In conclusion, age and CMV serostatus both contribute to the decrease in CD45RA+ CD27+ CD4+ T cells during ageing but the increase in CD45RA− CD27− and CD45RA+ CD27− T cells in old individuals is primarily the result of CMV infection. We next investigated whether the increase in CD45RA− CD27− and CD45RA+ CD27− CD4+ cells in CMV-seropositive donors only occurred within CMV-specific CD4+ T cells or also in those that are specific for different persistent viruses. To do this, we first identified virus-specific populations by intracellular IFN-γ staining after stimulation with lysates of virus-infected cells for

18 hr (see Supplementary Information, PD98059 mouse Fig. S1a).15 Background responses detected in unstimulated cells (negative control) were subtracted from those detected in stimulated samples. Only responses > 0·02% above background were considered positive. The IFN-γ secretion after stimulation with viral lysates was specific because no cytokine production was observed when CMV lysate was used to stimulate CD4+ T cells from CMV-seronegative donors as described previously.15 Y27632 We found that in CMV-seropositive donors, there was a significantly higher proportion of CMV-specific CD4+ T cells compared with T cells that were specific

for other persistent viruses such as VZV, HSV EBV or mycobacterial antigens (tuberculin PPD) (see Supplementary Information, Fig. S1b). We next investigated whether the increased proportion of CD45RA− CD27− and CD45RA+ CD27− CD4+ T cells in CMV-seropositive donors (Fig. 1c) was only the result of changes

within the CMV-specific T-cell population. We found that there were significantly more CD45RA− CD27− and CD45RA+ CD27− CD4+ T cells in CMV-seropositive donors compared with CMV-seronegative donors (Fig. 2a,b). However, although the majority of CD45RA− CD27− and CD45RA+ CD27− CD4+ T cells in CMV-seropositive donors were Ceramide glucosyltransferase CMV-specific, there was also a higher proportion of CD45RA− CD27− and CD45RA+ CD27− CD4+ T cells specific for the other viruses in CMV-seropositive subjects (Fig. 2b,c). Similar results were observed in both young and old donors (data not shown). This result reinforces the idea that CMV infection influences directly the composition of the CD4+ T-cell pools. Furthermore, our results indicate that CMV infection may have a global effect on driving the differentiation of other antigen-specific CD4+ T cells. This confirms our previous observations where the relative expression of CD28 and CD27 instead of CD45RA and CD27 was used to identify CD4+ T cells at different stages of differentiation.15 Several reports on CD8+ T cells suggest that the CD45RA+ CD27− subset is terminally differentiated17,22 with limited capacity for self-renewal.