditioning required for T-cell-depleted grafts [16, 17]. Donor mice for the congenic group were C57BL/6 (CD45.2, H-2^b), whereas CB6F1 (CD45.2, H-2^b/d) mice were used as donors for the allogeneic group. TCD-BM cells were prepared using anti-CD90.2 microbeads (Miltenyi Biotec, 130-121-278) for T-cell depletion. Recipient mice were intravenously infused with 5 × 10⁶ TCD-BM cells from C57BL/6 donors in the congenic model or from CB6F1 donors in the allogenic model on the day of irradiation. Spleens were harvested 4 weeks after transplantation for FACS analysis.

For the aging mouse model, thymus and spleen tissues were collected from young (9-week-old) and middle-aged (48-week-old) C57BL/6 mice and subjected to FACS analysis.

Preparation of single-cell suspensions from spleen and thymus

Spleen and thymus tissues were isolated from mice and mechanically dissociated using the plunger end of a syringe, followed by filtration through a 40 µm cell strainer to generate single-cell suspensions. Cells were washed with PBE buffer (PBS supplemented with 2% FBS and 2 mM EDTA). Red blood cells in both spleen and thymus samples were lysed using ACK lysis buffer (Solarbio, R1010). Cells were subsequently washed, and cell number and viability were determined by trypan blue staining before downstream flow cytometry, cell sorting, and single-cell RNA sequencing .

Sorting splenic T cells for scRNA sequencing

Spleens were collected from recipient mice at 4, 8, and 24 weeks after transplantation, and single-cell suspensions were prepared. Splenic T cells were first enriched by negative selection using the Mouse Pan T Cell Isolation Kit II (Miltenyi Biotec, 130-095-130). Cells were subsequently stained with anti-CD45.2 antibody and sorted on a BD FACS Aria III (BD Biosciences). CD45.2⁺GFP⁺ and CD45.2⁺GFP⁻ T-cell populations were separated based on endogenous GFP fluorescence. Specific gating strategies are provided in Figure S1. Post-sort purity was consistently > 95%. Sorted cells were resuspended in D-PBS supplemented with 0.04% BSA and loaded onto the 10x Genomics Chromium platform for single-cell library preparation and downstream scRNA-seq analysis.

Flow cytometry

Single-cell suspensions prepared from murine spleens, thymus tissues, and human PBMCs were washed with PBE buffer. Cells were stained with surface antibodies together with Zombie Yellow viability dye for 30 min at 4 °C, washed, and subsequently analyzed by flow cytometry. Data acquisition was performed on a Cytek Northern Lights (Cytek Biosciences), and data were analyzed using FlowJo software v10. Detailed antibody information is provided in Supplementary Table S1.

In murine spleens, CD4⁺ naïve T cells were defined as CD3⁺CD4⁺CD44⁻CD62L⁺ cells, whereas CD4⁺ RTEs were defined as CD3⁺CD4⁺CD25⁻Qa-2⁺CD24⁺ cells. CD61⁺ and Ly6C⁺ CD4⁺ naïve T-cell subsets were defined as CD61⁺ or Ly6C⁺ CD3⁺CD4⁺CD44⁻CD62L⁺ cells, respectively. Corresponding CD8 subsets were defined using the same gating strategy, with CD8 substituted for CD4. In murine thymus, CD61 expression was analyzed within CD4⁺CD69⁻H-2B⁺ and CD8⁺CD69⁻H-2B⁺ populations. In human PBMCs, CD4⁺ naïve T cells were defined as CD3⁺CD4⁺CCR7⁺CD45RA⁺ cells, among which CD31⁺ cells were defined as CD4 RTEs and CD61⁺ cells as CD61⁺ CD4⁺ naïve T cells. CD8⁺ naïve T cells were defined as CD3⁺CD8⁺CCR7⁺CD45RA⁺ cells, among which CD38⁺ cells were defined as CD8 ⁺ RTEs and CD61⁺ cells as CD61⁺ CD8 naïve T cells.

sjTREC quantification by qRT-PCR

PBMCs were collected from young and older individuals, and precision count beads (BioLegend; 424902) were added for absolute cell counting. Genomic DNA was extracted using the QIAamp DNA Blood Mini Kit (Qiagen, 51104) according to the manufacturer’s instructions, and DNA concentration was measured using a Qubit fluorometer. sjTREC levels were quantified by TaqMan real-time PCR using absolute quantification, with TRAC used as the reference gene. Serial dilutions of a sjTREC-TRAC plasmid standard were used to generate standard curves. Each sample was analyzed in triplicate. PCR reactions were performed in a total volume of 25 µL containing genomic DNA, 2×TaqMan Universal PCR Master Mix (Thermo Fisher Scientific, 4304437), and sjTREC- or TRAC-specific primers and probes. Reactions were run on an ABI 7500 Real-Time PCR System using the following cycling conditions: 50 °C for 2 min, 95 °C for 10 min, followed by 45 cycles of 95 °C for 15 s and 60 °C for 1 min. sjTREC copy numbers were calculated from standard curves and normalized to TRAC copy numbers using the formula: mean sjTREC quantity ÷ (mean TRAC quantity/2) × the absolute count of the target T-cell subset/mL, followed by conversion to sjTREC copies per microliter of peripheral blood. Primer and probe sequences are listed in Supplementary Table S2.

scRNA-seq by 10X Genomics

FACS-sorted CD45.2⁺GFP⁺ and CD45.2⁺GFP^- splenic T cells were used to generate single-cell RNA-seq libraries with the Chromium Single Cell 5' platform (10× Genomics, Pleasanton, CA, USA) according to the manufacturer’s instructions. Sorted cells were washed with 0.04% BSA in D-PBS and resuspended at 700–1200 cells/µL, ensuring a viability of ≥ 85%. Single-cell capture, reverse transcription, and barcoded cDNA purification were performed in the microfluidic device, followed by PCR amplification of cDNA and construction of 5′ gene expression libraries. Libraries were sequenced on an Illumina NovaSeq platform to generate 150 bp paired-end reads.

Raw FASTQ files were processed with Cell Ranger v7.1.0 and aligned to the mm10 reference transcriptome (refdata-gex-mm10-2020-A) to generate a count matrix for downstream single-cell analysis.

scRNA-seq analysis

Single-cell RNA-seq data were analyzed using Seurat (v4) and performed on n = 3 mice per group at each post-transplant time point and one pre-HSCT sample (The “Pre-HSCT” sample consists of GFP⁺ splenic T cells from the same cohort of Rosa26-CAG-EGFP donor mice used for transplantation, sorted and sequenced immediately prior to infusion). For the full T-cell object, quality control filters included nFeature_RNA 200–5000, nCount_RNA < 25,000, and mitochondrial gene percentage < 5%. Doublets were identified and removed using scDblFinder. Batch effects across samples were corrected using Harmony, and the Harmony-corrected embeddings were used for neighbor graph construction, clustering, and UMAP. Data were normalized with SCTransform, and highly variable features were identified. Principal component analysis (PCA) was performed, followed by neighbor graph construction and clustering (resolution = 0.5). UMAP was used for visualization. CD4 and CD8 subsets were separately extracted and independently normalized, dimensionally reduced, clustered, and visualized by UMAP. CD4⁺ naïve T-cell subsets were further analyzed using pseudotime analysis with Slingshot, with the cluster exhibiting the highest stemness score (calculated as described below) used as the root. Stemness scores were calculated using a custom gene set (listed in Supplementary Table S3) via the AddModuleScore function in Seurat. Differential gene expression analysis was performed using FindAllMarkers (Wilcoxon rank-sum test, min.pct = 0.25, logfc.threshold = 0.25), and functional enrichment analyses (GO/KEGG) were conducted using clusterProfiler.

Public dataset analysis

Public mouse spleen single-cell RNA-seq data (syn2225543) were downloaded and analyzed using Seurat (v4). Only spleen cells (Spleen_Aged, Spleen_Young) were retained from this dataset. Standard quality control filters were applied, retaining cells with nFeature_RNA between 200 and 5000. Data were normalized (NormalizeData), highly variable features were identified (FindVariableFeatures), and scaled (ScaleData). Downstream analysis included principal component analysis (PCA), neighbor graph construction (FindNeighbors), clustering (FindClusters, resolution = 0.5), and UMAP dimensionality reduction (RunUMAP). Expression of marker genes (Itgb3, Ly6c1) was visualized and compared between age groups.

Raw 10X Genomics single-cell RNA-seq data from human PBMCs and cord blood (GSE157007) were downloaded, and the RNA expression matrices were extracted to create Seurat objects (v4). Standard quality control filters were applied, retaining cells with nFeature_RNA between 200 and 3000 and mitochondrial gene percentage < 10%. Data were normalized using the LogNormalize method, highly variable features were identified (FindVariableFeatures, nfeatures = 2000), and data were scaled (ScaleData). PCA was performed for dimensionality reduction, followed by batch correction using Harmony (group.by.vars = “sample”). Neighbor graphs were constructed (FindNeighbors, dims = 1:10), clustering was performed (FindClusters, resolution = 0.4), and UMAP was used for visualization (RunUMAP, dims = 1:10).

Public microarray gene expression data from human peripheral blood mononuclear cells (GSE63117; platform GPL7020, NuGO array) were downloaded from the Gene Expression Omnibus using the GEOquery R package. Normalized expression values and platform annotation provided by the original submitters were retrieved directly (AnnotGPL = TRUE). Probes were mapped to gene symbols using the accompanying platform annotation; probes without an annotated gene symbol were excluded, probes mapping to multiple gene symbols were assigned to the first listed symbol, and where multiple probes mapped to the same gene, expression values were averaged. Only baseline (T0, pre-caloric-restriction) samples from young (n = 10) and old (n = 9) healthy men were retained for downstream analysis, avoiding confounding by the subsequent dietary intervention.

Single-sample gene set enrichment analysis (ssGSEA) was performed using the GSVA R package (method = “ssgsea”, kcdf = “Gaussian”) on the normalized expression matrix, using a gene set representing naïve T-cell, RTE, and CD61-related transcriptional features (Supplementary Table S3), yielding an enrichment score for each sample. Pearson correlations between ITGB3 expression and signature scores were calculated within each age group. Correlation coefficients were compared between age groups using Fisher’s Z transformation, and partial correlation analyses were performed controlling for a T-cell abundance score.

Statistical analyses

Data are presented as mean ± SEM, except for non-normally distributed data (e.g., sjTREC), shown as median with interquartile range. Two-group comparisons used a two-tailed unpaired t-test (Welch’s t-test for unequal variances; Mann-Whitney U test for non-normally distributed data; Wilcoxon rank-sum test for single-cell data). Multiple-group comparisons used the Kruskal-Wallis test with pairwise Wilcoxon tests and Benjamini-Hochberg correction. Correlations were assessed by Pearson’s or Spearman’s correlation as indicated, with between-group differences in correlation evaluated by Fisher’s z-transformation. Single-cell subset frequencies were calculated from pooled cells. P < 0.05 was considered significant (*P < 0.05, **P < 0.01, ***P < 0.001); absence of asterisks indicates no significant difference. Analyses were performed using GraphPad Prism (v10.6.1) and R (v4.5.1).

Results

GFP-based lineage tracking reveals preferential thymus-dependent reconstitution of CD4⁺ and CD8⁺ naïve T cells after HSCT

To distinguish peripheral homeostatic expansion from thymus-dependent naïve T-cell reconstitution after transplantation, we established a congenic GFP-labeled mouse transplantation model. Lethally irradiated CD45.1 recipients received GFP⁺ splenic T cells from Rosa26-CAG-EGFP donors together with GFP⁻ TCD-BM from CD45.2 donors. Donor-derived cells were identified as CD45.2⁺, and GFP⁺ and GFP⁻ fractions were resolved within this compartment. GFP⁺ cells represent peripherally expanded donor T cells, whereas CD45.2⁺ GFP⁻ cells are generated de novo from donor HSPCs following thymic development in the recipient. CD45.2⁺ donor cells accounted for 92.5% ± 2.1% (n = 3), 87.9% ± 4.3% (n = 3), and 89.0% ± 4.3% (n = 3) of splenic T cells at 4, 8, and 24 weeks after transplantation, respectively, demonstrating stable and near-complete donor chimerism throughout the observation period. (Figure S1A). Recipient splenic T cells were isolated at 4, 8, and 24 weeks post-transplantation for single-cell RNA sequencing, with parallel validation by flow cytometry (Fig. 1a).

Fig. 1.

GFP-based lineage tracking reveals preferential thymus-dependent reconstitution of CD4⁺ and CD8⁺ naïve T cells after HSCT. a Schematic of the fully congenic GFP-labeled transplantation model. Lethally irradiated (8 Gy) recipient mice received GFP⁺ donor splenic T cells together with GFP⁻ T-cell-depleted bone marrow (TCD-BM). GFP⁺ T cells represent peripheral expansion-derived cells, whereas GFP⁻ T cells are derived from donor hematopoietic progenitors undergoing thymic-dependent development. Splenic T cells were collected at 4, 8, and 24 weeks post-transplantation for single-cell RNA sequencing (scRNA-seq) and flow cytometric analysis. An aliquot of GFP⁺ donor splenic T cells was additionally sorted and sequenced immediately prior to infusion, from the same cohort of donor mice, and is referred to as the baseline (Pre-HSCT) sample. b UMAP visualization of post-transplant splenic T cells resolved by scRNA-seq into multiple distinct clusters: CD4 naïve (Tnaïve), CD4 effector memory (Tem), CD4 regulatory (Treg), CD4 effector (Teff), CD4 exhausted (Tex), CD4 Th17, CD8 naïve (Tnaïve), CD8 Tcm, CD8 exhausted (Tex), and CD8 Teff. Expression of canonical lineage and functional markers (Cd4, Cd8a, Ccr7, Lef1, Sell, Tcf7, Foxp3, Pdcd1, Lag3, Ifng, Gzma, Gzmk) is shown, validating cluster annotation. scRNA-seq data were derived from n = 3 mice per group (GFP⁺ and GFP⁻) at each post-transplant time point, together with a single baseline (Pre-HSCT) sample of donor GFP⁺ splenic T cells collected prior to transplantation. A total of 118,275 cells passed quality control. c, d UMAP plots showing the temporal dynamics of peripheral expansion-derived (GFP⁺, c) and thymus-dependent (GFP⁻, d) T cells at 4, 8, and 24 weeks post-transplantation. In c, the baseline (Pre-HSCT) sample represents the T0 state of the same donor T-cell population subsequently tracked as GFP⁺ cells. In d, this sample is shown alongside GFP⁻ populations as a steady-state reference depicting the composition of mature, unmanipulated splenic T cells, Stacked bar plots show subset proportions calculated from pooled cells across all mice at each time point. e Flow cytometric validation of T-cell subset composition among GFP⁺ (left) and GFP⁻ (right) CD4⁺ and CD8⁺ T cells at indicated time points. Representative contour plots (gated on CD44 versus CD62L) are shown; stacked bar plots show the proportions of Tnaïve, Tcm, Tem, and Treg. Data are presented as mean ± SEM (n = 3 mice per time point for post-transplant groups; n = 10 mice for normal C57BL/6), from one independent experiment. f Flow cytometric analysis of GFP⁺ (peripheral reconstitution) and GFP⁻ (thymus-dependent central reconstitution) fractions within CD4⁺ naïve T and CD8⁺ naïve T cells at 4, 8, and 24 weeks post-transplantation. Left panels show representative FACS plots of CD4⁺ and CD8⁺ naïve T cells with GFP labeling. Right panels quantify the proportions of GFP⁺ and GFP- cells over time for both CD4⁺ and CD8⁺ naïve T-cell populations. Data are presented as mean ± SEM (n = 3 mice per time point) from one independent experiment

Single-cell transcriptomic analysis identified major post-transplant T-cell populations based on lineage- and function-associated marker genes, including CD4+ Tnaïve (Lef1, Sell, Tcf7, Ccr7), CD4 Tem (Il7r, Klf2), CD4 Treg (Foxp3, Ikzf2, Tnfrsf18), CD4 Teff (Nkg7, Ccl5), CD4 Tex (Pdcd1, Lag3, Tox), CD4 Th17 (Rorc), CD8 Tnaïve (Lef1, Sell, Tcf7, Ccr7), CD8 Tcm (Tcf7, Il7r), CD8 Teff (Gzma, Klrg1), and CD8 Tex (Pdcd1, Lag3, Tox) cells (Fig. 1b; Figure S1B, C). We next examined the temporal distribution of T cells reconstituted through peripheral expansion (GFP⁺) or the thymus-dependent central pathway (GFP⁻). In the peripheral expansion pathway, both CD4⁺ and CD8⁺ T cells underwent lymphopenia-induced homeostatic proliferation and acquired effector- and memory-like phenotypes, as indicated by the expansion of effector and memory subsets accompanied by contraction of the naïve compartment compared with the baseline (Pre-HSCT) donor T cells (Fig. 1c). In contrast, thymus-dependent central reconstitution predominantly generated naïve T cells in both the CD4⁺ and CD8⁺ compartments (Fig. 1d). Notably, centrally reconstituted CD4⁺ naïve T cells progressively differentiated into effector and memory subsets at later time point after HSCT (24 weeks), whereas CD8⁺ naïve T cells largely maintained their naïve phenotype over time (Fig. 1d).

We further validated these findings by flow cytometry. Consistent with the single-cell transcriptomic results, peripherally expanded CD4⁺ and CD8⁺ T cells preferentially differentiated into effector and memory subsets after transplantation, with this phenomenon being particularly pronounced in CD8⁺ T cells. In contrast, thymus-dependent central reconstitution predominantly generated naïve CD4⁺ and CD8⁺ T cells. Centrally reconstituted CD4⁺ naïve T cells gradually acquired effector and memory phenotypes at later stages, whereas CD8⁺ naïve T cells remained relatively stable in the naïve state (Fig. 1e). Because naïve T cells could arise from both peripheral homeostatic expansion and thymus-dependent de novo generation, we further gated on naïve T cells and analyzed their cellular origin. The results demonstrated that the majority of naïve T cells were derived from the thymus-dependent central pathway (Fig. 1f). Collectively, these data suggest that peripherally expanded T cells undergo substantial differentiation after HSCT, whereas a large fraction of post-transplant naïve T cells arises from the thymus-dependent central pathway in this GFP-labeled transplantation model.

Single-cell analysis reveals heterogeneous naïve T-cell reconstitution and identifies Itgb3- and Ly6c-associated naïve T-cell states

To further investigate whether naïve T cells generated through thymus-dependent central reconstitution and peripheral homeostatic expansion converge toward similar transcriptional states, we performed subclustering analysis of CD4⁺ and CD8⁺ naïve T cells. CD4⁺ naïve T cells were further classified into Ccr7-naïve, Itgb3-naïve, Ly6c-naïve, and Sox4-naïve T cells (Fig. 2a, Figure S2A).

Fig. 2.

Single-cell analysis reveals heterogeneous naïve T-cell reconstitution and identifies Itgb3- and Ly6c-associated naïve T-cell states. UMAP visualization of CD4⁺ T cells post-transplantation, showing distinct subclusters: Ccr7- naïve T, Itgb3- naïve T, Ly6c- naïve T, Sox4- naïve T, Tem, Treg, Teffector, Tex, Th17. b UMAP plots of CD4⁺ T cells showing expression of selected marker genes. Top row: markers used for annotation of CD4⁺ T-cell subclusters (Ccr7, Lef1, Sell, Tcf7, Il7r, Foxp3). Bottom row: RTE-related markers (S1pr1, Pecam1, Cd38, Itgae) and Itgb3/Ly6c1 expression across naïve T-cell subclusters. Color intensity represents expression levels in individual cells. c UMAP plots of CD4⁺ T-cell subclusters at 4, 8, and 24 weeks post-transplantation and pre-HSCT, separated into GFP⁻ (thymus-dependent) and GFP⁺ (peripheral expansion-derived) cells. Right, stacked bar plots showing subcluster proportions calculated from pooled cells at each time point. d Top differentially expressed genes for each CD4⁺ naïve T-cell subcluster, identified using FindAllMarkers. Red dots indicate genes with adjusted p-value < 0.01. e Stemness score (left) and Pecam1 expression (right) across the four CD4⁺ naïve T-cell subsets (Ccr7-, Itgb3-, Ly6c-, and Sox4- naïve T), calculated on a per-mouse basis (n = 3 mice). Box plots show the median and interquartile range; dots represent individual mice. Overall differences were assessed by the Kruskal-Wallis test, and pairwise comparisons by two-sided Wilcoxon rank-sum tests with Benjamini-Hochberg correction (**p < 0.01, ***p < 0.001, ****p < 0.0001). f Stemness score, RTE signature score, and Pecam1 expression between Itgb3⁺ and Itgb3⁻ CD4⁺ naïve T cells; the pie chart shows their relative proportions. Comparisons were performed using the two-sided Wilcoxon rank-sum test (****p < 0.0001). g UMAP visualization of CD8⁺ T-cell subclusters post-transplantation: Ccr7- naïve T, Ly6c- naïve T, Tcm, Teff, and Tex. h UMAP plots of CD8⁺ T cells showing expression of selected marker genes. Top, markers used for subcluster annotation (Ccr7, Lef1, Sell, Tcf7, Gzma, Gzmk); bottom, RTE-related markers (S1pr1, Pecam1, Cd38, Itgae) and Itgb3/Ly6c2 expression. Color intensity represents per-cell expression level. i UMAP plots of CD8⁺ T-cell subclusters at 4, 8, and 24 weeks post-transplantation and normal state, separated into GFP⁻ and GFP⁺ cells. Right, stacked bar plots showing subcluster proportions calculated from pooled cells at each time point. j UMAP visualization of T cells from spleens of young and aged mice (1,290 and 1,015 cells, respectively), with expression of selected genes (Sell, Lef1, Il7r, Ly6c1, Tcf7, Ccr7, Cd44, and Itgb3) shown below. Data were obtained from a publicly available dataset (syn22255433). k Bar plots and pie charts summarizing the mean expression and proportion of Itgb3⁺ and Ly6c1⁺ cells in young versus aged spleens, derived from the publicly available dataset in (j). As individual-animal information was not available in the original dataset, statistical testing was not performed; differences are shown descriptively to illustrate the trend toward reduced Itgb3 and Ly6c1 expression with aging. For panels a-i, scRNA-seq data are from the same dataset as Fig. 1 (n = 3 mice per group at each post-transplant time point)

Classical naïve T cell markers, including Ccr7, Lef1, Sell, and Tcf7, were consistently expressed in these CD4⁺ naïve T cell subsets, and classical RTE markers, including S1PR1 (encoded by the S1pr1 gene) and CD31 (encoded by the Pecam1 gene), were also highly expressed in these naïve T cells (Fig. 2b). However, previous studies identified that RTE marker Cd38 ⁴ was not highly expressed in any of the naïve compartment (Fig. 2b). Longitudinal UMAP analysis revealed dynamic remodeling of the CD4⁺ naïve T-cell compartment after transplantation. At the latest time point examined (24 weeks), the naïve compartment most closely resembled that of the normal steady state, being predominantly composed of Ly6c-associated naïve T cells, whereas at earlier time points it consisted mainly of Itgb3- and Ccr7-associated naïve T cells (Fig. 2c). Consistent with the previous study [4], Sox4-naïve T cells were enriched among centrally reconstituted T cells (Fig. 2c). Consistently, analysis of GFP composition across CD4⁺ clusters showed that all four naïve subsets were predominantly derived from the GFP⁻ (thymus-dependent) compartment, in clear contrast to the effector- and memory-associated clusters (Tem, Treg, Teff, Th17), which were predominantly GFP⁺ (Figure S2B). Besides, although these naïve CD4⁺ T- cell subsets consistently expressed classical naïve and RTE markers, they exhibited distinct transcriptional profiles (Fig. 2d, Figure S2C). Direct comparison of GFP⁺ and GFP⁻ CD4⁺ naïve T cells revealed that Itgb3, Sox4, and Ccr7 were significantly upregulated in the GFP⁻ fraction (Figure S2D). Genes upregulated in GFP⁻ cells were enriched for GO terms including lymphocyte differentiation, T cell differentiation, and negative regulation of cell migration. These findings indicate that the expression of Itgb3, Sox4, and Ccr7 is associated with the thymus-dependent reconstitution pathway.

We evaluated the functional state of four naïve T-cell subsets (Ccr7-, Itgb3-, Ly6c-, and Sox4-naïve T cells) using transcriptional stemness scoring. Stemness scoring analyses revealed that Itgb3-naïve T cells exhibited the highest stemness-related scores among all naïve T-cell subsets (Fig. 2e). Pseudotime analysis further showed that Itgb3-naïve T cells were positioned along a trajectory toward effector-memory-like (Tem) differentiation states (Figure S2E). We next assessed a recently defined RTE-associated transcriptional signature across the four naïve T-cell subsets. The RTE signature score was broadly comparable among Sox4-, Ccr7-, and Itgb3-naïve T cells and was lowest in Ly6c-naïve T cells (Figure S2F). However, the inclusion of the corresponding defining marker, Sox4, in the RTE signature may have contributed to the observed score in Sox4-naïve T cells (Supplementary Table S3). Expression of the classical RTE marker Pecam1 (CD31) was highest in Itgb3-naïve T cells among all subsets, further supporting Itgb3-naive T cells are enriched for classical recent thymic emigrant-associated features and suggesting that Itgb3/CD61 may complement established RTE markers (Fig. 2e). TCR repertoire analyses demonstrated that all four naïve T-cell subsets were predominantly composed of unique clonotypes. These subsets displayed highly comparable TCR diversity, consistent with the expected polyclonal nature of naïve T cells (Figure S2G). These findings indicate that Itgb3/CD61 expression defines a transcriptionally distinct naïve T-cell state characterized by higher stemness and RTE-associated features, without evidence of clonal restriction or altered TCR repertoire diversity.

To further characterize Itgb3 (CD61)-associated naïve T cells, we next extracted Itgb3⁺ and Itgb3⁻ CD4⁺ naïve T-cell subsets for comparative analyses. Itgb3⁺ naïve T cells accounted for approximately 25.1% of the CD4⁺ naïve T-cell compartment (Fig. 2f). Consistent with our previous findings, Itgb3⁺ naïve T cells exhibited significantly higher stemness-related scores compared with Itgb3⁻ naïve T cells (Fig. 2f). In parallel, Itgb3⁺ naïve T cells also showed significantly higher RTE signature scores, supporting their enrichment in thymus-associated transcriptional programs. In addition, Pecam1 (CD31) expression was significantly higher in Itgb3⁺ naïve T cells, further supporting the association of this subset with classical recent thymic emigrant-related features (Fig. 2f). Despite these transcriptional differences, TCR repertoire analyses revealed no significant difference in TCR diversity between Itgb3⁺ and Itgb3⁻ naïve T cells. Both subsets displayed highly comparable normalized Shannon evenness (Figure S2G). These findings further indicate that Itgb3 expression defines a transcriptionally distinct naïve T-cell state characterized by enhanced stemness and RTE-associated features without altering TCR repertoire diversity.

Consistent with our finding in CD4⁺ naïve T cells, we also identified substantial heterogeneity within the CD8⁺ naïve T-cell compartment. CD8⁺ naïve T cells were further divided into Ccr7-naïve and Ly6c-naïve subsets (Fig. 2g, Figure S2H). Classical naïve T-cell markers, including Ccr7, Lef1, Sell, and Tcf7, were consistently expressed in these CD8⁺ naïve T-cell subsets. The classical RTE-associated marker CD103 (encoded by the Itgae gene) was highly expressed in the Ccr7-CD8⁺ naïve T cell subset (Fig. 2h). A previous study has shown that CD38 is associated with thymic output in both CD4⁺ and CD8⁺ T cells [4]. Although Cd38 expression was not enriched in CD4⁺ naïve T cells (Fig. 2b), we observed high Cd38 expression in CD8⁺ naïve T cells, particularly within the Ly6c- CD8⁺ naïve T subsets (Fig. 2h). Longitudinal UMAP analysis further revealed distinct reconstitution patterns of CD8⁺ naïve T-cell subsets during peripheral and central reconstitution. Ly6c-CD8⁺ naïve T cells were preferentially detected among peripherally expanded cells, whereas Ccr7-CD8⁺ naïve T cells were detected in both peripherally expanded and centrally reconstituted compartments (Fig. 2i, Figure S2I). Consistently, direct comparison of centrally reconstituted (GFP⁻) and peripherally expanded (GFP⁺) CD8⁺ naïve T cells showed upregulation of Ccr7 and Itgae in the GFP⁻ fraction and of Ly6c2 in the GFP⁺ fraction, with GFP⁻-upregulated genes enriched for lymphocyte and T-cell differentiation programs (Figure S2J).

To further evaluate whether Itgb3 and Ly6c are associated with thymic output, we analyzed a public single-cell dataset of splenic T cells from young and aged mice. Feature mapping showed that Itgb3 and Ly6c1 expression was primarily observed within the naïve T-cell cluster and exhibited a distribution pattern similar to canonical naïve T-cell genes, including Ccr7, Lef1, Sell, Tcf7, and Il7r (Fig. 2j). Moreover, the average expression levels of both Itgb3 and Ly6c were higher in young mice than in aged mice (Fig. 2k). Together, these results suggest that, although these populations share a conventional naïve T-cell identity, naïve T cells at different stages of immune reconstitution display distinct transcriptional features. The enrichment of Itgb3 and Ly6c within the naïve T-cell compartment, together with their higher expression in young mice with preserved thymic activity than in aged mice, supports an association between these marker-defined naïve T-cell states and thymic functional status. Thus, these findings nominate Itgb3- and Ly6c-associated naïve T-cell states for further evaluation in the context of thymus-dependent immune reconstitution after HSCT.

CD61- and Ly6C-associated naïve T cells are associated with thymic functional status in transplantation and physiological thymic involution mouse models

To further evaluate whether the candidate markers identified by single-cell analysis were associated with thymic functional status, we examined CD61- and Ly6C- naïve T-cell subsets in transplantation and physiological thymic involution mouse models. To exclude the confounding effects of GVHD and focus specifically on central reconstitution of T cells, we transplanted only TCD-BM from congenic or allogeneic donors into irradiated recipient mice. At 4 weeks post-transplantation, we harvested spleens from both groups and compared the T-cell reconstitution between the two groups (Fig. 3a). Conventional thymic output-associated parameters, including naïve T cells and RTEs, were analyzed together with CD61- and Ly6C- naïve T-cell subsets. The gating strategy is shown in Figure S3A.

Fig. 3.

CD61- and Ly6C-naïve T cells are associated with thymic functional status in transplantation and physiological thymic involution mouse models a Experimental design illustrating congenic and haploidentical transplantation. Lethally irradiated (13 Gy) CD45.1 recipient mice received T-cell-depleted bone marrow from either C57BL/6 (congenic) or CB6F1 (haploidentical) donors. Splenocytes were collected at 4 weeks post-transplantation for flow cytometric analysis (n = 6 mice per group). b Flow cytometric analysis of splenic CD4⁺ and CD8⁺ T cells in congenic and haploidentical recipients at 4 weeks post-transplantation. Left, representative FACS plots showing gating of naïve T cells, RTE, and Ly6C-naïve T cells, together with representative CD61 histograms (FMO control shown in grey) used to quantify CD61 mean fluorescence intensity (MFI) in CD4⁺ and CD8⁺ naïve T cells. Right, bar and violin plots quantifying the proportions (percentage of parent) and absolute numbers of naïve T cells (of CD4⁺/CD8⁺), RTEs (of naïve), Ly6C-naïve T cells (of naïve), and CD61-naïve T cells (of naïve), together with scatter plots of CD61 MFI in CD4⁺ and CD8⁺ naïve T cells. Data are presented as mean ± SEM (n = 6 mice per group). Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05, **p < 0.01, ***p < 0.001). c Experimental design for the aging model. Young (9-week-old, n = 6) and middle-aged (48-week-old, n = 5) C57BL/6 mice were analyzed for splenic and thymic T-cell populations by flow cytometry. d Absolute numbers of thymic CD61⁺ CD4single-positive (CD4sp) and CD8single-positive (CD8sp) cells in young and middle-aged mice. Data are presented as mean ± SEM (n = 6 young, n = 5 middle-aged). Statistical significance was determined using a two-tailed unpaired t-test (**p < 0.01). e Flow cytometric analysis of splenic CD4⁺ and CD8⁺ T cells in young and middle-aged mice. Left, representative FACS plots showing naïve T cells, RTEs, Ly6C-naïve T cells, and CD61-naïve T-cell populations, together with representative CD61 histograms (FMO control shown in grey) used to quantify CD61 MFI in CD4⁺ and CD8⁺ naïve T cells. Right, bar and violin plots quantifying the proportions and absolute numbers of naïve T cells (of CD4⁺/CD8⁺), RTEs (of naïve), Ly6C-naïve T cells (of naïve), and CD61-naïve T cells (of naïve), together with scatter plots of CD61 MFI in CD4⁺ and CD8⁺ naive T cells

In the transplantation model, both CD4⁺ and CD8⁺ naïve T cells were more abundant in congenic recipients than in haploidentical recipients, indicating more efficient naïve T-cell reconstitution under the congenic condition (Fig. 3b). In contrast, RTE fractions showed relatively limited differences : CD4⁺ and CD8⁺ RTEs were comparable between the two groups (Fig. 3b). These findings suggested that RTE frequencies showed relatively modest differences between the two transplantation settings at this early time point. Among the candidate marker-defined subsets identified by single-cell analysis, Ly6C-CD4⁺ and CD61-CD8⁺ naïve T cells were reduced in haploidentical recipients, whereas Ly6C-CD8⁺ and CD61-CD4⁺ frequencies did not differ. CD61 MFI, a gate-independent readout measured against tissue-matched FMO controls (Figure S3B), showed the same pattern with CD61-associated naive T- cell frequencies (Fig. 3b). These findings suggest that selected CD61-naïve T cells are associated with more favorable naïve T-cell reconstitution in this transplantation model.

To determine whether the association between these naïve T-cell states and thymic activity extends beyond the transplantation setting, we next examined an independent physiological context in which thymic output declines. Thymus and spleen were collected from young (9-week-old) and middle-aged (48-week-old) mice for flow cytometric analysis (Fig. 3c, Figure S3C). In C57BL/6 mice, thymic involution begins shortly after sexual maturity and progresses continuously. Mice at 48 weeks of age therefore represent an intermediate stage of thymic decline rather than complete thymic atrophy, allowing assessment of marker-defined naïve T-cell subsets under partially reduced thymic activity. In the thymus, absolute numbers of CD61-CD4sp and CD61-CD8sp thymocytes were significantly lower in middle-aged compared with young mice (Fig. 3d). Gating was defined using a tissue-matched FMO control (Figure S3D). At the level of frequency, the reduction reached significance in CD61-CD8sp but not CD61-CD4sp thymocytes (Figure S3E). In contrast, Ly6C-CD4sp and Ly6C-CD8sp thymocytes showed no substantial differences in absolute numbers between young and middle-aged mice (Figure S3F). These findings further support an association between CD61-expressing thymocyte populations and age-related thymic functional decline.

We then examined whether age-associated thymic involution was reflected in peripheral splenic T cells. Middle-aged mice showed reduced frequencies and absolute numbers of both CD4⁺ and CD8⁺ naïve T cells (Fig. 3e). In contrast, conventional RTE frequencies showed relatively modest changes, although the absolute number of CD8⁺ RTEs was reduced. The frequency of Ly6C-CD4⁺ naïve T cells decreased in middle-aged mice, whereas Ly6C-CD8⁺ naïve T cells showed no significant change. However, the absolute numbers of both subsets were reduced. Notably, CD61-associated changes were more evident in the CD4⁺ compartment, with significant reductions in CD61-associated naïve T-cell frequency, absolute number, and CD61 MFI, whereas CD61-CD8⁺ naïve T cells showed no consistent age-related changes (Fig. 3e, Figure S3G). These findings suggest that reduced CD61 expression in naïve T cells, especially within the CD4⁺ naïve compartment, is associated with age-related thymic involution.

Together, these findings indicate that marker-defined naïve T-cell subsets display lineage- and context-dependent patterns across transplantation and aging models. CD61-associated and Ly6C-associated CD4⁺ naïve T-cell subsets may therefore provide complementary phenotypic information to RTE analysis for evaluating thymic functional status, rather than replacing established RTE markers.

Human transcriptomic datasets reveal naïve T-cell abundance and CD61-associated signatures as correlates of age-related thymic activity.

To explore whether thymic-output-associated features identified in mouse models were also associated with age-related thymic activity in humans, we analyzed public scRNA-seq data from cord blood and peripheral blood samples from young and old individuals (Fig. 4a). After dimensionality reduction and annotation, T cells were classified into major populations, including naïve T cells (LEF1, SELL, TCF7, CCR7, IL7R), CD4 Tcm (CD4, IL7R, EZR), CD8 Tcm (CD8A, IL7R, EZR), CD8 Tem (GZMB, NKG7), Treg (FOXP3), and exhausted T cells (TOX, NEAT1) (Fig. 4b, c). Classical naïve T-cell markers, including CCR7, LEF1, SELL, TCF7, and IL7R, were highly expressed in the naïve T-cell compartment. Classical RTE markers, such as CD31 (encoded by the PECAM1 gene) and CD38, were also detected in subsets of naïve T cells (Fig. 4b). Comparison across age groups showed a progressive decline in naïve T-cell abundance, with the highest proportion in cord blood, an intermediate level in young individuals, and the lowest level in older individuals (Fig. 4d), consistent with previous studies [18]. Besides, RTE-associated transcriptional signature scores within naïve T cells showed a similar age-associated pattern, with higher scores in cord blood and young individuals and lower scores in older individuals (Fig. 4e). RTE signature score represents a transcriptional program associated with naïve T-cell and thymic output-related gene expression (Supplementary Table S3). These results support an association between naïve T-cell abundance, RTE-associated transcriptional features, and age-related thymic activity in healthy human cohorts.

Fig. 4.

Human transcriptomic datasets support naïve T-cell abundance and ITGB3(CD61)-related signatures as features associated with age-relatedthymic activity. a Schematic of the study design. Cohort 1: publicly available single-cell RNA-seq data (GSE157007) from cord blood (n = 3), young adults (n = 3), and old adults (n = 6). Cohort 2: a clinical PBMC microarray dataset (GSE63117) from young (n = 10) and old (n = 9) individuals. b Single-cell RNA-seq analysis of Cohort 1. Left, UMAP showing T-cell subclusters (Naive T, CD4 TCM, CD8 TCM, CD8 TEM, Exhausted T, CD4 Treg). Right, expression of key marker genes, including RTE-associated genes and ITGB3, across cells. A total of 128,291 cells passed quality control. c Dot plot of marker genes used to annotate T-cell subclusters. Dot size indicates the proportion of cells expressing each gene; color represents the average expression level. d UMAP plots of T-cell subclusters for cord blood, young, and old groups (top), with stacked bar plots showing subcluster proportions (calculated from pooled cells) in each group (bottom). e Naïve T-cell proportions (left) and RTE signature scores within naïve T cells (right) across cord blood, young, and old groups. Data are presented as mean ± SEM (n = 3 cord blood, 3 young, 6 old). Overall differences were assessed by the Kruskal-Wallis test, with pairwise comparisons by Wilcoxon rank-sum tests and Benjamini-Hochberg correction (***p < 0.001). f ssGSEA enrichment scores for a naïve T/RTE/ITGB3 gene set in samples from microarray dataset, comparing young and old groups (n = 10 young, 9 old). Box plots show the median and interquartile range. Statistical significance was determined using Welch’s t-test (unequal variances) (*p < 0.05; p = 0.022). g Pearson correlations of ITGB3 expression with naïve T signature score and total T-cell signature score in young and old groups (n = 10 young, 9 old). Left, scatter plots with Pearson correlation coefficients; right, bar plots of correlation coefficients. h Pearson correlations of ITGB3 with individual naïve T markers (CD3D, CCR7, IL7R, LEF1, CD27) in young and old groups (n = 10 young, 9 old). Left, scatter plots with Pearson correlation coefficients; right, bar plots of correlation coefficients. The naïve T/RTE/ITGB3 gene dataset, human RTE gene dataset, and human stemness gene dataset were listed in Supplementary Table S3.

Next, we examined whether candidate marker-associated features identified in our mouse models showed similar age-associated patterns in human datasets. As Ly6C lacks a clear human ortholog, we focused on CD61 and examined whether CD61-related features were associated with age-related thymic activity in humans. However, ITGB3/CD61 expression was sparsely detected in the scRNA-seq dataset (Fig. 4b), likely due to limited sequencing depth and dropout events. We therefore analyzed public human peripheral blood microarray dataset to examine whether CD61-related transcriptional features were associated with age-related thymic activity (Fig. 4a). Using a curated gene set incorporating naïve T-cell, RTE-associated, and ITGB3-related features (Supplementary Table S3), we calculated a composite ssGSEA score. A composite naïve T/RTE/CD61-associated transcriptional signature, defined by ssGSEA-based enrichment of a curated gene set, was significantly higher in young individuals compared with old individuals in human peripheral blood samples (Fig. 4f), consistent with an immune state associated with relatively preserved thymic activity in younger populations. We next examined the association between ITGB3 expression and naïve T-cell transcriptional programs. ITGB3 expression showed a modest positive association with the naïve T-cell signature score, with this association being more apparent in young individuals than in old individuals (Fig. 4g). Consistently, ITGB3 expression showed a tendency toward positive associations with canonical naïve T-cell genes, including CCR7, CD27, IL7R, and LEF1, particularly in young individuals (Fig. 4h). Collectively, these findings suggest that ITGB3/CD61-related transcriptional features are associated with naïve T-cell programs and are enriched in younger individuals with relatively preserved thymic function.

Naïve T cells and CD61⁺ naïve T-cell subsets are associated with age-related thymic activity in primary human peripheral blood

Having observed associations between ITGB3/CD61-related features and age-related thymic activity in public human transcriptomic datasets, we next examined these features in primary human peripheral blood samples from healthy individuals. Peripheral blood was collected from 23 young individuals (9–20 years, median age 16 years) and 18 old individuals (55–73 years, median age 61 years). PBMCs were isolated for parallel analyses, including qRT-PCR quantification of sjTRECs and flow cytometric assessment of naïve T cells, classical RTEs, and CD61⁺ naïve T-cell subsets (Fig. 5a). sjTRECs are episomal DNA by-products generated during TCRα rearrangement that do not replicate during cell division and are therefore widely used as molecular indicators of recent thymic output and thymic function [19]. The gating strategy for CD4⁺ and CD8⁺ naïve T cells, CD31⁺ CD4⁺ naïve T cells, CD38⁺- CD8⁺ naïve T cells, and CD61⁺naïve T-cell subsets is shown in Figure S4A.

Fig. 5.

Naïve T cells and CD61⁺ naïve T-cell subsets are associated with age-related thymic activity in primary human peripheral blood. a Schematic of the study design. PBMCs were collected from young (9–20 years, median 16; n = 23) and old (55–73 years, median 61; n = 18) individuals. Samples were analyzed by qRT-PCR for sjTREC levels and by flow cytometry for naïve T cells, CD31⁺ or CD38⁺ recent thymic emigrants (RTEs), and CD61⁺ naïve T cells. sjTREC quantification and correlation analyses (panels e and g) were performed on a subset of 14 individuals (n = 6 young, 8 old) for whom matched sjTREC and flow cytometry data were available. b-c Flow cytometric analysis of naïve T cells and RTEs. Left, representative dot plots from young and old individuals; right, quantification of naïve T (% of CD4⁺/CD8⁺) and CD31⁺CD4⁺ (CD4 RTE) or CD38⁺CD4⁺ (CD4 RTE) or CD38⁺CD8⁺ (CD8 RTE) naïve T cells (% of Tnaïve). Data are presented as mean ± SEM (n = 23 young, 18 old). Statistical significance was determined by two-tailed unpaired t-test (***p < 0.001). Representative plots show samples approximating the group mean. d Flow cytometric analysis of CD61⁺ naïve T cells within CD4⁺ and CD8⁺ naïve T cells. Left, representative dot plots from young and old individuals; right, quantification of CD61⁺ cells (% of CD4⁺ and CD8⁺ naïve T) and CD61 mean fluorescence intensity (MFI) in CD4 and CD8 naïve T cells. Data are presented as mean ± SEM (n = 23 young, 18 old). Statistical significance was determined by two-tailed unpaired t-test (*p < 0.05, ***p < 0.001). e sjTREC levels measured by qRT-PCR in PBMCs from young and old individuals (n = 6 young, 8 old). Data are presented as mean ± SEM. Statistical significance was determined by a two-tailed unpaired Student’s t-test (***p < 0.001). f Correlation between sjTREC levels and T-cell subsets was assessed using linear regression in matched samples (n = 14; young n = 6, old n = 8). Scatter plots show associations between sjTREC and CD4⁺/CD8⁺ naïve T cells, CD31⁺CD4⁺ and CD38⁺CD8⁺ naïve T-cell subsets, and CD61⁺ naïve T-cell subsets, with coefficients of determination (R²), regression lines, and 95% confidence intervals indicated. g Pearson correlation analysis of CD61⁺ naïve T cells with CD31⁺ CD4⁺ naïve T cells or or CD38⁺ CD4⁺ naïve T cells or CD38⁺ CD8⁺ naïve T cells (n = 6 young, 8 old). h Sorting strategy and sjTREC quantification in CD61⁺ and CD61⁻ naïve T cells from healthy human PBMCs. Left, representative FACS plots showing CD61 FMO controls and the corresponding sorting gates used to isolate CD61⁻ and CD61⁺ populations within CD4⁺ and CD8⁺ naïve T cells. Right, sjTREC copy number per 50,000 sorted cells in paired CD61⁻ and CD61⁺ CD4⁺ (top) and CD8⁺ (bottom) naïve T cells, determined by qPCR following genomic DNA extraction. Bars indicate mean values and lines connect paired samples from the same donor (n = 3 healthy donors). i Average proportion of naïve T cells and their subsets within total CD4⁺ T cells (left) and total CD8⁺ T cells (right) at 6 and 12 months post-transplant.

Flow cytometric analysis showed that young individuals had higher frequencies of CD4⁺ and CD8⁺ naïve T cells than older individuals (Fig. 5b). Consistently, both CD31⁺CD4⁺ naïve T cells (CD4 RTEs), CD38⁺CD4⁺ naïve T cells (CD4 RTEs) and CD38⁺CD8⁺ naïve T cells (CD8 RTEs) were also enriched in the young group (Fig. 5c), consistent with relatively preserved thymic activity in younger individuals. We then focused on CD61 expression within the naïve T-cell compartment. The frequency of CD61⁺ cells within CD4⁺ naïve T cells was higher in young individuals than in older individuals. In addition, CD61 MFI in CD8⁺ naïve T cells was also increased in the young group (Fig. 5d), indicating age-associated differences in CD61 expression across naïve T-cell subsets. In agreement with these cellular findings, sjTREC levels were higher in the young group than in the old group, providing molecular support for higher recent thymic output in young individuals (Fig. 5e).

Moreover, we further evaluated the relationship between sjTREC levels and different naïve T cell subsets. CD61 expression in CD4⁺ naïve T cells was positively correlated with sjTREC abundance (R² = 0.50, P = 0.00482), although this correlation was slightly weaker than that observed for CD31⁺CD4⁺ naïve T cells (R² = 0.58, P = 0.00151). The association between CD61⁺CD8⁺ naïve T cells and sjTREC levels (R² = 0.37, P = 0.0221) was weaker than that observed for CD38⁺CD8⁺ naïve T cells (R² = 0.52, P = 0.00378) (Fig. 5f).

Moreover, CD61 expression in naïve T cells was positively correlated with both naïve T-cell and RTE (CD31⁺CD4⁺ naïve T cells, CD38⁺CD4⁺ naïve T cells, and CD38⁺CD8⁺ naïve T cells) frequencies, and negatively correlated with age in both the CD4 and CD8 T-cell compartments (Fig. 5g, Figure S4B, C). Correlation matrix analysis further showed that CD61⁺ naïve T-cell frequencies were strongly correlated with conventional naïve T-cell and RTE measures and moderately correlated with sjTREC levels, with the inverse association with age being more pronounced in the CD4⁺ compartment (Figure S4D). In multivariate regression analyses using CD61⁺ naïve T-cell frequency and sjTREC levels as joint predictors of age, CD61 remained independently associated with age after adjustment for sjTREC levels in the CD4⁺ compartment (partial r = -0.49), whereas no significant independent association was observed in CD8⁺ T cells (partial r = -0.06) (Figure S4E). These findings suggest that CD61 provides complementary information to conventional naïve T-cell and RTE markers for assessing thymic activity.

To further directly examine the association between CD61 expression and recent thymic output, we sorted CD61⁺ and CD61⁻ naïve T cells from healthy donors and quantified sjTREC levels in each population. In both the CD4⁺ and CD8⁺ compartments, CD61⁺ naïve T cells consistently exhibited higher sjTREC content than their paired CD61⁻ counterparts across all three donors (Fig. 5h). These findings provide direct cell-sorted evidence supporting the association between CD61 expression and thymic output. We also enrolled an independent cohort of allo-HSCT recipients and performed longitudinal analyses of peripheral blood samples to assess RTE reconstitution after transplantation. The baseline characteristics and clinical outcomes of the allo-HSCT cohort are summarized in Supplementary Table S4. The results showed that total naïve T-cell frequencies progressively increased over time, reaching a higher level at 12 months post-transplant compared with earlier time points (6 months) (Fig. 5i). Importantly, CD61⁺ naïve T cells, together with classical thymic-output-associated subsets, including CD31⁺ CD4⁺ naïve T cells and CD38⁺ CD8⁺ naïve T cells, exhibited a coordinated expansion at 12 months after HSCT (Fig. 5i), suggesting that CD61⁺ naïve T cells are associated with post-transplant immune reconstitution dynamics.

Taken together, these data from healthy human peripheral blood samples support naïve T cells and classical RTEs as established peripheral indicators of thymic activity and suggest that CD61 enrichment within the naïve T-cell compartment provides complementary information associated with a thymus-associated immune state.

Discussion

The thymus plays a central role in maintaining T-cell diversity and immune surveillance throughout life. Although thymic activity progressively declines with age, residual thymic function remains important for immune homeostasis and can contribute substantially to immune reconstitution following HSCT. However, the cellular features that distinguish thymus-dependent naïve T-cell regeneration from peripheral homeostatic expansion remain incompletely understood. In particular, the phenotypic and transcriptional states associated with newly generated thymus-derived naïve T cells during post-transplant immune reconstitution have not been systematically characterized.

In this study, we used a GFP-labeled congenic transplantation mouse model to directly distinguish peripherally expanded T cells from de novo generated thymus-derived T cells during immune reconstitution. By integrating longitudinal single-cell transcriptomic, TCR repertoire, and flow cytometric analyses, we generated a comprehensive map of naïve T-cell regeneration after HSCT. Our results demonstrate that naïve T cells generated through peripheral expansion and thymus-dependent regeneration are not simply equivalent populations with different origins, but instead occupy distinct transcriptional states during immune reconstitution. Furthermore, we identified multiple naïve T-cell states enriched within the thymus-derived compartment, including Itgb3 (CD61)-high and Ly6c-high populations, suggesting that thymus-dependent regeneration is accompanied by dynamic remodeling of the naïve T-cell compartment rather than the generation of a single homogeneous RTE population.

T-cell reconstitution after HSCT occurs through two complementary pathways: an early phase dominated by peripheral expansion of mature donor-derived T cells and a later phase characterized by thymus-dependent de novo T-cell generation [5, 20]. Previous studies have consistently shown that CD4⁺ T cells rely more heavily on thymus-dependent regeneration, whereas CD8⁺ T cells undergo more rapid peripheral expansion during the early post-transplant period [5, 21–23]. Our previous work further demonstrated that these lineage-specific differences are particularly evident within the memory and effector compartments, where CD8⁺ memory and effector T cells are largely maintained through peripheral expansion, whereas their CD4⁺ counterparts depend more extensively on thymus-dependent central reconstitution [13]. In the present study, we further showed that naïve T cells from both the CD4⁺ and CD8⁺ lineages are predominantly regenerated through the thymus-dependent pathway. Furthermore, we found that thymus-derived naïve T cells undergo substantial transcriptional remodeling during immune recovery. Notably, the transcriptional landscape of naïve T cells gradually converged toward a steady-state profile at later post-transplant time points, suggesting progressive maturation and adaptation of newly generated T cells after thymic egress. This interpretation is consistent with previous studies showing that recent thymic emigrants continue to mature in the periphery for several weeks after leaving the thymus rather than immediately acquiring a fully mature naïve phenotype [24, 25].

Among the thymus-associated naïve T-cell states identified in this study, Ly6C and CD61 exhibited distinct biological characteristics. Ly6C has previously been associated with heterogeneity within murine naïve T-cell populations and has been linked to differences in self-reactivity, interferon responsiveness, and maturation status [26–30]. Consistent with these observations, Ly6C-high naïve T cells in our study displayed transcriptional features more closely resembling long-term steady-state naïve T cells, suggesting that Ly6C may mark a maturation-related state within the murine naïve T-cell continuum. However, because Ly6C lacks a clearly defined human ortholog, its translational relevance remains limited. Therefore, Ly6C should be viewed primarily as a mouse-specific feature associated with naïve T-cell maturation and immune reconstitution.

In contrast, ITGB3/CD61 is conserved between mice and humans, allowing evaluation across species. CD61 is a β3 integrin involved in cell adhesion, migration, and cellular interactions within tissue microenvironments [31–33]. Although integrin-mediated signaling has been implicated in thymocyte development and peripheral T-cell function [34], the role of CD61 in naïve T-cell biology remains poorly understood. In our study, CD61 expression was enriched in thymus-derived naïve T cells generated during the early phase of post-transplant immune reconstitution and was associated with transcriptional features consistent with a less differentiated state. CD61⁺ naïve T cells displayed elevated stemness-associated transcriptional programs and a strong association with established RTE-related features. Collectively, these findings suggest that CD61 identifies a naïve T-cell state associated with thymus-dependent regeneration.

Several established markers have been used to assess RTEs and thymic function, including CD31, CD38, CD103, PTK7, CR1, CR2, and Sox4-associated transcriptional programs [1, 4]. Previous studies have reported that SOX4 is enriched in human RTEs and that CD38^hi naïve T cells transcriptionally overlap with SOX4-associated RTE-like populations [1, 4]. In our mouse model, we observed a Sox4-associated naïve T-cell state. However, CD38 expression was not consistently enriched across both CD4⁺ and CD8⁺ naïve T-cell compartments. This discrepancy may reflect species differences, the inflammatory and lymphopenic environment after HSCT, differences in sampling time points, and the limited sensitivity of scRNA-seq for low-abundance or transiently expressed genes. Moreover, CD38 expression can be influenced by T-cell activation and metabolic states, potentially limiting its stability as an RTE marker during post-transplant immune reconstitution. CD31 and CD103 have also been widely used to assess thymic-output-associated naïve T-cell populations, although both markers have context-dependent limitations. CD31/PECAM1 is primarily used to identify TREC^hi “thymic naïve” cells within the human CD4⁺ naïve T-cell compartment and is therefore more informative in CD4⁺ T cells; however, its expression can also be affected by aging, proliferation, and peripheral homeostatic expansion [35, 36]. CD103/ITGAE was initially reported to associate with TREC^hi naïve CD8⁺ T cells [37]. However, as a canonical integrin involved in tissue residency and epithelial adhesion, CD103 is not specific for RTEs [38]. More recently, CD103 expression in circulating naïve CD8 T cells was shown to contribute to cytokine-driven acquisition of memory-like and effector programs [39], suggesting that CD103 may reflect tissue-adaptive and functional states rather than stable thymic output. Similarly, PTK7, CR1, and CR2 have been proposed as markers of human RTEs, but their expression is also influenced by developmental stage, age, and cellular context, highlighting the complexity of defining thymic output using any single marker.

Consistent with previous studies, our analyses confirmed that many of these conventional markers remain associated with thymic activity in both murine and human settings. However, our findings also suggest that thymus-dependent immune reconstitution cannot be fully captured by a single phenotypic marker. Instead, different markers likely reflect overlapping but non-identical aspects of T-cell biology, including thymic output, thymic egress, peripheral maturation, homeostatic adaptation, and functional differentiation. Within this framework, CD61 should not be viewed as a replacement for established indicators of thymic output. Rather, CD61 appears to identify a distinct naïve T-cell state associated with thymus-dependent regeneration and may provide complementary information regarding the heterogeneity of the regenerating naïve T-cell compartment. Future studies integrating phenotypic markers, TCR repertoire analyses, and direct sjTREC quantification in purified cell populations will be required to more precisely define the relationships among these markers and their respective contributions to monitoring thymic recovery after HSCT.

Several limitations of this study should be acknowledged. First, because CD61⁺ naïve T cells constitute only a small fraction of the peripheral naïve T-cell compartment, particularly in adults, direct sjTREC quantification in sorted CD61⁺ and CD61^- naive T-cell subsets was feasible in only a limited number of donors. Although CD61⁺ naive T cells showed higher sjTREC content than their paired CD61^- counterparts, this analysis was based on only three healthy donors and therefore requires validation in larger independent cohorts. Second, functional studies were not performed to determine whether CD61 plays a causal role in thymocyte maturation, thymic egress, peripheral survival, or T-cell function. Therefore, CD61 should currently be regarded as a phenotypic marker associated with thymus-dependent regeneration rather than a functional mediator. Third, although we evaluated the translational relevance of CD61 in healthy young and old individuals and performed preliminary validation in longitudinal HSCT recipients, the current cohort size is insufficient to establish a definitive relationship between CD61-associated thymic recovery and long-term clinical outcomes. Larger prospective longitudinal studies in HSCT recipients, ideally incorporating serial sjTREC measurements and established RTE markers, will be required to determine whether CD61 provides additional value for monitoring thymic recovery after transplantation and whether such monitoring is associated with improved clinical outcomes.

In summary, our study systematically characterized the cellular and molecular features of naïve T-cell regeneration following HSCT and revealed distinct transcriptional states associated with thymus-dependent and peripheral pathways of immune reconstitution. We identified CD61- and Ly6C-associated naïve T-cell states enriched within the thymus-derived compartment and demonstrated that ITGB3/CD61 is associated with thymus-related features in both mouse and human datasets. Our findings highlight previously unrecognized heterogeneity within regenerating naïve T cells and provide a framework for investigating thymus-dependent immune regeneration and peripheral signatures associated with immune recovery after HSCT.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1. ^{(131.5KB, pdf)}

Supplementary Material 2. ^{(3.2MB, pdf)}

Author contributions

The authors declare no potential conflicts of interest.F. Q. Zhang and H. D. Guo conceived the study. F. Q. Zhang, X. Y. Jiang, S. Fan, B. X. Wang, Z. G. Wu, Q. Zhang, J. R. Zhou, Z. Y. Zhang, J. N. Tang, Z. L. Xu, and Y. Wang performed the experiments. F. Q. Zhang and H. D. Guo wrote the manuscript. H. D. Guo and X. J. Huang provided a critical review of the manuscript. H. D. Guo and X. J. Huang supervised the study. H. D. Guo and X. J. Huang provided funding support. All authors reviewed and provided feedback on the manuscript.

Funding

This work was partly supported by grants from the Major Program of the National Natural Science Foundation of China (No. 82293630), the National Natural Science Foundation of China (No. 82570265 and 82300243), the Beijing Natural Science Foundation General Project (No. 7252142).

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding authors upon reasonable request. Any publicly available datasets are cited in the manuscript.

Declarations

Ethics approval and consent to participate

The study protocol involving animal and clinical experiments was approved by the Ethics Committee of Peking University People’s Hospital (Approval No. 2023PHE027 for animal experiments; Approval No. 2023PHB098-001 for clinical experiments) and conducted in accordance with the Declaration of Helsinki and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Competing interests

The authors declare no competing interests.