Human CD4⁺ Memory T Cells Are Preferential Targets for Bystander Activation and Apoptosis¹

Tcells of heterologous specificity, or bystander T cells, were once thought to make a major contribution to the immune response to infection; however, key studies in the late 1990s revealed that, in fact, the vast majority of responding T cells were specific for the invading pathogen (1), and the contribution of bystander T cells was minimum (2, 3). Today, the idea of a bystander mechanism of T cell activation remains a highly controversial one. The very concept seems at odds with the otherwise highly specific and heavily regulated responses that are characteristic of the adaptive branch of the immune system. The existence of bystander activation has considerable implications, not only for the normal functioning of the immune system, but also as a contributor to pathological conditions seen in diseases such as HIV and autoimmunity (reviewed in Ref. 4).

In this study, bystander T cell activation is defined as T cell activation to produce phenotypic or functional changes via a mechanism that is independent of specific TCR stimulation. There is much evidence that cytokine stimulation (5, 6, 7, 8, 9, 10, 11, 12, 13) or cross-linking of certain membrane-bound receptors (14, 15) is capable of activating T cells in the absence of TCR ligation. Few studies, however, address whether physiological levels of these cytokines produced during an immune response do actually induce bystander T cell activation. The vast majority of data has to date been acquired from murine model systems (2, 5, 6, 16, 17, 18, 19). Importantly, those studies that have been conducted in in vivo models using heterologous viral infections have often failed to rule out a contribution from TCR cross-reactivity, which may exist even between apparently unrelated peptide Ags (20). Thus, despite the capability of exogenous cytokine alone to stimulate T cells, it may be argued that this is of little relevance, because Ag-specific activation is responsible for the majority of the response to infection, with the remainder attributable to cross-reactive TCR responses.

It was the aim of the current study to develop an in vitro system to allow us to investigate the existence of bystander T cell activation, building upon previous work in the field to answer questions about whether this occurs in human T cells, and in the absence of the possibility for cross-reactivity.

In this study, we have shown that, in our system, bystander activation, mediated by soluble factors, does occur within the human T cell population, preferentially affecting the memory CD4⁺ T cell subset. Resting, directly activated, and bystander-activated T cells displayed distinct gene expression profiles. Furthermore, cells activated via a bystander mechanism undergo an elevated level of apoptosis in comparison with their resting counterparts. These findings provide an insight into the possible mechanisms behind attrition of existing memory populations during heterologous infections, as has been observed to occur in elderly CMV carriers (21).

PBMC were isolated from healthy donors by collection in heparin, or obtained from the National Blood Service as buffy coats, and separated by density gradient centrifugation using Lymphoprep (Axis Shield).

Cells were washed once in PBS containing 1% FCS and 0.1% sodium azide (FACS wash), and incubated for 20 min on ice with a combination of Abs from CCR7 allophycocyanin (R&D Systems), CD25 FITC (DakoCytomation), CD3 allophycocyanin (BD Biosciences), CD38 allophycocyanin (BD Biosciences), CD4 allophycocyanin (BD Biosciences), CD45RO FITC (DakoCytomation), CD69 allophycocyanin (BD Biosciences), CD69 FITC (DakoCytomation), Vβ13.1 PE (Serotec; Beckman Coulter), and Vβ17 PE (Beckman Coulter). Cells were washed again in FACS wash, and resuspended in PBS containing 1% formaldehyde (FACS fix). Annexin V staining was conducted according to the manufacturer’s instructions. Briefly, cells were stained with Ab, as described, washed in PBS, resuspended in annexin V buffer (BD Biosciences), and incubated with annexin V (BD Biosciences) for 15 min at room temperature before immediate acquisition without fixing. Data were acquired on a FACSCalibur flow cytometer or on a CyAn ADP flow cytometer, and analyzed using Summit software (DakoCytomation). For cell sorting (by MACS or FACS), cells were washed with RPMI 1640 medium supplemented with 100 μg/ml penicillin, 100 U/ml streptomycin, 2 mM glutamine, and 10% FCS (R10) before and after incubation on ice with Abs. For flow sorting, stained cells were passed through Cell Trics filters (Partec) to ensure a single-cell suspension before sorting on a Moflo machine (DakoCytomation).

PBMC were washed three times in warm sterile PBS, and resuspended at 2 × 10⁷ cells/ml in PBS. CFSE (Sigma-Aldrich) was dissolved in DMSO to a stock concentration of 5 mM and stored in single aliquots at −20°C. The stock solution was diluted to a working concentration of 5 μM in PBS, and an equal volume of this working solution was added to the cell suspension, resulting in a final concentration of 2.5 μM CFSE. After thorough mixing, the cell suspension was incubated at 37°C for 10 min, with frequent shaking. The labeling reaction was then quenched with the addition of ice-cold FCS for 1 min. The mixture was topped up to 10 ml with cold sterile PBS, and centrifuged at 1500 rpm for 5 min. CFSE-labeled PBMC were then washed a further three times in warm R10, and the resuspended cells were transferred into a clean tube after each wash. After the final wash, cells were counted and set up in culture.

MACS microbeads were used according to the manufacturer’s instructions. Briefly, PBMC were stained for Vβ13.1 or Vβ17 PE. Stained cells were washed twice in cold R10 and the supernatant was aspirated off, before resuspension in MACS buffer at 10⁸ cells/0.8 ml. MACS anti-PE microbeads (Miltenyi Biotec) were added at 0.2 ml/10⁸ cells, and the cell suspension was incubated at 4°C for 15 min. Cells were washed in MACS buffer, and applied to MS columns. The Ab-positive fraction was isolated on a miniMACS separator, and washed twice in H10 (as R10, but supplemented with 10% human AB serum instead of FCS). Assessment of four samples showed no significant difference in purity of MACS-isolated Vβ13.1 T cells when cultured for 5 days in the presence or absence of staphylococcal enterotoxin B (SEB)⁴ (mean percentage Vβ13.1-positive 86.8 and 87.1, respectively, p = 0.832), indicating no outgrowth of other Vβ family-expressing T cells.

All cell culture was conducted in either R10 or H10. For standard bystander activation assays, PBMC were resuspended for a final concentration of 1 × 10⁶ (for some experiments, 2 × 10⁶ cells/ml were used). SEB (Sigma-Aldrich) was added to a final concentration of 2.5 μg/ml (for some experiments, 5 μg/ml was used). For control unstimulated wells, an equivalent volume of carrier alone was added. Cells were incubated at 37°C in the presence of 5% CO₂ for 5 days, before harvesting and staining for surface markers.

For assessment of direct stimulation over a 5-day period, Vβ13.1 and Vβ17 T cells were isolated by MACS microbead separation, as described above, and cultured in H10 in the presence of 2.5 μg/ml SEB or an equivalent volume of carrier. On day 5, cells were harvested and stained for CD25 and CD69 expression.

For Transwell assays, Vβ13.1 cells were flow sorted and placed in the top well of a Transwell insert (0.4-μm polyester membrane pores, 24-well plate, top well capacity 100 μl; Corning). Whole autologous PBMC were placed in the bottom well at between 2 and 3 × 10⁶ cells/ml with 2.5 or 5 μg/ml SEB or an equivalent volume of carrier. An aliquot of sorted Vβ13.1 cells was resuspended in 100 μl of supernatant taken from day 5 of an unstimulated PBMC culture, with the addition of 2.5 or 5 μg/ml SEB or an equivalent volume of carrier. This served as a negative control for direct activation of purified Vβ13.1 T cells with SEB. Cells were harvested on day 5 and stained for CD25 and CD69 expression.

To assess the relative contributions of naive and memory populations, PBMC were flow sorted into Vβ13.1⁺CD45RO⁺ (memory) and Vβ13.1⁺CD45RO⁻ (naive) populations. The sorted naive and memory Vβ13.1 T cells were stimulated in Transwell, as described above, and cells were harvested on day 5 and stained for CD69 expression. Purified naive and memory Vβ13.1 T cells cultured alone in supernatant taken from day 5 of an unstimulated PBMC culture, in the presence or absence of SEB, served as negative controls. All SEB was batch tested to ensure no direct activation of sorted Vβ13.1 T cells before inclusion in this study.

Vβ13.1⁺ CD25⁺ T cells were flow sorted, and total cellular RNA was extracted using the RNeasy Minikit (Qiagen), according to the manufacturer’s instructions. Reverse transcription was conducted using oligo(dT) primers and avian myeloblastosis virus reverse transcriptase. Negative controls without RNA template and without avian myeloblastosis virus reverse transcriptase were also conducted. Seminested PCR was then conducted using Hotstar Taq polymerase (Qiagen). The first round of amplification used forward primer, 5′-CCA CCA TGG TCC AGT GAA TGC TGG TGT CAC TC-3′, from the V region of the β-chain genes 13.1, 13.4, and 13.6, and the reverse primer, 5′-GGA GAT CTC TGC TTC TGA TGG CTC-3′, from the C region of the β-chain gene. The second round of amplification was conducted using an internal forward primer, 5′-CCA CCA TGC ACA GAG GAT TTC CCG CTC AGG C-3′, and the same reverse primer. Negative controls without template were also conducted. PCR products from the second round of amplification were column purified using the QIAquick PCR Purification Kit (Qiagen), according to the manufacturer’s instructions. The product was then cloned using the TOPO TA Cloning Kit For Sequencing (Invitrogen), according to the manufacturer’s instructions. Colonies were screened by PCR for presence of the correct insert, and successful colonies were grown in 6 ml of Luria-Bertani with 100 μg/ml ampicillin. The plasmid DNA was extracted using the Qiagen Miniprep Kit, according to the manufacturer’s instructions. Samples were sequenced using M13 forward and reverse primers.

PBMC were isolated from buffy coats. Vβ13.1 and Vβ17 T cells were isolated using MACS microbeads, as described previously. Vβ13.1⁺ T cells were plated out in the top well of Transwell inserts (0.4-μm polyester membrane pores, 12-well plate; Corning), at 1 × 10⁶ cells/ml, 500 μl/well, and cocultured with 1.5 ml of whole autologous PBMC plated in the lower well at 1.3 × 10⁶ cells/ml in H10. Vβ17⁺ T cells were plated out in 24-well plates at 1 × 10⁶ cells/ml, 250 μl/well, mixed with 750 μl of whole PBMC at 1.3 × 10⁶ cells/ml in H10. All wells were treated with 2.5 μg/ml SEB. Cultures were incubated for 5 days at 37°C in the presence of 5% CO₂, and then the top wells of Transwell cultures and whole mixed wells were harvested and stained. Cells were sorted by FACS into bystander-activated (CD3⁺, Vβ13.1⁺, CD25⁺), resting (CD3⁺, Vβ13.1⁺, CD25⁻), and directly activated (CD3⁺, Vβ17⁺, CD25⁺) populations, all of which were gated on lymphocytes by forward and side scatter, and subsequently on CD3 allophycocyanin⁺ cells. Cells were collected into R20 medium (as R10, but containing 20% FCS) and kept on ice until RNA could be extracted. As controls for this experiment, residual MACS-isolated Vβ13.1 cells were resuspended in 200 μl of day 5 supernatant from an unstimulated PBMC culture, and plated out at 100 μl/well in a 96-well round-bottom plate with 2.5 μg/ml SEB or an equivalent volume of carrier. These cells were harvested on day 5 and stained for activation markers CD25 and CD69 to confirm that no direct activation of isolated Vβ13.1 cells by SEB had occurred. The same control was performed for the remaining Vβ17 cells to show that direct activation does occur under these conditions if the correct TCR is present. In addition, whole PBMC were subjected to a standard bystander activation assay, as described above.

Total cellular RNA extraction was conducted using the RNeasy Micro Kit (Qiagen), according to the manufacturer’s instructions, omitting the 15-min room temperature incubation with DNase I. RNA was eluted into 14 μl of RNase-free water, and 3-μl aliquots of RNA were separated for quality control testing by the Agilent Bioanalyser 2100 and Nanodrop ND-1000. RNA was transferred immediately for storage at −80°C.

All the procedures and hybridization were performed according to the Genechip expression technical manual (Affymetrix), as previously reported (22, 23). Briefly, a two-cycle cDNA synthesis kit (Affymetrix) was used to convert 50 ng of RNA to double-stranded cDNA. Biotin-labeled cRNA probes were generated by in vitro transcription. Fragmented cRNA (15 μg) was used in a 300-μl hybridization mixture that contained spiked controls (Affymetrix), 0.1 mg/ml herring sperm DNA (Promega), and 0.5 mg/ml acetylated BSA (Invitrogen). A 200-μl aliquot of this hybridization mixture was used on each chip, with a total of 12 high-density oligonucleotide Human HG U133 Plus 2.0 arrays (Affymetrix) used to analyze four biological replicates of three conditions (resting, bystander activated, and directly activated). Chips were incubated at 45°C for 16 h rotating at 60 rpm. Following hybridization, processing of the arrays was conducted on a Genechip Fluidics Station 450, according to the recommended protocols (EukGE-WS2v5; Affymetrix), of double staining and posthybridization washes. Fluorescent images were captured using gene Array Scanner 3000 (Affymetrix). All of the experiments were designed and information was compiled in compliance with Minimum Information About a Microarray Experiment guidelines (24). This allows the microarray data to be correctly interpreted and independently verified.

Data analysis was performed by GeneSpring 7.3 (Agilent Technologies) and the R environment for statistical computing (25) using the Affy and Limma packages (26). Affymetrix.cel files were preprocessed with Robust MultiChip Analysis (27). For each gene, the normalized intensity in each bystander-activated or directly activated sample was paired with the normalized intensity in the corresponding resting sample and analyzed with Limma. A linear model was fitted to a paired sample design matrix using Limma, and differentially expressed transcripts identified for each pairwise comparison using a moderated t statistic, which employs an empirical Bayes method to borrow information across genes. Values of p were adjusted for multiple testing using Benjamini and Hochberg’s method (28) to control the false discovery rate.

For pathways analysis, data were analyzed through the use of Ingenuity Pathways Analysis (Ingenuity Systems, www.ingenuity.com). Briefly, a dataset containing gene identifiers and corresponding log ratios of expression (M values) was uploaded into the application. A B value (log odds of differential expression) of >3 and p value of <0.005 were set to identify genes whose expression was significantly differentially regulated. Genes that met these criteria were used in the generation of networks and pathways, and in functional analysis.

The array data have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) 6 and are accessible through GEO Expression Omnibus Series accession number GSE13738 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13738).

Data in charts are means with error bars representing SEMs. Values of p were generated by two-tailed Student’s t tests.

An in vitro system was designed whereby human PBMC were stimulated with SEB, a superantigen with a well-characterized specificity (29), and responses were monitored among the directly activated population of Vβ17⁺ T cells and the nonspecific or bystander population of Vβ13.1⁺ T cells. There are numerous advantages to this system: a large proportion of the T cell repertoire is directly stimulated by SEB; the Vβ13.1 subset is well characterized as nonresponsive to SEB (29, 30, 31, 32); this nonresponsive bystander subset represents a sizeable proportion of T cells in most donors, facilitating the identification of rare events; finally, and perhaps most importantly, the V region of the TCR β-chain is the sole determinant of superantigen specificity, and as such, the relative promiscuity of the TCR peptide recognition site is irrelevant.

We hypothesized that the bystander activation event would occur secondary to an initial specific activation event, perhaps indirectly via interactions with accessory cells, and may therefore be substantially delayed. To detect bystander activation at the earliest possible time point (which is necessary to avoid deterioration of cells kept in culture conditions in the absence of exogenous growth factors for an extended period), early activation marker up-regulation was selected as the readout.

We sought further confirmation that activation events observed among SEB-stimulated Vβ13.1 T cells were of a genuine bystander nature. Manufacturers of the anti-Vβ13.1 Ab (clone IMMU222) could not rule out binding to Vβ13.4 and Vβ13.6. To exclude the possibility that the bystander events were the result of nonspecific staining, the bystander-activated T cells were cloned and sequenced. Whole PBMC were stimulated with SEB, and bystander-activated CD25⁺ Vβ13.1⁺ T cells were sorted by FACS. RNA from these cells was reverse transcribed and amplified by seminested PCR, using primers designed to amplify, with equal efficiency, the sequences for Vβ13.1, Vβ13.4, and Vβ13.6 (as defined by the sequence alignments generated by Arden et al. (33)). The PCR products were cloned, and 30 colonies were selected for sequencing. As shown in Fig. 1, all corresponded to the Vβ13.1 consensus sequence, with the exception of a few point mutations whose presence can likely be attributed to the use of Taq polymerase, which lacks proofreading 3′-5′ exonuclease activity.

Furthermore, it was confirmed that purified Vβ13.1 T cells were not directly activated by SEB. MACS bead-sorted Vβ13.1 T cells cultured for 5 days with SEB demonstrated no significant up-regulation of activation markers, whereas the positive control Vβ17 population did. Fig. 2^,A shows CD25 expression by these purified T cell populations. By contrast, when whole PBMC were cultured in the presence of SEB, at day 5 of culture, Vβ13.1 bystander T cells displayed significant increases in CD25, as shown in Fig. 2^,B. There was no significant decrease in the Vβ13.1 population size over the course of the 5-day culture period within a subset of samples analyzed (data not shown). Analysis of CD69 expression demonstrated a similar trend, although expression levels were consistently lower than those of CD25 (data not shown). As expected, virtually all Vβ17 T cells in the PBMC population responded to direct SEB stimulation by up-regulating CD25 (supplementary Fig. 1).⁵

An alternative explanation for this observation is that interactions with foreign serum proteins could be responsible for direct activation of the bystander population; however, this seems unlikely, because bystander T cell activation was also observed when the experiment was conducted using only human serum in the medium, as shown in the right-hand panel of Fig. 2 B.

We next attempted to determine whether bystander T cell activation elicited any functional responses. Cell division was observed among a small proportion of the Vβ13.1 subpopulation of T cells after stimulation of CFSE-labeled PBMC with SEB in five of seven samples, and a representative example is shown in Fig. 2 C. This failed to reach significance (p = 0.135). These results are indicative of a partial state of activation induced by indirect stimulation of bystander T cells, which may be below the threshold required for reliable induction of functional responses.

To characterize the mechanism by which bystander T cell activation occurs in this system, the bystander assay was conducted in Transwell. Sorted Vβ13.1 T cells were cocultured in Transwell with whole PBMC in the presence of SEB. As is shown in Fig. 3 A, up-regulation of CD25 was observed among the Vβ13.1 T cells. The identity of the soluble factors involved is unknown; however, a contribution from IL-15 and IL-23, which are known to induce TCR-independent activation of certain T cell subsets, was ruled out by neutralization assays (data not shown).

Down-regulated expression of the TCR complex at the cell surface is one measure of TCR-mediated activation. As a further indication that our observations are truly due to a bystander mechanism of activation, TCR down-regulation, as measured by the change in mean fluorescence intensity (MFI) of an anti-CD3 Ab relative to that observed on resting T cells, was not observed among bystander-activated Vβ13.1 T cells cultured in Transwell. Down-regulation of the TCR was, however, observed to occur consistently among directly activated Vβ17 T cells, with decreases in the range of 19.78–53.62%. Although TCR down-regulation following direct activation was observed within all individuals tested, this did not reach statistical significance (p = 0.0616) due to the high level of variation in the magnitude of the effect within the population. Mean CD3 MFI of the different subpopulations is displayed in Fig. 3 B.

Elevated CD45RO expression was observed among the bystander-activated T cell population (data not shown). To determine whether bystander activation is selective for memory T cells, Vβ13.1 T cells were sorted into naive CD45RO⁻Vβ13.1⁺ and memory CD45RO⁺Vβ13.1⁺ populations, and cultured in Transwell with whole PBMC, in the presence or absence of SEB. As can be seen in Fig. 4, although expression levels are low due to the use of the more transiently expressed CD69 marker, the memory T cell subset consistently makes the most substantial contribution to bystander activation; a very small, but significant proportion of naive T cells also became CD69 positive. No significant increase in CD69 expression was observed among control populations (data not shown).

Further investigation also demonstrated a slight elevation of CD4 expression levels among the bystander-activated T cell population (mean expression 87.62%, compared with a resting cell population mean of 78.37%; n = 6; p = 0.0081), as well as decreased expression of CCR7 (data not shown).

A genome-wide oligonucleotide microarray analysis of bystander-activated T cells allows verification that this subpopulation of T cells possesses a distinct gene expression profile to directly activated and resting cells, further validating the SEB/Vβ13.1 system for identifying bystander activation events. The wealth of information generated by this analysis also permits us to draw on the data for insights into the possible mechanism and functional outcomes of bystander T cell activation.

The microarray data generated passed all recommended quality control assessments. Genes differentially expressed (with a cutoff of B > 3, adjusted p value < 0.005) between directly activated and resting cells (4198 transcripts) can be viewed in Supplementary Table i,⁵ and those differentially expressed between bystander-activated and resting cells (1539 transcripts) can be viewed in Supplementary Table ii,⁵ ranked according to significance. Differentially expressed genes (p < 0.005) are displayed as a heatmap in Fig. 5 A, in which three distinct gene expression patterns, which cluster well into resting, bystander-, and directly activated groups, can clearly be seen.

There is good agreement between the genes found to be differentially expressed by directly activated T cells and previously characterized gene expression changes in response to T cell activation (34, 35, 36), including up-regulation of IL 2Rα (IL2RA), v-myb myeloblastosis viral oncogene homologue (avian), CTLA-4, granzyme B, member 4 of the TNFR family (TNFRSF4), and member 18 of the TNFR family (TNFRSF18), and down-regulation of IL-6R, IL-7R, and Kruppel-like factor 2 (lung). Expression changes of selected genes were independently validated at the protein level, as is summarized in Supplementary Table iii.⁵

Preliminary analysis reveals patterns of gene expression that substantiate some of our previous findings: in accordance with our observation that only directly activated T cells down-regulate the TCR from the cell surface, the transcript for intersectin 2 was found to be up-regulated by directly activated, but not bystander-activated T cells. Intersectin 2 is known to interact with Wiskott-Aldrich syndrome protein for TCR down-regulation (37). Furthermore, substantial differential regulation of several genes involved in apoptosis was observed in both the bystander- and directly activated cell populations. As shown in Fig. 5^,B, when we performed annexin V staining of bystander-activated T cells, we found that there is indeed an elevated level of apoptosis occurring within this population, as compared with resting T cells. As expected, directly activated T cells also displayed increased annexin V staining, to a much greater extent. Further subdivision of these apoptotic populations into early and late stages of apoptosis was conducted using 7-aminoactinomycin D costaining, as is shown in Supplementary Fig. 2.⁵

We used the pathway drawing function of Ingenuity Pathways Analysis to display gene expression changes related to apoptosis, as is shown in Figs. 6 and 7 (pathways were constructed with reference to selected literature (38, 39)). In agreement with our findings of elevated apoptosis within bystander-activated T cells and to an even greater extent within the directly activated population, directly activated T cells up-regulated more proapoptotic genes than the bystanders. Intriguingly, of the up-regulated genes that were unique to directly activated cells, the majority were associated with the intrinsic mechanism of cell death (the proapoptotic genes BCL2 antagonist of cell death, voltage-dependent anion channels, HtrA serine peptidase 2, caspase 9, the antiapoptotic genes baculoviral inhibitor of apoptosis repeat-containing 5 (survivin), and myeloid cell leukemia sequence 1 (BCL2 related)) or parts of the machinery shared by the intrinsic and extrinsic pathways (caspase 3), with the exception of Fas ligand. Conversely, several components of the extrinsic pathway were uniquely down-regulated (member 10A of the TNFR family (TNFRSF10A), member 25 of the TNFR family (TNFRSF25), and BH3-interacting domain death agonist). Within the bystander-activated population, however, all uniquely up-regulated genes belonged to the extrinsic pathway (FAS, TNF, TNFR-associated factor 1, and BH3-interacting domain death agonist). These findings are indicative of a death receptor-mediated mechanism of apoptosis at work within the bystander-activated T cell population, whereas directly activated T cells undergo apoptosis via the already well-characterized intrinsic mechanism of growth factor withdrawal (40, 41). Although further investigation is required to confirm that distinct mechanisms of apoptosis are in action, the hypothesis is supported by the differing levels of susceptibility observed between these two populations by annexin V staining.

In this study, we have shown the development of a reliable in vitro assay for the detection of bystander T cell activation; of key importance to this was the establishment of the specificity of the anti-Vβ13.1 Ab and the nonreactivity of Vβ13.1 T cells to direct SEB stimulation. Subsequently, several lines of evidence supported the successful elimination of TCR ligation: directly activated, but not bystander-activated T cells were observed to down-regulate the TCR; direct activation induced up-regulation of phenotypic markers, whereas bystander activation occurred preferentially in phenotypically distinct subsets; the two subsets had distinct gene expression profiles; many of the genes involved in cytoskeletal rearrangement and TCR down-regulation that were up-regulated in directly activated cells were not differentially regulated within the bystander-activated population.

To our knowledge, this is the first demonstration of bystander activation in a human T cell system activated by a primary TCR stimulation that excludes the possibility of TCR cross-reactivity and does not involve the use of exogenous growth factors.

Few studies have ruled out cross-reactivity as a potential explanation for apparent bystander activation events; this has become increasingly problematic with the discovery that pathogens as dissimilar as influenza and HCV exhibit epitope cross-reactivity (42). One study that did attempt to eliminate the possibility of cross-reactivity was conducted by Ehl et al. (2). An elegant transgenic mouse system was used to demonstrate bystander-activated cytotoxicity: TCR-transgenic mice, whose T cell repertoire consisted of 90% T cells specific for a lymphocytic choriomeningitis virus (LCMV) epitope, were infected with vaccinia virus, and subsequent killing assays demonstrated LCMV-specific cytotoxicity. Similarly, in vitro stimulation with allo-Ag elicited LCMV-specific responses. Significantly, a cold target competition assay (using an excess of vaccinia virus-infected unlabeled targets) failed to reduce the lysis of LCMV-infected targets, and transgenic T cells developed on a RAG knockout background failed to respond to allostimulation. Similarly, a study by Nogai et al. (43) demonstrated activation of murine memory T cells in response to up-regulated costimulatory molecules following LPS stimulation. Activation was dependent upon contact with accessory cells, but was shown to be TCR independent because addition of cyclosporin A failed to abrogate the response. By contrast, in the present study, Transwell assays indicated that bystander activation was not dependent upon cell-cell contact. However, in agreement with the findings of Nogai et al., phenotypic assessment revealed that bystander T cell activation preferentially affects memory T cells, and this observation is consistent with the lower threshold for activation of memory T cells as compared with naive T cells (44). A previous study examining the effects of cytokine combinations on human CD4 T cells found that differences in the responsiveness of naive and memory cells to some cytokines could be attributed to differential cytokine receptor expression (45). Furthermore, cytokine-stimulated human T cells display differing functional characteristics when compared with TCR-stimulated human T cells (46), a finding that has been supported by our observations of differing gene expression profiles.

Decreased CCR7 expression indicates that many of the bystander-activated T cells would be retained in the periphery. The finding that bystander activation displays a slight bias toward the CD4⁺ T cell subset suggests that IL-15, a factor that is induced secondary to type I IFNs during viral infection, and that acts directly on CD8⁺ T cells to induce proliferation and cytotoxicity (5, 6, 7, 13, 47), is not responsible. Indeed, Ab neutralization of IL-15 in a standard bystander activation assay failed to inhibit the bystander response (data not shown).

A major source of conflict is the diverging of ideas about the potential outcomes of bystander T cell activation. There are two chief schools of thought: bystander T cell activation results in effector functions (5, 7, 12, 13, 48), which may be damaging (in the case of immunopathology) or helpful (contributing to resolution of infection (49, 50, 51, 52)); alternatively, bystander T cell activation simply results in apoptosis and attrition of the existing repertoire to make way for the newly expanded, directly activated T cells. There is considerable evidence that bystander T cells undergo attrition. In mice, CD8⁺ memory T cells generated in response to one viral infection are reduced in frequency on subsequent infection with heterologous virus (53), and this may be mediated by apoptosis induced by type I IFNs (54). The observation that the frequency of EBV-specific memory T cells gradually increases in CMV-seronegative, but not in CMV-seropositive individuals, whereas the CMV-specific memory pool significantly increases with age, is indicative of a similar mechanism of attrition within human CD8⁺ T cell populations (21). The great majority of studies in the area of T cell memory attrition focus on CD8⁺ T cells, and there is some evidence that CD4⁺ T cells do not undergo a similar attrition process (55). The findings are mixed, however, with one group using a bacterial infection model of bystander memory T cell attrition to demonstrate that both CD4⁺ and CD8⁺ memory T cells were reduced in frequency after heterologous infection (56).

In the present study, elevated annexin V staining among bystander-activated T cells indicates increased susceptibility of these cells to apoptosis. Analysis of apoptotic pathway genes in the microarray indicated that unlike directly activated T cells, whose uniquely up-regulated genes were predominantly involved in the intrinsic pathway of apoptosis, bystander-activated T cells uniquely up-regulated transcripts related to death receptor-mediated apoptosis. Transcripts involved in prosurvival NF-κB signaling (57) were also up-regulated among bystander-activated T cells, and it is possible that the balance of these prodeath and prosurvival signals determines the outcome for the bystander population.

Overall, the data support the proposal of Jiang et al. (19) that bystander activation induces a standby state in T cells, which prepares the cells for a rapid response on encounter with cognate Ag. In the absence of such encounter, the cells undergo apoptosis, leaving space for the newly expanded Ag-specific T cells. An additional benefit of this model is that a larger Ag-specific response to a more severe infection would induce greater bystander activation, and ultimately clear a larger space to be occupied by the newly generated Ag-specific T cells. One discrepancy lies in the fact that Jiang et al. found apoptosis occurring among the bystander population only, with Ag-specific T cells apparently protected from death, whereas our study showed even greater susceptibility to apoptosis among directly activated cells; this discrepancy may be due to analysis of these cell populations at different time points. Furthermore, in vivo conditions may provide additional survival signals.

Further work is required to confirm our hypotheses regarding the diverging mechanisms of apoptosis acting on the bystander- and directly activated populations. It would also be of key interest to identify the soluble factors involved in bystander T cell activation. Finally, it is important to assess whether this mechanism makes a contribution to pathogenesis in diseases such as HIV. However, this is complicated by the absence of a unique surface marker for bystander-activated T cells. As such, it would be beneficial to subdivide further the phenotypic characteristics of the subset that is preferentially bystander activated; fluctuations within a highly characterized subset may serve as a useful surrogate indicator of bystander T cell activation, for example, at different stages of HIV infection.

Taken together, the results suggest that in the SEB system, soluble factors produced during the primary Ag-specific response are responsible for bystander activation of a small proportion of predominantly memory CD4⁺ T cells that would normally be located in the periphery. Bystander-activated T cells are susceptible to elevated levels of apoptosis, which may be a mechanism for creating immunological space to be occupied by the newly expanded Ag-specific T cells. This may erode pre-existing memory T cell pools and could account for reduced responses to infections such as influenza in elderly individuals who commonly have large CMV-specific expansions (21).

We thank Dr. Simon Brackenridge for assistance with R packages, and Ann Atzberger and Craig Waugh for cell sorting.

The authors have no financial conflict of interest.