The origin and function of human double negative (DN) TCR-αβ+ T cells is unknown. They are thought to contribute to the pathogenesis of systemic lupus erythematosus because they expand and accumulate in inflamed organs. In this study, we provide evidence that human TCR-αβ+ CD4− CD8− DN T cells can derive from activated CD8+ T cells. Freshly isolated TCR-αβ+ DN T cells display a distinct gene expression and cytokine production profile. DN cells isolated from peripheral blood as well as DN cells derived in vitro from CD8+ T cells produce a defined array of proinflammatory mediators that includes IL-1β, IL-17, IFN-γ, CXCL3, and CXCL2. These results indicate that, upon activation, CD8+ T cells have the capacity to acquire a distinct phenotype that grants them inflammatory capacity.
In healthy humans, a small fraction of peripheral blood T cells lacks CD4 and CD8. The origin and function of these cells, known as double negative (DN)3 T cells, is still mostly unknown. Because DN T cells are defined by exclusion (i.e., as CD3+ non-CD4 non-CD8 T cells), the term refers to several T cell populations that are infrequent in the peripheral blood including limited-repertoire cells such as NKT cells, mucosal associated invariant T cells (1), and TCR-γδ+ cells along with conventional TCR-αβ+ cells (2). In addition, TCR-αβ+ DN T cells with suppressive functions have been described in humans (3) and mice (4, 5).
Deficiency of Fas (CD95) or its ligand (FasL; CD178) cause dramatic accumulation of DN T cells in the context of the autoimmune lymphoproliferative syndrome, a lymphoproliferative disease with certain autoimmune features (6). In mice, the Faslpr and Faslgld mutations (that disrupt Fas and FasL expression, respectively) lead to an analogous disorder and act as autoimmune accelerators in certain genetic backgrounds such as MRL (7, 8). Patients with systemic lupus erythematosus (SLE), a chronic autoimmune disease, have increased numbers of TCR-αβ+ DN T cells in the peripheral blood (9, 10). However, Fas and FasL expression and function are preserved (11). Hence, DN cell expansion in the context of lupus is probably caused by different mechanisms. Mice with lupus-like manifestations display similar expansion of DN T cells, although, as in humans, DN cell accumulation is maximal in the absence of Fas or FasL (12, 13).
Several lines of evidence suggest that TCR-αβ+ DN T cells play a role in the pathogenesis of SLE. In mouse models, the accumulation of DN T cells is associated with the development and severity of autoimmune disease (14). TCR-αβ+ DN T cells promote the production of Ig and anti-DNA Abs by B cells (10, 15, 16). Further, TCR-αβ+ DN T cells represent a large component of the kidney-infiltrating T cells observed in patients with lupus nephritis and represent a major source of IL-17 in patients with SLE (17).
DN T cell expansion in the context of SLE may result from increased in vivo stimulation. In a previous communication, we reported that TCR-αβ+ DN T cells from healthy individuals and patients with SLE accumulate when purified T cells are stimulated in vitro suggesting that T cell activation drives DN T cell expansion (17). Prior work in humans and mice support this notion because DN T cells exhibit higher levels of activation-induced markers than CD4 and CD8 single-positive cells (9, 18, 19). A study performed in mice bearing a MHC class I-restricted transgenic TCR showed that Ag encounter and cell activation led CD8 T cells to become DN. This phenomenon, however, depended on the absence of Fas because it was only observed in Fas-deficient mice (20). Appearance of DN T cells has also been observed in B6 nude mice after the transfer of purified syngeneic CD4+ T cells (21). Thus, although expansion of TCR-αβ+ DN T cells is a phenomenon associated with T cell activation in in vitro and in vivo settings, their origin and function remain poorly defined.
Although known to be thymic-dependent (22), the origin of TCR-αβ+ DN T cells is mostly unknown. Studies in mice have shown that the CD8 locus is demethylated in DN T cells indicating previous CD8 expression (23, 24). In an attempt to characterize their origin and function, several reports have analyzed the Vα and β genes used by TCR-αβ+ DN T cells. An analysis of Vβ gene usage revealed that TCR-αβ+ DN T cells from patients with autoimmune lymphoproliferative syndrome express a Vβ gene repertoire that resembles that of their CD8+ counterparts (25). Accordingly, Vα usage was found to be similar between DN and CD8+ T cells suggesting they have a common origin and function (26). In contrast, other studies performed in healthy subjects reported that TCR-αβ+ DN T cells have a skewed Vβ gene repertoire that differs from that of single positive T cells suggesting that they develop and/or proliferate in a Ag-driven fashion (19, 27, 28).
In this study, we present evidence that TCR-αβ+ DN T cells from normal human individuals differentiate exclusively from CD8+ T cells. We show that they exhibit gene expression and cytokine production profiles that distinguish them from CD4+ and CD8+ T cells. Our results suggest that a fraction of CD8+ T cells down-regulates CD8 upon activation and acquires a distinct phenotype that grants them an inflammatory capacity.
Materials and Methods
Blood samples were obtained from 21 healthy platelet donors from the Kraft Family Blood Donor Center (Dana-Farber Cancer Institute, Boston, MA). Total T cells were isolated by negative selection (RosetteSep, Stem Cell Technologies). T cell purity was always ≥96%. For some experiments, T cell subsets were further purified by cell sorting in the Flow Cytometry and Cell Sorting Facility of the Beth Israel Deaconess Medical Center (in a FACSAria flow cytometer, BD Biosciences). Post sorting cell purity was >99%.
For cell stimulation, 2 × 106 T cells were cultured in RPMI 1640 with 10% FCS in 12-well plates coated with anti-CD3 (5 μg/ml, clone HIT3a, BD Biosciences) and anti-CD28 (5 μg/ml, BD Biosciences). For some experiments, 2 × 106 PBMC were incubated in 12-well plates during 2 h at 37°C to allow monocyte adherence. Next, wells were washed twice with PBS and autologous CFSE-labeled T cells were added into the well or into a Transwell where T cells were separated from monocytes by a 0.4-μm pore permeable membrane (Corning). For detection of intracellular cytokines, 1 μl/ml brefeldin A (BD Biosciences) was added to the culture, along with PMA and ionomycin 3 h before harvesting. For the quantification of cell proliferation, T cells were labeled with CFSE (5 μM during 5 min) before stimulation.
The following Abs were used: anti-CD4-Pacific blue, anti-CD4-PE-Cy5, anti-CD8-allophycocyanin-Cy7, anti-CD8-PerCP, anti-TCR-αβ-FITC, anti-CD25-allophycocyanin-Cy7, anti-IFN-γ-FITC, anti-IL-1-PE, anti-IL-2-allophycocyanin, anti-IL-8-PE, and anti-IL-10-PE (BD Biosciences); anti-IL-17-Alexa Fluor 647 (eBiosciences).
Intracellular staining was performed with the BD Cytofix/Cytoperm kit, according to the instructions of the manufacturer. Samples were acquired in a LSRII flow cytometry (BD Biosciences). Analysis was performed with FlowJo v. 7.2.2 (Tree Star).
Real time PCR
RNA purification was performed with the use of the RNeasy Mini kit (Qiagen). cDNA was produced using oligo dT primers, from an equal amount of RNA. Specific primers were designed to amplify mRNAs from the following genes: CD8a, CXCL3, CXCL2, CCL22, ADAM12, LGALS3, IL15, IL10, IL8, IL1b, MEF2C, and Fosb. Detection was performed with SYBR green (SYBR Green I Master) in a LightCycler 480 System (Roche).
Gene expression analyses
To define the DN cell population, we performed a microarray analysis from cells isolated from four healthy individuals. The DN cell population is enriched in TCR-γδ cells. To study a pure and homogeneous population devoid of TCR-γδ, and to exclude regulatory T cells from the CD4+ population, we stained isolated T cells using the following Abs: FITC-labeled anti-TCR-αβ, PE-labeled anti-CD25, PE-Cy5-labeled anti-CD4, and APC-Cy7-labeled anti-CD8. Single lymphocytes were identified according to forward and side scatter characteristics and TCR-αβ+ CD25− cells were gated. CD4+ CD8−, CD8+ CD4−, and CD4− CD8− T cells were sorted (see Fig. 3A). Cell lysis and RNA extraction was performed immediately. RNA from each cell subset was pooled. RNA was amplified and hybridized to Human Genome U133 Plus-2 chips (Affymetrix) in the BIDMC Genomics and Proteomics Core. Raw data were normalized using RMAExpress version 1.0 (http://RMAExpress.bmbolstad.com) and analyzed using Nexus Expression version 1.0. Low intensity probes were filtered when their expression value was <8.0 in 70% of the samples. Microarray data have been deposited in NCBI’s Gene Expression Omnibus (29) and are accessible through GEO Series accession number GSE16130 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE16130).
Paired and nonpaired Student two-tailed t tests, as well as χ2 test were used. Results are expressed as the mean ± SEM, unless noted otherwise. p values ≤0.05 were considered significant. All the experiments were performed at least three times and the presented data are representative of at least two similar experiments.
DN T cells have distinct proliferation and death kinetics pattern
Patients with SLE, a chronic autoimmune inflammatory disease, exhibit abnormally high levels of TCR-αβ+ DN T cells in peripheral blood (10, 15). These cells may represent in vivo activated cells because stimulation of SLE T cells with anti-CD3 and anti-CD28 Abs in vitro results in expansion of DN cells (17). We initiated our experiments by stimulating T cells with anti-CD3 and CD28 Abs and, as expected, we noted a significant expansion of the DN cells (Fig. 1⇓A). To determine whether DN T cell expansion was due to a relatively augmented proliferative capacity, we labeled T cells with CFSE and quantified division rates of CD4+, CD8+, and CD4−CD8− (DN) T cell subsets. As shown in Fig. 1⇓, B and C, CD4+ and CD8+ T cells displayed a vigorous proliferative response (81 and 73% of the cells proliferated, respectively). In contrast, a significantly lower fraction of DN T cells (30% ± 7, p < 0.005) divided after a 5-day stimulation period.
Alternatively, DN T cell accumulation may result from enhanced resistance to activation-induced cell death. To consider this possibility, we stained T cells, which had been cultured in the presence or absence of plate-bound anti-CD3 and anti-CD28 Abs with propidium iodide (PI). A minute fraction of unstimulated CD4+ and CD8+ T cells were stained by PI (0.22% ± 0.03 and 0.54% ± 0.15, respectively). As expected, the proportion of dead CD4+ and CD8+ T cells increased after stimulation (1.91% ± 0.28 and 2.11% ± 0.29, respectively). In contrast, a significantly higher fraction of DN T cells were PI-permeable in both unstimulated (14.26% ± 3.75, p < 0.05) and stimulated (6.34% ± 0.68, p < 0.05) conditions (Fig. 1⇑, D and E).
Taken together, these results demonstrate that the expansion of DN T cells observed among in vitro cultured T cells is not the result of increased proliferation or decreased cell death. Further, they suggest that cell proliferation and death kinetics are different in DN T cells than in CD4+ and CD8+ T cells.
A fraction of CD8 T cells become DN upon stimulation
The fact that expansion and resistance to cell death could not account for the accumulation of DN T cells prompted us to examine whether other T cell subsets can become DN upon stimulation. DN T cells have been shown to appear in mice that bear a CD8-selecting transgenic TCR after in vivo cognate stimulation (20). To investigate whether CD4+ and/or CD8+ T cells down-regulate their coreceptor and become DN, we sorted CD4+ and CD8+ T cells from healthy individuals and stimulated them in plates coated with anti-CD3 and anti-CD28 Abs. Cultured CD4+ T cells remained CD4+ during the 5-day incubation period (Fig. 2⇓, A and B). In fact, CD4 expression increased in CD4+ T cells that showed signs of cell stimulation and division (increased cell size and dilution of CFSE; data not shown). In sharp contrast, a fraction of CD8+ T cells down-regulated CD8 expression. Although this phenomenon was observed in nonstimulated as well as in stimulated CD8+ T cells (anti-CD3 plus anti-CD28), its magnitude was significant only in stimulated CD8+ T cells (Fig. 2⇓, A and B). To confirm the down-regulation of the coreceptor CD8, we isolated total RNA from sorted CD8+ T cells after stimulation with plate-bound anti-CD3 and anti-CD28 Abs. Stimulation of CD8+ T cells in vitro resulted in a significant decrease in CD8α mRNA confirming the claim that CD8 is down-regulated in CD8+ T cells upon stimulation (data not shown). The possibility of transient CD8α mRNA down-regulation was excluded by quantifying CD8α mRNA in CD8+ and CD8− (DN) subsets obtained after stimulating sorted CD8+ T cells during 5 days (Fig. 2⇓C).
Activation-induced CD8 down-regulation was not observed in the entire CD8+ population suggesting that it occurs only in a fraction of the cells. To identify the factors that facilitate the generation of DN T cells, we stimulated CFSE-labeled T cells in the presence or absence of autologous monocytes. As expected, T cell stimulation increased the percentage of DN T cells (p = 0.005). Addition of monocytes to the culture further augmented the generation of DN T cells (p = 0.002). Cell contact rather than monocyte-derived soluble factors was necessary to obtain this effect, because when monocytes and T cells were separated by a permeable membrane the amount of DN T cells observed at the end of the stimulation period was not different from that of cultures of T cells only (Fig. 2⇑D).
These results indicate that a fraction of CD8+ T cells down-regulates the CD8 coreceptor upon stimulation. Such modification is facilitated by signals delivered by monocytes in a cell-to-cell fashion and is controlled, at least partially, at the level of RNA transcription.
DN T cells exhibit a distinct gene expression profile
To further define the origin of DN T cells, we sorted TCR-αβ+ CD25− T cells into CD4+, CD8+, and DN populations and performed RNA microarray analysis (Fig. 3⇓A). For this purpose, we considered the 11441 probes (from a total of 54613) that had a high expression level in any of the three cell subsets. The signal intensity from each probe was compared between CD4+ T cells and DN T cells, between CD8+ T cells and DN T cells, and between CD4+ and CD8+ T cells. We selected the probes whose expression level varied more than 2 (Log scale) in any of the three comparisons. One hundred and forty probes (corresponding to 119 genes) fulfilled this criterion (Fig. 3⇓B and supplementary Table I).4 Of the 119 genes, 88 were expressed in DN T cells, 39 in CD4+ T cells, and 51 in CD8+ T cells. Only 11 were expressed in all three populations. Thirty-eight percent (15 of 39) of the genes expressed in CD4+ T cells were also expressed in DN T cells. In contrast, the fraction of genes expressed in CD8+ T cells that were also present in DN T cells was significantly higher (72.5%, p < 0.0001). CD8+ T cells shared more genes with DN T cells than with CD4+ T cells (72.5 vs 35.3%, p < 0.0001). Among the selected genes, CD4+ T cells shared 38.5% with DN T cells and 46.1% with CD8+ T cells (p = 0.335).
These data, which are summarized in the Venn diagram depicted in Fig. 3⇑C, represent evidence supporting the concept that TCR-αβ+ DN cells are derived from CD8+ T cells.
The gene expression profile of DN T cells indicates an effector cell capacity
Previous work performed in transgenic mice has suggested that CD8 down-regulation represents an inactivation process of CD8+ T cells that fail to engage class I MHC molecules (30). This concept, however, is difficult to reconcile with the fact that DN T cells expand in certain inflammatory processes such as SLE (10, 17), as well as following in vivo (20) or in vitro stimulation (Fig. 1⇑). Further, DN T cells have been shown to be able to provide help to B cells (10), and in previous work we have shown that they produce inflammatory cytokines and are found in inflamed target organs of patients with SLE (17). Thus, DN T cells may represent a terminally differentiated CD8 effector cell subset. In accordance with this hypothesis, we found that the cytokines and chemokines whose expression levels varied ≥2 log when DN were compared with CD4+ and CD8+ T cells were primarily proinflammatory (i.e., IL-8, IL-1β, CXCL2, CXCL3). Real time PCR, performed to confirm the microarray results, demonstrated that CXCL3, CXCL2, ADAM12, LGALS3 (Galectin-3), IL10, IL8, IL1b, and the transcription factors MEF2C and Fosb, were expressed in significantly higher levels by DN T cells than by CD4+ and CD8+ T cells (Fig. 3⇑D). These results indicate that DN T cells indeed represent a CD8 effector cell subset able to induce local and systemic inflammation by the production of cytokines and immune cell recruitment.
DN T cells exhibit a unique cytokine production profile
To confirm the data obtained by RNA analysis and to compare the cytokine production profile of peripheral blood DN T cells with that of in vitro-generated DN T cells, we performed intracellular cytokine staining. Purified T cells were studied immediately after venipuncture and after 5 days in culture in the presence or absence of polyclonal stimulation (plate-bound anti-CD3 and anti-CD28 Abs). As shown in Fig. 4⇓A, a large fraction of freshly isolated DN T cells (30.5 ± 1.4) was IL-1+. In contrast, CD4+ and CD8+ T cells lacked IL-1 (p < 0.001). CD4+ T cells remained mostly IL-1 negative after the 5-day stimulation period. However, ∼20% of CD8+ T cells acquired the ability to produce the cytokine. IL-8 was produced by a small amount of freshly isolated T cells. However, the frequency of IL-8+ cells increased significantly, following stimulation, within the CD8 and DN T cell subsets (Fig. 4⇓B). Production of IL-10 exhibited a similar pattern to that of IL-1. In freshly isolated T cells IL-10 was produced mainly by DN T cells (p < 0.05). After stimulation, the frequency of IL-10+ cells increased in all the cell subsets and particularly so in DN T cells (Fig. 4⇓C).
These results confirm the concept that DN T cells are able to produce inflammatory cytokines IL-1, IL-8, and IL-10. Further, they indicate that a large fraction of DN T cells secretes IL-1 and IL-10 spontaneously. Although absent in resting conditions, the capacity to produce these cytokines is acquired by CD8+ T cells during in vitro stimulation to an extent similar to that of DN T cells. This suggests that a fraction of CD8 T cells undergoes differentiation into an effector subset that down-regulates CD8 and produces inflammatory cytokines.
CD8 down-regulation is associated with a distinct cytokine production profile
To determine whether CD8-derived DN T cells can produce proinflammatory cytokines, we sorted TCR-αβ+ CD25− CD8+ T cells from the peripheral blood of healthy donors and stimulated them with plate-bound anti-CD3 and anti-CD28 Abs for 5 days. At the end of the stimulation period, we performed intracellular staining to compare cytokine production between CD8+ and CD8− cells (Fig. 5⇓).
In agreement with previous results, a significantly higher fraction of cells that had down-regulated CD8 expression produced IL-1 (Fig. 5⇑, B and C; p = 0.03) and IL-10 (Fig. 5⇑, F and G; p = 0.05) than cells which maintained high levels of CD8 on their surface. CD8− cells showed a trend toward a higher IL-8 production than CD8+ cells, albeit the difference was not statistically significant (Fig. 5⇑, D and E).
In a previous report, we showed that DN T cells produced IL-17 (17). IL-17A and IL-17F expression was not significantly different between DN cells, CD4+, and CD8+ T cells in the microarray analysis. However, expression of IL-23R was significantly higher in DN T cells than in CD4+ (2.4 log) and CD8+ (2.1 log) T cells. As shown in Fig. 5⇑H, IL-17 production was significantly higher in CD8− than in CD8+ T cells, confirming the notion that DN T cells are also producers of this proinflammatory cytokine.
As previously reported, IL-2 production was observed in a significantly lower fraction of DN T cells as compared with CD8+ T cells (17). IFN-γ was detected in a large fraction of DN and CD8+ T cells (Fig. 5⇑).
Taken together, these results demonstrate that upon TCR-mediated activation, a fraction of CD8+ T cells down-regulates CD8 expression and acquires the ability to produce mostly proinflammatory cytokines and chemokines.
In this communication, we have addressed the origin and phenotype of human TCR-αβ+ DN T cells. The evidence presented demonstrates that they can derive from CD8+ T cells which undergo a phenotypic transformation that involves the down-regulation of CD8 as well as the acquisition of a distinct effector phenotype characterized by the production of proinflammatory cytokines. These cells, considered important in the pathogenesis of SLE, might represent an alternative differentiation pathway, which activated CD8 cells follow.
The origin of TCR-αβ+ CD4− CD8− T cells, a component of DN T cells, has been debated. Conflicting data obtained from Vβ and Vα gene usage studies probably reflect the heterogeneity of DN T cells and differences in the composition of the analyzed cells (25, 26, 19, 27, 28). Our results demonstrate that a considerable proportion of CD8 T cells lose the expression of the CD8 coreceptor in a process regulated at the transcriptional level. These results are congruent with studies performed in mice showing accumulation of DN T cells following Ag stimulation of transgenic CD8 T cells in vivo (20). We observed no diminution of CD4 expression in parallel CD4+ T cell cultures and thus we believe that in humans TCR-αβ+ DN T cells derive from CD8 T cells. It is unclear at this point whether DN T cells derive from a distinct CD8+ T cell subset or their development depends on the provided stimulation context. Cell division does not seem to be a requisite of DN differentiation because most in vitro-generated DN T cells lack evidence of CFSE dilution (Fig. 1⇑B). Additional work is needed to identify the monocyte-derived molecule(s) that induce CD8 T cells to become DN (Fig. 2⇑D).
CD8 participates in the binding of the TCR and the MHC-peptide complex. CD8 down-regulation is thus predicted to decrease significantly the affinity of the interaction (31) and the T cell capacity to bind to target cells. TCR-αβ+ DN T cells proliferate less and have a higher death rate than CD4 and CD8 single-positive cells. These observations, along with the fact that they produce proinflammatory cytokines and chemokines suggests that they are terminally differentiated cells capable of inducing inflammation mainly through the production of cytokines. This concept is congruent with the fact that they are found in inflamed tissues of patients with SLE (17).
Several findings presented herein and previously indicate that TCR-αβ+ DN T cells represent activated T cells (9, 19, 18). The cytokine-production capacity of freshly isolated DN T cells is comparable to that of DN T cells generated following in vitro activation. Upon in vitro stimulation, a fraction of CD8+ T cells acquires a similar cytokine production profile. This corroborates the claim that CD8+ T cells can differentiate into IL-1- and IL-10-producing DN T cells. Our data indicate that the small number of TCR-αβ+ DN T cells observed in peripheral blood of normal individuals represent in vivo-activated T cells that are probably short lived. Patients with SLE have increased numbers of activated T cells in peripheral blood (32). Increased T cell activation in the context of SLE has been proposed to occur as a consequence of self-Ag presentation or binding of anti-T cell Abs (33). The expansion of DN T cells observed in this disease represents the consequence of in vivo T cell activation and/or decreased removal of activated DN T cells.
The cytokine production profile of DN T cells is unique. Most of the cytokines and chemokines have local and systemic proinflammatory capacity. An exception is IL-10, a cytokine known to stimulate B cell Ab production and suppress T cell responses. TCR-αβ+ DN T cells have been shown to have suppressive capacity in humans (3) and mice (4, 5). It is unclear whether the suppressive and proinflammatory functions are indeed exerted by the same cells. It is possible that under certain conditions, the suppressive function of DN cells may dominate over the proinflammatory capacity. The circumstances involved in such regulation are at present obscure and warrant further study. In contrast, increased IL-10 production may contribute to the increased capacity of SLE DN cells to provide help to autologous B cells to produce autoantibody (34).
In summary, we show that human TCR-αβ+ DN T cells can derive from CD8 T cells and they acquire a distinct proinflammatory cytokine production capacity. We propose that DN cells represent a novel differentiation path for peripheral blood CD8+ cells that may participate in the normal immune response and, when occurring in excess, in the expression of autoimmune pathology, as it is the case with SLE (10, 17).
We are grateful to John Tigges, Vasilis Toxavidis, and Heidi Mariani for help with cell sorting.
The authors have no financial conflict 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.
↵1 This work was supported by National Institutes of Health Grants R01 AI42269 and R01 AI49954 and by the Mary Kirkland Center for Lupus Research at the Hospital for Special Surgery funded by Rheuminations.
↵2 Address correspondence and reprint requests to Dr. George C. Tsokos, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Avenue, CLS-937, Boston, MA 02115. E-mail address:
↵3 Abbreviations used in this paper: DN, double negative; SLE, systemic lupus erythematosus; FasL, Fas ligand; PI, propidium iodide.
↵4 The online version of this article contains supplementary material.
- Received May 15, 2009.
- Accepted July 28, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.