LAG3 and PD1 co-inhibitory molecules collaborate to limit CD8+ T cell signaling and dampen antitumor immunity in a murine ovarian cancer model

Abstract

The immune co-inhibitory receptors lymphocyte activation gene-3 (LAG3) and programmed cell death 1 (PD1) synergistically contribute to autoimmunity and tumor evasion. Here we demonstrate how they collaborate and interact to regulate T cell function. We first show that LAG3 and PD1 are co-expressed on both OVA-specific and non-specific T cells infiltrating murine ovarian tumors. Dual antibody blockade or genetic knockout of LAG3 and PD1 significantly enhanced T effector function and delayed tumor growth. LAG3 and PD1 co-localized in activated CD8+ T cells in vitro at the trans-Golgi vesicles, early/recycling endosomal compartments, lysosomes, and microtubule organizing center. Importantly, LAG3 and PD1 cluster with pLck at the immunological synapse. Reciprocal immunoprecipitation of T cell extracts revealed physical interaction between LAG3 and PD1. Mutational analyses indicate that the cytoplasmic domain of LAG3 is not absolutely required for its association with PD1, while the ITIM and ITSM of PD1 are necessary for its association with LAG3. Finally, LAG3 protein also associates with the Src-homology-2 domain-containing phosphatases (SHP1/2) which are known to be recruited by PD1 during T cell signaling. Our data indicate that the association of LAG3 with PD1 contributes to their rapid trafficking to the immunological synapse, leading to a synergistic inhibitory effect on T cell signaling.

Keywords: LAG3; PD1; T cell signaling; antibody blockade; ovarian cancer.

Conflict of interest statement

CONFLICTS OF INTEREST

The authors declare there is no conflict of interest.

Figures

Figure 1. CD8+ T cells from Lag3−/−Pdcd1−/− knockout mice exhibit enhanced effector phenotype

A. Enhanced cytokine production by CD8+ T cells from Lag3−/−Pdcd1−/− mice. Purified naive mouse CD8+ T cells were isolated from C57BL/6, Lag3−/−, or Pdcd1−/−, or Lag3−/−Pdcd1−/− (DBKO) mice and activated with plate-bound anti-CD3/B7–1. Cytokines were measured from culture supernatants collected at 6 h or 24 h (Granzym B) post activation with ELISA. The statistics from three experiments were analyzed with one way ANOVA and error bars represent standard deviation between experiments. *P < 0.05. B. Enhanced anti-ovarian tumor immunity in OT-1-Lag3−/−Pdcd1−/− mice. IE9mp1 cells (1 × 107 cells) were i.p. injected into wild-type OT-1 (OT1-WT, n = 17), OT1-Lag3−/− (n = 13), OT1-Pdcd1−/− (N = 15), and OT1-Lag3−/−Pdcd1−/− (n = 16) mice. Data are combined from repeated experiments, 3 to 5 animals per group. Data were analyzed using the Mantel-Cox log-rank test, p = 0.0001. C. Tumor growth monitored by measuring the abdominal circumvent surface. The ascitic fluid started accumulating at approximately 40 days in wild-type tumor-bearing mice post tumor implantation. Mice were euthanized when moribund or the circumference reached approximately 12 cm. D. Frequency of CD8+ and CD4+ TALs (ascites) and TILs (tumor) from tumor-bearing mice. TILs and TALs were isolated from late stage tumor-bearing mice (3–5 mice per group) and were prepared as described in the Material and Methods. Gating strategy is shown in Supplementary Figure 1. E. TILs and TALs from the Lag3−/−Pdcd1−/− mice contain significantly more cytokine producing cells upon SIINFEKL stimulation and F. Fewer CD25+ FoxP3+ cells. Error bars represent SD. Statistical significance was determined by Student's t-test. *P < 0.05; **p < 0.01; ***p < 0.001. Data shown are representative of 2 experiments.

Figure 2. Co-expression of PD1 and LAG3 correlates with more severe dysfunction of CD8+ T cells

A. Coexpression of LAG3 and PD1 in TILs from IE9mp1-tumor-bearing mice. Representative flow cytometry analysis of TILs stained with Live/Dead dye, mAbs to CD4, CD8, PD1 and LAG3. Dot plot analyses were gated on live cells, then on CD8 or CD4 and show percentages of single and double stained PD1- and LAG3-positive cells (see also Supplementary Figure 1). TILs were isolated from IE9mp1 tumors resected from wild-type (C57BL/6) mice 30–35 days post-implantation and stained for surface expression of the above mentioned protein and analyzed with flow cytometry. B. Pooled data of LAG3/PD1 expression on CD8+ or CD4+ TILs from wild-type (C57BL/6) mice. Data were obtained from 20 animals and analyzed as described in (A). C. Proliferation rate and function of PD1+, LAG3+ and PD1+LAG3+ TILs. CD8+ TILs from ascites and tumors were sorted by FACS or with CD8+ selection kit (Invitrogen), labeled with CFSE, and stimulated with plate-bound anti-CD3/CD28 antibodies for 4 days. CD8+ T cell proliferation was determined by dilution of CFSE; numbers in the FACS panels indicate the percentage of CFSElow cells. Percentage of cells producing multiple cytokines (IFN- γ and TNF-α) is indicated in the panel to the right. Data shown are representative of three experiments.

Figure 3. Dual blockade of LAG3 and PD1 synergistically enhance anti-tumor immunity

A. Combinatorial anti-LAG3/anti-PD1 treatments delay ovarian tumor growth. Tumor-bearing mice C57BL/6 were randomized and treated with isotype control, anti-PD1, anti-LAG3, or anti-PD1/LAG3 combination on days 10, 12, 14, 16, and 18 (200 μg per treatment). Mice were euthanized when they developed ascites and their abdominal circumference reached 12 cm in diameter or when moribund. Data represent combined 10 mice per group. Data were analyzed with the Mantel-Cox log-rank test, P = 0.01. B. Tumor growth monitored by measuring the abdominal circumvent surface. The ascitic fluid started accumulating at approximately 20-25 days post tumor implantation. Mice were euthanized when moribund or the circumference reached approximately 12 cm. C. Combinatorial anti-LAG3/anti–PD1 treatment inhibits EG7 tumor growth in mice. C57BL/6 were randomized on day 6 when tumor volumes were approximately 40-60 mm3, and treated with isotype control, anti-PD1, anti-LAG3, or anti-PD1/LAG3 combination on days 6, 8, 10, and 12. Tumor volume was determined as described in Materials and Methods. Data shown are representative of 3 experiments with 10 mice per group. D. Frequency of CD8+ and CD4+ TALs (ascites) and TILs (tumor) from ovarian tumor-bearing mice post antibody blockade treatment. Both CD8+ and CD4+ TILs were significantly elevated in dual antibody-treated group. Mice were treated with 4 doses of antibodies on days 10, 12, 14, 16. TILs and TALs were isolated on day 22-25. E. Frequency of cytokine producing CD8+ TILs is enhanced after dual antibody blockade treatment. TILs or TALs were isolated as in (D) and stimulated with PMA/Ionomycin in the presence of BFA for 5 hours. Cytokine producing cells were analyzed as described in Material and Methods. Gating for flow cytometry analysis is shown in Supplementary Figure 2B. F. Dual antibody blockade treatment decreases CD25+FoxP3+ cells. Data are representative of 3 independent experiments with 3-5 animals per group. Error bars represent SD. Statistical significance was determined by Student's t-test. *P < 0.05; **p < 0.01; ***p < 0.001. G. Dual antibody blockade enhances the proliferation rate of CD8+ TILs. TILs were pulled from 2 mice and sorted by CD8+ selection kit (Invitrogen or STEM cell technology) and CFSE dilution assay was performed as described in Figure 2C. Percentage of CD8+ T cells with diluted CFSE label is shown. Data shown are representative of three experiments.

Figure 4. Co-expression and co-localization of LAG3 and PD1 during T cell activation

A. LAG3 co-localized with PD1 during CD8+ T cell activation peaked at 48 h post activation. CD8+ T cells freshly isolated from OT-1 mice were activated with plate-bound anti-CD3/B7.1 for various times (24 h, 48 h, 72 h). Cells were stained for LAG3 (Cy3, red) and for PD1 (AlexaFluor 488; green), and analyzed using confocal microscopy. DAPI (blue) was used as counterstain for the nucleus. Examples of single-plane confocal images are shown. The percentage of T cells with LAG3 and PD1 co-localization was quantified by counting the cells with yellow signals derived from the merge of the red and green images in the overlays (N = 300) using the color picker tool in the ImageJ program and is indicated on the left of each panel. Data shown are representative of three independent experiments. B. LAG3 and PD1 expression and co-localization using ImageStream at 48 h, and 72 h after activation. The co-localization of LAG3 and PD1 was examined by looking at the similarity score (coincidence of both signals) determined by the software's algorithms. A histogram is plotted showing the similarity scores (R4) for the entire population, and a mean score calculated. The more to the right the curve is the more co-localized the proteins. Data shown are representative of two experiments.

Figure 5. Sub-cellular co-localization of LAG3 and PD1 in cytoplasmic compartments in activated CD8+ T cells

OT-1 T cells were activated and prepared as described in Figure 4A and stained with anti-LAG3 (CY3, red), anti-PD1 (AlexaFluor488, green), plus either A. anti-EEA1 (AlexaFluor647, blue) or B. anti-Rab11b (AlexaFluor647, blue), C. anti-TGN38 (DyLight649, blue), D. anti-LAMP1 (AlexaFluor647, blue), E. anti-tubulin- γ (AlexaFluor647, blue) and analyzed with confocal microscopy. Single plane images are representative of two independent experiments. Scale bar represents 5 μm. Co-localization is demonstrated by the white color derived from the merge of the red, green and blue channels. The percentage of T cells with LAG3 and PD1 co-localizing with the cellular compartments (N = 300) was performed as described in Figure 4A.

Figure 6. LAG3 and PD1 co-localization with pLCK, MTOC, EEA1, TGN38, and LAMP1 near the immunological synapse

A. LAG3 (red) and PD1 (green) concentrated at the synapse of conjugates between activated DCs and OT1 CD8+T cells. Splenic DCs from BL6 mice were isolated with CD11c+ beads and activated with GMCSF and LPS and pulsed with SIINFEKL for 20 h. DCs stained with Cell Tracker Blue (CTB) were mixed with activated T cells for 20 min to form conjugates. Conjugates with concentrated pLck (magenta) at the synapse were scored for the co-localization. B. Microtubule organization center (MTOC) polarized toward the synapse. Tubulin- γ (magenta) was used as a marker for MTOC. C. Clustering of EEA1, and D. TGN38, and E. LAMP1 near the LAG3 (red) and PD1 (green) co-localization at the synapse.

Figure 7. Immunoprecipitation reveals interaction between LAG3 and PD1 in restimulated T cells

A. LAG3 and PD1 association in rested and restimulated CD8+ T cells. Activated OT1 cells were rested in IL2 for 3 days and restimulated (2 × 107) with plate-bound anti-CD3/B7.1 for 3 or 5 min before lysis. Lysates were immunoprecipitaed with anti-LAG3 (upper panels or anti-PD1 (lower panels) antibodies or control IgG (lane 3, IgG). Immunoprecipitated complexes were detected by Western blotting first using a different anti-PD1 antibody (raised in goat, upper panel) or anti-LAG3 antibody (lower panel). Closed arrows indicate specific bands corresponding to LAG3 or PD1; open arrows represent IgG or nonspecific bands. Input extract (~0.5% of total lysate) was used as markers and loading control. B. LAG3 association with PD1 was abolished in CD8+ T cells isolated from Lag3−/− or Pdcd1−/− mice. CD8+ T cells were isolated from C57BL/6, Lag3−/−, and pdcd1−/− mice using CD8+-Dynal beads, activated with plate bound anti-CD3/B7.1 for 48 h, expanded with IL2, restimulated with anti-CD3/B7.1. Immunoprecipitation and Western blotting were performed as described in (A). C. LAG3 and PD1 association was reduced by blocking anti-PD-L1 and anti-PD1 antibodies. Activated OT1 cells were rested in IL2 for 3 days and re-stimulated (2 × 107) with OVA- and PD-L1-expressing tumor cells (IE9mp1, 3 × 105) for 3 (lane 1) or 5 minutes (lane 2–4). IE9mp1 cells (lane 3) or both OT1 and IE9mp1 cells (lane 4) were pretreated with blocking anti-PD-L1 (20 μg/ml) and anti-PD1 (20 μg/ml) antibodies for 2 h before restimulation and lysis. Immunoprecipitation and Western blotting were performed as described above. Data shown are representatives of three independent experiments. Densitometry measurement of the bands was performed using ImageJ. The amount of PD1 (top) or LAG3 (bottom) pulled-down by each other was normalized with that of the input and shown on the panels to the right. Error bars represent SD. *p < 0.05; **p < 0.01

Figure 8. LAG3 associates with PD1-SHP2 or -SHP1 interacting complex

A. Reciprocal immunoprecipitation of LAG3 and SHP2. A LAG3 antibody pulled down PD1 and SHP2 (left panels) and a SHP2 antibody pulled down PD1 and LAG3 (right panels). OT-1 T cells were activated with SIINFEKL or anti-CD3/B7.1 for 2 days and expanded with IL2 for 3 days. Rested cells (2 × 107) were washed and treated with pervanadate for 5 mins, lysed and immuno-precipitated with anti-LAG3- or anti-SHP2- or control IgG-conjugated Dynal beads. Western blots were probed sequentially with anti-PD1 and anti-SHP2 (left panels) or anti-LAG3 (right panels) antibodies. B. LAG3 association with SHP2 is confirmed in the Lag3−/− CD8+ T cells reintroduced with LAG3. Activated CD8+ T cells from Lag3−/− mice were transduced with a retrovirus vector expressing LAG3 (pLAG3), expanded with IL2 and stimulated with pervanadate for 5 min, lysed and immunoprecipitated with anti-LAG3-conjugated Dynal beads. Western blots were probed sequentially with anti-PD1 and anti-SHP2 antibodies. OT1 CD8+ T cells were used as a positive control. C. Reduced recruitment of SHP2 and SHP1 in Pdcd1−/− and Lag3−/− CD8+ T cells. Activated CD8+ T cells were rested in IL2 and restimulated with OVA- and PD-L1-expressing tumor cells (IFN-γ-treated IE9mp1) for 5 min. Lysates were prepared, immunoprecipitated, and analyzed as described in Figure 7C. Data shown are representative of two independent experiments. Quantitation of the amount of SHP2 (middle panels) and SHP1 (bottom panels) pulled-down by anti-PD1 or anti-LAG3 antibody was normalized with that of the input and shown on the panels to the right. Error bars represent SD. *p < 0.05; ***p < 0.001

Figure 9. Differential contribution of the cytoplasmic domains of LAG3 and PD1 in their interaction

A. LAG3 protein lacking cytoplasmic domain associates with PD1 with slightly reduced efficiency. Wild-type LAG3 (pLAG3) or LAG3 protein lacking the cytoplasmic domain (pΔCY), or pMIG vector control constructs were transduced into activated Lag3−/− CD8+ T cells. Sorted GFP+ cells were rested with IL2 for 2 days and restimulated with IE9mp1 for 5 min before lysis. Lysates were immunoprecipitated with anti-PD1 antibody and probe with anti-LAG3 (top panels) and then with SHP1 (lower panels) as described above. Densitometry analysis of the corresponding bands to calculate the ratio of pLAG3 versus pDCY was performed using the ImageJ software. Relative luminescence units were normalized with input levels and expressed as a ratio relative to pLAG3. Data shown are representative of three (LAG3) or two (SHP1) independent experiments. B. Mutations in the ITIM and ITSM of PD1 compromise its association with LAG3. Wild-type PD1 (pPD1) or mutant PD1with mutations in the ITIM and ITSM (pPDF1F2) or control pMIG vector constructs were transduced into activated Pdcd1−/− CD8+ T cells. Transductants were sorted and immunoprecipitation performed as described above. Data shown are representative of two independent experiments. Relative luminescence units were normalized with input levels and expressed as a ratio relative to pPD1. Error bars represent SD. *p < 0.05; ***p < 0.001.