Multiparameter flow cytometric analysis of CD4 and CD8 T cell subsets in young and old people
© Koch et al; licensee BioMed Central Ltd. 2008
Received: 20 May 2008
Accepted: 25 July 2008
Published: 25 July 2008
Skip to main content
© Koch et al; licensee BioMed Central Ltd. 2008
Received: 20 May 2008
Accepted: 25 July 2008
Published: 25 July 2008
T cell-mediated immunity in elderly people is compromised in ways reflected in the composition of the peripheral T cell pool. The advent of polychromatic flow cytometry has made analysis of cell subsets feasible in unprecedented detail.
Here we document shifts in subset distribution within naïve (N), central memory (CM) and effector memory (EM) cells defined by CD45RA and CCR7 expression in the elderly, additionally using the costimulatory receptors CD27 and CD28, as well as the coinhibitory receptors CD57 and KLRG-1, to further dissect these. Although differences between young and old were more marked in CD8 than in CD4 cells, a similar overall pattern prevailed in both. Thus, the use of all these markers together, and inclusion of assays of proliferation and cytokine secretion, may enable the construction of a differentiation scheme applicable to CD4 as well as CD8 cells, with the model (based on Romero et al.) suggesting the progression N→CM→EM1→EM2→pE1→pE2→EM4→EM3→E end-stage non-proliferative effector cells.
Overall, the results suggest that both differences in subset distribution and differences between subsets are responsible for age-related changes in CD8 cells but that differences within rather than between subsets are more prominent for CD4 cells.
Numerous studies have established that many parameters of immunity are decreased in elderly people and suggest that these are likely to contribute to their increased susceptibility to infectious disease and poor responses to vaccination [1–3]. In particular, the ability to control disease caused by novel pathogens is greatly compromised; responses to previously-encountered pathogens are, however, also eventually eroded in the very elderly . These findings could be explained in at least two mutually non-exclusive ways: 1) that each T cell from an elderly donor is in some way compromised in its function, or 2) the proportions of the different T cell subsets differ between young and old people, but the function of each cell type is the same regardless of donor age. There is evidence for both views in that single T cells from the elderly may, for example, show apparent defects in signal transduction and activation . However, most earlier studies examined mixed cell populations and apparent differences could have been due to different proportions of cells in the test populations. Studies with monoclonal populations have indicated age-associated changes at the single cell level , but these were associated with culture not chronological age and their relevance to the in vivo situation remains open.
Thus, the model of one subset of N, one of CM, and several of EM and TEMRA cells as defined by CD27 and CD28 expression does seem to hold up for CD8 cells from the young, as previously proposed . However, the situation is not the same for CD8 cells from the elderly. The proportion of CD27+CD28+ cells within the naïve subset is significantly lower than in the young and the fraction of CD27-CD28+ cells significantly greater (Figure 3A). The proportion of CD27-CD28+ but not CD27+CD28- naïve cells is also significantly greater in the elderly than the young. For CM cells, differences between old and young reach significance only for the decrease in CD27+CD28+ cells. EM cells have decreased CD27+CD28+ and increased CD27-CD28- double negative cells in the elderly, with no significant differences between either the CD27-CD28+ or CD27+CD28- subset. Within the TEMRA cells, there are no significant differences between young and old in any subset, consistent with most TEMRA cells being as fully differentiated as they can be, regardless of the chronological age of the donor.
The model applied to CD8 cells above is not well established for the CD4 subsets. Age-associated changes in N, CM, EM and TEMRA subsets are also less well-documented for CD4 than CD8 cells. Data shown in Figure 2 had confirmed a decrease in naïve cells in the very elderly and increased proportions of CM cells. It remained possible that differences between young and old within each of these populations might be more marked. However, data shown in Figure 3B suggest that this is unlikely to be the case. There were no significant differences between N, CM or TEMRA cells in young and old, although here the EM cells of the elderly did show significant reductions in CD27+CD28+ cells and presumably reciprocal increases in CD27-CD28- cells.
The final questions raised here were whether subdivisions of CD4 cells could be made in the same way as for CD8 cells, and whether these might reveal differences between the young and the elderly. Figures 4B and 4D show analogous data to Figures 4A and 4C but for CD4 rather than CD8 cells. The overall trend seems comparable: naïve cells contain very few but some KLRG1+CD57- cells but most cells are KLRG1-CD57-, whereas CM cells have slightly larger proportions of CD57-KLRG1+ but no CD57+ KLRG1- cells. Unlike CD8 cells, the CD4+ CM cells did show significant differences between young and old donors for all except the CD57+KLRG-1- cells. Data shown in Figure 5B in general support the same scheme for CD4 cells as applied to CD8 cells (compare Figure 5A for CD8). In this case there is a statistically significant difference between old and young EM1 cells, but unexpectedly with fewer CD57-KLRG1+ and more CD57-KLRG1- cells in the elderly than in the young. No other differences are statistically significant. Finally, Figure 6B shows data from CD4 TEMRA cells. Here, the differences between pE1/2 cells and the CD27-CD28- TEMRA cells are even more striking than seen for CD8 cells (compare Figure 6A). Thus, although neither pE1 nor pE2 cells contain CD57+KLRG1+ cells, the majority of cells in this most highly differentiated subset express both markers.
Differentiation stages of CD8 T cells can be ordered according to the expression of cell surface proteins, with a popular model describing naïve cells as CD45RA+ CCR7+ and memory cells as either CD45RA- CCR7+ (CM), CD45RA- CCR7- (EM) or CD45RA+ CCR7- (TEMRA). However, an absolute correlation between differentiation state and expression of these two surface molecules cannot be expected in such dynamic systems. Combinations of markers, however, may improve the ability to dissect out different subsets. Thus, naïve CD8 cells express CD27 and CD28, have long telomeres and extensive proliferative capacity . The use of these markers may also not be sufficient, but the use of polychromatic flow cytometry for the first time allows constellations of markers to be used together for a finer definition of subsets at the single cell level. Here, we have explored the utility of combining the above markers with two others thought to be expressed by late-differentiated T cells, namely, CD57 and KLRG-1. All these markers together can be used to define subsets of memory cells with different properties, which we apply here not only for CD8 cells, but also for CD4 cells, which have not been extensively investigated previously . Here, we have shown that subdivisions of EM and TEMRA cells on the basis of CD27 and CD28 expression can be further dissected by their expression of CD57 and KLRG-1, and moreover, that with the exception of CD57 expression, the same subsets can be defined in CD4 as well as CD8 cells. Finally, we have sought differences between young and old people within each of these newly-defined subsets.
We purposefully selected a heterogeneous cohort of young and old donors over a wide age range and from different European countries, so that any statistically significant differences found are likely to be robustly generally applicable, independently of genetic or environmental background. This is likely to be one reason for the large inter-individual variation observed in many of the tested parameters. Thus, it is all the more striking that the expected [22, 23] age-associated reduction in the naïve cell population is clearly seen in CD8+ cells, as well as the increase in TEMRA cells (Figure 2). An age-associated reduction in naïve CD4+ cells was also observed, although this did not reach significance, and was accompanied by an increase in CM not TEMRA cells (Figure 2).
The expression of the costimulatory receptors CD27 and CD28 may be used to subdivide CD8 cells , demonstrating 4 populations of CD45RA- CCR7- EM cells and 3 of TEMRA (a CD27-CD28+ subset not being found in the latter). In general, in CD8 cells from young donors, our data confirm the presence of these subpopulations, although we found very few CD27-CD28+ or CD27+CD28- cells within either the EM or TEMRA subsets. Essentially the same pattern was found in cells from the elderly too, but interestingly within the EM but not within the TEMRA populations, there were significant differences between young and old for both CD27+CD28+ and CD27-CD28- cells (Figure 3A). This finding is consistent with the idea that subsets of TEMRA cells represent end-stage differentiated cells regardless of the age of the host, and which by definition cannot differentiate further.
Within the CD45RA+CCR7+ naïve CD8 cells, the elderly showed a reduction in CD27+CD28+ cells, an unexpected finding which could be explained by homeostatic proliferation of these cells with retention of naïve markers but loss of costimulatory receptors in the elderly. This would be likely to contribute to the poorer responses of naïve cells in old individuals, which has been demonstrated in TCR transgenic animal models, at least for CD4 cells . Alternatively, these cells may not be "fully" mature but may in fact have re-acquired CCR7 as well as CD45RA, thus appearing naive. There is a precedent in the literature for such an event . However, the picture with CD4 naïve cells was markedly different, with essentially all of them remaining CD27+CD28+, regardless of the age of the donor (Figure 3B). These data are consistent with the maintenance of a higher degree of both qualitative and quantitative integrity of the CD4 subset naïve potential than the CD8 with increasing donor age. However, as with CD8 cells, the CD4 subset also showed an increase of the CD27-CD28- EM3 cells at the expense of the CD27+CD28+ EM1 cells. This might also compromise the ability of previously activated effector memory cells to be reactivated, due to the lack of two major costimulatory receptors.
CD57 is a negative NK receptor also expressed on CD8 but not on CD4 cells, whereas KLRG-1 is a negative NK receptor expressed by both CD4 and CD8 cells . Both have been dubbed "senescence" markers, and their absence from "fully" naïve and CM cells from the old as well as the young (Figure 4) is consistent with this. Both markers together identify the latest state of differentiation within the respective subsets. On CD8 cells, all 4 possible phenotypes are present (Figure 5A), whereas CD57 is not expressed on CD4 cells, not even from old donors, with two exceptions: EM3 cells are the only EM subset expressing any CD57, again independent of the age of the donor, but only by a minority of this subset and only by those cells already expressing KLRG-1 (Figure 5B). This suggests that a very small proportion of effector memory CD4 cells can indeed express CD57, and that EM3 cells are the most differentiated subset thereof, as with their CD8 counterparts. The other CD4 subset which expresses CD57 is the most differentiated of all: the CD27-CD28- TEMRA subset, again only on cells which are KLRG-1+ as well (Figure 6B).
By labelling whole PBMC with the stable membrane dye CFSE, cell division after stimulation can be tracked over 4–6 or more divisions. Applying this technique to CD4 and CD8 cells from young individuals assessed by the same surface markers as described above, and assuming that the more differentiated a cell is the less it can proliferate, we are in the process of accumulating data which are consistent with the model derived from the differential expression of CD45RA, CCR7, CD27, CD28, CD57 and KLRG-1 (data not shown). Reciprocally, at least for CD8 cells, the accumulation of perforin and Granzyme A is also consistent with the same differentiation scheme (data not shown). This scheme supports the notion that T cells differentiate sequentially through the stages N→CM→EM1→EM2→pE1→pE2→EM4→EM3→E and that the major but not the only differences, between cells from young and old donors reside in the relative proportions of these different subsets present. Nonetheless, using CD27 and CD28, age-associated differences primarily within CD8 but not CD4 subsets can be discerned, whereas using CD57 and KLRG-1 reveals such differences predominantly within the CD4 but not the CD8 subsets.
Polychromatic flow cytometry is proving to be a powerful tool for examining changes in immune parameters in the elderly. The data provided here sought "lowest common denominator" changes in the distribution of both CD4 and CD8 subsets in different European populations over a range of ages from early middle aged to very old. By combining analyses using antibodies to CD27, CD28, CCR7, CD45RA, CD57 and KLRG-1, the results presented here confirm the robustness of the decrease in naïve and increase of late-differentiated CD8 cells with age, with a similar tendency in CD4 cells, in different populations with dissimilar genetic, nutritional and pathogen-exposure backgrounds. Combined with functional assays, this approach will facilitate the analysis of age-associated immune alterations in humans in unprecedented detail.
In order to encompass a broad (although exclusively Caucasian) population, cryopreserved PBMC were collected from several different European countries with a mean age of 40 (53% female) or 87 (66% female). Because this study was aimed at determining the most robust and reproducible differences between old and young donors, they were not rigorously selected for health status, nutrition, infection etc. They were from Italy, Sweden, Bulgaria and Germany, thus encompassing multiple genetic and environmental backgrounds. These donors were overtly healthy but were not SENIEUR-compliant, in order to study a more representative group of elderly. This approach obviously makes it more difficult to detect significant differences within such a heterogeneous group, but it is our belief that when significant differences do emerge in such a study, they are likely to be of more basic import than those seen only in highly selected subgroups. Thus, although infection with herpes viruses, especially CMV, is known to alter certain immune parameters, even this was also not taken into account in this study. Hence, we argue that any differences observed in this very heterogeneous group of subjects are more likely to reflect basic age-related alterations than influences of genetics, nutrition, infection etc.
Direct immunofluorescence was performed with pre-titrated anti-CD4-PacificBlue, CD28-AlexaFluor700 (BioLegend, Biozol, Eching, Germany), CD8-APC-Cy7, CD27-APC CCR7-PE-Cy7 (Becton Dickinson, Heidelberg, Germany) and CD45RA-PE-Cy5.5 (Invitrogen, Karlsruhe, Germany). CD57 was from Immunotools, Friesoythe, Germany. For indirect immunofluorescence, anti-human KLRG1 (clone 13A2), kindly provided by Prof. H.P. Pircher, Freiburg, was used as primary antibody. As secondary antibody, Pacific Orange-goat anti-mouse IgG (Invitrogen) was used. For blocking, human immunoglobulin GAMUNEX (Bayer, Leverkusen, Germany) or mouse serum (Caltag/Invitrogen, Karlsruhe, Germany) were used. The cell viability was determined with Ethidium monoazide (EMA) (Invitrogen). All staining steps were performed in PFEA buffer (PBS, 2% FCS, 2 mM EDTA and 0.01% Azide).
For each experiment, cells or mouse or hamster/rat κ-chain Comp Beads (Becton Dickinson) were stained with corresponding fluorochrome-labelled antibodies and incubated for 20 min at 4°C in the dark. As unstained negative controls we used negative Comp Beads (Becton Dickinson). After washing with PFEA, the cells or beads were resuspended in 200 μl PFEA and measured using an LSR-II flow cytometer and the acquisition software BD FACSDiva (Becton Dickinson). The spectral overlap between all channels was calculated automatically by the BD FACSDiva software, after measuring negative and single-colour controls. For additional data analysis, Flowjo 7.2.2 (Treestar Inc, San Carlos, CA) was used. Careful control experiments indicated that cryopreservation and thawing of the PBMC did not alter the pattern of staining observed for the major subsets studied here (Larbi et al., manuscript in preparation).
All statistical analysis were performed with Graphpad Prism 4.03 and 5.0. Non-parametric Mann-Whitney U test was used for comparison of two independent groups.
This work was supported by the European Commission (QLK6-02283, "T-CIA"; LSHG-CT2007-036894, "LifeSpan") and the Deutsche Forschungsgemeinschaft (Pa 361/11-1 and SFB 685-B4). We thank Lilly Weddel, Karin Haehnel and Arnika Rehbein for technical assistance and Prof. D. Wernet for making buffy coats and leukopheresates available from the Tübingen Blood Bank. We thank Profs. A. Wikby (Jönköping, Sweden) and E. Mariani (Bologna, Italy) for providing samples in the context of EU projects. We are very grateful to Prof. H. P. Pircher, Freiburg, for the generous gift of KLRG-1 antibody.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.