CD6 modulates thymocyte selection and peripheral T cell homeostasis

The role of CD6 in T cell development was assessed by generating CD6^−/− C57BL/6N mice in which exons coding for the extracellular region of CD6 were deleted by homologous recombination (Fig. 1 A). Flow cytometry analysis of LN CD4⁺ and CD8⁺ T cells from CD6^−/− mice showed a complete lack of CD6 surface expression as compared with CD6^+/+ and CD6^+/− littermates (Fig. 1 B). Analysis of thymus from sex- and age-matched CD6^−/− and CD6^+/+ littermates showed no significant differences in total cell numbers (Fig. 1 C, left). Although no differences were found in the frequency of NKT (CD161⁺CD3⁺) and γδ T cells (not depicted), as well as of DP or double-negative (DN) thymocyte subsets, a significant decrease was observed for CD4⁺SP cells (Fig. 1 C, middle). A detailed analysis of CD8⁺SP cells showed a reduction in the frequency of mature CD8⁺SP (TCRβ⁺CD24⁻), together with an increase in immature CD8⁺SP (TCRβ⁻CD24⁺; Fig. 1 C, right). These findings suggest the involvement of CD6 in the final steps of thymocyte selection. Consistent with that view, CD6 expression correlated with that of the positive selection marker CD69 in DP thymocytes (Fig. 1 D). A similar situation has been documented for CD5, a negative modulator of TCR signaling (Tarakhovsky et al., 1995) whose expression is developmentally regulated in a TCR signal- and avidity-dependent manner (Azzam et al., 2001). A significant increase in the frequency of DP cells undergoing selection (CD69⁺DP) and a concomitant reduction in recently selected CD69⁺CD4⁺SP and CD69⁺CD8⁺SP was found in CD6^−/− mice as compared with CD6^+/+ mice (Fig. 1 E). These results suggest that, rather than a defect in positive selection per se, CD6 deficiency triggers an increase in the negative selection of CD4⁺ and CD8⁺SP cells, the final product of the DP to SP transition.

Analysis of spleen and LN from CD6^−/− mice revealed total cell numbers (Fig. 2 A, left) and percentage of T (both CD4⁺ and CD8⁺), B (CD19⁺; Fig. 2 A, right), γδT, NK, and NKT cells (not depicted) similar to those found in CD6^+/+ littermates. Surface levels of CD3, CD4, CD8, and CD45 from peripheral CD6^−/− T cells were also similar to those of CD6^+/+ cells (Fig. 2 B). However, a slightly reduced CD5 expression was observed on both CD4⁺ and CD8⁺ spleen T cells from CD6^−/− mice compared with CD6^+/+ littermates (Fig. 2 C). This was also true for thymocytes, from early to late maturational stages (Fig. 2 C). The intriguing fact is that the analysis of CD5^−/− mice revealed increased CD6 expression in both CD4⁺ and CD8⁺ spleen T cells but not thymocytes. These reciprocal changes observed in both CD6^−/− and CD5^−/− mice suggest a functional cross-talk in the regulation of CD6 and CD5 surface expression that deserves further exploration.

Analysis of spleen and LN also revealed that CD6^−/− mice present significantly increased percentages of effector/memory CD4⁺ (CD4⁺T_EM and CD44^hiCD62L^lo) and central memory CD8⁺ (CD8⁺T_CM and CD44^hiCD62L^hi) T cells (Fig. 2 D). Increased percentages of cells with T reg cell phenotype (CD4⁺CD25⁺FoxP3⁺) were also observed (Fig. 2 E). Interestingly, this increase could not be detected in the thymus, suggesting that CD6 deficiency does not affect the generation of thymic T reg cells. Collectively, the data indicate that CD6 deficiency favors the expansion of antigen-experienced peripheral T cell subsets generated under high TCR signal strength conditions (Lanzavecchia and Sallusto, 2005).

Increased frequency of effector/memory T cells has been attributed to homeostatic proliferation, also known as lymphopenia-induced proliferation (Jameson, 2002). Such an increase could result from the decreased thymic output observed in the absence of CD6 (Fig. 1 C). However, the lower BrdU incorporation observed in spleen and LN CD8⁺T_CM and CD4⁺T_EM cells from CD6^−/− mice (Fig. 2 F) argued against the possibility of increased homeostatic proliferation. Instead, the previously reported role of CD6–ALCAM interactions in optimizing T cell activation and proliferative responses (Gimferrer et al., 2004; Hassan et al., 2004; Zimmerman et al., 2006) could feasibly account for the low in vivo proliferation observed.

To determine whether the defect in developing and mature lymphocytes observed in CD6^−/− mice is T cell–intrinsic, competitive mixed BM chimera experiments were performed. CD3-ε–deficient mice were used as recipients of BM cell mixes (1:1) from CD6^+/+ (WT45.1) mice with that of CD6^−/− (KO45.2) or CD6^+/+ (WT45.2) mice (Fig. 3 A). Control mice injected with only CD6^+/+ or CD6^−/− BM cells showed no significant differences in total cell numbers from primary and secondary lymphoid organs (unpublished data), indicating that absence of CD6 did not affect BM engraftment and/or thymic seeding. In the mixed chimera control group (WT45.2/WT45.1), a competitive advantage for CD45.2 cells was observed as visible by CD45.2/CD45.1 ratios >1 (Fig. 3 B). However, this advantage was modulated in the experimental group (KO45.2/WT45.1), where the CD45.2/CD45.1 ratio was substantially reduced. In the thymus, this effect was observed for all major thymocyte developmental subsets, though it reached statistical significance only for DN and CD4⁺SP and CD8⁺SP. This indicates that CD6^−/− T cells present a competitive disadvantage from very early stages of their development in regard to their CD6^+/+ counterparts. In the periphery, a significant reduction of CD6^−/− SP CD4⁺ and CD8⁺ T cells was also apparent in LNs (Fig. 3 C) and spleen (not depicted). Importantly, such a disadvantage was restricted to T cells, as the percentage of CD6^−/− CD45.2 B cells in both the LN (Fig. 3 C) and spleen (not depicted) was not affected as compared with the control group. Overall, these results indicate that CD6 deficiency confers a selective disadvantage to developing T cells in a cell autonomous manner and that, in a competitive setting, the presence of peripheral CD6^−/− T cells was also reduced. As observed in CD6^−/− mice, (Fig. 2, D and E), increased frequencies of CD6^−/− T reg and CD8⁺T_CM cells could be detected in peripheral lymphoid organs of KO45.2/WT45.1 mixed-BM chimeras compared with their CD6^+/+ counterparts from control WT45.2/WT45.1 mixed chimeras (Fig. 3 D), indicating that such phenotypic alterations are also T cell intrinsic.

The aforementioned results suggest that CD6 deficiency alters thymocyte selection by lowering the threshold for negative selection and decreasing the number of cells reaching the CD4⁺SP and CD8⁺SP stage. Considering that TCR repertoire readjustments may occur and complicate result interpretation in mice with polyclonal T cell populations, we further addressed the role of CD6 in intrathymic T cell section using CD6^−/− mice in which all T cells bear a single specific TCR. For that purpose, CD6^−/− mice were crossed with either OT-I TCR mice, which produces MHC class I–restricted, OVA-specific CD8⁺ T cells, or with Marilyn mice, whose CD4⁺ cells are MHC class–II restricted for the male HY antigen. CD6^−/− OT-I mice showed a reduction in the number of thymic CD8⁺SP cells expressing the TCR transgene, which reached statistical significance in peripheral organs (Fig. 4 A). Interestingly, CD6^−/− OT-I mice also showed an increased proportion of peripheral CD8⁺ T_CM cells (CD44^hiCD62L^hi) compared with their CD6^+/+ counterparts (Fig. 4 B). In Marilyn mice, female CD6^−/− showed a reduced number of thymic CD4⁺SP cells expressing the TCR transgene, which again reached statistical significance in spleen and LNs (Fig. 4 C). Collectively, these data indicate that CD6 regulates TCR signaling on developing and antigen-experienced T cells.

This result is reminiscent of the well-acknowledged negative regulatory role of CD5 in T cell development and activation (Tarakhovsky et al., 1995). The CD5 receptor is closely related to CD6 both structurally and genetically (Soldevila et al., 2011) and its expression is developmentally regulated by TCR avidity, contributing to the fine tuning of TCR signaling (Azzam et al., 2001). Consequently, CD5 surface levels on thymocytes and peripheral T cells parallel the avidity of their TCRs, as demonstrated by the higher levels of Nur-77 found in CD5^hi peripheral T cells compared with CD5^lo (Fulton et al., 2015). Nur-77 levels are considered a sensitive readout of TCR signaling strength and are also found to be elevated in CD6^hi peripheral T cells compared with CD6^lo counterparts (Fig. 4 D), further supporting the role of CD6 as a putative negative regulator under high-avidity TCR conditions.

The fate of developing thymocytes is determined by the TCR signal strength resulting from interaction with MHC–self peptide complexes presented by thymic APCs (Palmer, 2003). Such a signal is negatively modulated by CD5, and as a consequence CD5^−/− T cells are hyperresponsive to TCR cross-linking (Tarakhovsky et al., 1995). The analysis of TCR-induced Ca²⁺ flux showed that CD6^−/−, similar to CD5^−/−, displayed higher intracellular [Ca²⁺] than CD6^+/+ cells (Fig. 5 A). Further analyses showed that CD6^−/− DP, but not CD4⁺SP or CD8⁺SP, were responsible for the increased Ca²⁺ responses observed (Fig. 5 B). Consistent with this, no differences in Ca²⁺ responses were observed for spleen CD4⁺ or CD8⁺ T cells from CD6^−/− and CD6^+/+ mice (unpublished data). The increased Ca²⁺ responses observed in CD6^−/− DP thymocytes could not be attributed to the partial CD5 down-regulation observed in these cells, as Ca²⁺ fluxes from CD5^+/− mice, which display similar CD5 surface levels to those observed in CD6^−/− (Fig. 5 C, left), did not differ from those observed with CD6^+/+ thymocytes (Fig. 5 C, right). Moreover, CD5^+/− mice did not show other alterations found in CD6^−/− mice, such as reduced frequency of thymic CD4⁺SP cells or increase of peripheral CD8⁺T_CM cells (Fig. 5 D). Thus, the phenotype observed for CD6^−/− mice, in terms of Ca²⁺ flux, decrease in thymocyte maturation and increase of memory phenotype in the periphery, is caused by CD6 deficiency and not to the reduced expression of CD5.

Overall, Ca²⁺ measurements indicate that CD6 behaves as a negative regulator of TCR signaling in immature (DP) but not mature (SP) thymocytes nor peripheral T cells. However, in vitro CD3-mediated stimulation of CD6^−/− CD4⁺ and CD8⁺ peripheral T cells resulted in slightly increased proliferative responses, which reached statistical significance for CD4⁺ cells (Fig. 6 A) but disappeared under CD28 co-stimulatory conditions. In contrast, CD5^−/− T cells, used for comparative purposes, showed significantly higher proliferative responses under both stimulatory conditions (Fig. 6 A). This indicates that, under ligand-independent conditions, CD6 has the capacity of negatively regulating T cell activation. However, it should be mentioned that under ligand-dependent conditions (e.g., MLR) the missing CD6–ALCAM interaction could result in suboptimal cell–cell contacts and net lower proliferative responses of CD6^−/− T cells (unpublished data).

The increased proportion of peripheral T reg cells led us to assess their functional status. As shown in Fig. 6 B, CD6^−/− spleen CD4⁺CD25⁺ T cells exhibited lower suppressive activity in vitro compared with CD6^+/+ counterparts, indicating that CD6 expression is required for full T reg cell functionality. These results are in agreement with the fact that T reg cell function has been linked to correct SLP-76–mediated signaling (Schmidt et al., 2015), and that CD6 constitutes a key element of the SLP-76 interactome (Roncagalli et al., 2014). Thus, CD6 deficiency would prevent T reg cells from proper recruitment of SLP-76, an essential signaling effector for their function.

To further explore the possibility that CD6 deficiency may result in dysregulated immune responses to auto-antigens in vivo as a consequence of defective T reg cell function and/or defective modulation of TCR signaling, we evaluated the susceptibility of CD6^−/− mice to CIA, a well-established model of T cell–mediated autoimmune disease resembling rheumatoid arthritis. As illustrated by Fig. 6 C, CD6^−/− mice showed an earlier CIA onset and a significantly increased CIA incidence and clinical score than their CD6^+/+ littermates. This exacerbated CIA form was accompanied by a significantly enhanced joint expression of TNF and IL-6 transcripts and a reduced expression of FoxP3 (Fig. 6 D). Increased expression of IL-17A and IL-1β transcripts were also observed, but these differences did not reach statistical significance (Fig. 6 D). No differences regarding IFN-γ, IL-10, and TGF-β transcript levels were found (unpublished data). These results are compatible with CD6 deficiency potentiating an induced Th17 autoimmune response to collagen in the context of a dysfunctional T reg cell activity. This is particularly relevant in the light of a study correlating polymorphisms in the human CD6 gene with susceptibility to certain autoimmune diseases, such as rheumatoid arthritis or multiple sclerosis (Kofler et al., 2016). Although the functional consequences of CD6 polymorphism associated to rheumatoid arthritis has not been analyzed, carriers of the multiple sclerosis risk allele (rs17828933^GG) show altered CD4⁺ T cell responses and changes in the proportion of full-length and CD6ΔD3 isoforms in both CD4⁺ and CD8⁺ cells (Kofler et al., 2016), but their consequences on disease susceptibility merit further study.

This study represents the first in vivo characterization of CD6 deficiency. The results show that CD6 acts as a negative modulator of TCR-mediated signaling, setting the threshold for the final steps of T cell selection and maturation, and likely for peripheral T cell activation. The molecular mechanisms by which CD6 restrains TCR signaling are speculative, as CD6 signaling pathway is still relatively unknown. Given the reported association of CD5 and CD6 at the T cell surface (Gimferrer et al., 2003) and the regulatory role of CD6 on CD5 Tyr phosphorylation (Castro et al., 2003), CD5 could mediate such a negative effect. This situation would be analogous to that reported for CD2, which transduces inhibitory signals through its association with CD5 (Teh et al., 1997; Castro et al., 2002). Alternatively, CD6 could directly recruit downstream negative signaling effectors (e.g., protein Tyr phosphatases). The possibility that, apart from decreasing the threshold for negative selection, the absence of CD6 impacts thymocyte selection in other manners cannot be excluded. Indeed, CD6 has been related to resistance to apoptosis in different experimental settings. Accordingly, CD6 is up-regulated both in DP cells undergoing positive selection and thymocytes resistant to in vitro induced-apoptosis (Singer et al., 2002), and CD6 ligation protects chronic lymphocytic leukemia cells from IgM-induced apoptosis (Osorio et al., 1997). Moreover, CD6 expression protects leukemic T cells from galectin-induced apoptosis (Escoda-Ferran et al., 2014). Thus, further investigation into the role of CD6 as a pro-survival protein is required.

The CD6–ALCAM interaction is certainly a relevant driver of the observed effects on T cell development and peripheral T cell homeostasis. However, it has been observed that interactions between developing thymocytes (CD6⁺) and TECs (ALCAM⁺) are not sustained by long-term synapses, but rather by transient and repeated contacts with different TECs that maintain NFAT transcription and allow positive selection (Ebert et al., 2008). According to this model, the CD6–ALCAM interaction might be somewhat dispensable in the thymus, where the negative role of CD6 would be more evident.

The generation and maintenance of antigen-experienced T cells, such as CD4⁺T_EM, CD8⁺T_CM, and T reg cells, in CD6^−/− mice has been linked to a certain level of TCR signal strength and/or co-stimulation (Lanzavecchia and Sallusto, 2005). This fits with the view of CD6 as a negative regulator of thymic and peripheral TCR-mediated signaling. Indeed, absence of CD6 led to increased negative selection in the thymus and increased activation in response to self- or environmental antigens in the periphery, as shown by expansion of T cell subsets with memory and regulatory phenotypes. These data are reminiscent of those obtained from the analysis of CD5^−/− mice, where expansion of both memory and regulatory (Ordoñez-Rueda et al., 2009) T cells are also observed. Thymic generation of T reg cells requires high-avidity interaction of TCR with self-peptide–MHC complexes. Accordingly, increased frequency of T reg cells in thymus and periphery were previously reported for CD5^−/− mice, the suppressive functionality of which is debatable (Dasu et al., 2008; Ordoñez-Rueda et al., 2009). However, although increased proportions of T reg cells can be seen in thymus and periphery of CD5^−/− mice, they could only be detected in peripheral lymphoid organs of CD6^−/− mice. This would indicate that CD6 is particularly relevant for regulating the generation of extrathymic T reg cells during immune responses, rather than during thymic T cell selection. Furthermore, the ex vivo analysis of CD6^−/− spleen T reg cells showed a lower suppressive activity, further indicating that CD6 is also relevant to T reg cell functionality. The in vivo biological meaning of our findings is supported by the fact that CD6^−/− mice present an exacerbated form of an inducible autoimmune disease model (CIA).

Thus, the CD6-deficient mice described here represent a new and useful tool to fully understand the multifaceted mechanisms by which CD6 exerts its regulatory functions in different physiological and pathological contexts. Importantly, these mice can also provide valuable insights for the CD6-based therapies that are currently under assessment for clinically relevant diseases such as psoriasis, rheumatoid arthritis, or lymphoproliferative disorders (Hernández et al., 2016).

CD6-deficient mice were generated in C57BL/6N background as previously described (Skarnes et al., 2011) through the Mouse Biology Program (University of California, Davis, Davis, CA) using targeting vector DPGS00142_B_G12 from the trans-NIH Knock-Out Mouse Project (KOMP), and obtained from the KOMP Repository. The gene-targeting vector was designed to delete exons 3–6 from the cd6 gene coding for the three SRCR extracellular domains and the stalk region. It was successfully used to obtain a recombinant ES cell clone (JM8A1.N3) that gave rise to chimerical mice after injection into pseudo-pregnant female mice. After germ-line transmission of the mutant allele, denoted as C57BL/6N-A^tm1Brd, a homozygous (CD6^−/−) colony was established. Absence of CD6 protein expression on T cells was confirmed by flow cytometry. The CD6^−/− mice were viable and fertile, and their lifespans and behaviors were equivalent to their WT littermates. CD5-deficient mice (Tarakhovsky et al., 1995) on C57BL/6 background were provided by C. Raman (University of Alabama, Birmingham, AL). Mice were maintained at the animal facilities of the Universitat de Barcelona School of Medicine (Barcelona, Spain) under specific pathogen–free conditions.

CD6^−/− and CD6^+/+ mice were bred with MHC class I–restricted, OVA-specific OT-I (Clarke et al., 2000) and MHC class II–restricted male HY-specific Marilyn (Scott et al., 2000) TCR-transgenic lines at the Centre d’Immunologie Marseille-Luminy (CIML; Marseille, France) and maintained in SPF conditions. CD3ε-deficient (CD3ε^Δ5/Δ5) mice displaying thymocyte development arrest at DN3 stage (Malissen et al., 1995) used as hosts in BM chimera experiments were also kept at CIML. All the experiments were carried out in accordance with the guidelines of the Comité Étic d’Experimentació Animal from Universitat de Barcelona. Sex- and age-matched (8–12 wk) mice were used in all the experiments.

Cytosolic Ca²⁺ signal was determined by two different approaches. In the case of fluorescence microscopy assessment of TCR-induced store operated calcium release (SOCE) over time, cells were loaded with 4.5 µM Fura-2 AM (30 min) as previously described (Carreras-Sureda et al., 2013). Cytosolic [Ca²⁺] increases are presented as the ratio of emitted fluorescence (510 nm) after excitation at 340 and 380 nm, relative to the ratio measured before cell stimulation (Fura-2 AM ratio 340/380). All experiments were performed at room temperature, and cells were bathed in a solution containing 140 mM NaCl, 5 mM KCl, 1.2 mM CaCl₂, 0.5 mM MgCl₂, 5 mM glucose, 10 mM Hepes (300 mosmol/l, pH 7.4). Ca²⁺ entry was achieved using 1 µg/ml of anti-CD3ε mAb (145.2C11; BD), and cross-linking was performed using goat anti–Armenian hamster IgG (MP Bio). In the case of flow cytometry analysis, thymus and spleen cells were loaded for 20 min at 37°C with Indo-1 AM (Molecular Probes), and then stained with fluorescent-labeled anti-CD4 (RM4-5, PE; BD) and anti-CD8 (53–6.7, APC; BD). Basal fluorescence was recorded with an LSRII-UV (BD) before addition of 2.5 µg biotinylated anti-CD3 (145.2C11; BD). Cross-linking of anti-CD3 mAb was induced by the addition of streptavidin (10 µg/ml; Thermo Fisher scientific). Calcium fluxes were determined on gated CD4⁺ and CD8⁺ cells. As loading control, cells were stimulated with Ionomycin (1 µg/ml; Sigma-Aldrich).

Cell staining was done with fluorescently labeled mAbs to CD5 (53–7.3), CD4 (RM4-5), CD8α (53–6.7), CD24 (M1/69), TCR-Vα2 (B20.1TCR-Vβ6 (RR4-7), TCRγ (GL-3), and TCRβ (H57-597) were purchased from BD; CD62L (MEL-14), CD44 (IM7), CD69 (H1.2F3), CD3 (145.2C11), and FoxP3 (3G3) were obtained from TONBO Bioscience; CD25 (PC61.5), CD19 (6D5), CD161 (PK136), and CD6 (OX129) were purchased from BioLegend; and Nur77 (12.14) was obtained from eBioscience; all were used according to the manufacturer’s instructions. FACS analyses were done on a LSRII 564 or FACS Canto II system (BD) using FlowJo software (TreeStar). Cell viability was evaluated either with SYTOX Blue (Life Technologies) or 633- or 405-nm Viability dyes (Life Technologies). Apoptotic cells were detected by using Annexin V-FITC apoptosis detection kit (Immunostep).

Before use, 8-wk-old CD3ε^Δ5/Δ5 recipient mice received one single dose of irradiation (300 rads) to eliminate partially endogenous immune cells and to allow efficient engraftment. Mice were then injected i.v. with 2 × 10⁶ BM cells from CD6^+/+ (CD45.1) and from CD6^−/− (CD45.2) mice, to generate mixed chimeras. As control groups, recipient CD3ε^Δ5/Δ5 mice (CD45.2) were injected with 4 × 10⁶ BM cells from either CD6^+/+ or CD6^−/− mice. Mouse chimeras were kept on antibiotic-containing water (0.2% Bactrim; Roche) during the whole experiment and euthanized at 12 wk for FACS analyses on gated CD3⁺ or B220⁺ lymphocytes.

For the detection of proliferating cells in vivo, CD6^−/− and CD6^+/+ littermate mice were administered for 1 wk with BrdU (0.8 mg) into drinking water, changing water every 2 d. BrdU⁺ cells were visualized by FACS with BrdU Cell Proliferation Assay kit (BD) following the manufacturer’s instructions.

CD4⁺ and CD8⁺ T cells were negatively isolated from spleen and LN (Dynabeads; Life Technologies) and incubated for 15 min with CellTrace CFSE Cell Proliferation kit (Life Technologies), and then cultured in complete RPMI medium (10% FCS, 10 mM Hepes, 1 mM sodium pyruvate, 50 µM β-Mercaptoethanol, and 5 mM Penicillin-Glutamine-Streptomycin; Lonza) in 96-well plates coated with increasing concentrations of purified anti-CD3 mAb (145.2C11; TONBO Bioscience) alone or in combination with 1 µg/ml soluble anti-CD28 mAb (37.51; TONBO Bioscience) for 72 h at 37°C, 5% CO₂.

For T reg cell suppression assays, CD4⁺CD25⁻ (T conv cells) and CD4⁺CD25⁺ (T reg cells) T cells were electronically sorted from spleen specimens and, upon CFSE-labeling of the former, co-cultured for 3 d at different ratios in 96-well plates coated with anti-CD3 mAb plus soluble anti-CD28. Proliferative responses were analyzed by determining the percentage of CFSE^lo cells by FACS.

Induction and assessment of CIA was done in 8–12-wk-old male mice, as previously described (Postigo et al., 2011). In brief, mice were immunized once at the base of the tail with 150 µg of chicken collagen type II dissolved at a concentration of 2 mg/ml in 0.05 M acetic acid and emulsified with CFA containing 4 mg/ml of Mycobacterium tuberculosis in a final volume of 150 µl. Evaluation of arthritis clinical severity was performed weekly according to a graded scale of 0–3 as follows: 0, no inflammation (normal joint); 1, detectable local swelling and/or erythema; 2, swelling in >1 joint and pronounced inflammation; 3, swelling of the entire paw and/or ankylosis. Each paw was graded, and the scores were summed (with a maximum possible score 12 per mouse). The expression of arthritogenic cytokines and FoxP3 in the joints was explored before and 8 wk after CIA induction by quantitative real time-PCR, as described previously (Postigo et al., 2011). Results (in triplicate) were normalized to GAPDH expression and measured in parallel in each sample.

We are indebted to Elise Bergot and to the Cytomics core facility of the Institut d'Investigacions Biomèdiques August Pi i Sunyer for technical help.

This work is supported by grants from: Worldwide Cancer Research (14-1275) and Fundació La Marató TV3 (201319-30) to F. Lozano; Spanish Ministerio de Economía y Competitividad (Plan Nacional I+D+i) co-financed by European Development Regional Fund “A way to achieve Europe” to F. Lozano (SAF2013-46151-R), J. Merino (SAF2012-34059), R. Merino (SAF2014-55088-R), and R. Vicente (SAF2014-52228-R); IPT2011-1527-010000 (associated to Fibrostatin SL) to J. Merino; and Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale, European Research Council (FP7/2007–2013 grant no. 322465), and ANR grant Basilic to B. Malissen and M. Malissen. M. Orta-Mascaró, I. Simões, and E. Carreras are recipients of fellowships from Ministerio de Economía y Competitividad (BES-2011-048415), Fundação para a Ciência e a Tecnologia (SFRH/BD/75738/2011), and European Community Seventh Framework Program (BIOTRACK, FP7/2007/2013;229673), respectively. National Institutes of Health grants to Velocigene at Regeneron Inc. (U01HG004085), and the CSD Consortium (U01HG004080) funded the generation of gene-targeted vectors and ES cells for 8,500 genes in the KOMP Program and archived and distributed by the KOMP Repository at UC Davis and CHORI (U42RR024244).

The authors declare no competing financial interests.