CD155 blockade enhances allogeneic natural killer cell-mediated antitumor response against osteosarcoma

Abstract

Background: Allogeneic bone marrow transplant (alloBMT) is curative for hematologic malignancies through the graft-versus-tumor (GVT) effect but has been ineffective for solid tumors like osteosarcoma (OS). OS expresses CD155 which interacts strongly with inhibitory receptors TIGIT and CD96 but also binds to activating receptor DNAM-1 on natural killer (NK) cells. CD155 has never been targeted after alloBMT. Combining adoptively transferred allogeneic NK (alloNK) cells with CD155 blockade after alloBMT may enhance a GVT effect against OS.

Methods: Murine NK cells were activated and expanded ex vivo with superagonist interleukin (IL)-15/IL-15Rα. AlloNK and syngeneic NK (synNK) cell phenotype, cytotoxicity, cytokine production, and degranulation against CD155-expressing murine OS cell line K7M2 were assessed in vitro. Mice bearing pulmonary OS metastases underwent alloBMT and alloNK cell infusion with anti-CD155 either before or after tumor induction, with select groups receiving anti-DNAM-1 pretreated alloNK cells. Tumor growth, graft-versus-host disease and survival were monitored, and differential gene expression of lung tissue was assessed by RNA microarray.

Results: AlloNK cells exhibited superior cytotoxicity against CD155-expressing OS compared with synNK cells, and this activity was enhanced by CD155 blockade. CD155 blockade increased alloNK cell degranulation and interferon gamma production through DNAM-1. In vivo, CD155 blockade with alloNK infusion increased survival when treating OS that relapsed after alloBMT. No benefit was seen for treating established OS before alloBMT. Combining CD155 and anti-DNAM-1 pretreated alloNK did not affect survival and tumor control benefits seen with CD155 blockade alone. RNA microarray showed mice treated with alloNK and CD155 blockade had increased expression of cytotoxicity genes and the NKG2D ligand H60a, whereas mice treated with anti-DNAM-1 pretreated alloNK cells resulted in upregulation of NK cell inhibitory receptor genes. Whereas blocking DNAM-1 on alloNK abrogated cytotoxicity, blocking NKG2D had no effect, implying DNAM-1:CD155 engagement drives alloNK activation against OS.

Conclusions: These results demonstrate the safety and efficacy of infusing alloNK cells with CD155 blockade to mount a GVT effect against OS and show benefits are in part through DNAM-1. Defining the hierarchy of receptors that govern alloNK responses is critical to translating alloNK cell infusions and immune checkpoint inhibition for solid tumors treated with alloBMT.

Keywords: Bone Cancer; Natural killer - NK.

Figure 1. IL-15 expansion of murine NK cells results in activated murine NK. B6 and BALB/c NK cells cultured with IL-15/IL-15Rα conjugate with additional media and cytokine repletion every 2–3 days are shown. (A) Cell count and (B) fold expansion at days 0, 7, 9, 12, and 14. Percentage of (C) B6 and BALB/c NK cells with marker expression at day 0 and day 12 shown with individual experimental replicates plotted with mean and SEM. (D) Fold change of marker expression MFI at day 12 compared with day 0 shown for B6 NK cells and BALB/c NK cells. Mean with SEM shown for experimental replicates (n=5). Two-sided two-sample t-tests were performed (*p<0.05, **p<0.01, ***p<0.001, ****p>0.0001). IL, interleukin; MFI, median fluorescence intensity; NK, natural killer.

Figure 2. AlloNKs exhibit potent anti-OS activity enhanced by CD155 blockade. (A) Proportion of K7M2 OS expressing CD155 and CD112 shown. (B) Proportion of K7M2 OS expressing Fas and TRAIL R2 shown. Individual values (n=5) plotted with mean and SEM. EAE murine alloNK and synNK expanded for 12 days were plated with mKate2-expressing K7M2 OS at an effector:target ratio of 2.5:1 with or without anti-CD155 antibody in the presence of green fluorescent caspase 3/7-substrate. Green and red fluorescence objects were measured by Incucyte real-time analysis. %Lysis was calculated based on target alone and staurosporine-induced max lysis conditions. OS lysis shown for (C) alloNK and synNK with or without anti-CD155 antibody blockade at E:T 2.5:1 (n=5 per group), (D) alloNK with anti-CD155 antibody or anti-CD155 Fab at E:T 2:1 (n=3 per group), (E) synNK with anti-CD155, anti-DNAM-1, or anti-CD155 and anti-DNAM-1 blockade at E:T 2.5:1 (n=5 per group), and (F) alloNK with anti-CD155, anti-DNAM-1, or anti-CD155 and anti-DNAM-1 blockade at E:T 2.5:1 (n=5 per group). Supernatant was collected and analyzed for IFNγ release by ELISA with samples in triplicate. Concentration was calculated based on a 5-parameter curve fit interpolated from the standard curve. IFNγ concentration shown for (G) alloNK and synNK (H) synNK with anti-CD155, anti-DNAM-1, or anti-CD155 and anti-DNAM-1 blockade, and (I) alloNK with anti-CD155, anti-DNAM-1, or anti-CD155 and anti-DNAM-1 blockade. Mean with SEM shown for technical replicates. Two-sided two-sample t-tests and one-way ANOVA and Kruskal-Wallis with Dunn’s multiple comparisons tests were performed (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001). alloNK, allogeneic natural killer; ANOVA, analysis of variance; EAE, ex vivo activation and expansion; E:T, effector:target; IFNγ, interferon gamma; OS, osteosarcoma; synNK, syngeneic natural killer.

Figure 3. Degranulation is the main killing mechanism used by alloNKs against OS. EAE murine alloNK and synNK cells were plated with K7M2 OS at an E:T ratio of 2.5:1. Cells were incubated for 30 min, followed by addition of monensin and brefeldin A, and incubated for an additional 4 hours. NK cells were collected, fixed and permeabilized, stained with surface antibodies and analyzed by flow cytometry. Cells were gated on B6 (NK1.1⁺) and BALB/c (CD49b⁺) cells and CD107a MFI was calculated and gated on CD107a⁺ cells perforin⁺, granzyme B⁺, FasL⁺, and TRAIL⁺ cells. CD107a MFI and percentage of cells expressing CD107a for alloNK and synNK cells (A), synNK cells with CD155 axis blockade (B), and alloNK cells with CD155 axis blockade (C) are shown. MFI and percentage of NK cells expressing perforin, and granzyme B (D) and FasL and TRAIL (E) are shown. Mean with SEM shown for experimental replicates (n=5). Two-tailed two-sample t-tests and one-way ANOVA with Kruskal-Wallis with Dunn’s multiple comparisons tests were performed (*p<0.05, **p<0.01). alloNK, allogeneic natural killer; ANOVA, analysis of variance; EAE, ex vivo activation and expansion; E:T, effector:target; MFI, median fluorescence intensity; NK, natural killer; OS, osteosarcoma; synNK, syngeneic natural killer.

Figure 4. AlloBMT and single or multiple dose CD155 blockade with alloNK effectively treat relapsed OS pulmonary metastases. (A) Schematic showing alloBMT followed by K7M2 OS inoculation and single dose or multiple doses of alloNK infusion with CD155 axis blocking antibodies. (B) Serum levels of mouse IL-15/IL-15R⍺ pre- and post-BMT and single dose alloNK and CD155 axis blockade treatment at time points post alloBMT (n=6–15 mice per time point) (C) EFS for mice treated with alloBMT and single dose alloNK with isotype control or anti-CD155 blockade are shown at time points post alloBMT (n=8 or 10 mice per group). (D) Clinical GVHD scores for mice treated with alloBMT and single dose alloNK with isotype control or anti-CD155 blockade are shown (n=8 or 10 mice per group). (E) EFS for mice treated with alloBMT and multiple doses of alloNK with isotype control or anti-CD155 blockade are shown at time points post alloBMT (n=3 mice per group). (F) Clinical GVHD scores for mice treated with alloBMT and multiple doses of alloNK with isotype control or anti-CD155 blockade are shown (n=3 mice per group). Bars and points with error bars depict mean with SEM. Log-rank test performed and student’s t-tests were performed (*p<0.05). alloBMT, allogeneic bone marrow transplant; alloNK, allogeneic natural killer; BMT, bone marrow transplant; EFS, event-free survival; GVHD, graft-versus-host disease; IL, interleukin; NK, natural killer; OS, osteosarcoma.

Figure 5. CD155 blockade with alloNK shows limited efficacy against established OS disease after alloBMT. (A) Schematic showing K7M2 OS inoculation, alloBMT, and alloNK infusion with CD155 axis blocking antibodies. (B) Blood serum levels of mouse IL-15/IL-15R⍺ pre-BMT and post-BMT and alloNK and CD155 axis blockade treatment (n=7–8 mice per time point) (C) EFS for mice treated with alloBMT and alloNK with isotype control, anti-DNAM-1 blockade, anti-CD155 blockade, or anti-DNAM and anti-CD155 blockade are shown (n=7–8 mice per group). (D) Clinical GVHD scores for mice treated with alloBMT and alloNK with isotype control, anti-DNAM-1 blockade, anti-CD155 blockade, or anti-DNAM and anti-CD155 blockade are shown (n=7–8 mice per group). (E) Tumor burden measured by IVIS shown as total flux (photons/s) for mice treated with alloBMT and alloNK with isotype control or anti-CD155 blockade (n=6 per group). (F) Recovery of infused alloNK expressing CD45.1 from spleens harvested from mice treated with alloBMT and alloNK with isotype control or anti-CD155 blockade (n=3 mice per time point). Bars and points with error bars depict mean with SEM. alloBMT, allogeneic bone marrow transplant; alloNK, allogeneic natural killer; BMT, bone marrow transplant; EFS, event-free survival; GVHD, graft-versus-host disease; IL, interleukin; IVIS, in vivo imaging system; NK, natural killer; OS, osteosarcoma.

Figure 6. CD155 blockade and DNAM-1 blockade with alloNK treatment change gene expression in lung tissue of OS tumor-bearing mice post alloBMT. (A) Volcano plots of differential expression shown for treatment groups. A horizontal cut-off is drawn at adjusted p=0.05. Two vertical cutoffs are drawn at log2(fold change) at −0.6 and 0.6. (n=3 experimental replicates per group). (B) Mice treated with alloNK and anti-CD155 versus alloNKa nd IgG2a, (C) Mice treated with alloNK and anti-DNAM-1 over alloNK and IgG2a and (D) mice treated with alloNK, anti-DNAM-1, and anti-CD155 over alloNK and anti-CD155. (E) Multidimensional scaling distances between endogenous genes quantified from each NanoString dataset shown. The distance between samples on this plot corresponds to log-fold change in gene expressions between samples (n=3 experimental replicates per group). alloBMT, allogeneic bone marrow transplant; alloNK, allogeneic natural killer; Ccl3, chemokine ligand 3; Ccl9, chemokine ligand 9; Gzma, Granzyme A; Lamp3, lysosomal-associated membrane protein 3; NK, natural killer; OS, osteosarcoma.

Figure 7. DNAM-1 blockade increases NKG2D expression in synNK and alloNK cells and has a greater impact on cytotoxicity compared with NKG2D. EAE murine alloNK and synNK were plated with K7M2 OS at an E:T ratio of 2.5:1. NK cells were collected and analyzed by flow cytometry. NKG2D MFI and percentage of NKG2D⁺ NK cells shown for (A) alloNK and synNK (B) synNK with anti-CD155, anti-DNAM-1, or anti-CD155 and anti-DNAM-1 blockade, and (C) alloNK with anti-CD155, anti-DNAM-1, or anti-CD155 and anti-DNAM-1 blockade. Mean with SEM shown for experimental replicates (n=5). EAE murine alloNK and synNK were plated with mKate2-expressing K7M2 OS. %Lysis was calculated based on target alone and staurosporine-induced max lysis conditions. OS lysis shown for (D) alloNK and synNK with combinations of anti-CD155, anti-DNAM-1, and anti-NKG2D blockade at E:T 2.5:1 (n=5 per group). Two-tailed two-sample t-tests and one-way ANOVA with Kruskal-Wallis with Dunn’s multiple comparisons tests were performed (*p<0.05, **p<0.01). alloNK, allogeneic natural killer; ANOVA, analysis of variance; EAE, ex vivo activation and expansion; E:T, effector:target; MFI, median fluorescence intensity; NK, natural killer; OS, osteosarcoma; synNK, syngeneic natural killer.