Sara van den Berg

149 6 T-cell dynamics in CMV 2D ). The expression of Bcl-2 was significantly higher for CMV-specific CD8 + T CM and T EM cells compared to bulk CD8 + T CM and T EM cells of CMV+ individuals ( Figure 2D ). Finally, we performed a PC analysis on these markers for CMV-specific CD8 + T-cells and the different memory CD8 + T-cell subsets (explaining 72.5% of variance with two PCs). We confirmed that CMV-specific CD8 + T-cells mostly resemble bulk CD8 + T EMRA cells and, to a lesser extent, bulk CD8 + T EM cells ( Figure 2E ). Ten individuals selected for quantification of T-cell dynamics using heavy water labelling The previously described markers gave a snapshot of ongoing proliferation and apoptosis resistance of different T-cell subsets. However, to address the question whether T-cells of CMV- and CMV+ individuals differ in their production and loss rates, and if CMV-specific CD8 + T-cells accumulate due to accumulation of long-lived cells, we quantified the in vivo production and loss rates of different T-cell populations using heavy water ( 2 H 2 O) labelling. The principle of 2 H 2 O labelling is that the 2 H-atoms present in the body are incorporated via de novo DNA synthesis by cells undergoing cell division and lost when cells die, differentiate, or migrate to another body compartment [25]. Ten individuals were selected to participate in such a 2 H 2 O labelling study ( Figure 3A ). We hypothesized that any potential differences in T-cell production and loss rates between CMV-specific CD8 + T-cells and bulk memory CD8 + T-cells in CMV+ and CMV- individuals would be most evident in individuals with the highest CMV-specific CD8 + T-cell responses. We therefore selected the five CMV+ participants with the highest absolute numbers of CMV-specific CD8 + T-cells ( Supplementary Figure 3 ). Next, five CMV- participants were selected based on matching for age and sex to the CMV+ group ( Table 1 ). Percentages of memory and naive T-cells in these selected ten individuals were similar to those in the total group of participants ( Supplementary Figure 4A,B ). Based on the longitudinal data of each sorted T-cell subset per individual during the 5 week up- and on average 59 week (range 49-70) down-labelling phase of the 2 H 2 O labelling study, the average production rate of each T-cell subset was estimated by fitting a mathematical model to the deuterium enrichment data ( Supplementary Figure 5,6 ). T EM and T EMRA cells were sorted together as T EM/EMRA for the assessment of production rate, to enable a fair comparison with CMV-specific T-cells. In earlier deuterium labelling studies, we used quite broad definitions to separate memory (CD45RO + ) and naïve (CD45RO - CD27 + ) T-cell subsets. Here, we refined our sorting strategy to determine the production rates of different CD4 + and CD8 + T-cell subsets in all older adults. We sorted truly naive (T TN ) cells (CD45RO - CD27 + CCR7 hi CD95 - ) to exclude contamination with CD95 + T-cells, which are known to be highly proliferating [22] and include stem cell memory T-cells (T SCM ) [34]. For total memory and naïve T-cell populations, we assumed that the number of cells in the population remained constant over time, and thus that production rates equal loss rates. The median T TN production rate was 0.00065 per day (23.9% per year) for CD4 + T TN cells and 0.00067 per day (24.3% per year) for CD8 + T TN cells ( Figure 3C and Table 2 ), indicating that naive T-cells have a life expectancy of approximately 4 years. The