Rescue of Tumor-Infiltrating Lymphocytes from Activation-Induced Cell Death Enhances the Antitumor CTL Response in CD5-Deficient Mice

CD5 is a 67-kDa transmembrane glycoprotein that belongs to the highly conserved scavenger-receptor cysteine-rich superfamily (1, 2). It is constitutively expressed on thymocytes and mature T lymphocytes (3), and is associated both physically and functionally with the TCR/CD3 complex (4). Accumulating evidence indicates that CD5 acts as a negative regulator of TCR signaling in thymocytes and mature T lymphocytes, and that T cells lacking CD5 are more responsive to TCR stimulation (5, 6). More recently, it has been demonstrated that CD5 is rapidly recruited and colocalized with the TCR/CD3 complex at the immune synapse (7), and that it inhibits TCR signaling in T lymphocytes interacting with APC without influencing conjugate formation (8). Furthermore, it has been reported that CD5-mediated inhibition of TCR signaling does not require the CD5 extracellular domain, but only its cytoplasmic tail (9), in which a pseudo-ITAM is likely to play a role (8, 10). Therefore, interaction of CD5 with its ligand does not seem to be necessary for TCR signaling inhibition. Various potential ligands for CD5 have been described (11–15), but the true identity of the physiologically relevant ligand remains to be determined. It has also been reported that conserved fungal components bind to membrane-bound CD5 to induce cytokine release and that a soluble CD5 ectodomain protects mice from zymosan-induced septic shocklike syndrome (16). Regulation of CD5 expression in T lymphocytes is not well understood (17, 18), but it has been reported that CD5 expression levels on thymocytes and peripheral T cells are proportional to the affinity/avidity of the TCR interaction with peptide-MHC (p-MHC) (19). Accordingly, our previous results indicated that human CD8⁺ T lymphocytes infiltrating low-p-MHC–expressing tumors display decreased CD5 expression levels, correlated with increased antitumor reactivity (20). In contrast, it has been reported that in vivo delivery of specific peptides by dendritic cells induces increased expression of CD5 on activated T cells, resulting in peripheral T cell tolerance (21). Moreover, chronic exposure of either CD8⁺ or CD4⁺ T cells to auto-Ag results in upregulation of CD5 surface expression, rendering them anergic (22, 23).

CD5 is a negative regulator of T cell activation, and thus plays a critical role in preventing activation-induced cell death (AICD) (24). AICD, an apoptotic pathway triggered at least in part by the death receptor CD95 (APO-1, Fas) and its natural ligand (CD95L, FasL) following T cell hyperactivation, controls expansion of activated T lymphocytes after TCR engagement and induces T cell tolerance (25). We have previously reported that CD5 promotes prosurvival signals and protects T lymphocytes from TCR activation-dependent apoptosis triggered by recognition of the specific target (26, 27). Indeed, we have demonstrated that human CTL clones mediate antitumor responsiveness, which is inversely proportional to CD5 expression levels (20), and that CD5 prevents T cell overactivation, leading to AICD by regulating FasL expression (27). In addition, one of our laboratories has reported that mice lacking CD5 display significantly delayed onset and decreased severity of experimental autoimmune encephalomyelitis, associated with an increased frequency of apoptotic activated T cells (26). In the current study, we investigated the antitumor T cell response in CD5-deficient and wild-type (WT) C57BL/6 mice and we examined the contribution of CD5 to regulating establishment and expansion of B16F10 syngeneic melanoma. Our results indicate that CD5^−/− mice develop a stronger antitumor immune response than CD5^+/+ mice, which is associated with tumor infiltration by T lymphocytes with a more activated phenotype and enhanced cytokine and cytotoxicity functions. We also show that this antitumor response is transient, and drops as a consequence of an increase in AIDC of tumor-infiltrating CD8⁺ T lymphocytes. More importantly, inhibition of FasL-dependent apoptosis using soluble human CD95-Fc rescues tumor-infiltrating T lymphocytes (TIL) from AICD, resulting in improvement in the tumor-specific immune response, and thus persistent protection of CD5 knockout (KO) mice. Therefore, control of AICD of TIL may prevent T cell tolerance to malignant cells and contribute to the design of a more successful anticancer immunotherapy strategy.

C57BL/6 mice were purchased from Charles River Laboratories (Arbresie, France). CD5^−/− mice were backcrossed into C57BL/6 for 10 generations in one of our laboratories (26). All animals were housed at the Institut de Cancérologie Gustave Roussy animal facility and treated in accordance with institutional animal guidelines.

B16F10 melanoma and Lewis lung carcinoma (LL/2) cell lines (H-2^b) were purchased from the American Type Culture Collection. Tumor cells were grown in DMEM/F-12 medium (Seromed, Biochrom KG) supplemented with 10% FCS, 2 mM l-glutamine, 1 mM sodium pyruvate, and antibiotics (50 U/ml penicillin and 50 μg/ml streptomycin).

Anti-mouse CD2 (RM2-5), CD3 (145-2C11), CD4 (RM4-5), CD5 (53-7.3), CD8 (5H10), TCRαβ (Η57-597), TCRγδ (GL3), CD27 (LG.3A10), CD28 (37.51), CD62L (MEL-14), CD45RA (14.8), CCR7 (4B12), CD25 (PC61 5.3), CD69 (H1.2F3), and CD44 (IM7) mAb as well as hamster and rat isotopic controls were purchased from Invitrogen. Anti-Fas (15A7), anti-granzyme B (16G6), and anti-Foxp3 (FJK-16s) mAb were purchased from eBioscience.

Phenotypic analyses were performed by direct immunofluorescence using a FACSCalibur flow cytometer, as described (20). For intracytoplasmic Foxp3 and granzyme B expression, TIL were fixed using PBS containing 2% formaldehyde, and cell membrane was then permeabilized using PBS supplemented with 0.5% BSA and 0.2% saponin.

Six- to 12-wk-old mice (n = 5–10 per group) were inoculated s.c. with 3 × 10⁵ B16F10 cells. Tumor volume was measured using a caliper twice per week and estimated with the following formula: (width)² × length × 0.5 (cm³), according to (28).

For in vivo FasL neutralization, mice were inoculated i.v. with AdmFas-Fc (29) 7 d after B16F10 injection and followed up for tumor progression. AdCO1 empty virus was used as a negative control (10¹¹ viral particles per mouse). Plasma soluble Fas (sFas)-Fc levels were measured at day 10.

For TIL isolation, B16F10 tumors from 5–10 mice per group were surgically removed at indicated time points, weighed, and mechanically dissociated. T lymphocytes from each pool of tumors were then positively selected using anti-CD90.2 mAb-coated dynabeads according to the standard immunoselection protocol recommended by the manufacturer (Dynal, Invitrogen). Single-cell suspensions were stained with anti-CD3 mAb. Up to 95% of purified cells were CD3⁺. Isolated TIL were either used directly for phenotypic and functional analyses or cultured for 2–3 wk in complete RPMI 1640 medium supplemented with 10% FCS, 2 mM l-glutamine, 1 mM sodium pyruvate, 100 U/ml rIL-2, and antibiotics in the presence of irradiated splenocytes from B16F10-inoculated mice. To evaluate the number of CD3⁺/CD4⁺ and CD3⁺/CD8⁺ TIL per gram of tumor, the percentage of the respective subset was multiplied by the total number of CD90.2⁺ cells and divided by 100. The obtained number was then divided by the mass (in gram) of the 5–10 resected tumors.

The cytotoxic activity of TIL was measured in conventional overnight [⁵¹Cr]-release assay with B16F10 and LL/2 tumor cell lines used as targets. Three thousand target cells/well were used at indicated E:T ratios (30).

Secretion of cytokines and chemokines by TIL was measured using the Multiplex Bead Immunoassay, according to the manufacturer’s protocol (Invitrogen). Supernatants from overnight cocultures of TIL stimulated with B16F10 cells or activated by immobilized anti-CD3 mAb (145-2C11; BD Biosciences) were collected, and IFN-γ, MIP-1α, and MIP-1β were evaluated.

AICD was measured by flow cytometry using an annexin V–FITC apoptosis detection kit (BD Pharmingen). Briefly, TIL were isolated, as described above (including 5–10% of B16F10 stimulator cells), and incubated for 15 min in 5 μl annexin V labeling buffer with 5 μl 7-aminoactinomycin D (7AAD). Cells were analyzed within 1 h using a FACSCalibur flow cytometer. For CD4 and CD8 apoptotic TIL, CD3⁺/CD4⁺ or CD3⁺/CD8⁺ cells were gated and then analyzed for annexin V and 7AAD labeling, respectively.

For quantification of FasL expression, total RNA from 10⁶ TIL was extracted using a modified guanidine isothiocyanate phenol/chloroform method (TRIzol reagent; Gibco BRL/Life Technologies). Real-time quantitative RT-PCR analysis was then performed by TaqMan, according to the manufacturer’s recommendations, using oligonucleotide primers and a fluorescent probe for the mouse FasL gene (Applied Biosystems). The amount of RNA samples was normalized by amplification of an endogenous control (18S).

Statistical analyses were performed using Student t test with Prism software. Statistical significance was set at p < 0.05.

To investigate the role of CD5 in the antitumor immune response, we followed up tumor progression in CD5^−/− and CD5^+/+ C57BL/6 mice inoculated with B16F10 melanoma cells. Results indicated that B16F10 melanoma grew faster in WT than in CD5-deficient mice and that it flared up starting from day 16 after tumor cell injection (Fig. 1A). To assess the role of T cell infiltrates in control of cancer development, tumors from CD5^−/− and CD5^+/+ mice were removed at different time points, weighed, and dissociated. TIL were then isolated using anti-CD90.2 mAb-bound beads, counted, and phenotyped by flow cytometry for CD3, CD4, and CD8 expression. Results showed that both tumors were highly infiltrated by CD3⁺ lymphocytes at days 10 and 13 after tumor cell injection, and that the number of TIL decreased starting from day 16, to reach very low levels at day 19 (Fig. 1B). Fig. 1B also shows that the number of CD8⁺ and CD4⁺ T cells per gram of tumor was similar in the two mouse strains. Similar numbers of CD4⁺/CD25⁺/Foxp3⁺ regulatory T (Treg) cells infiltrated the tumors from both mouse strains (Fig. 1C). These results indicated that control of tumor growth in CD5 KO mice was not correlated with increased numbers of TIL compared with WT mice, and that CD5^−/− T cells were more efficient at protecting animals from tumor burden than CD5^+/+ T cells.

To determine whether control of tumor growth in CD5-deficient mice was due to an increased T cell response to TCR stimulation in the absence of CD5, TIL from CD5 KO and WT animals were isolated at day 16 and phenotypically and functionally characterized. Fig. 2A indicates that most of the positively selected cells expressed CD3, CD2, and TCRαβ, and included similar percentages of CD4⁺ and CD8⁺ subsets in both mouse strains. Similar proportions of TIL from CD5 KO and WT mice also expressed CD45RA and CD62L. Moreover, results depicted in Fig. 2B show that most of the TIL displayed a CD44^high/CD62L^low phenotype, suggesting that they corresponded to effector/memory T cells. In contrast, only a few cells from CD5 KO mice expressed the CCR7 chemokine receptor compared with WT animals, and CD27 and CD28 expression varied from one experiment to another, but was often slightly higher in CD5-deficient TIL (Fig. 2A). Higher numbers of CD5^−/− TIL also expressed the T cell activation markers CD25 and CD69 as compared with CD5^+/+ cells.

Next, experiments were conducted to assess whether TIL were able to kill the B16F10 melanoma cell line ex vivo and to secrete cytokines following stimulation with the specific target. Results shown in Fig. 3A indicate that, whereas TIL freshly isolated from CD5 KO mice mediated a cytotoxic activity against the B16F10 melanoma, those isolated from WT mice were inefficient (left panel). Moreover, in vitro short-term cultures from CD5^−/− TIL, but not CD5^+/+ TIL, isolated at days 10 and 16 were able to lyse the B16F10 target, but not the syngeneic LL/2 lung tumor target used as a negative control (data not shown). Similar results were obtained with CD8⁺ T cells sorted at day 16 and cultured in the presence of irradiated autologous feeder cells and IL-2 (Fig. 3A, right panel), suggesting an anergic status of CD5^+/+ TIL toward B16F10 target cells, which express low levels of MHC class I (MHC-I) molecules (data not shown). Failure of CD5^+/+ T cells to kill B16F10 cells (31) did not result from a defect in their lytic potential, because CD8⁺ T cells infiltrating WT mouse tumors expressed levels of granzyme B similar to those of CD5 KO mouse tumors (Fig. 3B). This defect did not also appear to result from a Treg inhibitory effect because tumors from CD5 WT and KO mice were equally infiltrated by CD4⁺/CD25⁺/Foxp3⁺ TIL (Fig. 1C) and CD5⁺/CD4⁺/CD25⁺ T cells were slightly less potent to inhibit CD4⁺/CD25⁻ T cell proliferation than CD5⁻/CD4⁺/CD25⁺ (data not shown) (32). With regard to activation-induced cytokine production, data shown in Fig. 3C indicate that TIL isolated from CD5^−/− mice secreted much higher levels of IFN-γ (left panel), MIP-1α, and MIP-1β (right panel) after stimulation with B16F10 or immobilized anti-CD3 mAb than TIL isolated from WT C57BL/6 mice. These results suggested an anergic status of CD5⁺/CD8⁺ TIL toward B16F10 tumor cells and demonstrated that CD5-deficient TIL had a higher activation capacity and stronger reactivity than CD5^+/+ TIL.

It has been widely documented that strong TCR-mediated activation leads to programmed T cell death (reviewed in Ref. 25), and that CD5 contributes to the regulation of specific T cell susceptibility to AICD triggered by tumor cells (27). Thus, we sought to determine the role of AICD in control of activated tumor-specific TIL expansion and its involvement in tumor escape from the immune system. We therefore isolated TIL from CD5 KO and WT mice at day 16 after tumor inoculation and compared cell survival of CD4⁺ and CD8⁺ T lymphocyte subsets using double staining with annexin V and 7AAD. For this purpose, CD3⁺/CD4⁺ or CD3⁺/CD8⁺ T cells were gated and analyzed for annexin V and 7AAD staining. As shown in Fig. 4, Exp 1, annexin V⁺/7AAD⁻ apoptotic cells comprised 60 and 54% of the CD4⁺ T cell subset from CD5^−/− and CD5^+/+ mice, respectively. For the CD8⁺ subset, results indicated that, whereas only 44% of TIL from WT mice were apoptotic, ≤73% of TIL from CD5-deficient mice underwent apoptosis (Fig. 4, Exp 1). These data demonstrated that CD4⁺ subpopulations displayed similar susceptibility to programmed cell death in the two mouse strains, and that CD8⁺/CD5⁻ TIL were more susceptible to AICD triggered by B16F10 cells than CD8⁺/CD5⁺ TIL. They also suggest that CD8⁺ and CD4⁺ subsets are differentially activated within the MHC-I⁺/MHC-II⁻ B16F10 tumor and that the CD8⁺ T cell subset may include higher numbers of tumor-reactive cells than the CD4⁺ subset.

AICD is triggered, at least in part, by the interaction of the death receptor Fas with its ligand FasL, which is induced on the T cell surface after hyperactivation (33). Experiments were thus performed to determine whether the Fas/FasL pathway is involved in tumor-mediated AICD of TIL. We first analyzed the expression of CD95 and CD95L on freshly isolated CD5^−/− and CD5^+/+ TIL by flow cytometry and quantitative RT-PCR, respectively. Results indicated that TIL from both mouse strains expressed similar levels of Fas, but failed to express TNF (TNFR) and TRAIL (DR5) receptors (Fig. 5A). In contrast, FasL mRNA was more strongly expressed in CD5^−/− than in CD5^+/+ TIL, as revealed by TaqMan quantitative PCR analysis (Fig. 5B). This suggests a role for CD5 in control of tumor-mediated T cell AICD by modulating FasL induction, presumably through inhibition of TCR signaling.

We therefore analyzed the effect on tumor growth in CD5-deficient mice of an adenovirus encoding a murine sFas (AdmFas-Fc), previously described for its capacity to neutralize CD95L in vivo (29). For this purpose, we injected i.v. AdmFas-Fc or an AdCO1 empty virus control into B16F10-bearing CD5 KO animals at day 7 after tumor cell inoculation and followed up tumor progression and T cell infiltration. Results revealed a considerable delay in tumor growth in mice inoculated with AdmFas-Fc compared with the AdCO1 empty virus control (Fig. 6A). To assess whether sFas-Fc–induced tumor growth regression was due to rescue of TIL from tumor-mediated AICD, we collected tumor specimens from CD5 KO mice, injected either with AdmFas-Fc, an AdCO1 empty vector, or a PBS control, and the number of CD4⁺ and CD8⁺ TIL was evaluated. Results indicate that tumors from AdmFas-Fc–inoculated mice included a much higher number of CD4⁺ and CD8⁺ TIL than tumors from animals injected with either PBS or the AdCO1 empty virus control. There were ∼3-fold more CD4⁺ T cells and 4-fold more CD8⁺ T cells in AdmFas-Fc–treated mice than in PBS- or AdCO1-treated animals (Fig. 6B). These results strongly suggested that TIL underwent Fas-dependent AICD within the tumor microenvironment and that rescue of hyperactivated CD8⁺ T cells from apoptosis by targeting FasL directly or indirectly participated in controlling tumor growth.

In the present work, using a CD5 KO mouse model, we show delayed growth of the B16F10 melanoma, which correlated with tumor infiltration by highly reactive tumor-specific CD8⁺ T lymphocytes. Indeed, CD5^−/− TIL display a more highly activated phenotype and enhanced ex vivo antitumor cytotoxicity and cytokine responses than CD5^+/+ counterparts. Thus, consistent with the inhibitory role of CD5, the absence of CD5 expression lowered the T cell activation threshold, thereby resulting in enhanced stimulation of tumor-specific T cells within the tumor microenvironment. Nevertheless, our results also indicated that TIL from CD5 KO mice were unable to sustain a persistent antitumor immune response required for cancer eradication. This was associated with increased apoptosis of activated CD8⁺ T lymphocytes and may thus have been due to insufficient expansion of tumor-reactive T cells with respect to tumor cell proliferation.

Although CD5-deficient mice have normal numbers of T cells and develop strong immune responses to a variety of Ag (34), the TCR repertoire selected in their thymus is affected (5). Thus, although similar numbers of CD4⁺ and CD8⁺ T cells infiltrated tumors in both mouse strains, the protection provided by the absence of expression of CD5 against B16F10 melanoma may simply reflect a higher frequency of tumor Ag-specific T cell precursors in CD5 KO mice. Alternatively, because many of the melanoma Ag correspond to self-Ag, thymocytes expressing TCR that bind with high affinity to self p-MHC may undergo negative selection in the absence of CD5, presumably because the intensity of the signals transduced by these TCR is too strong and is above the threshold for positive selection (35, 36). Thus, CD5^−/− mice may display a lower frequency of tumor Ag-specific T lymphocytes. To assess the extent of these two opposing outcomes in the situation of CD5 deficiency, future studies will attempt to compare tumor Ag-specific TCR repertoires in TIL and PBL from CD5 KO and WT mice.

Our results also indicate that the absence of CD5 enhances T cell activation, but also substantially promotes AICD of CD8⁺ T cells at later time points in tumor development. This dramatic T cell death most likely accounts for T cell tolerance to malignant cells. CD5-deficient mice also exhibit an enhanced T cell response to myelin oligodendrocyte glycoprotein, followed by decreased severity of experimental autoimmune encephalomyelitis correlated with elevated frequencies of apoptotic activated T cells (26). High expression levels of CD5 in WT mouse TIL might well prevent AICD of CD8⁺ T cells, but a concomitant reduction in positive signaling reduces the sensitivity of T lymphocytes to a level below the threshold for tumor cell killing. Accordingly, it has been reported that a fraction of myelin basic protein-reactive T cells was able to avoid AICD in response to a superagonist, but these lymphocytes had reduced sensitivity to the Ag that correlated with elevated levels of CD5 (37). Similarly, our previous results indicated that human tumor-reactive CD5^high CTL survived longer than CD5^low CTL and that CD5 protects T cells from TCR activation-dependent apoptosis triggered following interaction with autologous target cells (27). Our present data further corroborate that CD8⁺/CD5⁻ TIL are more susceptible to TCR-mediated apoptosis than CD8⁺/CD5⁺ TIL and that AICD is most likely involved in tumor escape from immune response control. In contrast, our data also indicate that CD4⁺/CD5⁻ TIL display sensitivity to AICD similar to that of CD4⁺/CD5⁺. This may be associated with the tumor-specific TCR repertoire of the CD4⁺ TIL subset compared with the CD8⁺ subset, and the reactivity of the two T cell subpopulations to the MHC-I⁺/MHC-II⁻ tumor.

AICD plays an essential role in contraction of activated T cells after eradication of pathogens. Accumulating evidence supports the notion that regulated expression of FasL on activated T cells is a critical event in control of AICD, and thus of the adaptive immune responses. Accordingly, polymorphism in FasL has been reported to contribute to increased apoptosis of TIL and risk of cervical cancer (38). Death of specific TIL has also been attributed to tumor-induced AICD (39, 40) and involves the Fas/FasL pathway (40). Our previous studies indicated that human CD5 protects T cells from AICD through regulation of FasL surface expression (27). In this study, we provide further evidence for a role of CD5 in promoting CD8⁺ T cell survival through modulation of FasL expression consequent to TCR signaling inhibition. Expression of CD95L is restricted to a few cell types, including T cells, macrophages, and cells of the testis. It is not present in resting T cells, but is highly expressed upon T cell activation. Therefore, blocking FasL (e.g., by sFas-Fc) should mainly prevent AICD of activated T cells. Our results indicate that systemic administration of CD95L-neutralizing AdmFas-Fc protects TIL from Fas-mediated apoptosis, resulting in sustained protection of CD5 KO mice from tumor burden. Therefore, selective rescue of activated T lymphocytes by targeting FasL may offer an opportunity for treatment of solid tumors. Current studies are aimed at immunizing B16F10-bearing WT C57BL/6 mice with a combination of AdmCD5-Fc and AdmCD95-Fc to concomitantly enhance antitumor CTL activity and protect tumor-reactive T cells from AICD.

AICD of antitumor T cells may be a major impediment in achieving robust, long-lasting CTL responses following active specific vaccination or after adoptive cell transfer in cancer immunotherapy. Persistence of effector T lymphocytes is a hallmark of successful T cell-mediated therapy, and modulation of AICD might be critical for achieving more successful cancer vaccines. Prolonging CTL survival so as to enhance the therapeutic benefits of current tumor immunotherapy approaches has been highlighted in both experimental tumor models (41) and clinical trials (42). Therefore, defining the molecular bases of the tumor microenvironment necessary for sustaining an effective antitumor immune response is critical for the design of more potent antitumor treatments. Present results further emphasize that modulation of CD5 expression on TIL and control of AIDC of tumor-specific CTL through regulation of the FasL pathway result in potentiation of intratumoral T cell responses and may contribute to the design of more successful anticancer immunotherapy.

We thank P. Romero, G. Bismuth, A. Prévost-blondel, S. Karray, O. Duc, and A. Jalil for helpful discussions.

The authors have no financial conflicts of interest.