Peripheral blood samples were obtained from systemically healthy individuals who did not suffer from diseases known to affect the immune system. Overall, our cohort included 60 donors, age range 21–84, 28 males and 32 females. The number of donors used for each individual experiment is indicated in the figure legends. Purification of PBMCs from heparinized blood was performed by density gradient centrifugation (Lymphoprep, Stemcell). Freshly purified PBMCs were washed with RPMI 1640 medium supplemented with 100 U/mL penicillin, and 100 μg/mL streptomycin (P/S, both Sigma-Aldrich) and finally resuspended in medium additionally supplemented with 10% fetal calf serum (FCS, Sigma-Aldrich, complete medium).

Flow cytometry experiments were performed in accordance with the “Guidelines for the use of flow cytometry and cell sorting in immunological studies” (Cossarizza et al., 2021). Immunofluorescence surface staining was performed by adding a panel of directly conjugated antibody to freshly prepared PBMCs. Dead cells were excluded from the analysis using a fixable viability dye (Zombie Aqua™ Fixable Viability Kit, Biolegend) or DAPI. After surface staining, cells were fixed and permeabilized using the Cytofix/Cytoperm kit (BD Pharmingen) for TNF, IFNγ, and granzyme B (GrzB) or the eBioscience Foxp3/Transcription Factor Staining Buffer Set (ThermoFisher Scientific) for Tbet and incubated with intracellular Abs. To analyze TNF and IFNγ expression, PBMCs were previously stimulated for 4 h at 37 °C 5% CO₂ with 30 ng/mL PMA and 500 ng/mL ionomycin in the presence of 10 μg/mL Brefeldin A (BFA, Sigma–Aldrich) or for 16 h with 1 μg/ml anti-CD3 + 10 μg/mL BFA. GrzB expression was assessed in unstimulated cells. Proliferation was measured after stimulation with 1 μg/ml anti-CD3 or 1 μg/ml PHA for 4 days. Samples for flow cytometric analysis were measured using a BD LSR Fortessa X‐20 (BD Biosciences), and analysis was performed with FlowJo software (FlowJo, version 10.8.1). Antibodies used for this study are shown in table 1.

Cellular senescence in naïve CD8⁺ T cells was assessed using the Cellular Senescence Detection Kit‐SPiDER‐βGal (Dojindo Molecular Technologies). Bafilomycin A1 was reconstituted in 30 μL dimethyl sulfoxide (DMSO, Sigma Aldrich) and SPiDER‐βGal in 20 μL DMSO (Pangrazzi et al., 2025). PBMCs were incubated in complete medium with a 1:500 dilution of bafilomycin A1 for 1.5 h at 37 °C 5% CO₂ before the addition of 1:1000 dilution SPiDER‐βGal for another 30 min. PBMCs were washed with PBS and afterwards incubated with surface Abs at 4 °C diluted in PBS.

Total mitochondrial content was assessed in naïve CD8⁺ T cell subsets within PBMCs using 100 nM MitoTracker Deep Red FM and mitochondrial ROS with 100 nM reduced MitoTracker Red CM-H2-XRos (both ThermoFisher Scientific). Mitochondrial membrane potential was assessed using 0.5 μg/ml JC-1 (ThermoFisher Scientific). Derived PE/FITC mean fluorescence ratio was calculated using Flowjo and was normalised against Mitotracker Deep Red (DR) intensity. PBMCs were incubated in the presence of each dye as well as surface Abs for 20min at 37 °C 5% CO₂ and afterwards washed with PBS. Negative control samples (i.e., PBMCs stained with JC-1 in the presence of 5 µM FCCP to induce mitochondrial membrane potential depolarization) as well as positive controls (i.e., PBMCs stained with 5 µM oligomycin to induce mitochondrial membrane potential hyperpolarization) were additionally included. After the incubation, cells were washed with PBS and measured immediately at the flow cytometer.

To assess intracellular ROS, PBMCs were incubated with the fluorescent dye dihydroethidium (DHE, Sigma-Aldrich) at a concentration of 50 nM in complete RPMI for 20 min at 37 °C 5% CO₂. Cells were washed in PBS and measured at the flow cytometer.

Proliferation was measured after labelling of PBMCs with Cell Proliferation Dye eFluor™ 450 (CPD450, ThermoFisher Scientific) and stimulation with either 1 μg/mL anti-CD3 Ab or 1 μg/mL PHA (BD Biosciences) for 4 days. After incubation, cells were stained with surface Abs and measured by FACS. Results were displayed using the proliferation index (PI, calculated as Σ n > 1 P n 2 n /   Σ n > 0 P n 2 n ) and percentage of proliferated cells (i.e., number of proliferating cells/total number of cells) * 100.

Apoptosis and DNA damage were assessed after incubation of PBMCs with 30 μg/mL etoposide (Sigma–Aldrich) for 1 day at 37 °C 5% CO₂. Apoptotic cells were assessed after intracellular staining with caspase three or alternatively with 400 nM Apotracker green (Biolegend) in the presence of surface Abs and DAPI. Early apoptotic cells were identified as Apotracker⁺ DAPI⁻ and late apoptotic cells as Apotracker⁺ DAPI⁺. DNA damage was quantified after intracellular staining with anti-γH2AX pS139-FITC (REA502, Miltenyi) Ab.

Statistical significance was assessed using nonparametric Friedman test followed by Dunn’s post hoc test, and Spearman correlations, as indicated in the figure legends. A p-value <0.05 was considered significant. Both analyses were performed using the GraphPad 10.1 software. Cluster analysis and related statistics were performed using the JMPpro17 software (SAS Institute Inc., Cary, NC, 2023). Unsupervised clustering was carried out using the ‘hierarchical clustering’ function embedded in JMP environment. The elbow method was used to determine the ideal number of clusters (Thorndike, 1953). Prior to the clustering, the sphericity assumption was confirmed using a Bartlett’s test. The effect size and significant impact of immunological parameters on the clustering model was assessed using a multiple regression model controlling false-discovery rate. Differences in the frequency distributions of cell populations within clusters were evaluated using Fisher’s exact test. Means ± SD are shown in each graph.

To investigate the differentiation state of naïve CD8⁺ T cell subsets, the expression of key T cell differentiation markers was measured in naïve CD8⁺ T cells defined using CD45RA in combination with CCR7 (N_CCR7), CD28 (N_CD28), or CD27 (N_CD27), as well as in phenotypically most “true-naïve”-like CD45RA⁺CCR7⁺CD28⁺CD27⁺CD57⁻ (N_TN) cells within PBMCs from donors of different age (Figure 1). The gating strategy used to define the four subsets is reported in Supplementary Figure S1. The expression of the chemokine receptor CX3CR1 increases during T cell differentiation (Zwijnenburg et al., 2023). When we measured CX3CR1 levels in the four naïve subsets, it was the highest in the N_CCR7 subpopulation, and it significantly decreased in N_CD28, N_CD27 and N_TN cells (Figure 1a). Representative FACS plots showing the levels of CX3CR1 in the four subpopulations is shown in Supplementary Figure S2. The lowest expression was found in the N_TN subset, although they were particularly low also in N_CD27. Natural killer (NK) markers are known to increase with T cell differentiation and show the highest levels in terminally differentiated T cells (Pangrazzi et al., 2020; Koh et al., 2023). The expression of CD16, CD94, NKp80 and CD56 was assessed in the four naïve CD8⁺ T cell subsets and was higher in the N_CD28 cells compared to the other naïve subsets, slightly lower in N_CCR7, and further decreased in the N_CD27 and N_TN populations (Figures 1B–E). The NK inhibitory receptor CD158 (CD158b/CD158e1), which belongs to the killer-cell immunoglobulin-like receptor (KIR) family, was absent on N_TN cells, while low but still detectable numbers of CD158⁺ cells were found in the other naïve subsets (Figure 1f). Relatively high levels of CX3CR1, CD16, and NKp80 were described in stem-like memory (T_SCM, CD45RA⁺CD27⁺CCR7⁺CD28⁺CD95⁺) cells (Supplementary Figure Sa-c). Thus, N_CCR7, N_CD28, N_CD27 and N_TN show unique expression of differentiation markers.

The production of cytokines and effector molecules is associated with T cell differentiation. We therefore investigated whether the expression of TNF, IFNγ, and granzyme B (GrzB) may differ between the N_CCR7, N_CD28, N_CD27 and N_TN subsets (Figure 2). In both PMA/Ionomycin and anti-CD3 stimulated cells, TNF expression was highest in N_CCR7 cells, intermediate in N_CD28 and N_CD27 and lowest in the N_TN population (Figure 2a). Similar results were observed when IFNγ, and GrzB levels were assessed (Figures 2b,c). Higher expression of TNF, IFNγ, and GrzB was present in the T_SCM subpopulation (Supplementary Figures S3d-f). Taken together N_CCR7, N_CD28, N_CD27 and N_TN CD8⁺ T cells are different regarding their expression of differentiation markers and T cell cytokines, with N_CD27 being more like N_TN cells.

We next measured the expression of parameters related to mitochondrial functionality, oxidative stress, and senescence in N_CCR7, N_CD28, N_CD27 and N_TN CD8⁺ T cells (Figure 3). Total mitochondrial content, assessed using mitotracker DR, was highest in N_CD28 and progressively decreased in N_CCR7, N_CD27 and N_TN subsets (Figure 3a). Mitochondrial membrane potential, a key indicator of mitochondrial health and activity, was evaluated using JC-1 intensity, normalised on mitochondrial content. While N_CD28 cells exhibited the lowest JC-1 intensity, it was progressively higher in the N_CCR7, N_CD27 and N_TN subsets (Figure 3b). The highest JC-1 values were found in N_TN cells. Both mitochondrial and intracellular ROS were elevated in N_CD27 cells, intermediate in N_CCR7 and N_CD28, and lowest in the N_TN population (Figures 3c,d). Similarly, SA-β-gal activity was higher in the N_CCR7, decreased in N_CD28 and N_CD27 and significantly lower in N_TN cells (Figure 3d). Elevated SA-β-gal activity was present in T_SCM cells (Supplementary Figure S3g). In parallel, the senescence marker p21 was expressed at comparable levels in the N_CCR7, N_CD28, N_CD27 subpopulations, but it was reduced in the N_TN subset (Figure 3e). In summary, the four naïve subsets are unique regarding the functionality of mitochondria and markers of oxidative stress and senescence.

We next assessed whether naive CD8⁺ T cells may respond differently to DNA damage (Figure 4). To achieve this aim, PBMCs were stimulated for 1 day with etoposide, and markers of apoptosis were assessed within N_CCR7, N_CD28, N_CD27 and N_TN CD8⁺ T cells (Figure 4). We first measured the frequency of early and late apoptotic cells using the combination of Apotracker and DAPI, and therefore Apotracker (Apo)⁺ DAPI⁻ and Apo⁺ DAPI⁺ populations were defined (Figure 4a). The frequency of both early and late apoptotic cells was the highest in N_CCR7, intermediate in N_CD28 and N_CD27 and the lowest in N_TN CD8⁺ T cells, indicating that N_TN may be more resistant to DNA damage-induced apoptosis. Similarly, the levels of cleaved caspase 3 decreased in N_TN cells, although no statistically significant differences were found between N_CCR7, N_CD28, and N_CD27 cells (Figure 4b). γH2AX levels, indicating the presence of DNA double-strand breaks (Podhorecka et al., 2010), were measured in the four subsets, and N_CCR7 and N_TN cells showed the highest and the lowest γH2AX levels respectively. We next measured proliferation of the subpopulations of interest within PBMCs after stimulation with anti-CD3 Ab or PHA for 4 days (Figure 5). Both PI and frequency of proliferated cells were again the lowest in N_CCR7, intermediate in the N_CD28 and N_CD27 subsets, while N_TN showed the highest proliferation rate in comparison to the other populations. Results were similar for the stimulation with anti-CD3 and PHA.

Taken together, although the response to DNA damage and proliferative capacities differ across the subsets, the N_CD27 subset appears most closely related to N_TN CD8⁺ T cells.

We then used an unsupervised hierarchical clustering approach to delineate discrete patterns of markers across the entire dataset including the data shown in Figures 1–4 (Figure 6a). In this way, we identified two theoretical clusters of cells (Figures 6a,b). By analysing the relative weight of each marker onto the clustering model, we observed that cluster one showed significantly lower expression of the differentiation markers CX3CR1, CD16, CD56 and NKp80, as well as decreased TNF, IFNγ, SA-ß-gal, and apoptosis. Conversely, cluster two showed an increased level of the same markers. Interestingly, significant differences were additionally driven by apoptosis markers (Figure 6a). Next, cells from the four original subpopulations were identified within the theoretical groups obtained using the clustering algorithm (Figure 6c). Our data show that N_CD27 and N_TN were mostly assigned to cluster 1, whereas N_CCR7 and N_CD28 were almost exclusively represented within cluster 2 (Figure 6c). This data was confirmed using the correspondence analysis, showing N_CD27 and N_TN and cluster one in the opposite quadrant as N_CCR7 and N_CCR8 and cluster 2 (Figure 6d). Altogether these findings show a remarkable similarity in the phenotype of N_CD27 and N_TN, as well as a similar scenario for N_CCR7 and N_CD28 subpopulations.

We next investigated whether the expression of all markers measured in the four naive CD8⁺ T cells may change with aging (Figures 7, 8; Supplementary Figure S4). Overall, the expression of differentiation markers and pro-inflammatory/effector molecules increased with age within all subsets, although most significant results were seen for N_CD28 and N_CD27 cells (Figure 7). Furthermore, mitochondrial membrane potential was reduced in all subpopulations with age while ROS levels increased. Importantly, in all subsets, highly significant correlations were observed between SA-ß-gal expression and age. Furthermore, the proliferation potential decreased with aging, with the most prominent differences observed for N_CCR7, N_CD28, and N_CD27 cells. Finally, reduced apoptosis was described, although no correlations between γH2AX levels and aging were identified. We next divided the donors into three age groups: young (≤35 years), middle-aged (36–69 years) and old (≥70 years) and investigated the expression of CX3CR1, CD94, Tbet, intracellular ROS levels and SA-ß-gal in N_CCR7, N_CD28, N_CD27 and N_TN CD8⁺ T cell subsets (Figure 8). The most significant differences were observed between the young and old groups within the N_CCR7, N_CD28 and N_CD27 subsets, while the middle-aged donors generally resembled the younger group. Although certain trends were present, no significant differences were detected in the N_TN subset.

In summary, impaired phenotype and functionality (including the acquisition of an activated phenotype and loss of quiescence) were observed during aging in naive CD8⁺ T cells. Although less pronounced, some age-related changes were also identified in the N_TN subset.

Although a small proportion (∼10%) of CX3CR1⁺, CD16⁺, or NKp80⁺ cells was detected within the N_CCR7 and N_CD28 subsets, this observation requires further investigation. While our experimental data support the presence of these cells, additional validation is needed to determine whether this reflects a true biological phenomenon or subtle technical limitations related to donor variability or gating parameters. Furthermore, the lack of alternative gating strategies to define naïve T cells (such as CCR7/Fas-based approaches) represents a limitation of the current study.

This study relies primarily on phenotypic analyses using well-established surface markers to define naïve CD8⁺ T cell subsets. Despite comprehensive characterization, these markers may not fully capture the underlying heterogeneity or functional diversity within the naïve compartment. Moreover, potential contamination with partially differentiated or non-naïve cells in some subsets highlights the limitation of surface marker-based definitions. While our hierarchical clustering provided additional resolution, deeper insights could be obtained through high-dimensional approaches such as single-cell RNA sequencing or epigenetic profiling, which were beyond the scope of this study.