LXR/ApoE Activation Restricts Innate Immune Suppression in Cancer

Abstract

Therapeutic harnessing of adaptive immunity via checkpoint inhibition has transformed the treatment of many cancers. Despite unprecedented long-term responses, most patients do not respond to these therapies. Immunotherapy non-responders often harbor high levels of circulating myeloid-derived suppressor cells (MDSCs)-an immunosuppressive innate cell population. Through genetic and pharmacological approaches, we uncovered a pathway governing MDSC abundance in multiple cancer types. Therapeutic liver-X nuclear receptor (LXR) agonism reduced MDSC abundance in murine models and in patients treated in a first-in-human dose escalation phase 1 trial. MDSC depletion was associated with activation of cytotoxic T lymphocyte (CTL) responses in mice and patients. The LXR transcriptional target ApoE mediated these effects in mice, where LXR/ApoE activation therapy elicited robust anti-tumor responses and also enhanced T cell activation during various immune-based therapies. We implicate the LXR/ApoE axis in the regulation of innate immune suppression and as a target for enhancing the efficacy of cancer immunotherapy in patients.

Keywords: ApoE; LRP8; LXR; MDSC; cancer; clinical trial; immune therapy; myeloid; nuclear hormone receptor; tumor immunology.

Figure 1. LXR agonist treatment robustly suppresses tumor growth and progression across a broad set of mouse and human tumors

(A–C) Tumor growth by 1 × 10⁶ SKOV3 ovarian cancer cells subcutaneously injected into NOD SCID (A) or RAG (C) mice. Following tumor growth to 5–10 mm³ (A) or 40–50 mm³ (C) in volume, mice were fed control chow or chow supplemented with GW3965 (100 mg/kg/day) (A) or RGX-104 (100 mg/kg/day) (C); insets represent growth curves for individual tumors. (B) Survival of mice bearing SKOV3 tumors shown in (A) (n ≥ 5). (D–F) Tumor growth by 1 × 10⁶ U118 glioblastoma cells subcutaneously injected into NOD SCID mice. Following tumor growth to 5–10 mm³(D) or 200–250 mm³ (F) in volume, mice were fed a control chow or a chow supplemented with GW3965 (100 mg/kg) (D) or RGX-104 (100 mg/kg) (F). (E) Survival of mice bearing U118 tumors shown in (D) (n ≥ 5). (G) Tumor growth by 2.5 × 10⁵ GL261 glioblastoma cells subcutaneously injected into C57BL/6 mice. Following tumor growth to 5–10 mm³ in volume, mice were fed control chow or chow supplemented with GW3965 (100 mg/kg) (n ≥ 5). (H–I) Tumor growth (H) and metastasis (I) by 2 × 10⁵ LLC lung cancer cells subcutaneously injected into C57BL/6 mice. Following tumor growth to 5–10 mm³ (H) or 30–40 mm³ in volume (I), mice were fed control chow or chow supplemented with GW3965 (100 mg/kg) (H) or RGX-104 (100 mg/kg) (I). (I) Quantification of macroscopic metastatic nodules in H&E-stained lungs extracted at day 15 (n ≥ 5). (J) Tumor growth and bioluminescence quantification of lung metastasis of 1 × 10⁶ MDA468 breast cancer cells subcutaneously injected into NOD SCID mice. Following tumor growth to 5–10 mm³ in volume, mice were fed control chow or chow supplemented with GW3965 (100 mg/kg) (n ≥ 4). (K) Quantification and exemplary images of macroscopic metastatic nodules in H&E-stained lungs extracted 71 days after subcutaneous injection of 1 × 10⁶ ACHN renal cancer cells; mice were fed control chow or chow supplemented with GW3965 (100 mg/kg) when tumors reached 5–10 mm³ (n ≥ 4). (L) Tumor growth by 5 × 10⁴ Renca renal cancer cells subcutaneously injected into syngeneic C57BL/6 mice. Following tumor growth to 5–10 mm³ in volume, mice were fed a control chow or a chow supplemented with GW3965 (100 mg/kg). Survival of mice bearing Renca tumors is shown (n ≥ 5). (M) Tumor growth by 5 × 10⁴ B16F10 cells subcutaneously injected into C57BL/6 mice. Following tumor growth to 5–10 mm³ in volume, mice were fed control chow or chow supplemented with GW3965 (100 mg/kg) (n ≥ 5). (N) Tumor growth by 5 × 10⁵ MC38 colon cancer cells subcutaneously injected into C57BL/6 mice. Following tumor growth to 5–10 mm³ in volume, mice were fed a control chow or a chow supplemented with RGX-104 (100 mg/kg) (n ≥ 6). (O) Tumor volume at day 16 by 5 × 10⁵ MC38 cells injected subcutaneously into C57BL/6 or NSG mice. Following tumor growth to 5–10 mm³ in volume, mice were fed a control chow or a chow supplemented with RGX-104 (100 mg/kg) (n ≥ 6). Data represent mean ± s.e.m. See also Figure S1.

Figure 2. LXR agonism reduces tumor-infiltrating and systemic myeloid-derived suppressor cells

Percent tumor-infiltrating immune cells of total CD45⁺ tumor-infiltrating lymphocytes (TILs), A–D, in B16F10 tumors in mice treated with control or GW3965 (100 mg/kg) administered in chow when tumors reached 5–10 mm³ in volume. Flow cytometry analysis was performed 14 days after tumor injection (n = 6). (A) Foxp3+ T regulatory cells, (B) CSF-1R+ tumor-associated macrophages, (C) total granulocytic myeloid-derived suppressor cells, and (D) total myeloid lineage cells. Representative plots show CD11b⁺ Gr-1^high granulocytic MDSC populations. (E) Percent granulocytic (left) and monocytic (right) MDSCs of CD45⁺ TILs in B16F10 tumors grown in mice treated with control or RGX-104 (80 mg/kg) administered I.P when tumors reached 5–10 mm³. Flow cytometry analysis was performed 13 days after tumor injection (n ≥ 8). Representative contour plots show Ly6G⁺ granulocytic and Ly6C⁺ monocytic populations. (F) Mean tumor volume of subcutaneous B16F10 tumors described in (E) (n ≥ 5). (G) Quantification of tumor-infiltrating Gr-1⁺ cells in B16F10 tumors removed at day 11 on treatment with control or GW3965 (100 mg/kg). Representative images of Gr-1⁺ immunofluorescence in B16F10 tumor sections on treatment with control (top) or GW3965 (bottom). Five sections were imaged per tumor to achieve an average number of tumor-infiltrating Gr-1⁺ cells per high-power field (n = 6). Scale bar =143 microns. (H) Correlation between percent tumor-infiltrating G-MDSCs and tumor volume (n = 17). (I–J) Quantification of tumor-infiltrating Gr-1⁺ cells in Ovcar (I) and GL261 (J) tumors removed at day 81 (I) and 27 (J) on treatment with control or GW3965 (100 mg/kg). Five sections were imaged per tumor to achieve an average number of tumor-infiltrating Gr-1⁺ cells (n = 5). (K) Percent granulocytic MDSCs of total CD45⁺ TILs in B16F10 tumors grown in LXRαβ^−/− mice treated with control or RGX-104 (100 mg/kg) when tumors reached 5–10 mm³ (n ≥ 6). (L) Percent circulating granulocytic MDSCs of total CD45⁺ lymphocytes in peripheral blood of B16F10 (left) or U118 (right) tumor-bearing mice treated with control or RGX-104 (100 mg/kg) (n = 5). (M) Percent granulocytic MDSCs of total CD45⁺ splenocytes (n = 5). Representative spleens of control or GW3965-treated mice after 10 days of treatment (right). (N) Transwell MDSC differentiation assay. Bone marrow cells were cultured with B16F10 melanoma cells and GMCSF for 6 days. On day 3, RGX-104 (2uM) was added to the culture. Mean number of Gr-1^high CD11b⁺ cells per 50uL of the culture solution is shown as assessed by flow cytometry on day 6. Data represent mean ± s.e.m. See also Figure S2.

Figure 3. LXR agonism increases tumor-infiltrating activated CD8⁺ and CD4⁺ T cells

(A) Suppressive properties of splenic MDSCs isolated from tumor bearing mice treated in vivo with control or RGX-104 (100 mg/kg) chow for 48 hours—as assessed by CD8⁺ T cell activation (IFN-γ expression) and proliferation (BV dilution) after co-culture in vitro. Representative contour plots show IFN-γ expression and BV fluorescence of CD8⁺ cells (n = 4). (B) Percent IFN-γ and Granzyme B (GZM-B) expressing, activated CD8⁺ T cells of total CD45⁺ TILs from B16F10 tumors of control or GW3965-treated mice (100 mg/kg) after 10 days of treatment (n = 6). Representative contour plots show percentages of double positive cells. (C) Percent PD-1⁺ CD8⁺ T cells of total tumor-infiltrating CD8⁺ T cells from B16F10 tumors of control or GW3965-treated mice (100 mg/kg; 10 days) (n = 6). (D) Correlation between tumor-infiltrating CD8⁺IFN-γ⁺GZM-B⁺ T cells and G-MDSCs, reported as percentages of total CD8⁺ and CD45⁺ TILs, respectively (n = 20). (E) Percent CD4⁺IFN-γ⁺TNFα⁺ T cells of total CD45⁺ TILs from B16F10 tumors treated for 10 days with control or GW3965 (100 mg/kg) (n = 6). Data represent mean ± s.e.m. See also Figure S3.

Figure 4. LXR agonist treatment promotes MDSC apoptosis in vivo and in vitro

(A) Percent Gr1⁺ CD11b⁺ G-MDSCs of total CD45⁺ lymphocytes in bone marrow of GW3965-treated (100 mg/kg) mice as compared to controls after 10 days of treatment (n = 5). (B) Percent Annexin⁺ 7AAD⁺ G-MDSCs of total CD45⁺ bone marrow cells after GW3965 treatment (100mg/kg) for 10 days as compared to controls (n = 5). (C) Percent circulating G-MDSCs of total CD45⁺ lymphocytes in RGX-104-treated (100 mg/kg) mice as compared to controls after 8 days of treatment (left) and percentage of cleaved caspase-3⁺ (CC3+) GMDSCs of total circulating G-MDSCs from B16F10 tumor bearing mice treated for 8 days with either control or RGX-104 (100m/kg) (right) (n = 8). Contour plots show population of cleaved caspase-3⁺ MDSCs. (D) Percentage of labeled, adoptively transferred G-MDSCs of total CD45⁺ splenocytes from recipient mice treated for 48 hours with GW3965 or control after adoptive transfer. Representative plots show labeled CD11b⁺Gr-1⁺ MDSCs from control and GW3965 treated recipient mice (n = 5). (E) Annexin⁺ 7AAD⁺ granulocytic (left) and monocytic (right) CD11b⁺Gr-1⁺ MDSCs of total CD45⁺ splenic lymphocytes. Representative contour plots show double positive cells gated on CD11b⁺ Gr1^high G-MDSCs (n = 5). (F) Percent Annexin⁺ 7AAD⁺ G-MDSCs of total CD11b⁺ Gr1⁺ G-MDSCs in B16F10 tumors (n = 5). (G) Quantification of caspase-3⁺ Gr-1⁺ cells in the spleens of B16F10 tumor-bearing mice after 12 days of control or RGX-104 (100 mg/kg) treatment. 5 sections per spleen were imaged to calculate an average number of double positive cells per high-power field (n=5). Representative immunofluorescence images of splenic sections stained for Gr-1 and cleaved caspase-3 treated with control (top) or RGX-104 (100 mg/kg) (bottom). Scale bar =29 microns. (H) Percent CD11b⁺Gr-1⁺ MDSCs alive after 3 hours of treatment in vitro with RGX-104 (2uM). MDSCs were isolated from B16F10 tumor-bearing mice (n = 4). (I) Percent cleaved caspase-3⁺ Gr1⁺ CD11b⁺ MDSCs from wild type (WT) or LXRαβ^−/− B16F10-tumor bearing mice treated in vitro with 1uM RGX-104 or vehicle control after a 6-hour culture (n = 4). Data represent mean ± s.e.m. See also Figure S4.

Figure 5. ApoE/LRP8 signaling regulates MDSC survival downstream of LXR agonism

(A) Percent tumor-infiltrating granulocytic (left) and monocytic (right) MDSCs of total CD45⁺ tumor-infiltrating lymphocytes in B16F10 tumors in Apoe^−/− mice treated with control or GW3965 (100 mg/kg) (n ≥ 4). Representative plots show CD11b⁺Gr1^high granulocytic and CD11b⁺Gr1^int monocytic populations. (B) Mean tumor growth of control- or GW3956-treated ApoE-depleted B16F10 tumors injected into Apoe^−/− mice (n = 4). (C) Percentage of labeled, adoptively transferred ApoE deficient (Apoe^−/−) MDSCs of total CD45⁺ splenocytes from ApoE^−/− recipient mice treated for 48 hours with GW3965 or control after adoptive transfer (n = 5). (D) Percent splenic granulocytic (left) and monocytic (right) CD11b⁺Gr-1⁺ myeloid cells of total CD45⁺ splenic lymphocytes in WT and Apoe^−/− mice (n = 5). (E) Percent CD11b⁺Gr-1⁺ myeloid cells of total CD45⁺ lymphocytes in the bone marrow of WT and Apoe^−/− mice (n = 5). (F) Mean tumor growth of GL261 (left) and B16F10 (right) grown in WT and Apoe^−/− mice (n = 5). (G) Quantification (left) and representative immunofluorescence staining (right) of tumor-infiltrating Gr-1⁺ cells in GL261 tumors grown in WT and Apoe^−/− mice for 27 days (n = 6). 5 sections per tumor were imaged to calculate an average number of Gr-1⁺ cells per high-power field (n = 5). Scale bar =143 microns. (H) Percent circulating Gr-1⁺ CD11b⁺ G-MDSCs of total CD45⁺ cells after 7 days of tumor growth in WT or Apoe^−/− mice (n = 5). (I) Percent cleaved caspase-3⁺ (CC3+) Gr1⁺ CD11b⁺ splenic MDSCs from WT B16F10-tumor bearing mice treated in vitro with 5uM BSA or recombinant mouse ApoE protein (rmApoE) after a 6-hour culture (n = 4). (J) Percent cleaved caspase-3⁺ (CC3+) Gr1⁺ CD11b⁺ splenic MDSCs from LRP8^−/− B16F10-tumor bearing mice treated in vitro with 5uM BSA or recombinant mouse ApoE protein (rmApoE) after a 6-hour culture (n = 4). Data represent mean ± s.e.m. See also Figure S5.

Figure 6. LXR agonism provides additive anti-tumor efficacy when combined with immunotherapy

(A) Therapy regimen schematic (top) and individual tumor volumes of subcutaneous B16F10 tumors. Treatment cohorts include control without adoptive T cell transfer (ACT), RGX-104 (50mg/kg) without ACT, control with ACT and gp100 vaccination, and RGX-104 (50 mg/kg) with ACT and gp100 vaccination. 2 × 10⁶ CD8⁺ T cells from Pmel mice were transferred per recipient (n ≥ 5). (B) Survival data at 34 days of B16F10 tumor-bearing mice treated with ACT and gp100 vaccination, RGX-104 (50mg/kg), or ACT and gp100 vaccination in combination with RGX-104 (50 mg/kg) as compared to control (n ≥ 5). (C) Therapy regimen schematic (top), mean tumor volume (left), and individual tumor growth curves (right) of subcutaneous B16F10 tumors. Treatment cohorts include RGX-104 (100 mg/kg) with control antibody isotype-matched to anti-PD-1, anti-PD-1, and anti-PD-1 with RGX-104 (100 mg/kg) as compared to control (control antibody isotype-matched to anti-PD-1). All cohorts received ACT of 2×10⁶ CD8⁺ T cells from Pmel mice per recipient, as well as gp100 vaccination 3 days after tumor injections (n = 7). (D) Therapy regimen (top) and mean tumor volume of subcutaneous Lewis Lung Carcinoma (LLC) tumors. Treatment cohorts include RGX-104 (100 mg/kg) with control antibody isotype-matched to anti-PD-1, anti-PD-1, and anti-PD-1 with RGX-104 (100 mg/kg) as compared to control (control antibody isotype-matched to anti-PD-1) (n = 7). (E) Therapy regimen schematic (top), mean tumor volume (left) and individual tumor growth curves (below) of subcutaneous B16F10 tumors treated with anti-PD-1 monoclonal antibody or anti-PD-1 with RGX-104 (50 mg/kg) (n = 5). (F) Percent TCRαβ⁺IFN-γ⁺GZM-B⁺ expressing, cytotoxic T cells of total CD45⁺ TILs from B16F10 tumors treated with anti-PD-1 or anti-PD-1 in combination with RGX-104 (50 mg/kg) as described in (E) (n = 5) (G) Therapy regimen (top) and quantification (left) and representative immunofluorescence images of tumor-infiltrating Gr-1⁺ cells within subcutaneous B16F10 tumors. Treatment cohorts included Gvax with control antibody isotype-matched to anti-PD-1, Gvax with anti-PD-1, Gvax with anti-PD-1 and GW3965 (100 mg/kg) as compared to control (with control antibody isotype-matched to anti-PD-1) (n = 6). Scale bar =143 microns. (H) Mean tumor volumes of B16F10 subcutaneous tumors from the cohorts described in (G). Cohorts included Gvax with control antibody isotype-matched to anti-PD-1, Gvax with anti-PD-1, and Gvax with anti-PD-1 and GW3965 (100mg/kg) (n = 6). Data represent mean ± s.e.m.

Figure 7. RGX-104 depletes MDSCs and activates CD8⁺ T cells in human cancer patients

(A) Percent of granulocytic MDSCs (G-MDSC) of total circulating cells in the peripheral blood of cancer patients relative to healthy volunteers (n ≥ 6). Representative plots demonstrating the G-MDSC population in healthy volunteers compared to cancer patients. (B) Percent of G-MDSCs of total circulating cells measured weekly from six patients treated with two 28-day cycles of RGX-104 (administered once daily for three weeks, then off for one week) – week 0 corresponds immediately prior to treatment initiation. For some patients, data is not available for the entire two cycles due to lack of blood samples or treatment termination (n = 6). (C) Representative plots demonstrating the G-MDSC population in a colorectal cancer patient (top) and a renal cancer patient (bottom) treated with RGX-104 at week 0 (pre-treatment) compared with 2 weeks after therapy initiation. (D–E) Percent G-MDSCs (D) and M-MDSC (E) of total circulating cells in 6 patients treated with RGX-104 at week 0 compared to weeks 2–3 on therapy (n = 6). (F) Percent peak change in CD8⁺ T cells that express GITR of total CD8⁺ T cells in the circulation of patients treated with RGX-104 at week 0 compared to weeks 1–4 of the therapy cycle (n = 6). (G) Representative plot showing the population of PD-1⁺GITR⁺ double-positive CD8⁺ T cells (activated PD-1⁺ CTLs) in the circulation of a patient treated with RGX-104 at week 0 (pre-treatment) and at week 2 of therapy. (H) Percent peak change in CD8⁺ T cells that are double positive for PD-1⁺GITR⁺ in the circulation of patients treated with RGX-104 at week 0 compared to weeks 1–4 of the therapy cycle (n = 6). Data represent mean ± s.e.m. See also Figure S6.