Age-associated alterations in inducible gene transcription in human CD4⁺ T lymphocytes

Introduction

Human aging is associated with chronic low-grade inflammation marked by increased levels of pro-inflammatory cytokines IL-1, IL-6 and TNFα [1-3]. Elevated levels of these inflammatory markers are powerful risk factors for accelerated decline of physical and cognitive function, and mortality. Because genes encoding these cytokines are targets of the transcription factor NF-κB, it has been suggested that dysregulation of this factor is a central feature of the aging process [4, 5]. The most prominent role of NF-κB is as an inducible transcription factor [6, 7]. In the immune system, NF-κB is activated in response to antigen receptor signals, to Toll-like receptor signals and to a variety of cytokines. Additionally, reactive oxygen species (ROS) are potent activators of NF-κB in diverse cell type [8, 9]. The connection to ROS provides another possible link between NF-κB and aging [10]. Biochemical evidence for alteration of NF-κB during aging was first obtained in rodents where NF-κB DNA binding was shown to be elevated in tissues from older animals [11-14]. More recently, elevated basal NF-κB has also been detected in primary human dendritic cells, endothelial cells, and in skeletal muscle [15-17]. However, the effect of increased basal NF-κB on gene expression has not been examined in most cases.

Surprisingly few studies have coordinately examined the effect of age on NF-κB induction and NF-κB-dependent inducible gene expression. In their pioneering studies, Ponnapan and colleagues showed that TNFα-induced NF-κB DNA binding activity was impaired in T lymphocytes isolated from the elderly [18, 19]. However, NF-κB-dependent changes in gene expression were not examined in these studies. More recently, Huang et al. found that CD4⁺ T cells from older mice produced higher levels of pro-inflammatory IL-17 and IL-6 in response to TCR signaling [20]. This effect was reduced by anti-oxidants and pharmacological inhibitors of canonical NF-κB activation, leading to the proposal that NF-κB responses were increased with age. The contrasting observations in these papers could be due to differences in the activating stimuli (TNFα versus anti-CD3 and anti-CD28), differences between murine and human cells, or differences in responses between CD4⁺ and total T cells. Because immune impairment remains a key health issue in the elderly, it is imperative to clarify molecular mechanisms that contribute to age-associated reduction in immune responses.

Several earlier studies, mostly with murine lymphocytes, have shown that TCR signaling is defective in T cells obtained from older mice [21, 22]. These defects include reduced synapse formation and TCR recruitment to plasma membrane rafts, and reduced activation of cytoplasmic MAP kinase signaling. However, few studies have connected the consequences of age-associated TCR signaling defects to changes in inducible gene expression. Han et al. studied the TCR responsiveness of total CD3⁺ T cells from young and old mice [23]. They found that IFNγ was induced to significantly higher levels in cells from old mice, whereas MIP-1α and MIP-1β (CCL3 and CCL4, respectively) induction was somewhat higher in the older cells. Analysis of the CD4⁺ compartment by Yung and colleagues showed that certain chemokines (such as RANTES, MIP-1α and MIP-1β) and chemokine receptors (such as CCR2, CCR5 and CXCR5) were expressed at higher levels in unstimulated cells from old mice [24, 25]. However, long-term stimulation via the TCR showed comparable changes in cells from young or old mice in these studies. Recently, Mirza et al. carried out a comprehensive analysis of inducible gene expression in naïve murine CD4⁺ T cells obtained from young and old mice [26]. Genes and pathways that were affected with age included those associated with cell cycle, cell death, and inflammation. Chemokine genes that were up-regulated with age included CCL3 and 4, CCL1 and 17 and CXCL9 and 11. Cytokine genes that were up-regulated with age included IL-4 and IL-10 and IFN-γ. Despite the variations, these studies demonstrate that TCR signaling defects translate into gene expression defects in T lymphocytes from old mice. Corresponding studies of human aging are lacking. Here we use biochemical assays and genome-wide RNA profiling to provide the first analysis of age-associated alterations in TCR-inducible gene expression in human CD4⁺ cells. Subjects for these studies were selected from the Baltimore Longitudinal Studies on Aging (BLSA), which has strict health criteria for enrollment eligibility. Additionally, cells were obtained from individuals not included in the BLSA but selected according to the same strict criteria for healthy status. Because of the exceptional health status of this population, changes seen in these populations can be confidentially attributed to aging per se.

Time-dependent response to T cell receptor stimulation showed that NF-κB induction was not sustained in CD4⁺ T cells from older donors. Gene expression analysis revealed a dichotomy in the regulation of NF-κB target genes. Activation of most target genes was not sustained in the older cohort, thereby closely recapitulating the biochemical studies. However, a subset of target genes, including IL-6, were hyper-activated in cells from older donors. In addition, activation or repression of numerous other genes was affected in the older cohorts. These observations define the functional consequences of dysregulated TCR signaling that accompany human aging and may, in part, explain the simultaneous occurrence of low grade chronic inflammation coupled with reduced immune responses in the elderly.

Results

We purified CD4⁺ T cells from 31 donors of age range 25-81 years (see Figure S1 and Table S1), and activated them with plate-bound anti-CD3 antibodies for 2h and 4h prior to preparation of nuclear and cytoplasmic extracts, and RNA. As controls, we used cells incubated for 4h at 37° in the absence of anti-CD3. We used anti-CD3 as an activating stimulus rather than IL-1 or TNFα in order to gain insight about age-associated attenuation of antigen-specific immune responses; the times of activation were chosen to limit the analysis to primary effects of TCR signaling. We first analyzed the response of the transcription factor NF-κB to TCR activation. Nuclear extracts were fractionated by SDS-PAGE, transferred to PVDF membranes and assayed for p65/RelA induction by immunoblotting (Figure 1A). To normalize between extracts we examined several ubiquitous proteins such as SP1, β-actin, TBP. We found that all of these were suitable for comparing extracts from one individual, but varied too much to compare between individuals (Figures S2A, S2B). To get an absolute measure of nuclear p65/RelA expression, we used a standard set of p65/RelA-expressing extracts (Figure S2C).

We compared nuclear p65/RelA induction in cells obtained from donors <65 years (Y) to ≥65 years (O). In the younger cohort, TCR activation resulted in a time-dependent increase of nuclear p65/RelA in both the 0-2 hour and 2-4 hour intervals (Figure 1B, Table S2, S3, S4). However, in the older group nuclear p65/RelA levels were not significantly different between 2h and 4h of activation. The same trend was also noted when total nuclear p65/RelA levels were compared between these age groups (Figure S3A). Because Y and O cells induced nuclear p65/RelA equally at 2h but not at 4h, we conclude that NF-κB activation in response to TCR stimulation is not effectively sustained in older individuals. During this activation time-course we did not observe nuclear induction of Rel.

It is well-recognized that CD4⁺ T cell subsets in peripheral blood change with age [27]. Notably, the proportion of naïve CD4⁺ T cells (expressing CD45RA antigen) decrease and the proportion of memory CD4⁺ T (expressing CD4RO antigen) increase with age. These changes stabilize within the third decade of life [28]. Accordingly, we did not observe significant differences in the RA/RO ratio in our cohort (Figure S1E). However, to circumvent possible effects of changing CD4⁺ T cell subsets we also compared NF-κB responsiveness in the ≥65 cohort with cells obtained from 50-64 year old donors. We found that NF-κB induction increased between 2h and 4h of activation even in this middle-aged group (M) (Figures 1C, S3B). We conclude that the observed difference in NF-κB responsiveness between the 50-64 age group and the ≥65 group reflects a true reduction in the response of the CD4⁺ cells obtained from older donors.

To examine the transcriptional response to anti-CD3 treatment, we analyzed RNA obtained from these cells using Illumina microarrays. PAGE analysis of the normalized data sets showed that the ‘NF-κB-induced’ pathway was the most prominent one induced after 2h or 4h anti-CD3 treatment regardless of age (Figure S4A), thereby validating the choice of activation times for functional analysis of NF-κB responses. Within the NF-κB induced pathway, cytokine and chemokine genes were amongst the most highly activated genes at both time points, and in both subject groups (Figure S4B, S4C). The pattern of nuclear RelA induction predicted that NF-κB-dependent gene expression would continue to increase between 2h and 4h of activation in the younger, but not older, participants. To test this, we evaluated pathway changes that occurred between 2h and 4h of activation. Though, 17 out of the top 20 pathways that increased between 2h and 4h of activation remained the same in the two age groups, we found that the ‘NF-κB induced’ pathway increased significantly only in the young group (Figure 2A) but not the old group (Figure 2C). Pathway analysis comparing 50-64 year old donors to the ≥ 65 group showed the same pattern (Figure 2B). We conclude that NF-κB transcriptional function is not sustained in activated CD4⁺ T cells from the elderly, thereby correlating closely with the biochemical assays of NF-κB induction.

To critically test the idea that NF-κB activation was not sustained in the elderly, we identified putative NF-κB target genes that were induced between 2h and 4h of activation. We found that in the younger and middle-age cohorts, 18 genes within this pathway were significantly up-regulated between 2h and 4h, whereas only 10 genes were significantly up-regulated in the older group (Figure 2A-C). Prominently missing from the older group were genes encoding CXCL10, TNFAIP2, CCL2, CXCL5, and IL-1α. Moreover, several genes that continued to be up-regulated in the older group had lower Z ratios than the same genes in the younger group. These genes included CXCL11 (Z-ratio(O) = 8.02; Z-ratio (Y) = 10.54); GBP1 (Z-ratio(O) = 6.14; Z-ratio(Y) = 8.51), and SERPIN B2 (Z-ratio (O) = 3.6; Z-ratio (Y) = 4.51). Thus, more than half of “NF-κB-induced” genes that were up-regulated between 2h and 4h in the younger group were attenuated in the older group. Together, this set of genes identified the transcriptional consequences of the lack of continued NF-κB activation in response to TCR stimulation in CD4⁺T cells from older donors.

Interestingly, not all NF-κB target genes were reduced in expression between 2h and 4h of activation in the older cohort. Of particular interest, expression of IL-6, CXCL1, CXCL11, and IL-15 continued to increase in CD4⁺ T cells from older individuals. Continued up-regulation of these genes in the absence of sustained NF-κB induction revealed regulatory differences between classes of NF-κB target genes. While expression of many target genes correlated well with NF-κB inducibility, others deviated considerably. This dichotomy was especially apparent in samples from older donors. We note that many of the discordantly regulated genes have been previously associated with aging. Serum IL-6 levels increase in the elderly [2, 3], yet several genome-wide analyses of aging tissues from humans and mice failed to uncover the source of systemic IL-6 [29-34]. Our observations suggest that inefficient down-regulation of IL-6 during T cell immune responses in the elderly may be one source of age-associated IL-6. Along with IL-6, CXCLl is one of the cytokines associated with senescence-associated secretory program (SASP) [35, 36]. It is intriguing that TCR activation of CD4⁺ T cells from older individuals induces some aspects of SASP. Finally, elevated IL-15 levels have been noted in age-associated osteoarthritis and very old individuals [37, 38].

To more broadly identify age-associated changes in response to TCR signaling we compared the pattern of inducible gene expression in cells from the Y and O age groups activated for 2h or 4h. At both activation time points, many genes were differentially expressed in the two groups; the top 50 genes up-or-down regulated in the ≥ 65 group compared to the ≤ 65 group are shown in Figure 3. Only a small proportion of these genes fell in the NF-κB induced category showing that many other pathways were also affected in cells from older donors. Putative NF-κB target genes that were consistently up-regulated in cells from older individuals included the cytokines IL-1α and IL-6. Additionally, up-regulation of GZMH, which is normally restricted to NK and CD8 cells, in cells from older individuals was interesting from the perspective that aged lymphocytes are known to express NK cell-specific cell surface receptors [39, 40]. Thus, older CD4 ⁺ T cells express several aspects of the NK cell gene program. The combination of NK receptors and GZMH may permit cytolytic activity of such cells.

We further probed the nature of age-associated changes in TCR inducible gene expression. One perspective was provided by investigating the induction characteristics of genes whose expression was significantly different between Y and O cells after 2h activation (Figure 4, 2O-2Y). Amongst genes that were over expressed in the O group after 2h anti-CD3 treatment (Figure 4, 2O-2Y), we found 23 genes that were significantly induced in both Y and O groups (Figure 4, red box); our interpretation is that these genes are hyper-induced in the O group. Conversely, amongst genes whose expression was attenuated in the O group after 2h of anti-CD3 treatment, we found 12 genes that were repressed in both groups (Figure 4 green box); we infer that these genes are hyper-repressed in the O group. Additionally, we found genes that were induced in Y but not in O and thereby fell in the down-regulated category in a comparison of O and Y after 2h activation (Figure 4, blue box). Conversely, some genes that were turned down in response to TCR signaling in Y but not in O fell in the up-regulated category in the comparison of O and Y after 2h activation (Figure 4 yellow box). Thus, age-associated alterations in TCR signaling are reflected in complex transcriptional responses that affect inducible gene activation as well as repression. Similar categories could be seen after 4h anti-CD3 treatment (Figure S5).

Discussion

Our studies provide the first view of age-associated alteration of inducible gene regulation in response to TCR signaling in human CD4⁺ T cells. We used short-term anti-CD3 activation to focus on direct consequences of TCR signaling and, in particular, on the effects of age on NF-κB induction and NF-κB dependent gene expression. Using cells obtained from cytapheresis of subjects enrolled in the Baltimore Longitudinal Study on Aging, we show that NF-κB RelA nuclear translocation was induced similarly in CD4⁺ T cells from young and old subjects, but induction was not sustained in cells from older individuals. The biochemical data was reflected in reduced induction of putative NF-κB target gene transcription in the older cohort. However, mRNA levels of a subset of NF-κB target genes continued to increase in the elderly even after nuclear RelA levels reached a plateau. These observations indicate that age-associated NF-κB dysregulation differentially impacts different sets of NF-κB target genes. Importantly, we found that NF-κB target genes that were over-expressed in the older group included genes that have been implicated in age-associated chronic inflammation such as IL-6, IL-1 and IL-15, and those involved in cellular senescence such as CXCL1. In the course of the biochemical studies, we also showed that standard normalization proteins, such as actin or the transcription factors SP1 and TBP, vary sufficiently between subjects that they cannot be used to compare between individuals.

Although our studies were conducted with a mixed population of CD4⁺ naïve and memory cells, we believe that the age-associated changes noted here do not reflect differences in responsiveness of naïve and memory cells. Firstly, the changes in putative NF-κB-dependent genes were apparent in a comparison between middle-aged and older donors where the population of naïve and memory cells are very similar. Secondly, the genes that we propose to be differentially regulated with aging were not identified as being differentially expressed in naïve and memory CD4⁺ T cells in earlier studies [41, 42]. We therefore propose that dysregulation of genes such as IL-1, IL-6 and IL-15 reflect true age-associated changes in CD4⁺ T cell responses. Whether these changes occur in one or both subsets of CD4⁺ T cell remains to be determined.

We propose that the differential effect on NF-κB target genes may, in part, explain the co-existence in older individuals of a background mild pro-inflammatory state and reduced ability to mount inflammatory immune responses. We suggest that those NF-κB target genes that require continued NF-κB activation contribute to effective immune response, for example, via activation of chemokines and their receptors. Reduced expression of such genes would result in impaired immunity in the elderly. At the same time, some NF-κB target genes such as IL-6, IL-1, and IL-15 appear not to be exclusively dependent on NF-κB levels and, in some cases, are even hyper-induced in cells from older individuals. We suggest that this form of dysregulation may contribute over time to chronic inflammation seen in the elderly. An important corollary of the hypothesis is that understanding the regulatory differences between the two kinds of NF-κB-dependent gene transcription may provide insights into the origins of age-associated chronic inflammation. Additionally, our observations implicate CD4⁺ T cells as one source of age-associated inflammatory cytokines.

The majority of genes that were differentially expressed in activated cells from older individuals were not classifiable as NF-κB target genes. Clearly, inducible gene expression mediated by transcription factors other than NF-κB is also affected by age. The altered patterns of gene expression arose in different ways. For example, genes that were up-regulated in O compared to Y after 2h activation included genes that were induced to higher levels in O than in Y (hyper-induced) or genes that were inducibly suppressed in Y but not in O. Conversely, genes that were down- regulated in O compared to Y after 2h activation included genes that were repressed more in O than in Y (hyperrepressed) or genes that were activated in Y but not in O. The aggregate of these different kinds of changes results in widespread alterations of gene expression in TCR-activated CD4⁺ T cells from older individuals. It will be important to identify which changes are most important in contributing to the overall attenuation of CD4⁺ T cell responses in the elderly.

Finally, it is interesting to compare our results in human CD4⁺ T cells with those of earlier studies that examined age-associated inducible gene expression in murine T cells, particularly with regard to inflammation-related genes. Despite the use of different T cell populations, up-regulation of the chemokines CCL3 and CCL4 and the cytokine IFN-γ was noted in several of the murine studies [23, 24, 26]. Additionally, up-regulation of the cytokines IL-4, IL-10, and IL-17 was noted in a different subset of mouse studies [20, 26]. However, these genes did not score as being differentially expressed in human CD4⁺ T cells from older individuals at least at the activation time points used here. Instead, we found IL-1 α, IL-6 to be prominently up-regulated in human CD4⁺ T cells from older individuals. A common feature between the mouse and human studies was the differential up-regulation of CXCL9 and CXCL11 in the older cohorts. These differences highlight the importance of continued analysis of human samples for the study of aging.

Methods

Supplementary Materials

Supplemental File