Degeneration of the thymus and severe contraction of the T cell repertoire with aging suggest that immune homeostasis in old age could be mediated by distinct effectors. Therefore, receptors expressed on T cells as they undergo senescence in vitro, as well as those displayed by circulating T cells during normal chronologic aging, were examined. Monitoring of T cells driven to senescence showed de novo induction of CD56, the prototypic receptor of NK cells. Analysis of fresh T cells in peripheral blood showed an age-dependent induction of CD56. These unusual T cells expressed high levels of Bcl2, p16, and p53, and had limited, or completely lost, ability to undergo cell division, properties consistent with senescence. CD56 cross-linking without TCR ligation on CD56+ T cells resulted in extensive protein phosphorylation, NF-κB activation, and Bax down-regulation. CD56 cross-linking was also sufficient to drive production of various humoral factors. These data suggest that the immunologic environment in old age is functionally distinct, rather than being a dysfunctional version of that seen at a young age. CD56+ T cells are unique effectors capable of mediating TCR-independent immune cascades that could be harnessed to enhance protective immunity in the elderly.
Chronologic aging is characterized by impaired production of new T cells due to the involution of the thymus (1). Antigenic challenge through life leads to the progressive contraction of the pool of naive T cells (2). Consequently, the peripheral T cell repertoire of elderly persons (aged ≥70 years) consists predominantly of oligoclonal memory cells, and the overall T cell diversity is significantly reduced relative to that seen in younger persons (3, 4, 5). The number of circulating T cells is nevertheless maintained through life (6), a phenomenon that is attributable in part to increased steady-state homeostatic Ag-independent proliferation in adult life (7, 8). However, homeostatic expansion along with clonal proliferation due to persistent or cyclical exposure to Ag lead to telomere erosion that limits the overall mitotic capacity of T cells (5, 9, 10). There are also functional defects of the TCR with aging. Experimental studies indicate that certain subsets of T cells of elderly people are unable to sustain TCR signal transduction, resulting in blunted cell- and Ab-mediated responses (11, 12). All these immunologic alterations are thought to underlie age-related decline in Ag-specific immunity (generally referred to as “immunosenescence”), epitomized by the yearly high morbidity and mortality to influenza among the elderly (13, 14).
Despite a deficiency of TCR-driven immunity with aging, vigorous immune responses in a segment of the elderly population have been documented (15, 16, 17, 18). Whether this subgroup of elderly persons shares distinct genetic factors is unknown. Vaccine studies suggest a potential role of particular subsets of T cells, such as those that retain ability to produce IL-2, IL-4, and IFN-γ (19, 20, 21). Comparative phenotypic studies of leukocytes also suggest pervasive changes in the T cell repertoire with advancing age through up- or down-regulation of receptors other than the TCR (22, 23, 24). To explore the significance of these observations, we undertook studies to characterize functional subsets of T cells with aging. We hypothesized that modulation of expression of receptors on T cells with aging is a way to maintain immune homeostasis despite a prevailing deficiency, or inefficiency, of TCR-driven immunity in old age. A prediction of this hypothesis is that induction of novel receptors imparts effector functions that would be independent of the TCR, or at least be costimulatory for those T cells that retain a functional TCR. To explore this hypothesis, we revisited an in vitro senescence system that we and others have used previously to track the conversion of highly proliferating CD28+ T cells into senescent CD28null cells (25, 26). We focused on the induction of receptors normally expressed by NK cells because of reported increased transcription of NK receptors in CD28null T cells generated in vitro (27). We verified the relevance of these in vitro studies by cross-sectional analyses of fresh leukocytes from community-dwelling persons. We document here the increasing levels of expression of CD56, the prototypic NK receptor (28), on T cells with advancing age. Experimental studies were conducted to examine whether CD56+ T cells have features of senescence, and to demonstrate CD56 as a functionally competent receptor for T cells.
Materials and Methods
Study cohort and collection of biological specimens
Adults of both sexes (ages ≥18 years; n = 158) were recruited from the Pittsburgh, PA, and the Rochester, MN, areas. For the purposes of this study, “elderly” persons refer to those ≥70 years of age (n = 43; mean age 85.1 ± 4.6, age range 70–93). These individuals were recruited from prescreening of volunteers who were undergoing routine physical and medical examination. They were also enrolled from survivors of an ongoing, long-term, large-cohort, epidemiological study of aging who have been clinically monitored for the last 18 years (29, 30, 31). Exclusion/inclusion criteria for these community-dwelling, ambulatory, acute illness-free, exceptionally aging elderly subjects were described previously (32, 33). The remaining adults (age range 18–69 years; n = 115) were also volunteers from the community who were screened free of any medically indicated immunosuppression, chemotherapy and anti-inflammatory medications, or chronic inflammatory condition. At enrollment, all subjects were free of acute medical problems. All subjects provided written informed consent. Waste de-identified cord blood samples (n = 20) were also obtained from Magee Women’s Hospital, University of Pittsburgh Medical Center. Studies involving human subjects were approved by institutional review boards of the Mayo Clinic (Rochester, MN) and the University of Pittsburgh.
Mononuclear cells were isolated from blood and cord blood by standard isopycnic centrifugation over Ficoll-Hypaque (GE Healthcare), or by direct lysis of erythrocytes using the VitaLyse reagent (BioE).
Cell phenotypes were examined by multicolor flow cytometry. Cell suspensions were stained with fluorochrome-conjugated Abs (from BD Biosciences and eBioscience) to Ags attributed to T cells (CD3, CD4, CD8, CD28), NK cells (CD16, CD56, GL183), macrophage/monocyte (CD14), and B cells (CD19, CD20). In specific experiments, TCR αβ usage was also determined using Abs to five of the most commonly expressed TCRs in the population, namely, AV12.1, BV2, BV5, BV6S7, and BV8 (34). In other assays, nuclear and cytoplasmic staining with Abs to p16, p53, and Bcl2, molecules associated with cell senescence (35, 36, 37), were conducted using the appropriate staining kit (BD Biosciences). For the latter, cells were first stained for cell surface Ags, followed by fixation/permeabilization, and nuclear/cytoplasmic Ag staining.
For each cytometry experiment, control cells were incubated with the appropriate species-specific Ig isotype conjugated with the same fluorochrome, and used to measure intrinsic cell fluorescence. Single Ab staining for control cells was also done for each of the indicated Ags being measured to ascertain the fluorescence spectrum of each of the fluorochromes used.
Raw cytometric data were acquired using the LSRII cytometer (BD Biosciences). Cytometer efficiency and signal compensation were established for each experiment using calibration beads (Spherotec) for each fluorochrome used. Analyses of cell populations were done offline using the FlowJo software (Tree Star). Live single cells were electronically discriminated using forward/side scatter profiles and height versus width of the forward scatter. Signal intensities of Ag-specific staining were normalized against a compensation matrix constructed from signals of the calibration beads and control cells.
Cell culture, isolation of fresh T cells and NK cells, replicative senescence system, and CFSE assay
Primary T cells from blood were isolated using the RosetteSep system (StemCell Technologies). For activation assays (see below), T cell preparations were enriched for CD56+ cells by negative selection using the EasySep system (StemCell Technologies) to remove contaminating cells expressing CD16, CD19 (or CD20), and CD14. These cell isolation procedures were conducted using the protocols described by the manufacturer. Subsequent bioassays (see below) were conducted using T cell preparations that were ≥15% CD56+ as verified by flow cytometry.
Primary NK cells from blood were similarly purified using the EasySep system by the positive selection of CD16+CD3− cells. Purity and the level of CD56 expression of NK cell preparations were also verified by flow cytometry. Bioassays (see below) were conducted using NK cell preparations that were ≥42% CD56+ and ≥99% CD3−.
Culture of primary T cells, establishment and analysis of an in vitro T cell senescence system, and T cell proliferation assays using cells loaded with CFSE were conducted using procedures we described previously (24, 25, 40). Bioassays with primary NK cells were done using an NK-specific culture media (39).
These assays were performed by two strategies. The first strategy involves incubation of T cells (or NK cells as appropriate) in plate-immobilized IgG, anti-CD56 (C218; Beckman Coulter), anti-CD3 (OKT3; Ortho Biotech), or a combination of anti-CD56 and anti-CD3. The indicated range of concentration of anti-CD3 used for plate-immobilization was based on our previous studies (41, 42). We had established that precoating culture plates with 50 μl of a working concentration of 50 μg/ml anti-CD3 was necessary to activate T cells without the need of other costimulatory factors. Precoating a working concentration of ≤1 μg/ml anti-CD3 was suboptimal and required costimulatory factors. A working concentration of 50 μg/ml anti-CD56 for plate-immobilization was empirically determined (data not shown) to be activating. Plate-immobilization of Abs was conducted in a bicarbonate buffer with pH 9.0 at 4°C, a condition that ensured stable attachment of Igs on neutral plastic substrates (41, 42, 43). Working solutions of coating Abs were added at 50- or 200-μl aliquots for 96- or 24-well plates, respectively. Subsequently, cell suspensions were culture-plated in a total volume of 250 μl or 1.0 ml in 96- or 24-well plates, respectively.
The second strategy is receptor ligation in solution following procedures analogous to immunostaining for flow cytometry. Briefly, cells were incubated on ice with a primary Ab (e.g., anti-CD3, anti-CD56), washed with cold PBS, and followed by incubation on ice with the secondary Ab. After extensive washing, cells were incubated in prewarmed culture media and incubated at 37°C for the indicated period. Based on previous studies (42, 43, 44), 0.1 μg/ml soluble anti-CD3 was used for activation. For anti-CD56, concentrations of 1.0–5.0 μg/ml of soluble Ab were empirically determined (data not shown) as activating. Cell activation by this strategy was examined by phosphorylation and by expression of activation markers by flow cytometry (see below).
Activation assays were performed in either short or longer incubation periods. A short 15-min incubation period was used for molecular assays of cellular activation. One assay is a measurement of protein phosphorylation by Western blotting using specific Abs to phospho-tyrosine or phospho-serine/threonine (4G10 and 44–006, respectively; Millipore). Protein bands in immunoblots were visualized either by the SuperSignal chemiluminescent reagent (Pierce) or by the Odyssey infrared fluorescent imaging system (LI-COR Biosciences). Densitometry was conducted using the Odyssey software or the ImageQuant TL software (Amersham Biosciences). All Western blotting experiments were conducted using equal protein content of cell lysates and by probing blots with anti-GAPDH (MAB374; Millipore) as control.
Protein phosphorylation assays for freshly isolated CD56+ T cells were examined by a flow cytometry-based procedure described previously (45). As described above, cells were incubated on ice with soluble anti-CD3, anti-CD56, IgG isotype control, or combinations thereof, or in media alone. Cells were washed with cold PBS and then incubated on ice with Cy5-conjugated anti-mouse IgG (BD Biosciences) to cross-link cells prestained with anti-CD3 and/or anti-CD56. After washing, prewarmed culture media was added to the cell pellet and incubated at 37°C for 15 min. Cellular activation was stopped by the addition of 2% phosphate-buffered paraformaldehyde and washed, followed by incubation in cold 75% methanol to permeabilize cells. Cells were washed extensively, and incubated with FITC-conjugated anti-phosphotyrosine 4G10 Ab or with IgG2b isotype control (Millipore). As positive control, an aliquot of cells was incubated with 5 mM hydrogen peroxide in PBS (46) and processed similarly for anti-phosphotyrosine staining. Unstimulated cells preloaded with FITC-phosphotyrosine (Millipore) were used as an additional negative control to the IgG-incubated cells. Cell populations were analyzed by flow cytometry using cytometer calibration and data collection procedures described above.
Activation of primary T cells and NK cells was also examined by the expression of activation markers CD25 and CD69. This was conducted by flow cytometry with the inclusion of appropriate fluorochrome-conjugated Abs to these two markers in the multicolor cytometry procedure described above.
Cell lysates were also used in protein hybridization of commercially available membrane arrays to identify substrates of various cell-signaling pathways (Hypromatrix). By this strategy, we found (data not shown) that lysates of cells incubated in anti-CD56 alone contained activated forms of several proteins including Bcl2 and I-κBα. These results were verified by Western blotting using new batches of cell lysates. Relevance of the Bcl2 pathway was examined by immunoblotting with Abs to Bcl2 (Clone 100; R&D Systems) and to Bax (Cell Signaling Technology). Relevance of NF-κB pathway was verified by immunobloting with Abs to I-κBα, p65, and p52 (Cell Signaling Technology).
NF-κB activation was further verified in gel-shift assays using procedures we had described previously (47). In this case, nuclear extracts were prepared separately following the activation procedure described above. These extracts were used in DNA-binding reactions containing radiolabeled double-stranded oligonucleotides corresponding to a consensus NF-κB-binding sequence AGTTGAGGGGACTTTCCCAGGC (48), and 1 μg/ml anti-p65 Ab or an IgG isotype (Santa Cruz Biotechnology). Nuclear extracts of cells stimulated for 30 min with 10 ng/ml TNF-α (R&D Systems), an NF-κB activator (48), were used as positive control. As system control, similar assays were performed on the identical set of extracts with an oligonucleotide probe for Sp1 (47). DNA-protein complexes were fractionated by standard nondenaturing PAGE and visualized by autoradiography.
Activation assays involving 24–48-h incubation of cells were used to examine cytokine production. Approximately 5 × 105 CD56+ T cells was added to appropriate Ig-coated tissue culture plates and left overnight in a tissue culture incubator without exogenous growth factors or any other form of stimulation. Culture supernatants were harvested and cytokine content was examined by multiplex analysis using the Luminex system as we described previously (40).
For adhesion assays, ∼5 × 105 cells was added to appropriate Ig-coated plates (described above) and incubated for 30 min at 37°C in a tissue culture incubator. Nonadherent cells were gently aspirated and transferred to microfuge tubes. The wells were gently washed several times with PBS to further remove nonadherent cells. The nonadherent cell aspirate was also washed and centrifuged. This cell pellet and the adherent cells in the wells were separately lysed with 500 μl of standard Western lysis buffer. Protein content of the lysates was determined using a protein assay reagent kit (Bio-Rad). Cell adhesion was calculated as the percentage protein content of the adherent cell lysate over the sum protein content of both adherent and nonadherent cell lysates.
Quantitative data were analyzed nonparametrically using SPSS software. The relationship of age with the frequency of CD56+ T cells, and with CD28null T cells, and also the relationship between these two T cell phenotypes for the entire population cohort were examined independently by regression analysis. Statistical significance of the regression lines, the regression coefficients, and the slopes was indicated by p values ≤0.01. As an alternative method of analysis, subjects were first grouped by age-decades, and the relationship between the frequency of CD56+ and CD28null T cells with each increment decade of life was examined by Kruskall-Wallis-ranked ANOVA, where p < 0.01 was also considered significant. Comparison between two age-groups was examined by least squares difference or by the Tukey statistic, and p < 0.01 was considered significant.
For the bioassays, paired comparisons between any two groups, Mann-Whitney U test, χ2 test, Dunn’s method, or Student’s t test were conducted as appropriate, and p < 0.05 was considered statistically significant. Differences in the levels of production of multiple cytokines in response to anti-CD56 or anti-CD3 (or both) or to IgG were examined by ANOVA where p < 0.05 was considered significant. Dependence of cytokine production on the concentration of anti-CD3 used as activating stimulus was examined by regression analysis where statistical significance was assessed by estimation of 95% confidence interval of the regression curve.
CD56 expression on T cells undergoing replicative senescence in vitro
Whereas clonal proliferation of T cells is a transient but requisite event for mounting Ag-specific immune responses, accumulation and persistence of clonal populations of T cells are features of chronologic aging (3, 4, 5). In the latter case, subsets of clonal T cells have limited, or have completely lost, the capacity to divide, suggesting cells that are in advanced stages of senescence (9, 10).
To further examine properties of senescent T cells, we revisited an in vitro senescence system that we and others have used to monitor phenotypic changes that T cells undergo during their proliferative lifespan (25, 26). This involved the repeated stimulation of CD3+CD56−CD28+ T cells freshly isolated from PBMC of young donors (age of 18–22 years) by a polyclonal approach with soluble anti-CD3 and irradiated allogeneic PBMC every 15 days. In recent work (40), we showed that this senescence system consistently yielded T cell lines that lacked expression of CD28 but that had acquired high levels of expression of CD56, the classical NK receptor (28). These in vitro-generated CD28nullCD56+ T cells of either CD4 or CD8 subsets had lost their ability to divide following subsequent stimulation (40), a property consistent with senescent somatic cells (35). In a standard CFSE proliferation assay, such nondividing CD28nullCD56+ T cells were distinguished as cells that remained CFSEhigh even after 5 days of stimulation (40).
Fig. 1 shows additional features of nondividing CD28nullCD56+ T cells generated in vitro. They expressed high levels of p16, a mitotic inhibitor (35), which likely explain their inability to further divide following subsequent stimulation. Large proportions of p16+CD56+ T cells emerged in the cultures between the 5th and 10th stimulation cycle depending on the donor. More importantly, the resulting CD56+CD28null T cell lines showed overrepresentation of cells expressing particular TCR α- or β-chains. As depicted, cells bearing BV5 and AV2, two TCRs that normally comprises ≤3% of circulating T cells (34), were overrepresented in the CD56+CD28null T cell lines generated through this in vitro system. These expansions of TCR α- and β-chains were observed despite a nonselective, polyclonal activation approach using a combination of soluble anti-CD3 and irradiated allogeneic feeder cells derived from different donors during each stimulation cycle. These findings were consistent for seven independent experiments.
Age-dependent accumulation of CD56+ T cells in vivo
To verify the biological relevance of the in vitro senescence system, we conducted a cross-sectional analysis of T cell phenotypes in fresh PBMC. Because it is not yet known whether the molecular machineries for the gain of CD56 and for the loss of CD28 are mutually exclusive or are coordinately regulated, we examined independently the frequency of CD56+ and of CD28null T cells. Fig. 2 A shows the results of analysis of 178 community-dwelling adults and neonates. CD56+ T cells were abundantly found in adults, but were considerably less frequent among neonates. The entire population cohort showed an overall pattern for the progressive accumulation of CD56+ T cells with advancing age, as indicated by highly significant linear regression in both CD4 and CD8 compartments. However, there were higher proportions of CD56+ T cells in the CD8 compartment relative to the CD4 compartment, demonstrated by a higher regression coefficient and slope of the regression line for CD8 T cells. With age criteria, there was evident rise in the abundance of CD56+ T cells around the age of 50 years.
The increasing levels of expression of CD56 with advancing age occurred concomitantly with the progressive loss of CD28 expression consistent with previous studies (47, 49). Fig. 2 A also shows progressive accumulation of CD28null T cells with age as evidenced by highly significant regression lines for both the CD4 and CD8 compartments. Similar to the pattern for the gain of CD56 expression with age, the proportions of CD28null cells were higher among CD8 T cells relative to CD4 T cells. With age criteria, there was also a characteristic rise in the numbers of CD28null T cells around the age of 50 years. The regression coefficient and the slope of the regression line for the age-related increased frequency of CD28null T cells were highly significant.
When population analysis was conducted by grouping subjects into age-groups by decade, frequency of CD56+ (and CD28null) T cells every 10 years of age was also significant (p < 0.0001 by Kruskall-Wallis ranked ANOVA). Age-group comparisons showed a statistically significant rise in the numbers of CD56+ or CD28null T cells beginning at the third decade of life (p < 0.001 by least squares difference or by the Tukey statistic). At the seventh decade of life onward, the numbers of these cells appeared to plateau. There was no difference in the frequency of CD56+ or CD28null T cells between males and females (data not shown).
Higher frequencies of CD28null and CD56+ T cells in the CD8 compartment relative to that of the CD4 compartment in vivo (Fig. 2,A) were reminiscent of our in vitro senescence system showing a more rapid loss of CD28 and gain of CD56 among CD8+ T cells compared with the CD4+ subset (Fig. 1) (40). Such differences in the relative abundance of CD28null and CD56+ T cells between the two compartments are further illustrated in Fig. 2 B. Irrespective of age, the regression of the frequency of CD56+ and CD28null T cells has a higher slope and a higher regression coefficient for the CD8 subset compared with the CD4 subset.
Usage of TCR αβ-chains by CD56+ T cells found in vivo was also examined. We focused on BV2, BV5, BV8, BVS67, and AV12.1 since these are the commonly expressed TCRs in the population, with each TCR representing ≤3% of the total T cells of young adults (34). Fig. 2 C shows that CD56+ T cells of persons ≥70 years of age have high frequencies of each TCR examined. AV12 and BV5 were the most dominant, occurring at ≥10% and ≥5%, respectively, for all T cells for all elderly subjects examined. BV2 was the least represented, occurring at the documented normal frequency of <3% as seen in younger persons (34). This skewed TCR αβ usage for CD56+ T cells is within the reported ranges of TCR αβ skewing for CD28null T cells among elderly people (49).
CD56+ T cells derived in vivo with advancing age have senescent properties
We examined whether CD56+ T cells represent a senescent subset of lymphocytes in vivo by the coexpression of CD56 with molecules associated with cell senescence, namely, p16, p53, and Bcl2. p16 is a direct inhibitor of mitosis that leads to irreversible arrest of cell division (35), whereas p53 is an indirect mitotic inhibitor through signaling of p21 expression (36). A role of Bcl2 in senescence could be inferred from findings about correlation between high levels of its expression and resistance to programmed cell death of CD28null T cells (41) and other senescent somatic cells (37, 50, 51).
Fig. 3 A shows typical cytograms demonstrating high levels of expression of p16, p53, and Bcl2 in CD56+ T cells of older adults ≥70 years of age. Coexpression of these molecules was found at frequencies of ≥30% of T cells in 50% of all elderly subjects examined. In young persons aged 18–25 yr (data not shown), Bcl2 was detected in less than half magnitude, and p16 and p53 were negligible.
To further examine senescence of CD56+ T cells, we performed proliferation assays. Elderly subjects who carried ≥30% CD56+ T cells were identified from the study cohort (Fig. 2,A). PBMC were loaded with CFSE and cultured with anti-CD3, irradiated allogeneic cells, and rIL-2. Fig. 3 B shows a typical proliferation profile of the entire T cell population following polyclonal activation. As expected, unstimulated T cells showed a single peak of high CFSE fluorescence. Following stimulation, about half of the T cells showed a spectrum of lower fluorescence intensities indicative of active cell division, but the other half retained the high fluorescence peak indicating a cell subset that underwent little or no proliferation. Phenotypic analysis showed that most CFSEhigh nondividing cells were CD56+ and CD28null T cells (37.4 and 35.8%, respectively). The CFSElow actively dividing cells were mostly CD28+ and CD56neg T cells (46.4 and 44.9%, respectively). These findings were consistent for five independent experiments.
CD56-mediated TCR-independent activation of CD56+ lymphoblastoid cell lines
We began to assess the functionality of CD56 by screening lymphoid cell lines to identify suitable cellular models. We found two lines, JT and NK92 (38, 39), that expressed high levels of CD56 (data not shown). Fig. 4 A shows results of adhesion assays using these cell lines. CD56+CD3neg NK92 cells adhered to an anti-CD56 substrate, but not to anti-CD3 or to IgG. CD56+CD3+ JT cells also did not adhere to anti-CD3, but adhered strongly to anti-CD56. Furthermore, CD56-mediated adhesion of JT cells was unaffected by the addition of anti-CD3 as expected since CD3 itself is not an adhesion molecule. Parenthetically, these adhesion assays were carried in a precoating concentration that we previously established for plate-immobilized anti-CD3 to optimally activate T cells without the need of costimulation (42, 52). These assays showed that the overall degree of CD56-mediated, CD3-independent adhesion of JT and NK92 cells was statistically significant (ANOVA; p < 0.001). Comparison of CD56-mediated adhesion to that of either IgG or anti-CD3 was also significant (p < 0.05 by t test or Mann-Whitney U test).
To evaluate whether adhesion of CD56+ JT and NK92 cells to specific Abs led to cellular activation, we examined the patterns of protein phosphorylation. Fig. 4, B and C, shows the results of immunoblotting for phospho-serine/threonine and phospho-tyrosine, respectively. Cells incubated in anti-CD56 alone had an overall higher level of phosphorylation compared with those cells incubated in the IgG isotype control. Several phosphoprotein bands (labeled as proteins “a” to “k” on the immunoblots) showed higher or lower intensity, indicating either increased phosphorylation or dephosphorylation, respectively, following ligation of CD56. For instance, phospho-serine/threonine band “a” that has a calculated size of ∼82 kDa (Fig. 4,B), was detected with higher intensity in immunoblots of cells incubated in IgG compared with cells incubated in anti-CD56 (t test, p < 0.05). Conversely, phospho-tyrosine bands “g” and “k” that have similar calculated sizes of ∼30 kDa (Fig. 4 C) had low or negligible amounts in unstimulated (IgG-incubated) cells, but were found in significantly higher amounts in cells stimulated with anti-CD56 (t test, p < 0.05).
Additionally, particular phosphoprotein bands differentiated JT cells from NK cells. For example, phospho-serine/threonine band “b”, which has a calculated size of ∼63 kDa (Fig. 4 B), was found in anti-CD56-stimulated, but not in IgG-incubated, JT cells. A similar size phospho-serine/thronine band was already present in unstimulated (IgG-incubated) NK92 cells, which then showed increased intensity following CD56 ligation (t test, p < 0.05).
For JT cells, significant differences in the patterns of phosphorylation elicited by ligation of CD56 and CD3 were observed. This was most evident in phospho-tyrosine blots in Fig. 4 C. Whereas CD3 ligation induced extensive phosphorylation as expected, the resulting phospho-tyrosine pattern was distinct from that seen with cells stimulated through CD56. For example, phospho-tyrosine bands “h” and “k” were found predominantly in CD56-stimulated cells. These phospho-tyrosine bands were found in negligible amounts in CD3-stimulated cells (t test, p < 0.05). Coligation of CD3 and CD56 resulted in an overall pattern of tyrosine phosphorylation that combined the patterns seen with the independent ligation of CD3 and CD56, rather than an amplification of the CD3 pattern. Instead, CD3 coligation appeared to be the amplifier of the CD56 pattern. This was exemplified by phosphoprotein bands “h” and “k” that showed increased intensity with CD3 coligation over that specifically induced by CD3 ligation alone (p < 0.05). These tyrosine (and the above serine/threonine) phosphorylation patterns were observed consistently in five independent experiments conducted.
CD56-mediated activation of primary CD56+ T cells
To validate the above observations with NK92 and JT cell lines, we conducted similar assays using primary NK cells and T cells freshly isolated from blood. For these experiments, we used CD56+-enriched cell preparations that were isolated by negative selection techniques. We had found empirically (data not shown) that positive selection either variably activated cells during isolation and/or that the selecting Abs tightly bound cells even after extensive washing and interfere with subsequent immunostaining for phenotypic analysis.
Fig. 5,A shows that freshly isolated NK cells bound anti-CD56 substrates at levels significantly higher levels than IgG substrates (p < 0.05). The levels of adhesion observed were equivalent to the proportion of CD56+ cells in these preparations of mixed NK cell phenotype; that is, ≥42% CD56+ as verified by flow cytometry. Fig. 5 B shows that this adhesion led to activation. Fresh NK cells cultured overnight in anti-CD56 substrates expressed high levels of CD69 expression, whereas those in IgG had little or no expression of this classic activation marker.
For primary T cells, we examined the CD56-mediated activation by flow cytometry-based assays due to the wide range in levels of expression of CD56 (Fig. 2,A). These assays have the advantage of requiring significantly smaller numbers of cells for multicolor analysis (∼0.5–1.0 × 105 CD56+ cells in a mixed population), compared with Western blotting (∼2 × 106 purified CD56+ cells). Fig. 5,C shows the results of anti-phosphotyrosine fluorescence assays. Cross-linking of CD56 with specific Abs resulted in high levels of phosphotyrosine. Reminiscent of phosphotyrosine immunoblotting results using the JT cell line in Fig. 4,C, phosphotyrosine fluorescence levels of primary CD56+-enriched T cell preparations cross-linked with anti-CD56 alone were equivalent to those seen for cells cross-linked with anti-CD3 (Fig. 5 C). Combined cross-linking of CD56 and CD3 also resulted in similar levels of phosphotyrosine fluorescence, indicating neither additive or amplification effect of coligation.
As expected, cells incubated in the IgG isotype control (followed by the secondary fluorochrome-conjugated anti-mouse Ig cross-linker) showed low or negligible phosphotyrosine fluorescence. The levels of CD56- and/or CD3-mediated anti-phosphotyrosine fluorescence were orders of magnitudes above that seen in the IgG-incubated cells (paired t test; p < 0.001). Specificity of these phosphotyrosine fluorescence measurements were validated by two sets of controls. Primary T cells incubated in hydrogen peroxide, a nonspecific inducer of phosphorylation (46), showed very high levels of phosphotyrosine fluorescence independent of receptor cross-linking, and control T cells loaded with phosphotyrosine showed negligible fluorescence.
We also examined whether ligation of CD56 on primary T cells led to overall cellular activation. Fig. 5 D shows the induction of CD25 and CD69 on CD56+-enriched CD4 and CD8 T cell subsets following overnight incubation in anti-CD56. The levels of CD25 and CD69 expression were equivalent to those seen with primary T cells stimulated with anti-CD3 alone or with both anti-CD3 and anti-CD56. As expected, cells incubated in the IgG control did not show any perceptible expression of either CD25 or CD69.
CD56-mediated activation of cell survival pathways
Protein phosphorylation and CD25/CD69 induction by the ligation of CD56 without the need of CD3/TCR coligation (Figs. 4 and 5) suggested a distinct CD56 signaling pathway. In empirical protein array experiments (data not shown), we found that lysates of CD56+ T cells incubated in anti-CD56 alone contained activated signaling substrates linked to the NF-κB pathway and to the Bcl2 family. To verify these observations, immunoblotting studies were conducted.
Fig. 6,A shows that CD56+ JT cells incubated in anti-CD56 resulted in an overall reduced amount, if not the complete loss, of I-κbα, the inhibitor of NF-κB (48), relative to that seen with cells incubated in IgG. There was also an increase in the amounts of p65 and p52. To verify that these results were indicative of NF-κB activation, gel shift assays were conducted. As shown in Fig. 6 B, nuclear extracts of cells stimulated with anti-CD56 contained proteins that bound an oligonucleotide containing the NF-κB-binding sequence (48). These DNA-protein complexes contained the p65 subunit as shown by supershifts in reactions containing anti-p65 Ab, but not IgG. The amount of NF-κB found in the nucleus following CD56 stimulation was comparable to that seen with cells stimulated with TNF-α, a known activator of NF-κB (48). Nuclear accumulation of NF-κB following CD56 ligation, or exposure to TNF-α, was specific. There were no differences in the levels of Sp1, a ubiquitous transcription factor (47), between activated and unstimulated cells.
Fig. 6 C shows the results of Western blotting for two Bcl2 family members. Cells incubated with anti-CD56 contained the 26-kDa monomer of Bcl2, as well as an increased amount of the 52-kDa dimer, the active form of Bcl2 that confers protection from apoptosis (53). Moreover, there was a concomitant reduction in the amounts, if not the complete loss, of the proapoptotic molecule Bax (54). Densitometric analysis of immunoblots showed a Bcl2:Bax ratio of 3:1 to 5:1 for cells stimulated with anti-CD56 compared with a 1:1 ratio for cells incubated in IgG (p < 0.05).
CD56-mediated production of humoral factors by CD56+ T cells
We examined the ability of T cells to produce cytokines in response to CD56 ligation. Fig. 7 A shows that CD56+ JT cells produced IL-2, IL-8, IL-13, MIP-1β, and TNF-α when incubated in anti-CD56 alone. CD56-mediated production of these cytokines was significantly higher than cells incubated in anti-CD3 alone. However, this CD3-dependent cytokine production was increased with the coligation of CD56. CD56-mediated production of IL-2, IL-8, IL-13, MIP-1β, and TNF-α by JT cells was specific, as there were no detectable amounts levels of IL-4, IL-6, and IL-17. Coligation of CD56 and CD3 also did not induce the latter three cytokines. Data shown were results of a 48-h assay and were not substantially different from 24-h assays.
To validate the above results, cytokine assays were conducted using CD56+ T cells freshly isolated from blood. Fig. 7 B shows that indeed CD56 ligation on these fresh CD56+ T cells was sufficient to induce production of IFN-γ, IL-2, IL-8, IL-13, MCP1, MIP-1β, and TNF-α. The overall levels of CD56-mediated production of these molecules were comparable to that seen in cells incubated in anti-CD3 alone. Incubation of CD56+ T cells in both anti-CD56 and anti-CD3 showed a somewhat additive effect in the production IFN-γ and TNF-α, but did not further increase the production levels for the other humoral factors. Ligation of CD56, CD3, or both on fresh CD56+ T cells did not induce production of IL-4, IL-6, or IL-17. The overall patterns of cytokine production by freshly isolated CD56+ T cells were statistically significant (p < 0.001, ANOVA).
Increases in the expression of NK-related receptors, such as CD56, on T cells of elderly persons relative to younger persons have been reported (55). In the present work, we document, in a cross-sectional analysis, the age-dependent increase in the numbers of CD56+ and CD28null T cells (Fig. 2,A). Although it remains to be investigated whether the gain of CD56 and the loss of CD28 are regulated by mutually exclusive, or coordinated, machineries, our data (Fig. 2,B) show positive regression of the frequencies of CD56+ and CD28null T cells with advancing age. Furthermore, we show that CD56+CD28null T cells are unequivocally derived from the repeated stimulation of CD56−CD28+ precursors in vitro (Fig. 1) (40). De novo expression of CD56 accompanies the loss of CD28 expression, a well-documented feature of T cells that are approaching the end stages of senescence (25, 26). Concordance of these in vivo and in vitro studies underscores that counterregulation of the expression of CD28 and CD56 is an important biological feature of the aging immune system. We have reported that the irreversible loss of CD28 expression on T cells is due to inactivation of the transcriptional initiator of the CD28 gene promoter through the inhibition of complex formation of the relevant trans-acting factors (56, 57). The basis for CD56 induction in T cells is unknown at this time. However, its expression in various leukemic cells has been associated with abnormally high levels of expression of unusual isoforms of the Runx1 transcription factor that drive CD56 gene promoter reporter constructs (58). Whether this Runx1 regulatory machinery regulates CD56 induction as T cells undergo senescence or during chronologic aging remains to be examined.
A previous study (9) reported that T cells of elderly persons have an overall decreased proliferative capacity compared with younger persons. Reduced T cell proliferation was thought to be associated with the loss of activation-related Ags. In the present work, we characterize a discrete, nondividing subset consisting predominantly of CD56+CD28null T cells (Fig. 3,B). We suggest that the large numbers of CD56+CD28null T cells (Fig. 2,A) likely contributed to the finding for the overall reduced rate of mitosis in that previous study (9). More importantly, our data (Fig. 3,A) show that these in vivo-derived CD56+CD28null T cells also express high levels of p16 and p53, two molecules known to promote cell senescence (35, 36). These in vivo findings recapitulate the senescent properties of CD56+ T cells generated in vitro (Fig. 1) (40). All these observations are consistent with reports that CD28null T cells are progeny of repeatedly activated precursors and have shortened telomeres and inactive telomerase that render them mitotically restricted (59, 60, 61).
Supporting the idea for their directional development from CD56negCD28+ precursors, CD56+CD28null T cells have clonotypic TCRs (Figs. 1 and 2,C). Although the antigenic specificities of CD56+ T cells in vivo remain to be examined, TCR specificities of oligoclonal senescent CD28null T cells have been documented to include persistent viruses such as CMV, suggesting a role for Ag in driving T cell senescence in vivo (62). Nevertheless, it is possible that senescent oligoclonal T cells may emerge without specific Ag-driven selection. Whereas anti-CD3 stimulation is a nonselective means of driving proliferation, we consistently observed that the resulting senescent CD56+CD28null T cells have restricted TCR-αβ usage (Fig. 1). Although the basis for this restricted TCR chain usage despite polyclonal CD3 activation is not yet known, we have reported that normal chronologic aging is accompanied by an overall severe contraction of TCR diversity (5). This is due to large populations of oligoclonal cells and to an overall increase in steady-state proliferation of T cells. Presumably, the latter is mediated by homeostatic cytokines derived from T cells, such as IL-2 and IL-15 (63). We and others have shown that IL-2 is required for the survival of senescent T cells (25). A specific role for IL-15 in the generation and maintenance of CD8+CD28null T cells has also been reported (64). Thus, it is likely that homeostatic cytokines and a lifetime of antigenic challenge, in concert, account for in vivo accumulation of senescent clonotypic CD56+CD28null T cells with age (Fig. 2, A and C).
Consistent with the general idea that senescence imparts novel functions to nondividing differentiated somatic cells (35), our data show that CD56 is a functionally competent receptor on T cells. Note that although CD56 is a well-recognized NK receptor, its relevance for NK cell function remains to be examined (28). Our data demonstrate that CD56 is an important adhesion molecule for both NK and T cells (Figs. 4,A and 5,A). CD56-mediated adhesion is also sufficient to elicit cellular activation as evidenced by extensive protein phosphorylation (Figs. 4, B and C,and 5,C). For T cells, we find that CD56 ligation produces a phosphorylation pattern that is distinguishable from that produced by ligation of the TCR/CD3 complex. This is highlighted by the differential induction of particular phosphoprotein bands. Coligation of CD56 and TCR/CD3 results in a phosphorylation pattern that combines those seen with the independent triggering of either receptor (Fig. 4 C). These observations suggest a distinct CD56 signaling pathway.
CD56-mediated TCR-independent activation of T cells is further indicated by the parallel down-regulation of Bax and up-regulation of Bcl2 (Fig. 6,C) in response to Ab ligation of CD56. Because CD56+ T cells already express high levels of Bcl2 (Fig. 3,A), this inverse expression of Bax and Bcl2 suggests an autoregulatory role of CD56 itself in promoting persistence and accumulation in vivo of CD56+ T cells (Fig. 2 A).
Functional relevance of CD56 in T cells is further demonstrated by the induction of the classic activation markers, CD25 and CD69 (Fig. 5, B and C). CD56 triggering also activates NF-κB (Fig. 6, A and B), a transcription factor that regulates expression of various genes, including Bcl2 family members and various humoral factors (65, 66). Indeed, our data show that CD56 triggering in either the JT lymphoid cell line (Fig. 7,A) or in freshly isolated blood CD56+ T cells (Fig. 7 B) is sufficient in eliciting production of NF-κB-regulated cytokines such as IL-2, IL-8, IL-13, IFN-γ, and TNF-α, and the chemokines MCP1 and MIP1β. The levels of CD56-mediated cytokine production are comparable to those seen with TCR triggering, albeit CD56 could also costimulate the TCR in the production of the same cytokines. These data show that CD56+ T cells are functionally versatile immune effectors.
Whether senescent CD56+ T cells are endowed with NK-like cytotoxic activity is not yet known. We note, however, that in vitro-generated, alloreactive, senescent CD4+ T cell clones that kill targets in a nonspecific manner have been reported (67). There is also mounting evidence for discrete subsets of CD4+ T cells expressing a variety of killer Ig-like receptors (KIRs)4 with aging (24). Some of these KIR+ T cells have been demonstrated to kill targets without MHC-restriction like NK cells do (68). Other KIR+ subsets have also been shown to exhibit MHC-restricted cytotoxicity just like classical T cells (68, 69). Whether KIRs and CD56 use similar or different cytotoxic mechanism(s) in senescent T cells remains to be investigated.
Similar CD56+ T cells in the intestinal mucosa have been recently reported (70). Whether these mucosal CD56+ T cells are derived from the same precursors as the circulating CD56+ T cells we report here (Fig. 2,A) is unclear at this time. However, studies indicate that mucosal T cells have different developmental origins and functions (71, 72). Mucosal CD56+ T cells appear to have a regulatory function by modulating CD2-dependent responses of other mucosal T cells through direct contact (70); however, it remains to be investigated as to whether CD56 itself is the triggering receptor responsible for this modulatory activity. Nonetheless, it is noteworthy that mucosal CD56+ T cells also have limited capacity for cell division and are important sources of IFN-γ similar to the circulating CD56+ T cells we report here (Figs. 3,B and 7 B). It would be of interest to examine whether the reduced mitotic capacity of mucosal CD56+ T cells could also be associated with senescence indicators, such as p16/p53 expression and telomere loss (35, 36, 59), or, alternatively, could be due to an intrinsic defect of TCR signaling unrelated to senescence of these mucosal cells. Irrespective of the lineages of CD56+ T cells in the mucosa and in circulation, these observations are consistent with the idea that induction CD56, an NK receptor, in T cells confers novel effector functions.
In summary, our data show that circulating CD56+CD28null T cells are quintessential senescent lymphocytes that accumulate in vivo with advancing age. Despite their limited, or complete lack of, mitotic capacity, they are functionally active. They are capable of eliciting CD56-mediated immune cascades in a TCR-independent manner or by coligation with the TCR. In the context of aging, this versatility of CD56+ T cells is advantageous. Where TCR signaling is intact, CD56 could amplify Ag-specific immune responses. And where TCR-driven immune cascades are defective, such as those reported for certain CMV-specific clonotypes in elderly persons (73), a TCR-independent CD56-mediated pathway could rescue immune responsiveness.
TCR-independent activation of T cells, however, could have a downside. We have recently reported the accumulation of similar senescent CD56+ T cells among patients with rheumatoid arthritis (40). In this case, there is an even higher frequency of CD56+ T cells among patients disproportionate with age, and that these cells colonize rheumatoid lesions. Since the incidence of autoimmune disorders, like rheumatoid arthritis, and other chronic illnesses increase with chronologic age (74), it will therefore be of interest to examine whether the relative abundance of CD56+ T cells in vivo (Fig. 2,A) and/or the types, levels, and interactions of the cytokines they produce (Fig. 7) influence health outcomes of normal aging. One scenario is whether there may be further subsets of these cells; one subset perhaps could have pathogenic effects in the setting of chronic disease.
Alternatively, the biological impact of senescent CD56+ T cells may depend on the prevailing physiologic milieu. Epidemiological studies document age-associated, low levels of up-regulation of TNF-α, IL-1, and IL-6 in circulation (75). Serum levels of these cytokines have been variably associated with many age-related maladies, such as frailty, sarcopenia, and dementia, as well as mortality. A key question, then, is whether the prevailing cytokine environment promotes, alters, or modulates the function of CD56+ T cells. Although they have an advantage of combining TCR-independent and TCR-driven immune functions (Figs. 4–7), the beneficial role of CD56+ T cells may depend on whether they could override negative effects of systemic low-grade elevation of cytokines or, alternatively, whether such a cytokine environment favors survival and/or functionality of these cells. We have proposed that CD56+ T cells and other NK-receptor bearing T cells are novel immune effectors in old age (76). Thus, it will be of interest to examine whether variation in the abundance of CD56+ T cells in the elderly population (Fig. 2 A) might indicate a spectrum of immune competence that could influence health outcomes in old age.
We thank Judith Kadosh for assistance in specimen collection; Cristina Iclozan, Zak Koeske, and R. Grant Mueller for technical contributions in the cellular assays; and Dr. Carter Ralphe and Michelle Witt for use of the Odyssey imaging system.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by investigator-initiated research Grants R01AG022379 (to A.N.V.), R01AG023629 (to A.B.N.), and P30AG024827 (to S.A.S.) and by the University of Pittsburgh Cancer and Aging Program (P20CA103730) from the National Institutes of Health, and by the Department of Pediatrics, Children’s Hospital of Pittsburgh. D.T.M. was supported by the Pittsburgh Medical Student Training Program in Aging Research (T35-AG026778). The Children’s Hospital of Pittsburgh Rangos Research Center is a National Center for Research Resources-supported facility (C06-RR14489).
Abbreviation used in this paper: KIR, killer Ig-like receptor.