Loss of ATG5 reduces CD4+ T cell activation and may boost anti-tumor immune responses

Atg5^fl/fl and OX40-Cre mice were described previously (26, 27) and have been kindly provided by Dr. Mizushima and bought from The Jackson Laboratory, respectively. Atg5^ΔOX40 mice were generated by crossing Atg5^fl/fl with OX40-Cre mice. To test for OX40-Cre activity, OX40-Cre mice were crossed with ROSA^mT/mG mice [Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J], which have been described previously (28). All mice were kept under specific pathogen-free conditions in the animal facility of the Helmholtz Center for Infection Research (HZI), Braunschweig (29). Animal experiments and breeding were performed following the guidelines of local and national authorities.

The B16 tumor model was performed as previously described (30). Briefly, 2 x 10⁶ B16 melanoma cells were subcutaneously injected into the right flank of the mice, and tumor growth was monitored by measuring the orthogonal tumor diameters. Mice were sacrificed if the mean tumor diameter reached 12 mm.

For surface marker staining, cells were resuspended in 100 μl FACS buffer (2% BSA in PBS) and incubated in the presence of the respective antibodies for 15 minutes at 4 °C in the dark. Afterwards, cells were washed twice with 500 μl FACS buffer and analyzed using an LSRII flow cytometer (BD Biosciences). Before surface staining, in the case of cell population analyses (excluding viability determinations), dead cells were excluded by incubating the cell suspension with LIVE/DEAD^® Fixable Blue Dead Cell Stain staining (LifeTechnologies) for 30 minutes at 4˚C in the dark and subsequently washed twice with PBS; afterwards Fc receptors were blocked by 15 minutes incubation with Fc-block (CD16/32) in FACS buffer at 4˚C and washed with FACS buffer. For intracellular proteins, staining was performed using Foxp3 Staining Buffer Set (Miltenyi, #130-093-142) according to the manufacturer's instructions. Antibodies used were: CD4-FITC (RM4-5, eBiosciences), CD4-PacificBlue (L3T4, BioLegend), CD8-PECy7 (53-6.7, Biolegend), CD25-PE-Cy7 (PZ61, BD Pharmigen), CD44-APC (IM7, Biolegend), CD62L-PerCPCy5.5 (MEL-14, eBioscience), CD134(OX40)-PE (OX86, Biolegend), CD134(OX40)-APC (OX86, Biolegend), CTLA-4-PE (UC10-4P9, eBioscience), Foxp3-AlexaFluor488 (FJK-16s, eBioscience), GITR-PECy7 (DTA-1, eBioscience), GITR-PE (DTA-1, BD Pharmigen), Helios-PE (22F6, Biolegend) and Nrp1-PE (3E12, Biolegend). Samples were acquired on LSR Fortessa or LSR II (Becton, Dickinson and Company) and analyzed by FlowJo software (Tristar).

Cells isolated from peripheral lymph nodes of Atg5^ΔOX40 and Atg5^fl/fl control mice were seeded in primary T cell medium, i.e. RPMI1640 supplemented with 10% FCS (PAA Laboratories), 50 μM β-mercaptoethanol, 50 μg/ml each of penicillin and streptomycin, 1% non-essential amino acids and 1 mM sodium pyruvate (all from Life technologies) and stimulated with Concanavalin A (1.5 μg/ml) and IL-2 (100 ng/ml). After 72 hours, cells were re-stimulated with pre-coated anti-CD3 (10 μg/ml), PMA (100 ng/ml) + Ionomycin (1μg/ml), CD95L (100 ng/ml), or left untreated for additional 24 hours. Subsequently, cells were stained with 7-AAD (Biolegend), CD4, and CD134 antibodies to determine cell viability in a Cytoflex LX (Beckman Coulter) flow cytometer.

CD4+ CD25− conventional T cells (Tcon) and CD4+ CD25+ regulatory T cells (Treg) were isolated from mouse spleen using the mouse CD4+CD25+ Regulatory T Cell Isolation Kit, (#130-091-041, Miltenyi). Tcon cells were stained with Cell Trace Violet according to the manufacturer´s protocol (C34557, ThermoFisherScientific). Afterwards, Tcon and Treg cells were co-cultured in primary T cell medium and stimulated using T Cell Activation/Expansion Kit, mouse (130-093-627, Miltenyi) supplemented with IL-2 (100 ng/ml). Treg: Tcon cells were seeded in the indicated ratios. After 72 hours, cells were harvested and stained with CD4-FITC (RM4-5, eBiosciences) and CD25-PE (Miltenyi) and analyzed for proliferation in a Cytoflex LX (Beckman Coulter).

OX40- and OX40+ cells were purified using a CD4+ CD25+ regulatory T cell isolation kit for magnetic separation (130-091-041; Miltenyi Biotec) according to the manufacturer's protocol, with the modification that the anti-OX40-PE (#119410, BioLegend) was used instead of the anti-CD25-PE provided in the kit. Cells were lysed by incubation in TPNE buffer (1x PBS add 300 mM NaCl, 1 mM EDTA, 1% Triton X-100) supplemented with protease inhibitors for 20 minutes on ice. After centrifugation, protein concentration was determined by BCA assay (ThermoFisher Scientific). Protein lysates were loaded onto 12% SDS-polyacrylamide gels for ATG5 or 15% SDS-polyacrylamide gels for LC3, blotted onto PVDF membrane (GE Healthcare), and detected by chemiluminescence. Antibodies used for Western blotting: ATG5 (12994S, Cell Signaling Technology), LC3A/B (4801S, Cell Signaling Technology), Beta-Actin (66009-1-lg, Proteintech).

Statistical analyses were performed by Mann-Whitney, one-way ANOVA (Tukey's multiple comparison post-test), or two-way ANOVA (Dunnett´s multiple comparison test) tests to determine statistical significance using GraphPad Prism (GraphPad Software Inc.). Standard error of the mean (SEM) is represented as error bars in the graphs.

OX40 is a member of the tumor necrosis factor receptor superfamily (TNFRS4, CD134) and is a secondary co-stimulatory molecule that is expressed mainly by activated T cells to prolong their activation and to sustain their proliferation (31). The Tnfrsf4 promoter has been used to drive Cre expression in activated CD4⁺ T cells and Foxp3⁺ Treg cells (27, 32). Crossing OX40-Cre^tg/wt mice with Rosa26-mT/mG^fl/wtOX40-Cre^tg/wt mice (Figure 1A), we verified that OX40-driven Cre expression is detectable in CD3⁺ T cells, but not CD19+ B cells (Figure 1B). Interestingly, we detected some Cre activity in Gr-1+ cells (Figure 1B), which is consistent with the fact that activated Th1 cells express this marker (33). Comparing T cell lineages, OX40-Cre was largely inactive in CD8+ and active in CD4+ T cells, with particularly high recombination efficiency in Foxp3⁺ CD4⁺ T cells (Figure 1C).

To investigate the role of autophagy in activated CD4⁺ T cells, OX40-Cre^tg/wt mice were bred with Atg5^fl/fl mice (26) to generate Atg5^ΔOX40 mice (Figure 2A). Deletion of ATG5 in OX40-expressing CD4+ T cells was confirmed by immunoblotting (Figure 2B). In line with ATG5 deficiency, LC3 conjugation to phosphatidylethanolamine was strongly impaired as LC3-I but hardly any LC3-II was detected in OX40-positive CD4+ T cells (Figure 2C). Consistent with the lack of Cre activity in the thymus, T cell development was not affected in Atg5^ΔOX40 mice since both the frequencies of thymocyte populations (Figure 2D) and the thymic cellularity (Figure 2E) were comparable to control mice. While the cellularity was only marginally affected in the spleen, the effect of autophagy-deficiency in OX40-expressing cells of Atg5^ΔOX40 mice resulted in a significant expansion of cell numbers in peripheral lymph nodes (pLN) (Figures 2F, G). Although not statistically significant, there was a tendency towards a higher cellularity in mesenteric LN (mLN) (data not shown). Accordingly, the lymph nodes of Atg5^ΔOX40 mice were enlarged, consistent with lymphadenopathy (Figure 2H).

When CD4⁺ and CD8⁺ T cell populations were analyzed, Atg5^ΔOX40 mice exhibited a reduction in the frequencies of both subsets (Figures 3A, B). However, the absolute numbers of helper and cytotoxic T cells were not significantly altered, although there was a slight tendency towards an enlarged T cell compartment (Figure 3C). This suggests that other cell populations were expanded in Atg5^ΔOX40 mice, which were found to be mostly CD19⁺ B cells (Figure 3D). In summary, Atg5^ΔOX40 mice exhibit normal T cell development but increased cellularity within peripheral lymph nodes, suggesting a lower threshold for the activation of the immune system.

The enhanced cellularity described above might result from enhanced activation of effector cells or impaired regulatory mechanisms. In that respect, autophagy has on the one hand been connected to activation of T cells (34), but has also been shown to down-modulate TCR-induced signaling (35). To dissect between these two possibilities, we injected anti-CD3 antibodies into Atg5^ΔOX40 and control mice to activate T cells in vivo. We observed an increased cellularity in secondary lymphoid organs of Atg5^ΔOX40 mice, which was most prominent in pLN (Figure 4A). Accordingly, the absolute numbers of CD4⁺ and CD8⁺ T cells were increased in Atg5^ΔOX40 mice upon anti-CD3 injection (Figure 4B). Thus, autophagy deficiency in CD4⁺ T cells appears not to affect T cell proliferation in general.

Subsequently, we investigated the activation status of CD4⁺ T cells by analyzing the expression of the activation markers CD69, CD25, and OX40 in mice injected with anti-CD3 or left untreated. T cell activation was not impaired in general since CD69 expression was normal (Figure 4C). In contrast, the late activation markers CD25 and OX40 exhibited impaired induction in CD4⁺ T cells of stimulated Atg5^ΔOX40 mice (Figures 4D, E). Although OX40 expression by CD4⁺ T cells was reduced in OX40-Cre^tg/wt mice compared to Atg5^fl/fl mice due to the knock-in of Cre into the Tnfrsf4 locus, OX40 induction was even further reduced in Atg5^ΔOX40 mice (Figure 4E).

Taken together, although the immune system appears to be activated in Atg5^ΔOX40 mice, ATG5 deficiency in T helper cells impairs CD4⁺ T cell activation. Therefore, the increased cellularity in the secondary lymphoid organs alludes to an altered threshold for immune cell activation, potentially due to deficient Treg suppression.

Treg cells are an essential part of peripheral tolerance in the immune system. We had a closer look again at the Rosa26-mT/mG^fl/wtOX40-Cre^tg/wt mice and compared the EGFP-positive with the EGFP-negative cell population from pLN. Interestingly, the former had a much higher expression of CD25 both in unstimulated and anti-CD3 stimulated mice and a higher expression of OX40 in stimulated mice, which is consistent with a Treg cell phenotype (Figure 5A). Indeed, most CD4⁺Foxp3⁺ Treg cells of Rosa26-mT/mG^fl/wtOX40-Cre^tg/wt mice were EGFP ⁺ suggesting that OX40-dependent loss of autophagy affects especially Treg cells (Figure 1C). CD4⁺Foxp3⁺ Treg frequencies and numbers were reduced in peripheral lymphoid organs of both unstimulated and stimulated Atg5^ΔOX40 mice (Figures 5B, C). Although several OX40 knockout studies have shown that OX40 expression is crucial for Treg development and function and that these mice have severely reduced mature Treg numbers (36 –38), the effect seen in Atg5^ΔOX40 mice was independent of reduced OX40 expression, since OX40-Cre^tg/wt mice did not show reduced Treg numbers in peripheral lymphoid organs (Figures 5B, C). Furthermore, when investigating functional Treg markers, the percentage of CTLA-4-expressing Treg cells was increased in peripheral lymphoid organs of Atg5^ΔOX40 mice (Figure 5D). Analyzing additional Treg markers, we found that Nrp-1 expression was reduced in Treg cells from peripheral lymphoid organs of unstimulated and especially of stimulated Atg5^ΔOX40 mice compared to Treg cells from control mice (Figure 5D).

Atg5 has been connected to the regulation of life and death of T cells (20, 22, 34). Therefore, we tested susceptibility to activation-induced cell death (AICD) in OX40-negative and OX40-positive T cells. To this end, CD4+ T cells were purified, activated in vitro and subsequently, after a short culture period, stimulated with anti-CD3, PMA plus ionomycin, or recombinant CD95L to induce cell death. While CD95L was the most potent death stimulus, no significant differences between OX40-negative and OX40-positive T cells and the two genotypes were observed (Figure 6A). We conclude that differences in cell death susceptibility did not account for the disturbed immune homeostasis in Atg5^ΔOX40 mice.

Next, we tested the suppressive capacity of Treg cells from Atg5^ΔOX40 and control mice. Conventional wild-type T cells were labeled with cell trace violet, activated via CD3 plus CD28, and incubated with Treg cells of conditional knockout or control mice. No differences in Treg suppressive activity could be detected in this in vitro setting (Figure 6B). Since Treg activity might be different in vitro versus in vivo, we went on to test Atg5-deficient Treg cells in Atg5^ΔOX40 mice in vivo.

Nrp-1 has been described as a marker for thymus-derived Treg cells (39, 40), but is also known to play a role in the migration of Treg cells to VEGF-expressing tumors, and this results in increased tumor growth due to Treg cells suppressing anti-tumoral responses (12). When injecting Atg5^ΔOX40 and control mice with B16 tumor cells, tumor growth was significantly delayed in Atg5^ΔOX40 mice (Figure 7A). CD4⁺Foxp3⁺ Treg cell frequencies were substantially reduced in all peripheral lymphoid organs analyzed as well as in the tumor (Figure 7B), whereas CD4⁺Foxp3^– conventional T cell frequencies were only slightly reduced in draining (dLN) and control lymph nodes (cLN) (Figure 7C). Furthermore, an increased frequency of intratumoral CD8⁺ cytotoxic T cells was seen in Atg5^ΔOX40 compared to control mice (Figure 7D), similar to what has been reported for mice in which Nrp-1 expression was deleted (12). In turn, the frequencies of CD8⁺ cytotoxic T cells were decreased in dLN and spleen, suggesting migration of these cells from the secondary lymphoid tissue towards the tumor. Lastly, intratumoral CD62L⁺CD44^– naïve CD4⁺ T cell frequencies were reduced, whereas activated CD62L^–CD44⁺ cell frequencies were increased in these mice (Figure 7E). CD62L/CD44 populations of CD4⁺ T cells derived from dLN, cLN, and spleen were normal (Figure 7E and data not shown). CD8⁺ T cells had decreased frequencies of CD62L⁺CD44^– naïve cells and increased frequencies of CD62L^–CD44⁺ activated cells (Figure 7E). Furthermore, whereas in dLN and cLN the CD62L⁺CD44⁺ memory cell frequencies were increased, this cell population was decreased in intratumoral CD8⁺ T cells (Figure 7E and data not shown). In summary, our data indicate that autophagy is a critical process for the activation of T cells and supports previous findings that have illustrated the importance of this biological process for the homeostasis and function of Treg cells.