CD1 molecules present lipid antigens to T cells. An intriguing subset of human T cells recognize CD1-expressing cells without deliberately added lipids. Frequency, subset distribution, clonal composition, naïve-to-memory dynamic transition of these CD1 self-reactive T cells remain largely unknown. By screening libraries of T-cell clones, generated from CD4+ or CD4−CD8− double negative (DN) T cells sorted from the same donors, and by limiting dilution analysis, we find that the frequency of CD1 self-reactive T cells is unexpectedly high in both T-cell subsets, in the range of 1/10–1/300 circulating T cells. These T cells predominantly recognize CD1a and CD1c and express diverse TCRs. Frequency comparisons of T-cell clones from sorted naïve and memory compartments of umbilical cord and adult blood show that CD1 self-reactive T cells are naïve at birth and undergo an age-dependent increase in the memory compartment, suggesting a naïve/memory adaptive-like population dynamics. CD1 self-reactive clones exhibit mostly Th1 and Th0 functional activities, depending on the subset and on the CD1 isotype restriction. These findings unveil the unanticipated relevance of self-lipid T-cell response in humans and clarify the basic parameters of the lipid-specific T-cell physiology.
Conventional TCR-α/β+ T lymphocytes recognize peptides presented by MHC molecules and are key players in the adaptive immune response. Thymic selection maximizes the generation of conventional mature T-cell repertoires specific for foreign Ags while minimizing autoreactivity 1. A key feature of the adaptive immune response is that newly generated conventional T cells are naïve and acquire an effector/memory phenotype upon Ag encounter 2.
There exists also an unconventional population of TCR-α/β+ T lymphocytes that are restricted for CD1 molecules and recognize self- and microbial lipid antigens 3. CD1 are non-polymorphic MHC class I-like molecules classified into three groups based on the sequence homology: group 1 comprises CD1a, CD1b and CD1c; group 2 CD1d; group 3 CD1e 4. The best characterized CD1-restricted T cells are the CD1d-restricted invariant Natural Killer T (iNKT) cells that express, in humans, an invariant Vα24-Jα18 TCR paired with Vβ11, together with NK-cell receptors 5–7. iNKT cells can be unequivocally identify through their peculiar TCR: they are overly autoreactive, display an innate-like (constitutive) effector/memory phenotype already at birth 8, unlike conventional T cells, and are divided in two main and functionally distinct CD4+ and CD4−CD8− double negative (DN) subsets 9, 10.
A second type of CD1-restricted T lymphocytes does not express the invariant TCR and is mainly restricted for group 1 CD1 3. Because of the lack of specific markers, this T-cell type has been investigated using sporadically isolated T-cell clones, which provided fundamental hints on the microbial lipids and lipopeptides Ags recognized by these T cells and their presentation pathways 11, 12. An intriguing group of CD1-restricted T-cell clones is fully activated by exposure to CD1-expressing antigen-presenting cells (APCs) in the absence of foreign lipid antigens 13. Consistent with their marked autoreactivity, group 1 CD1-restricted T cells recognize different cell-endogenous (self) glycosphingolipids 12, 14, 15. Dual reactivity against self- and microbial antigens has also been described for several group 1 CD1-restricted T-cell clones 16, suggesting the possibility that such T cells take part in both initial phase (via self-recognition) and later phase (via bacterial Ag recognition) of the adaptive response. Because of their peculiar reactivity, CD1 self-reactive T cells might play a role in infections, autoimmunity and cancer, in which both exogenous and endogenous Ags are involved. Despite this potential high relevance for the pathophysiology on the immune response, the lack of specific markers has hampered so far a general characterization of the CD1 self-reactive T-cell population. Critical aspects of physiology of these cells, such as their abundance, distribution within CD4+, CD8+ and DN subsets, adaptive- or innate-like dynamics between the naïve and effector/memory compartments, display of specialized effector functions have never been determined.
In this study, we directly addressed these issues by performing massive approaches that allow to overcome the absence of known markers specific for CD1 self-reactive T cells, namely: (i) screenings of libraries of T-cell clones, obtained from circulating CD4+ and DN subsets or from the naïve and effector/memory compartments of healthy adults and neonates; (ii) limiting dilution analysis (LDA) of total adult T cells. T-cell clones and lines were screened for the recognition of APCs engineered with CD1a, CD1b, CD1c or CD1d isotypes, determined by IFN-γ secretion, assuming that these APCs would present the whole repertoire of endogenous CD1-bound lipids.
High frequency of CD1-restricted self-reactive T cells in the CD4+ and DN subsets of healthy adult donors
Due to the absence of specific markers that unequivocally identify CD1-restricted self-reactive T cells, we generated large libraries of T-cell clones using random PHA cloning and analyzed each T-cell clone for CD1 self-reactivity. We assumed that unbiased T-cell cloning might provide a representative view of the in vivo distribution by calculating the number of CD1 self-reactive T-cell clones over the totality of the screened ones. CD4+ and DN T cells from 12 adult healthy donors were sorted and immediately cloned at 1 cell/well without any further enrichment (Fig. 1A). Since we included also CD1d as the target of our screening, to obtain the widest figure of the unknown CD1 self-reactive T-cell repertoire in humans, Vα24+ iNKT cells were gated out from the sorted cells to prevent their analysis. The CD1 restriction of growing T-cell clones was determined by the amount of IFN-γ released in response to C1R cells expressing one particular CD1 isotype, but not to C1R cells expressing the other isotypes or to WT C1R, and confirmed by inhibition with isotype-specific anti-CD1 mAbs (Fig. 1B). We screened 702 CD4+ and 574 DN T-cell clones and we found that the number of CD1 self-reactive T-cell clones was surprisingly high, accounting on average for 10.2 and 10.5% of all the tested CD4+ and DN clones, respectively (Fig. 2A, left panel). Because in the tested healthy adult donors, the mean frequencies of CD4+ and DN T cells were 70% (±8.6) and (2.1%±1.0) of the total peripheral blood T cells, respectively, we could estimate that among circulating TCR-αβ+ T cells there were about 7% CD4+ and 0.2% DN CD1 self-reactive T cells. T-cell clones restricted for each CD1 isotype were generated from each donor, with a prevalent recognition of one or two CD1 isotypes (Table 1). On an average, CD1a and CD1c were the most frequently recognized isotypes, accounting for the restriction of the vast majority of self-reactive T-cell clones in both subsets analyzed (Table 1 and Figure 2B). CD1d was recognized by fewer T-cell clones, whereas CD1b was the least recognized isotype: only 1/9 donors analyzed in CD4+ T cells and 3/7 donors in DN cells. The CD4+ and DN subsets of the same donor were often heterogeneous in terms of CD1 isotype restriction and could contain different CD1 self-reactive T-cell repertoires.
Table 1. Frequency of CD1 self-reactive T cells in the sorted CD4+and DN subsets from healthy adult PBMCs and neonate UCBLs
a) – indicates not done.
b) Number of tested T-cell clones.
c) Numbers indicate the % of T-cell clones reactive against the indicated CD1 isotype, determined as (number of IFN-γ producing-clones/total number of screened clones)×100.
As an internal control for our experimental conditions, we found that MHC-alloreactive CD4+ T-cell clones, defined as the T-cell clones that responded equally to both C1R WT and each single C1R-CD1-transfected cells, were on average 4.5% (range 1–10%) of the total screened CD4+ T-cell clones (Fig. 2A, right panel). This value was in the range of published alloreactive precursor frequencies estimated by LDA 17, suggesting that our experimental system was not introducing TCR specificity bias during clonal expansion.
The TCR Vβ and Vα gene usage of 9 CD4+ and 10 DN CD1a self-reactive T-cell clones from 5 donors, and of 5 DN CD1c self-reactive T-cell clones from two donors, was broadly diverse, even in the T-cell clones derived from the same donor and restricted for the same CD1 isotype, (Supporting Information Table 1). This ruled out that the observed high frequencies of CD1 self-reactive T cells could be the result of robust clonal expansion in vivo of relative few precursors, as previously observed in the case of one CD1a-restricted DN T clone 18.
Altogether, these data suggested that CD1 self-reactive T cells are a large fraction of human circulating T cells, are contained with comparable frequencies in the CD4+ and DN subsets and utilize a diverse TCR repertoire.
LDA confirm high frequencies of CD1c self-reactive T cells
To confirm with a different method the frequency obtained by screening the T-cell clone libraries, we sought to utilize LDA. We focused on CD1c self-reactive T cells because their frequency was less variable in both subsets of the most tested donors. T cells were screened against THP1-CD1c cells in the presence of MHC class I and II blocking mAbs, with or without anti-CD1c mAb. T-cell cultures were scored positive when the anti-CD1c mAb induced a significant reduction of IFN-γ release. The limiting dilutions performed with sorted CD3+ cells from three different donors yielded linear titration curves (Fig. 2C). Assuming a linear single-hit kinetics, the estimated CD1c self-reactive T-cell precursor frequencies were 1 in 176, 1 in 96 and 1 in 235 total T cells for donors L, G and O, respectively. These frequencies were slightly lower than that estimated by screening of T-cell clones from donors L and G. This might be caused by the fact that, in limiting dilutions, IFN-γ secretion might not be blocked by anti-CD1c mAb in the wells containing both CD1c-restricted and non-CD1c-restricted T cells reactive against THP1. These wells would be scored as negative causing an underestimation of the CD1c-restricted T-cell frequency.
Nonetheless, the observed frequencies remained very high and confirmed the unexpected large number of CD1c-reacting T cells in peripheral blood.
CD1 self-reactive T cells are present at birth and show a naïve/memory adaptive-like population dynamics
We next assessed whether CD1 self-reactive T cells exhibited a naïve/memory dynamics or, instead, a constitutive effector/memory phenotype already at birth. To this aim, we first determined the frequency of CD1 self-reactive T cells in newborns. By the same sorting and cloning strategy described above, we obtained libraries containing a total of 230 CD4+ and 193 DN T-cell clones from 6 different UCBL samples that were screened against the panel CD1-transfected C1R cells. The mean frequency of CD1-restricted self-reactive T-cell clones among CD4+ and DN T-cell clones derived from umbilical cord blood lymphocytes (UCBLs) was comparable with that found from adult PBMCs (Fig. 3A). CD1a and, to a greater extent, CD1c were again the most frequently recognized isotypes (Table 1 and Fig. 3B). However, neonatal self-reactive T cells showed a less biased recognition of CD1 isotype compared to adult cells.
We then compared the frequency of CD1 self-reactive T cells in the naïve and effector/memory compartments, identified by the reciprocal expression of the CD45RA or CD45R0 isoforms 19, which were sorted and cloned from CD4+ and DN T cells of three samples each of adult PBMCs and neonatal UCBLs (Fig. 3C). CD1 specificity of the 230 CD4+ and 193 DN T-cell clones obtained was assessed as described above. CD1 self-reactive T-cell clones were enriched in the CD45RA compartment in both CD4+ and DN T-cell subsets of both newborns and adults (Fig. 3D). Furthermore, the percentage of the CD1 self-reactive T-cell clones from the CD45R0 compartment increased in the adult PBMCs relative to UCBLs (Fig. 3D).
CD1 self-reactive T cells are thus present in adult and cord blood at comparable frequency and display a naïve/memory dynamics similar to that of conventional T cells of the adaptive immune system.
CD1 self-reactive T cells display Th1/Th0 functional phenotype and killing activity
To characterize the effector phenotype of CD1 self-reactive T cells, we evaluated the ability to produce different Th1 and Th2 cytokines by 10 CD4+ and 25 DN T-cell clones isolated from different donors, induced upon recognition of the C1R cells expressing the cognate CD1 isotype. As expected from their initial screening protocol, all the tested clones produced IFN-γ, irrespectively of their coreceptor usage and CD1-isotype restriction (Fig. 4A). Apart from IFN-γ, the cytokine pattern of the CD4+ and DN T-cell clones was not equivalent. GM-CSF was mostly produced by DN T-cell clones (Fig. 4A). TNF-α production was prevalent in the CD4+ (3 out of 5 tested clones) compared to the DN CD1 self-reactive T-cell clones (5 out of 25). A minor proportion of T-cell clones in both CD4+ and DN subsets, predominantly CD1c and CD1d self-reactive, also secreted the TH2 cytokines IL-4 and IL-5.
At different extents, two CD4+ and 5 DN self-reactive T-cell clones restricted for CD1a or CD1c were also able to kill in a CD1-dependent fashion the specific C1R-CD1-transfected cell (Fig. 4B).
Because CD1c is expressed by different types of DC present in circulation and in tissues 3, we tested a CD4+ and a DN CD1c self-reactive T-cell clone for the recognition of monocyte-derived DC. Both T-cell clones secreted IFN-γ in a CD1c-dependent manner upon coculture with DC (Fig. 4C).
These results underscored the heterogeneity of the CD1 self-reactive T-cell compartment also in terms of effector functions and suggested that diverse specialized subsets might exist within this particular population of T lymphocytes, apparently associated with the CD1 isotype restriction.
Using two different approaches, we found that the frequencies of CD1 self-reactive T cells are very large in all donors, in the range of 1/10–1/300 circulating T cells and comparable to that of MHC alloreactive T cells. These frequencies could be underestimated since both approaches used IFN-γ release as the revealing assay. The detection of other cytokines in the screening might reveal an even higher precursor frequency, better approximating the number of CD1-restricted T cells.
Self-reactive T cells specific for all the four CD1 isotypes were detected in adults, although with inter-individual variability. A predominance of T cells restricted for CD1a, followed by CD1c and CD1d, was observed, while CD1b was seldom recognized. This occurred in both CD4+ and DN subsets, suggesting that the selection of this autoreactive T-cell repertoire is not biased by the expression of the CD4 coreceptor. In line with this assumption, preliminary data suggest that high frequency of CD1a, CD1b and CD1c, but not CD1b, self-reactive T cells are also contained in the CD8+ subset. Furthermore, the polyclonal TCR repertoire that we have found expressed by CD1 self-reactive T-cell clones rules out that clonal expansion of few auto-reactive T-cell clones is responsible for the skewing observed in adults toward CD1a and CD1c restriction. We assume that CD1 self-recognition by T-cell clones is mediated by the TCR. Our unpublished experiments (Scelfo A. et al) show that non-CD1-restricted T cells acquired CD1a or CD1c self-reactivity, respectively, upon transduction with lentiviral vectors encoding the TCR-α/β chains expressed by the CD1a self-reactive DN F.20 and CD1c self-reactive DN 4.99 T-cell clones (listed in Supporting Information Table S1). Although only two T-cell clones were tested, the experiment supports the likelihood that CD1 recognition by self-reactive T-cell clones is mediated by TCR.
Although we have excluded CD1d-restricted Vα24+NKT cells from the analysis, we cannot formally rule out the possibility the part of the remaining CD1d self-reactive T-cell repertoire that we have identified comprised also the so-called type II NKT cells. These are CD4+ and DN CD1d self-reactive T cells identified in mice, where they do not express the invariant Vα14-Jα18 TCR though, like iNKT cells, they express NK1.1, display a constitutive effector/memory phenotype and can make IFN-γ 20. However, although human type II NKT cells have not yet been clearly identified, we do not find an enrichment of CD1d self-reactive T cells in the effector/memory compartments of neonates (data not shown), suggesting that we are probably not detecting type II NKT cells in our screenings.
The mechanism by which the CD1 self-reactive T-cell repertoire is generated remains unknown. The great inter-individual variability in the frequency of CD1 self-reactive T cells, which results in sporadic holes in the repertoire of CD1a, CD1c or CD1d-restricted T cells found in some donors, might be due to variations in the thymic development/selection of these cells. This is not completely unexpected, because also human iNKT cells exhibit a 1000-fold inter-individual variability in frequency (0–1%), which seems to be genetically controlled 21 and might be a common feature for all CD1 self-reactive T cells. Why CD1b self-reactive T cells are not detectable in the great majority of investigated adults is intriguing. One possibility could be that CD1b-restricted T-cell repertoire is poorly auto-reactive, as a result of a thymic selection process. Self-lipids that are qualitatively or quantitatively different from those presented by C1R cells would be required to activate cytokine production in these T cells. CD1b-restricted T cells could indeed be isolated from adult PBMCs when high concentrations of purified self-glycosphingolipids were added to the cultures 22.
However, we could detect CD1b self-reactive T cells at higher frequencies in newborns than adults, particularly in the CD4+ subset. More generally, we observed an evolution of the relative frequencies of self-reactive T cells restricted for each CD1 isotype going from neonates to adults. This would suggest that the CD1 self-reactive T-cell repertoire is dynamically shaped by integrated mechanisms that act not only in thymus, but also in the periphery, such as survival, expansion and tissue homing, which could contribute to modify the CD1 self-reactive T-cell repertoire between neonatal and adult blood, and possibly also among different adult donors.
The unexpected high frequency of CD1 self-reactive and potentially dangerous T cells raises questions about their functional control and physiological role. The control of CD1 auto-reactivity could be achieved at steady state at the level of activation threshold of CD1 self-reactive T cells, owed to the expression of inhibitory receptors, as described for conventional T cells 23, lack of agonist self-lipid Ags and costimulatory molecules in APCs 11, 24 and to suppression by CD4+CD25+Foxp3 Tregs 25. Conditions of tissue damage and inflammation that concomitantly counteract Treg suppression 25 and promote the upregulation of stimulatory self-lipids, CD1 molecules and inflammatory cytokines by activated APCs 24, 26 might be required to unleash CD1 self-reactivity. Once activated, CD1 self-reactive T cells could regulate both cell-mediated and humoral immune responses, as suggested by the heterogeneous TH1/TH0 cytokine pattern that we have found in our CD1 self-reactive T-cell clones. These T cells could have a positive role in host protection by assisting the induction of the conventional T-cell response against infectious pathogens 11. Moreover, their cytolytic capacity against CD1+ target cells could allow them to eliminate pathogenic cells in which the expression of CD1 and/or the lipid metabolism can be altered and modified, for example tumor cells 27. However, it is also possible that the CD1 self-reactive T-cell response may sustain autoimmunity, as suggested by the increased frequency of CD1 self-lipid reactive T cells in multiple sclerosis patients 22, or by the capacity of CD1c self-reactive DN T cells to help B cells producing pathogenic IgG in systemic lupus erythematosus 28. At the clonal level, we find different cytokines produced according to the CD4+ or DN phenotype or to the CD1 isotype restriction, suggesting some degree of functional specialization in the CD1 self-reactive T-cell compartment. It will be important to determine whether the cytokine pattern of CD1 self-reactive T cells is modified upon different Ag-priming conditions or in pathological conditions, as found in some T-cell clones from lupus patients 28.
The naïve/memory dynamics displayed by CD1 self-reactive T lymphocytes was not anticipated. It resembles that of MHC-restricted T cells of adaptive response and contrasts with the innate-like characteristics of the CD1d-restricted iNKT cells. iNKT cells acquire the constitutive effector/memory phenotype as the result of lineage-specific developmental cues that involve agonistic thymic selection by CD1d-expressing DP thymocytes 29. Human DP thymocytes express all the four CD1 isotypes 3 and could also select CD1 self-reactive T cells; however, the fact that the majority of CD1 self-reactive T cells are naïve at birth suggest that their thymic development is regulated differently from iNKT cells.
The observation that newly generated CD1 self-reactive T cells are naïve raises the question as to whether these cells require priming by self-lipids in order to differentiate into full effector cells. These stress conditions following infections and inflammation described above could provide adequate conditions for priming naïve T cells with self-lipids. Furthermore, in transgenic mice humanized for the human group 1 CD1, CD1-restricted T cells must be primed with mycobacterial lipid Ags in order to acquire effector functions 30. As the majority of CD1-restricted T cells induced by mycobacterial lipid were also self-reactive 30, these T cells could be primed also by exogenous Ag recognition.
While this manuscript was under review, the study by de Jong et al. 31 reported a remarkable high frequency of CD1a self-reactive T cells in healthy donors, accounting for the greatest majority of the CD1 autoreactive T-cell repertoire. The comparative frequencies of CD1 self-reactive T cells were obtained after one or two rounds of in vitro stimulation with autologous DC expressing all the four CD1 isotypes, followed by ELISPOT assayed against K562 cells, which display low surface levels of MHC, engineered with single CD1 isotypes. With our direct ex vivo cloning approach and LDA, we detect a similar prevalence of CD1a self-reactive T cells; however, we find higher frequencies of CD1c and CD1d self-reactive T cells compared to the published ones 31. The different cell culture systems utilized by ours and de Jong's study might account for the different frequencies of self-reactive T cells restricted for the various CD1 isoforms. Yet, both studies underscore the unexpected abundance of this unconventional T-cell repertoire in the human immune system.
The relevant size of the CD1 self-reactive T-cell compartment warrants future studies to define its potential roles in the immune response in healthy and disease conditions.
Materials and methods
Primary lymphocyte preparations
PBMCs and UCBLs were isolated by a Ficoll-Hypaque density gradient and immediately utilized. For UCBLs informed consent of mothers were obtained.
T-cell sorting and cloning
TCR-α/β+CD4+ and TCR-α/β+CD4−CD8− cells were sorted following staining with anti-TCR-α/β-PE, anti-CD8-FITC (Becton Dickinson), anti-CD4-QD (Sigma) mAbs, plus anti-Vα24-FITC mAb 5 to gate out iNKT cells. The gating strategies utilized in the sorting are described in Supporting Information Fig. 1. TCR-α/β+CD45RA and TCR-α/β+CD45R0 cells from the CD4+ or DN subsets were sorted from adult PBMCs or UCBLs, previously depleted of iNKT cells by immunomagnetic sorting with anti-Vα24 mAbs, by staining with anti-TCR-α/β-PE, anti-CD4-Cy5, anti-CD45R0 FITC, anti-CD45RA-bio mAbs and Streptavidin-APC (Becton Dickinson), or with anti-TCR-α/β PE, anti-CD4-Cy5 and anti-CD8-Cy5, anti-CD45R0-FITC, anti-CD45RA-bio mAbs and Streptavidin-APC. Subsets were sorted using a Beckman Coulter Moflo cell sorter equipped with a Cyclone automated cloning device, which distributed 1 sorted cell per well of 96 U-bottomed well plates containing 106 irradiated allogenic PBMCs and 1 μg/mL PHA in 200 μL of Glutamax RPMI (Gibco) supplemented with 5% NHS (Euroclone) (RPMI-NHS) and 200 U/mL rhIL-2, as described 5. Growing T cells were restimulated with the same protocol before assessing CD1 self-reactivity. Coreceptor expression of growing clones was verified by staining with CD4- and CD8-specific mAbs. Cells were acquired on a BD LSR-II cytometer and data analyzed with FloJo.
T-cell clones were tested for CD1-restricted recognition against: C1R B cell line transfected with each CD1 isotype or mock transfected 32 (provided by Dr. S. Porcelli, Albert Einstein College of Medicine, New York, NY), THP-1 acute myelo-monocytic leukemia cell lines transfected with CD1c cDNA (THP-1-CD1c) (M. Lepore and G. de Libero, data not shown), monocyte-derived DC prepared from PBMCs as described 33. The expression of the CD1 isotypes of all APCs utilized was monitored before experiments by flow cytometry. T-cell clones were cocultured at 1:10, 1:3 and 3:1 ratios with irradiated C1R-CD1, THP1 cells and DC, respectively for 48 h. CD1 recognition was blocked with 20 μg/mL of anti-CD1a (OKT6, ATCC CRL8019), anti-CD1b (WM25, Millipore), anti-CD1c (F10/2A3, provided by Dr. S. Porcelli) and anti-CD1d (42.1, BD) mAbs. Irrelevant mouse IgG (20 μg/mL) were used as the negative control. Cytokine production was assessed by the standard ELISA or Cytokine Beads Array (CBA).
CD3+ cells were sorted from PBMCs of healthy adult donors and seeded at 10, 25, 50, 100, 250 cells/well in 96 U-bottomed well plates and expanded as previously described for clones. After 14 days, replicated T-cell cultures were screened for recognition of THP1-CD1c: 48 cultures for 10 cell/well input, 50 cultures for 25 and 50 cell/well input, 40 cultures for 100 cell/well input, 35 cultures for 250 cell/well input were tested. THP1-CD1c cells increased the signal-to-noise ratio in the activation of T-cell clones compared to C1R-CD1c cells (data not shown), which was required to detect the CD1c self-reactive T-cell precursors at the higher cellular densities. To minimize differences in the T:APC ratio, the numbers of T cells and APC used for the assay was adjusted according to the initial T-cell input: 0.5×105 T cells from 10 cell/well input were cocultured with 7.5×103 ThP1-CD1c cells, 1×105 T cells from 25, 50 and 100 cell/well inputs were cocultured with 1.5×104 ThP1-CD1c cells; 3×105 T cells from 250 cell/well input were cocultured with 5×104 THP1CD1c cells. Cells were cultured in the presence of 20 μg/mL each of blocking HLA class I (W632) and HLA-DR (L243)-specific mAbs, with or without anti-CD1c mAb or mouse IgG isotype control. After 16 h, culture supernatants were assayed for IFN-γ content by the standard ELISA (Endogen). T cells were also cultured without APCs to evaluate the spontaneous IFN-γ release. T-cell cultures were considered positive for CD1c self-reactive T-cell precursors when the inhibition percentage of IFN-γ production in the presence of anti-CD1c mAb vs IgG isotype was twice the Coefficient of Variation % of the screening assay. The frequency of CD1c self-reactive T-cell precursors was calculated according to the Poisson distribution 34.
TCR V gene usage
Total RNA extracted from the selected T-cell clones was converted into cDNA and amplified by PCR utilizing oligonuclotide primers specific for the human TCR Vβ and Vα genes coupled to primers specific for Cβ and Cα, respectively, as previously described 35. The presence of the correct PCR products was assessed by gel electrophoresis.
This study was funded by grants from the Italian Association for Cancer Research (AIRC-5804) to G. C., the International Association for Cancer Research (AICR) to P. D., G. C. and G. D. L., the Italian Ministry of Health Programma Straordinario Ricerca Oncologica RFPS-2006-4-341763 to G. C., the Swiss National Fund No. 3100A0-122464/1 to G. D. L. Marco Lepore was supported by fellowships from the PhD Program in Molecular Medicine, University of Vita-Salute San Raffaele (Milano, Italy).
Conflict on interest: The authors declare no financial or commercial conflict of interest.