Immune checkpoints PVR and PVRL2 are prognostic markers in AML and their blockade represents a new therapeutic option

Abstract

Immune checkpoints are promising targets in cancer therapy. Recently, poliovirus receptor (PVR) and poliovirus receptor-related 2 (PVRL2) have been identified as novel immune checkpoints. In this investigation we show that acute myeloid leukemia (AML) cell lines and AML patient samples highly express the T-cell immunoreceptor with Ig and ITIM domains (TIGIT) ligands PVR and PVRL2. Using two independent patient cohorts, we could demonstrate that high PVR and PVRL2 expression correlates with poor outcome in AML. We show for the first time that antibody blockade of PVR or PVRL2 on AML cell lines or primary AML cells or TIGIT blockade on immune cells increases the anti-leukemic effects mediated by PBMCs or purified CD3⁺ cells in vitro. The cytolytic activity of the BiTE® antibody construct AMG 330 against leukemic cells could be further enhanced by blockade of the TIGIT-PVR/PVRL2 axis. This increased immune reactivity is paralleled by augmented secretion of Granzyme B by immune cells. Employing CRISPR/Cas9-mediated knockout of PVR and PVRL2 in MV4-11 cells, the cytotoxic effects of antibody blockade could be recapitulated in vitro. In NSG mice reconstituted with human T cells and transplanted with either MV4-11 PVR/PVRL2 knockout or wildtype cells, prolonged survival was observed for the knockout cells. This survival benefit could be further extended by treating the mice with AMG 330. Therefore, targeting the TIGIT-PVR/PVRL2 axis with blocking antibodies might represent a promising future therapeutic option in AML.

Fig. 1

PVR and PVRL2 are highly expressed on AML cell lines and primary CD33⁺ AML blasts. PVR and PVRL2 protein expression, as depicted by the percentage of CD33⁺ cells as well as median fluorescence intensity as the measure of expression intensity on AML cell lines (n = 9; a, b) and CD33⁺ AML blasts from untreated patients (n = 17; c, d). Black dashes represent the median

Fig. 2

Blocking of the TIGIT-PVR/PVRL2 axis increases the lysis of AML cell lines. HD-PBMC-mediated lysis, with subject to the blocking of PVR and PVRLs on AML cell lines TF-1 (a, n = 3), Molm-13 (b, n = 6), Kasumi-1 (c, n = 3), was measured after 24 h. The effect of blocking the receptor TIGIT on effector cells was examined for the cell lines TF-1 (d, n = 5) and MV4-11 (e, n = 3). Results are depicted as the mean ± SD fold changes (FC) of dead target cells, relative to the control without blocking antibodies. Measurements were performed in technical triplicates and for statistical analysis Mann–Whitney U-tests were performed (#p ≤ 0.05; *p ≤ 0.001)

Fig. 3

T-cell mediated lysis of the BiTE® antibody construct AMG 330 is significantly enhanced by additional administration of PVR and PVRL2 or TIGIT blocking antibodies. TF-1 (a, n = 3), Molm-13 (b, n = 6), Kasumi-1 (c, n = 6) cells were incubated with HD-PBMCs and AMG 330 in the presence or absence of blocking antibodies against PVR or PVRL2. Blocking the receptor TIGIT on immune cells showed similar results for the cell line TF-1 (d, n = 5) and MV4-11 (e, n = 3). Results are depicted as the mean ± SD fold change (FC) of dead target cells, relative to the control without blocking antibodies. Lysis is mediated via CD3⁺ cells, as comparing HD-PBMCs and purified CD3⁺ cells from the same donor showed comparable results using the cell line TF-1 (f, n = 2). Results are depicted as the mean ± SD of dead target cells of two independent experiments. Measurements were performed in technical triplicates, and for statistical analysis Mann–Whitney U-tests were performed (#p ≤ 0.05; *p ≤ 0.001)

Fig. 4

Blocking of the TIGIT-PVR/PVRL2 axis results in increased levels of Granzyme B secretion of immune cells. TF-1 (a–d) and MV4-11 (e, f) target cells were mixed with HD-PBMCs and incubated with blocking antibodies against PVR and PVRL2 or TIGIT in the presence or absence of the BiTE® antibody construct AMG 330. After 24 h, supernatants were harvested and human Granzyme B concentration was measured using ELISA. Results are depicted as the mean ± SD Granzyme B concentration of at least three independent experiments. For statistical analysis paired t-tests were performed (# p ≤ 0.05)

Fig. 5

Combined blocking of PVR and PVRL2 on primary AML blasts increases the cytotoxic effects of HD-PBMCs. Mononuclear cells containing at least 75% blasts from bone marrow aspirates of ten different, newly diagnosed AML patients were stained with CMFDA (CellTracker™), mixed with HD-PBMCs as effector cells, and incubated for 72 h. Blocking of PVR and PVRL2 alone could increase the specific lysis of primary blasts in four (a, b, c, f) of the ten analyzed patients, and augment the anti-leukemic effect of the BiTE® antibody construct AMG 330 in five of nine patients (a–e). The sample of patient (f) was CD33 negative, and therefore excluded from the AMG 330 experiment. Results are depicted as the mean of technical triplicates ± SD of dead target cells

Fig. 6

PVR and PVRL2 double-knockout cells recapitulate antibody effects in vitro and prolong the survival of NSG mice reconstituted with human T cells in vivo. By using CRISPR/Cas9, a polyclonal population of MV4-11 harboring double-knockout cells of PVR and PVRL2 was generated. Either MV4-11 wildtype or double-knockout cells were incubated with HD-PBMCs (a) or CD3⁺ cells (b) for 24 h without or with AMG 330. For statistical analysis, Mann–Whitney U-tests were performed (#p ≤ 0.05; *p ≤ 0.001, n = 3). c Immunodeficient NSG mice were transplanted with either MV4-11 wildtype (WT) cells or PVR- and PVRL2-double-knockout (KO) cells and reconstituted with human T cells. Treatment consisted of daily intraperitoneal application of either placebo (n = 13 for WT and n = 12 for KO) or 15 µg/kg AMG 330 (n = 12 for WT and n = 15 for KO). Log-rank tests were performed: WT placebo vs. KO placebo p < 0.001; WT AMG 330 vs. KO AMG 330 p < 0.001; WT placebo vs. WT AMG 330 p = 0.003; KO placebo vs. KO AMG 330 p = 0.027

Fig. 7

Impact of PVR and PVRL2 expression on clinical outcome. Microarray-based gene expression data of 290 AML patients (cohort B) were divided into low and high expressors and analyzed for OS. High expression of either PVR or PVRL2 correlated significantly with a shortened overall survival