• Open Access

Aging-related differences in basal heat shock protein 70 levels in lymphocytes are linked to altered frequencies of lymphocyte subsets


Prof. Dr Tony Mets, Gerontology and Geriatrics, Universitair Ziekenhuis Brussel, Vrije Universiteit Brussel, Laarbeeklaan 101, B-1090 Brussels, Belgium. Tel.: +32 2477 63 66; fax: +32 2477 63 64; e-mail: tmets@vub.ac.be


Cell stress responses are ubiquitous in all organisms and are characterized by the induced synthesis of heat shock proteins (Hsp). Previous studies as well as recent reports by our group have consistently suggested that aging leads to an increase in the basal levels of Hsp70. Here we extend these studies by examining the differential Hsp70 response of peripheral blood lymphocyte (PBL) subsets. It is well established that with aging, one of the major changes in the T cell pool is an expansion of T cells with the memory phenotype as well as those deficient for the CD28 molecule. To determine if alterations in the frequency of T cell subsets might be responsible for the observations, we have carried out a more comprehensive flow cytometric analysis of the various phenotypes of PBL under unstimulated conditions. Cells were obtained from 10 young and 10 elderly normal subjects. The basal Hsp70 levels in the various PBL phenotypes were comparable between young and elderly subjects. However, different patterns of Hsp70 response were noticed among the PBL subtypes, which were similar in both young and elderly subjects. In particular, the memory cell phenotypes produced more Hsp70 than the naïve phenotypes. These results suggest that aging-related changes in basal Hsp70 levels in PBL are linked to the altered frequency of lymphocyte subsets and not to increases in aged lymphocytes per se. In addition, the increase in Hsp70 can be interpreted as the result of a tendency towards more pronounced cellular differentiation in aging.


The heat shock response is a highly conserved, tightly regulated process, which is activated upon exposure to stressors including normal physiological stress, immune activation, heat, and other environmental challenges (Zou et al., 1998; Cotto & Morimoto, 1999; Jolly & Morimoto, 2000; Raz et al., 2001; Wallin et al., 2002; Njemini et al., 2003). This response to stress is characterized by the enhanced transcription of gene products encoding a family of proteins referred to as heat shock proteins (Hsp). An increasing array of investigations has revealed that the stress response is of great interest both in basic biology and in medicine. Biologically, the ability to survive and adapt to stress appears to be an essential defence mechanism for protection of cells, as cell stress responses are ubiquitous in all organisms. In medicine, evidence is mounting that the ability to survive and adapt to severe systemic physiological stress is critically dependent on the ability of cells to mount an appropriate compensatory stress response (Sonna et al., 2002; Heys et al., 2007). Moreover, the observation that induction of a stress response in one type of cell might benefit other cells deficient for the stress response (Guzhova et al., 2001) has raised the interesting possibility that cells involved in the stress response might represent useful targets for therapeutic manipulation.

Although several Hsp are involved in a variety of specific physiological processes, their general function as molecular chaperones in regulating protein folding is of particular importance in the context of cytoprotection. In the aging process, where damage to proteins is reported to be high, Hsp-mediated stress protection could be of special interest (Rattan, 1996; Berlett & Stadtman, 1997). During aging, a decreased production of Hsp upon stress has been described for various cell types, in vivo as well as in vitro (Liu et al., 1989; Wu et al., 1993). We have also provided evidence for an age-related decrease in the heat shock-induced production of Hsp70 in peripheral blood mononuclear cells (PBMC) in normal individuals (Njemini et al., 2002). Although the stress-induced levels of Hsp tend to decrease with age, the basal levels of several Hsp, including Hsp22, Hsp32, and Hsp70, have been reported to increase in association with aging (Maiello et al., 1998; King & Tower, 1999; Unno et al., 2000; Hirose et al., 2003). Recently, we have portrayed results indicative of an aging-related increase in the basal levels of Hsp32, Hsp70 and Hsp90 in PBMC (Njemini et al., 2005; 2007). These increases are more modest than those induced by heat shock, but, since aged organisms are continuously exposed to these elevated Hsp levels, the impact of this phenomenon might be more important than the decreased capacity to produce very high levels of Hsp after heat shock, which is an extremely rare event in humans. The origin of this aging-related increase in basal Hsp levels has largely remained unclear. Other mechanisms than a generalized reaction to damage might be responsible for the aging-related increase in basal Hsp levels.

T cells, which represent about 80% of human peripheral blood lymphocytes (PBL), can be divided into those that have yet to undergo specific antigen-dependent activation (naïve T cells, CD45RA+) and those that have been previously activated and differentiated (memory T cells, CD45RO+) (Akbar et al., 1988; Gray, 1993; Sprent, 1997; Young et al., 1997). During aging, studies have consistently demonstrated a progressive transition from the CD45RA+ naïve cell phenotype to the more differentiated CD45RO+ memory subset (Gabriel et al., 1993; Karanfilov et al., 1999; Fagnoni et al., 2000). Moreover, an expansion of the CD28 T cell subset with a concomitant decrease of CD28+ T cells, particularly in the CD8+ T cell population (Fagnoni et al., 1996; Nociari et al., 1999), has been reported in elderly subjects. CD28 is a major costimulatory molecule required for functional T cell activation (Sperling & Bluestone, 1996). Furthermore, CD28 T cells do not proliferate readily in vitro and have been proposed to represent the senescent phenotype of CD28+ T cells (Globerson & Effros, 2000; Pawelec et al., 2002). The accumulation of specific cell phenotypes with aging may contribute to aging-related changes in cell responses since various cell phenotypes have different activation requirements and might differ in the generation of activation products (Flurkey et al., 1992).

The CD45RA+ and CD45RO+ lymphocyte subpopulations differ in their functional activity, particularly in their requirements for activation and differentiation, as well as in their capacity to secrete cytokines (de Jong et al., 1991). CD45RO+ cells, both in CD4+ and CD8+ subsets, are the main producers of cytokines (Bettens et al., 1989; Hirohata & Lipsky, 1989; Salmon et al., 1989; de Jong et al., 1991). In addition, it has been reported that cytokine receptors are expressed preferentially on CD45RO+ lymphocytes (Kaech et al., 2003; McQuaid et al., 2003). This differential expression of cytokines and their receptors may instigate differences in the response of these cell subtypes to Hsp production (Stephanou et al., 1999). Another possible point of concern is the sensitivity for apoptosis of various T cell phenotypes. Functional studies have shown that CD45RO+ memory cells express lower levels of bcl2, an antiapoptotic gene product, than CD45RA+ naïve cells, making the former more susceptible to apoptosis (Akbar et al., 1993; Boudet et al., 1996). Noteworthy, the propagation of the apoptotic program involves cleavage of several vital proteins, which can serve as stress signals for the induction of Hsp. Thus, knowledge of the Hsp response of various cell phenotypes that characterize a pool may provide insights on the response to Hsp with age.

Despite the consistently reported variation in PBL phenotypes with age, studies on Hsp in aging have not addressed PBL at the level of their phenotypes. Here we report, in a study on young and elderly human volunteers, that various cell phenotypes in a cell pool exhibit different basal levels of Hsp70. We found that the memory cell phenotypes produced more Hsp70 than the naïve phenotypes. Taken together, our results favor the viewpoint that aging-related differences in basal Hsp70 levels are not brought about by changes in the lymphocytes per se, but instead reflect the increased prevalence of specific cell phenotypes with age.


Figure 1 pools the results from all participants, and shows the expression pattern of Hsp70 by PBL subtypes, under basal conditions. The CD4+ T cells had the highest levels of Hsp70 (p < 0.001 compared to all the other lymphocyte subtypes). On the contrary, B cells showed the lowest Hsp70 concentrations.

Figure 1.

Basal Hsp70 levels in various lymphocytes. Peripheral blood lymphocytes were surface stained with lymphocyte subset-specific markers followed by intracellular immunostaining for Hsp70. Results are representatives of 14 independent experiments with cells from 20 participants. The median and interquartile ranges of data are indicated. Statistical analyses were performed by a two-tailed paired t-test. Difference: *p < 0.001 compared to CD4+ cells and &p < 0.05 compared to CD8+ cells, †p < 0.001 compared to natural killer (NK) cells, ‡p < 0.05 compared to NK cells. Similar statistical significances were obtained when young and elderly participants were analyzed separately.

The data in Fig. 2 indicate that the levels of Hsp70 were significantly higher in CD45RO+ memory cells compared to CD45RA+ naïve cells in both the CD4+ (p = 0.002 and p = 0.002, for young and elderly subjects, respectively) and CD8+ (p = 0.030 and p = 0.051 for young and elderly subjects, respectively) T cell subsets.

Figure 2.

Hsp70 levels in lymphocyte subpopulations. Peripheral blood lymphocytes were surface stained with lymphocyte subset-specific markers followed by intracellular immunostaining for Hsp70. Naïve and memory cells were distinguished based on the expression of the CD45RA or the CD45RO surface marker, respectively. The Hsp70 expression levels of the various lymphocyte subsets were analyzed by three parameter fluorescence scattergrams and values were adjusted using isotype matched PE/FITC mouse IgG1 controls for nonspecific staining. Box plots depict the median and interquartile ranges of data from 10 participants and each column represents the results of seven independent experiments with cells from the subjects. The significance of differences between levels of Hsp70 in the lymphocyte subtypes is indicated on the plot.

There was no age-related difference in the Hsp70 concentration in any of the PBL subpopulations. However, we noted a more important interindividual variability in the concentrations of Hsp70 in the elderly group compared to the young group. The difference in variability between these groups was more striking for the CD4 (p = 0.001) subpopulation, particularly CD4+ cells expressing the CD45RO+ memory phenotype (p = 0.028) (Fig. 2). The differences in the interindividual variation for the other lymphocyte subsets were not significant.

There was no age-related difference in the frequencies of CD4+ and CD8+ cells. Figure 3 shows the age-related differences in the percentage of PBL subsets. When looking at all T cells together, elderly subjects had significantly higher percentages of CD45RO+ (p = 0.004), and significantly lower percentages of CD45RA+ cells (p = 0.001). Similar findings were observed when T cells were subtyped: CD45RA+ cells were less frequent, while CD45RO+ were more frequent, in both the CD4+ (p = 0.035 and p = 0.003, respectively) and the CD8+ (p < 0.001 for CD45RA+ cells) T cell pools from elderly compared to young subjects. In addition, the percentage of natural killer cells was significantly higher in the elderly subjects (p = 0.023) in contrast to the percentage of γδ T cells, which was lower (p = 0.035).

Figure 3.

Percentage of lymphocyte subsets in whole blood. FACS plots were gated on CD4+ T cells, CD8+ T cells, γδ, NK and B cells to depict the percentage of various lymphocyte phenotypes in the peripheral blood lymphocyte pool. Each column and bar represents the mean and SD of the results of seven independent experiments with cells from 10 subjects. Statistical analyses were performed by a two-tailed unpaired t-test and p-values are indicated.

The Hsp70 levels as well as the percentages of T cells in relation to their expression of the CD28 molecule are portrayed in Fig. 4. No difference was evident in the levels of Hsp70 between CD28+ and CD28 T cells. In addition, there was no age-related difference in the levels of Hsp70 with respect to the presence or absence of the CD28 receptor. Significantly lower percentages of CD28+ cells and higher percentages of CD28 cells were noticed in both the CD4+ T cells (p = 0.043 for both comparisons) and CD8+ T cells (p = 0.001 for both comparisons) in elderly compared to young subjects.

Figure 4.

Hsp70 concentrations and percentages of T cells in relation to their expression of the CD28 receptor. Peripheral blood mononuclear cells were isolated from whole blood as described in the method section. (A) Intracellular levels of Hsp70 (B) Percentage of cells in both the CD4+ and the CD8+ populations. Cells were analyzed by flow cytometry and the mean fluorescence intensities (MFI) as well as the percentages of cell are indicated. Each column represents the results of seven independent experiments with cells from 10 subjects. Statistical analyses were performed by a two-tailed unpaired t-test and p-values are shown in the figure.


This work demonstrates different Hsp70 concentrations in various lymphocyte subpopulations. Particularly, lower levels of Hsp70 were found in B lymphocytes compared to natural killer (NK) cell and T lymphocytes. The inducibility of Hsp70 has been reported to vary importantly depending on the state of B-cell activation (Spector et al., 1989). Hardy et al. (1997) as well as Spector et al. (1989) have demonstrated that resting B lymphocytes have low or absent levels of Hsp70 and that they respond to heat stress with only a minute increase in the amount of Hsp70 mRNA. This mitigated response to stress by resting B cells is attributed to the low cytoplasmic levels of the heat shock transcription factor 1 protein and particularly to its limited potency to bind the heat shock element in the promoter region of the Hsp70 gene in unactivated B cells (Hardy et al., 1997). Another possible explanation for the differences in Hsp70 responsiveness of lymphocyte subsets is reflected in the differential capability of these cells to secrete Hsp70. It has been reported that more than 70% of Hsp70 secreted by PBMC originates from B cells (Hunter-Lavin et al., 2004), meaning that resting B cells do not only synthesize the protein in reduced amount, but that they also secrete a large amount of their Hsp70.

We observed that CD4+ T cells from both young and elderly subjects had higher Hsp70 concentrations than the other PBL subsets (p < 0.001 for all comparisons). This observation was also evident when these cells were subdivided into the various phenotypes (Fig. 2; p < 0.001 for all comparisons with corresponding CD8+ phenotypes). Possibly, this difference is related to the low antioxidant status of CD4+ cells (Roozendaal et al., 2002). Ex vivo induction of oxidative stress (Pathak & Khandelwal, 2007), as well as in vivo oxidation through infection (Aukrust et al., 1995), have revealed a relatively marked intracellular glutathione antioxidant depletion in CD4+ compared to CD8+ subpopulation of cells. Notably, under normal physiological situations, oxidative damage to cells can result from reaction with oxygen radicals formed as a by-product of normal metabolism (Wheeler et al., 1995). Such damages, as has been demonstrated by various studies (Arata et al., 1995; Bechoua et al., 1999; Njemini et al., 2005), can result in the induction of the HSP gene family which protects proteins from denaturation. Accordingly, the limited availability of glutathione in CD4+ cells may lead to enhanced Hsp70 expression as a result of the potential accumulation of reactive oxygen metabolites (Schreck et al., 1991; Lautier et al., 1992; Chang et al., 1999). The difference in Hsp70 levels seen in this study may also be related to the nature of the T cell response with respect to the epitopes recognized and functional characteristics of the T cell subset. For example, differences between CD4+ and CD8+ T cells with respect to their cytokine profile are well known (Kroemer et al., 1993; Corrigan et al., 1995; Wahlstrom et al., 2001; Thien et al., 2005). Noteworthy, cytokines in various combinations have been reported to enhance the production of Hsp, depending on the type of cytokine and the Hsp under consideration (Stephanou et al., 1997; Stephanou et al., 1998).

In the present study, we found that the memory cell phenotypes exhibited more Hsp70 than the naïve phenotypes. It is well known that CD45RO+ memory cells and CD45RA+ naïve cells differ profoundly in the level of cytokine production (Bettens et al., 1989; Salmon et al., 1989; de Jong et al., 1991). Indeed, CD45RO+ cells produce significantly higher levels of cytokines including tumor necrosis factor-alpha (de Jong et al., 1991), interleukin-4 (IL-4; Bettens et al., 1989; Salmon et al., 1989) and interferon-gamma (Hirohata & Lipsky, 1989; Salmon et al., 1989). Moreover, the preferential expression of cytokine receptors comprising the IL-7, IL-12 and IL-18 receptors on CD45RO+ PBL has been reported (Kaech et al., 2003; McQuaid et al., 2003). This enhanced expression of cytokines and their receptors may lead to increased HSP70 gene transcription in memory cells (Stephanou et al., 1999). Alternatively, the higher Hsp70 concentrations in memory cells might reflect their higher susceptibility to apoptosis (Akbar et al., 1993; Boudet et al., 1996).

Consistent with other studies (Zeman et al., 1988; Gabriel et al., 1993; Fagnoni et al., 1996, 2000; Mariani et al., 1998; Karanfilov et al., 1999; Nociari et al., 1999), elderly subjects displayed significantly lower proportions of the CD45RA+ naïve T cells with a concomitant increase in CD45RO+ memory T cells, and NK cells, as well as CD28-deficient cells compared to young subjects. The CD28+ T cells and CD28 T cells displayed comparable levels of Hsp70. In addition, comparison of the production levels of Hsp70 in both the CD28+ and CD28 T cells from elderly and young subjects showed no significant difference (Fig. 4). This finding suggests that the aging-related increase in CD28-receptor-deficient T cells is not responsible for the difference in Hsp70 levels.

A primary objective of this study was to see whether aging-related differences in basal Hsp70 levels were associated with alterations in the frequencies of lymphocyte subsets. We found that the basal levels of Hsp70 in the various lymphocyte subpopulations were comparable in the young and elderly subjects. We show that the CD45RO+ memory cell phenotypes displayed more Hsp70 than the naïve CD45RA+ phenotypes in the entire population (p < 0.001 and p = 0.003, for CD4+ and CD8+, respectively). Furthermore, NK cells ranked second among the lymphocyte subpopulations, with regards to their Hsp70 levels. Because CD45RO+ T cells as well as NK cells (Zeman et al., 1988; Mariani et al., 1998) are up-regulated in the elderly subjects, changes in Hsp70 levels with age could indeed be linked to the altered frequencies of lymphocyte subpopulations.

Taken together, our results indicate that data obtained from pooled cell populations should be interpreted with caution. When aging-related changes are detected, either in vivo or in vitro, it is often assumed that they are present in the entire cell population under study. It is well known, however, that cell populations are rarely homogeneous. In vivo, cell populations are generally built up of several varieties of cells, which can be in various states of differentiation. In cell cultures in vitro, most cell populations are built up of precursor, intermediately and variously differentiated cells, and senescent cells (Mets & Verdonk, 1981a,b; Bayreuther et al., 1988). Although elderly organisms can retain cells with youthful characteristics that are fully functional (Smith et al., 2002), it is noteworthy that with aging, both in vivo and in vitro, important changes in the proportion of cellular subtypes can occur (Mets et al., 1983). As demonstrated by our study, some changes occurring during aging can be interpreted in terms of differentiation of cell types. Memory cells are, indeed, to be considered as a more differentiated cell type than the naïve counterparts from which they arise.

In conclusion, we have provided evidence suggesting that aging-related changes in Hsp70 concentration in lymphocytes could be linked to the frequencies of lymphocyte subtypes. We found that the more differentiated memory cell phenotypes had the highest Hsp70 concentrations. On the contrary, B cells showed the lowest Hsp70 concentrations. The differential presence of Hsp70 in various cell subsets may have a number of still unexplored implications and is a challenge for future research.

Experimental procedures


Ten apparently healthy young (five women and five men, aged between 22 and 25 years, mean age 23.3 years (SD 1.3)) and 10 apparently healthy elderly (four women and six men, aged between 71 and 86 years, mean age 77 years (SD 4.5)) volunteers, all living in the community, were included into the study after informed consent.

Reagents and antibodies

Phosphate-buffered saline (PBS) was purchased from Life Technologies (Gibco, Paisley, Scotland, UK). Bovine serum albumin (BSA) was from Roche (Boehringer, Mannheim, Germany). Lymphoprep was from Nycomed (Oslo, Norway). PE-CY5-labeled anti-CD4, PE-CY5-labeled anti-CD8, PE-CY5-labeled anti-CD3, PE-labeled anti-CD28, PE-labeled anti-CD19, PE-labeled anti-CD56, PE-labeled anti-CD45RA, PE-labeled anti-CD45RO, PE-labeled anti-γδ TCR, PE-labeled anti-mouse IgG1 control, PE-labeled anti-mouse IgG2A control, and FITC-labeled anti-mouse IgG1 control were from Becton Dickinson (Erembodegen, Belgium). The FITC-labeled monoclonal antibody against the inducible form of Hsp70 (clone C92F3 A-5, SPA-810) was from StressGen (Victoria, Canada). As reported by StressGen, the Hsp70 antibody recognizes the products of the inducible Hsp70 genes localized in the major histocompatibility gene region and there is no reactivity with the constitutive Hsc70. We have confirmed the specificity of this antibody using Western blotting (Njemini et al., 2007). The products were applied according to the manufacturers’ guidelines.

Cell preparation

Peripheral blood mononuclear cells were recovered as described previously (Njemini et al., 2003). Briefly, EDTA blood was diluted twice with PBS and carefully layered over lymphoprep. After centrifugation, the cells were isolated, washed twice in PBS containing 1% BSA (PBS/BSA) at 900 ×g for 3 min, and resuspended in PBS/BSA. Thereafter, the cells were stained for their intracellular levels of Hsp70.

Cellular staining

About 5 × 105 cells were double-stained with PE-CY5-labeled anti-CD4 (or PE-CY5-labeled anti-CD8) and PE-labeled anti-CD28 (or PE-labeled anti-CD45RA, or PE-labeled anti-CD45RO) at 4 °C. NK, B and γδ T cells were stained with PE-labeled anti-CD56, PE-labeled anti-CD19 and PE-labeled anti-γδ TCR antibodies, respectively. After 15 min cells were washed at 900 ×g for 3 min in 1 mL PBS/BSA and fixed at room temperature with 100 µL of PBS containing 4% paraformaldehyde for 15 min. After washing, the cells were permeabilized with 100 µL of PBS/BSA containing 0.1% saponin, and at the same time incubated with the Hsp70-specific antibody conjugated to FITC. The staining protocol included isotype matched controls for both surface (PE-labeled anti-IgG1 or anti-IgG2a) and cytoplasmic (FITC-labeled anti-IgG1) staining. Incubation was carried out for 15 min at room temperature. After washing, 500 µL of FACSFlow solution (Becton Dickinson, Immunocytometry System, San Jose, CA, USA) were added and the samples were analyzed using flow cytometry.

Flow cytometry analysis

The labeled samples were analyzed with a Coulter Epics XL-MCL three color flow cytometer (Coulter, Miami, FL, USA). Data acquisition was performed using the Coulter System II 3.0 software (Epics). The FITC, PE, and PE-CY5 dyes were excited with a 15 MW argon laser of wavelength 488 nm. Analysis was done in listmode for the green FITC fluorescence through a 525-nm filter, for the orange PE fluorescence through a 575-nm filter, and for the red PE-CY5 fluorescence through a 670-nm filter. The lymphocyte subpopulation was differentiated according to granularity and size in the forward vs. side scattergram and was gated. Data are represented as mean fluorescence intensity (MFI), corrected for background fluorescence with the negative control. Dead cells were excluded by electronic gating. A total of 3000 cells were collected for each lymphocyte subtype. The Hsp70-related immunofluorescence was quantified by calculating the MFI from histograms of logarithmic fluorescence intensity.

Statistical analysis

Statistical evaluation was performed using Prism version 3 and Analyse-it for Microsoft Excel. Column statistics (with statistical package Prism 3.0) was used to test the approximation of the population distribution to normality. Averages were compared using Student's t-test or analysis of variance. Continuous variables were compared by Pearson or Spearman correlation. For data that were not normally distributed, the nonparametric Wilcoxon signed rank test and the Mann–Whitney test were applied. To evaluate the interindividual variation, we used an index for diversity as proposed by Gorus et al. (2006). Briefly, we calculated for each participant the difference between individual mean MFI and group mean MFI of Hsp70. This difference was expressed as a percentage of the group's mean MFI and a t-test was performed to estimate the variation of Hsp70 levels in a group with reference to that group's mean Hsp70 level. A p-value (two-sided) of < 0.05 was considered as statistically significant.