Samples were obtained from systemically healthy individuals who did not receive immunomodulatory drugs or suffer from diseases known to influence the immune system, such as autoimmune diseases and cancer. None of them was frail or had symptoms of cognitive impairments. In all patients, the indication for surgery was osteorarthrosis. Further information about the donors included in the study is summarized in Table 1.

Hip replacement surgery was performed and bone from the femur shaft was harvested. A biopsy of substantia spongiosa osseum, which would otherwise have been discarded, was used to isolate BM mononuclear cells (BMMCs). BM biopsies were fragmented, washed once with complete RPMI medium (RPMI 1640 supplemented with 10% FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin; Invitrogen) and treated with purified collagenase (CLSPA, Worthington Biochemical; 20 U/ml in complete RPMI medium) for 1 h at 37°C. BM biopsies were then centrifuged and BMMCs purified by density gradient centrifugation (Ficoll-Hypaque). Purification of PB mononuclear cells (PBMCs) from heparinized blood was also performed by density gradient centrifugation.

RNA was isolated from purified BMMCs using the RNeasy Plus mini kit (Qiagen). First-strand cDNA synthesis was performed using a Reverse Transcription system (Promega). Quantitative RT-PCR experiments were performed using the LightCycler 480 System (Roche Diagnostics), 2X SYBR Green 1 Master (Roche Diagnostics), and β-actin as housekeeping gene for relative quantification of effector/memory cell survival factors. Sequence-specific oligonucleotide primers were designed using Primer3 software (25) and synthesized by MWG Biotech (Ebersberg, Germany). The following primers were used: IL-15FW 5′-ATTTTGGGCTGTTTCAGTGC-3′, IL-15RW 5′-TTACTTTGCAACTGGGGTGA-3′, IL-7FW 5′-GTAGCAATTGCCTGAATAATG-3′, IL-7RW 5′-GTTGTGCCTTCTGAAACT-3′.

Immunofluorescence surface staining was performed by adding a panel of directly conjugated Abs to BMMCs. After surface staining, cells were permeabilized using the Cytofix/Cytoperm kit (BD Pharmingen), and incubated with intracellular Abs. Labeled cells were measured using a FACSCanto II (BD Biosciences). Data were analyzed using Flowjo software. The following labeled Abs were used: IL-7Rα-PE (hIL-7Rα-M21), CD8-PeCy7 (RPA-T8), CCR7-FITC (150503), and CD28 BV421 (CD28.2) purchased from BD, CD122 (IL-2/IL-15Rβ)-APC (TU-27), CD45RA-PerCp (HI100), and KLRG-1-PeCy7 (2F1/KLRG1) purchased from Biolegend, CD3-APC-eFluor 780 (SK7) purchased from eBioscience.

The responsiveness of CD8⁺ T cell subsets to IL-15 and IL-7 was measured by quantifying the cells expressing the receptors CD122 (IL-2/IL-15Rβ) and IL-7Rα, respectively (26–28).

Immunofluorescence analysis of BM sections was performed, as described by Herndler-Brandstetter et al. (8). Formalin-fixed, paraffin-embedded 4-µm BM sections were deparaffinized in xylene and re-hydratated in ethanol. The slides were boiled in 0.01 M citrate buffer (pH 6) for 16 min in the microwave for epitope retrieval and allowed to cool for about 1 h at room temperature. Slides were blocked with 3% skim milk in TBS/Tween for 20 min at room temperature. Rabbit anti-IL-15 (1:200; ab55276; Abcam) and mouse anti-CD8 (1:50; C8/144B; Dako) Abs were incubated overnight at 4°C. After washing, the slides were incubated for 1 h at 4°C with biotinylated swine anti-rabbit Ab (1:300; E0431; Dako) and a goat anti-mouse Alexa Fluor 546 Ab (1:300; A11018; Molecular Probes). Following washing steps with TBS/0.1% Tween, the BM sections were stained with a streptavidine-Alexa Fluor 488 Ab (1:500; S11223; Molecular Probes) for 30 min at 4°C. The stained slides were analyzed using confocal microscopy with an m-Radiance confocal scanning system (Bio-Rad) attached to a Zeiss Axiophot microscope (Carl Zeiss). For the quantitative analysis of CD8⁺ T cells in close contact with IL-15–producing cells in the BM, pictures from different areas of the BM sections were analyzed. In total, 800–1,000 CD8⁺ T cells were analyzed from each donor to calculate the percentage of contact with IL-15⁺ cells.

Antibodies against CMV were determined in the serum of the donors included in the study using a commercially available ELISA Kit (Siemens).

The data obtained in the study follow a non-parametric distribution. Statistical significance was assessed by Spearman correlation analysis, Mann–Whitney test and Wilcoxon matched pairs test. A p-value of less than 0.05 was considered as significant.

The study was approved by the Ethics Committees of the “Klinikum Wels-Grieskirchen” (Austria). Written informed consent was received from participants prior to their inclusion in the study.

CD8⁺ T cells with a CD45RA⁺ CCR7⁻ T_EMRA phenotype have been shown to accumulate in the blood after CMV infection (29, 30). To assess whether T_EMRA cells are also enriched in the BM from CMV⁺ persons, we measured the levels of BM CD8⁺ T_EMRA cells in CMV⁻ and CMV⁺ persons (Figure 1). CD8⁺ T_EMRA cells were gated, as indicated in Figure 1A. Within the CD8⁺ CCR7⁻ T cell population, a subpopulation with an intermediate (TEMRAdim) and one with a bright (TEMRAbright) expression of CD45RA and a subset, which does not express CD45RA (T_EM) were defined. Higher percentages of both TEMRAdim and TEMRAbright cells were found in BMMCs compared to PBMCs (TEMRAdim p = 0.001; TEMRAbright p = 0.04, data not shown). While the size of the CD8⁺ T_EM and the CD8⁺ TEMRAdim populations in the BM was similar in CMV⁻ and CMV⁺ persons, the percentage of CD8⁺ TEMRAbright cells was higher in CMV⁺ persons (Figure 1B). No differences were observed in the numbers of CD8⁺ T_EM, TEMRAdim, and TEMRAbright cells when we compared younger (≤70 years) and older (>70 years) donors in both the CMV⁻ and the CMV⁺ group (n.s., data not shown).

CD8⁺ T_EMRA cells have been described to downregulate CD28 and to express senescence markers in the PB (31). CD8⁺ CD28⁻ T cells have also been observed to accumulate in the BM in old age (12, 32). To assess whether CMV affects the phenotype of CD8⁺ subsets in the BM, we analyzed CD8⁺ CD28⁻ T cells in the BM of CMV⁻ and CMV⁺ persons (Figure 1C). Increased frequencies of CD8⁺ CD28⁻ TEMRAbright cells were found in CMV⁺ compared with CMV⁻ persons, while no differences between the two groups were observed in CD8⁺ T_EM and CD8⁺ TEMRAdim cells. While CD8⁺ CD28⁻ TEMRAbright T cells correlated positively with age in CMV⁻ persons, no age-related changes were observed for CD28⁻ TEMRAbright cells in CMV⁺ persons and for T_EM and TEMRAdim cells in both the CMV⁺ and the CMV⁻ group (Figure S1 in Supplementary Material).

We then analyzed the expression of the senescence marker KLRG-1 (33) in BM CD8⁺ T cell subsets and compared the results in CMV⁺ and CMV⁻ persons (Figure 1D). The percentage of KLRG-1-expressing cells was relatively low in T_EM, high in TEMRAdim, and was even higher in TEMRAbright cells. No differences were observed when the CMV serostatus was considered or when younger (≤70 years) and older (>70 years) donors were compared (data not shown). These data suggest that CD8⁺ T_EMRA cells with a high expression of CD45RA, which frequently lack CD28 and express KLRG-1, increase in the BM of CMV⁺ persons.

IL-7 and IL-15 influence the survival and turnover of CD8⁺ T cells (34). They are of particular relevance for the interaction of T cells with stromal cells in the BM (8, 35). We, therefore, decided to study the expression of CD122 and IL-7Rα in CD8⁺ T cells. Specifically, we compared these parameters in the T_EM, TEMRAdim, and TEMRAbright subsets from CMV⁻ and CMV⁺ persons (Figures 2 and 3). The “typical” phenotype of a CD8⁺ TEMRAbright cell in CMV⁻ and CMV⁺ donors is shown in Figure 2. In the whole cohort (CMV⁺ plus CMV⁻ persons), CD122 was expressed on a great majority of cells, in all subsets, but on almost every cell in the CD8⁺ TEMRAbright population (Wilcoxon matched pairs test, p < 0.001 TEMRAbright vs. TEMRAdim, Figure 3A). No differences were found between CMV⁻ and CMV⁺ persons. Interestingly, when relating the expression of CD122 with age, in CD8⁺ T cell subsets from CMV⁻ and CMV⁺ persons, we found positive correlations in CMV⁺ persons, whereas negative correlations were seen among T_EM and TEMRAdim subsets (Figures S2A,B in Supplementary Material). No significant correlations with age were found in CD8⁺ TEMRAbright cells (Figure S2C in Supplementary Material). Lower percentages of IL-7Rα⁺ cells were observed in both CD8⁺ TEMRAdim and TEMRAbright cells from CMV⁺ persons when compared to their CMV⁻ counterparts (Figure 3B). In CMV⁻ persons, there was a positive correlation between IL-7Rα and age in CD8⁺ T_EM cells, while negative correlations were seen between CD8⁺ TEMRAdim and TEMRAbright cells in CMV⁺ persons (Figures S2D–F in Supplementary Material).

We then compared the expression of CD122 and IL-7Rα in CD8⁺ CD28⁺ and CD8⁺ CD28⁻ TEMRAbright cells (Figures 3C,D). Higher percentages of CD122⁺ cells were found in CD8⁺ CD28⁻ TEMRAbright cells compared to their CD8⁺ CD28⁺ counterparts (Wilcoxon matched pairs test, p = 0.01 in CMV⁻ and p = 0.002 in CMV⁺ persons; Figure 3C). No differences were seen between CMV⁻ and CMV⁺ persons. In contrast, the percentage of IL-7Rα⁺ cells was lower in CD8⁺ CD28⁻ TEMRAbright cells compared to CD8⁺ CD28⁺ TEMRAbright cells (Wilcoxon matched pairs test, p < 0.001 in both CMV⁻ and CMV⁺ persons; Figure 3D). When comparing CMV⁺ and CMV⁻ samples, reduced numbers of IL-7Rα⁺ cells were seen in CMV⁺ persons within the CD8⁺ CD28⁺ TEMRAbright subset; however, no differences were seen within the CD8⁺ CD28⁻ TEMRAbright population. In summary, our findings indicate that numbers of CD8⁺ TEMRAbright cells with high CD122 and/or low IL-7Rα expression increase in the BM of CMV⁺ persons. Thus, in CMV⁺ persons, the “typical” CD8⁺ TEMRAbright cell is most likely CD28⁻, CD122^hi IL-7Rα^low, and KLRG-1⁺; however, whether every cell in the TEMRAbright subset carries the full marker pattern is not yet known and is currently being investigated. In CMV⁻ persons, the phenotype of CD8⁺ TEMRAbright cells differs somewhat as these cells still express reasonably high levels of both CD28 and IL-7Rα.

Although it is not clear which role CMV-specific Abs play in latent CMV infection, high Ab titers are connected with CMV re-activation (36). We, therefore, determined whether CMV Ab titers correlated with the percentage of CD28⁺, CD122⁺, and IL-7Rα⁺ expression in BM CD8⁺ TEMRAbright cells. There was a positive correlation between Ab titers and the percentage of CD28⁻ cells in the CD8⁺ TEMRAbright population, but no relationship between Ab titers and CD122-expressing cells was observed (Figures 4A,B). In contrast, when the percentage of IL-7Rα⁺ cells in the CD8⁺ TEMRAbright population was assessed in relationship to CMV Ab titers, there was a negative correlation (Figure 4C). In addition, a ratio between the percentages of CD122⁺ cells and of IL-7Rα⁺ cells in the CD8⁺ TEMRAbright population was calculated (% CD122⁺ cells/% IL-7Rα⁺ cells). When this ratio was correlated with CMV Ab titers, a positive correlation was found (Figure 4D). Thus, our data indicate that CMV Ab titers in the serum correlate with the expression of CD28 and IL-7Rα as well as with the ratio CD122⁺/IL-7Rα⁺ in CD8⁺ TEMRAbright cells.

In order to assess whether the expression of effector/memory T cell survival factors in the BM differs in CMV⁺ persons, we measured the expression of IL-15 and IL-7 at the mRNA level in BMMCs from CMV⁻ and CMV⁺ persons of varying ages (Figure 5). Indeed, higher IL-15 mRNA levels were found when CMV⁺ persons were compared with their CMV⁻ counterparts (Figure 5A). IL-15 mRNA was 1.9 ± 0.1-fold higher in BMMCs from CMV⁺ compared to CMV⁻ persons. No difference in the expression of IL-7 mRNA was observed (Figure 5B). In a previous study, we demonstrated that IL-15 increased while IL-7 decreased during aging in the BM (12). We now confirm these data in CMV⁻ and CMV⁺ persons (Figures 5C,D). Again, a positive correlation for IL-15 and a negative one for IL-7 was found with age in both groups, CMV⁻ and CMV⁺ persons. Interestingly, CMV⁻ persons clustered differently from CMV⁺ ones when IL-15 mRNA expression was studied in correlation with age.

Interactions with IL-15-producing cells in the BM are required for the survival of effector/memory CD8⁺ T cells (7, 8). To assess whether the increased expression of IL-15 in CMV⁺ persons affects the number of interactions between IL-15-producing cells and CD8⁺ T cells, we studied BM sections and quantified CD8⁺ T cells in close proximity to IL-15⁺ BM cells in CMV⁻ and CMV⁺ persons (Figure 5E). 13.7 ± 2.7 and 27.1 ± 5.2% of CD8⁺ T cells, respectively, were interacting with IL-15-producing cells in CMV⁻ and CMV⁺ persons (Figure 5F). In summary, our findings indicate that IL-15 expression increases not only with aging but also with CMV infection, and is highest in old CMV⁺ persons. Additionally, the increased production of IL-15 in the BM may attract more CD8⁺ T cells into the close proximity of IL-15-producing cells.