Whole-genome CRISPR screening identifies molecular mechanisms of PD-L1 expression in adult T-cell leukemia/lymphoma

Abstract

Adult T-cell leukemia/lymphoma (ATLL) is an aggressive T-cell malignancy with a poor prognosis and limited treatment options. Programmed cell death ligand 1(PD-L1) is recognized to be involved in the pathobiology of ATLL. However, what molecules control PD-L1 expression and whether genetic or pharmacological intervention might modify PD-L1 expression in ATLL cells are still unknown. To comprehend the regulatory mechanisms of PD-L1 expression in ATLL cells, we performed unbiased genome-wide clustered regularly interspaced short palindromic repeat (CRISPR) screening in this work. In ATLL cells, we discovered that the neddylation-associated genes NEDD8, NAE1, UBA3, and CUL3 negatively regulated PD-L1 expression, whereas STAT3 positively did so. We verified, in line with the genetic results, that treatment with the JAK1/2 inhibitor ruxolitinib or the neddylation pathway inhibitor pevonedistat resulted in a decrease in PD-L1 expression in ATLL cells or an increase in it, respectively. It is significant that these results held true regardless of whether ATLL cells had the PD-L1 3' structural variant, a known genetic anomaly that promotes PD-L1 overexpression in certain patients with primary ATLL. Pevonedistat alone showed cytotoxicity for ATLL cells, but compared with each single modality, pevonedistat improved the cytotoxic effects of the anti-PD-L1 monoclonal antibody avelumab and chimeric antigen receptor (CAR) T cells targeting PD-L1 in vitro. As a result, our work provided insight into a portion of the complex regulatory mechanisms governing PD-L1 expression in ATLL cells and demonstrated the in vitro preliminary preclinical efficacy of PD-L1-directed immunotherapies by using pevonedistat to upregulate PD-L1 in ATLL cells.

Graphical abstract

Figure 1.

A CRISPR screen reveals PD-L1 regulating genes. (A) Cell surface expression of PD-L1 in ATLL cell lines, ALK⁺ ALCL cell lines, and T-ALL cell lines measured by flow cytometry. (B) Schematic design of the CRISPR library screen in the study. Two Cas9-transducing ATLL cell lines, ST1 and KK1, were used for identifying genes whose knockout decreases or increases PD-L1 expression, respectively. (C) Log₂ fold-changes (sorted/unsorted population) of genes enriched in ST1 expressing low level PD-L1 (left panel) and in KK1 expressing high level PD-L1 (right panel). (D) Top 10 log₂ fold change genes with 3 or 4 sgRNA log₂ fold changes ≥1.0 in ST1 expressing low level PD-L1 (left panel) and in KK1 expressing high level PD-L1 (right panel).

Figure 2.

STAT3 inhibition suppresses PD-L1 expression in ATLL cells. (A) Cell surface expression of PD-L1 in KK1 and ST1 ATLL cells transduced with the indicated sgRNA by flow cytometry. PD-L1 mean fluorescent intensity (MFI) was normalized by that of control sgRNA (sgAAVS1)-transduced cells. (B) Cell surface expression of PD-L1 in ATLL (KK1, Su9T01, and ST1) and ALK+ ALCL (DEL, Karpas299, and SUDHL1) cells treated with ruxolitinib for 24 hours by flow cytometry. MFI was normalized by that of dimethyl sulfoxide (DMSO)-treated cells. (C) Immunoblot analysis of p-STAT3, STAT3 and α-tubulin in ruxolitinib-treated ATLL and ALK⁺ ALCL cells. (D) Schema for the genomic locations targeted by sgRNAs for CRISPR/Cas9-mediated gene editing in the PD-L1 3′-UTR region. (E-F) Cell surface expression of PD-L1 in KK1 PD-L1 SV#7 cells and KK1 PD-L1 SV#15-12 cells by flow cytometry. The cells were treated with the indicated concentrations of ruxolitinib for 24 hours, and MFI was normalized by that of DMSO-treated cells in panel F. (G) Immunoblot analysis of PD-L1, p-STAT3, STAT3, and α-tubulin in KK1 PD-L1 SV#7 cells and KK1 PD-L1 SV#15-12 cells treated with or without ruxolitinib (312.5 nM) for 24 hours. Error bars represent the mean with standard error of the mean (SEM) of replicates. ∗P < .05; ∗∗P < .01, Welch 2-sample t test. All experiments were repeated at least 2 times.

Figure 3.

Neddylation inhibition increases PD-L1 expression in ATLL cells. (A) Cell surface expression of PD-L1 in KK1 and ST1 ATLL cells transduced with the indicated sgRNA by flow cytometry. (B) Immunoblot analysis of PD-L1, UBA3, NEDD8, NAE1, CUL3, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in sgAAVS1, sgUBA3, sgNEDD8, sgNAE1, and sgCUL3-transduced KK1 cells. Error bars represent the mean with SEM of replicates. ∗P < .05; ∗∗P < .01; Welch 2-sample t test. All experiments were repeated at least twice.

Figure 4.

Pevonedistat increases PD-L1 expression in ATLL cells. (A) Cell surface expression of PD-L1 in ATLL, ALK⁺ ALCL, and T-ALL cell lines treated with the indicated concentration of pevonedistat for 24 hours measured by flow cytometry. The y-axis represents MFI normalized by that of DMSO-treated cells. (B) mRNA expression of PD-L1 in KK1 and ST1 ATLL cells treated with pevonedistat for 24 hours as measured by the TaqMan gene expression assay. (C) Immunoblot analysis of PD-L1, pSTAT3, STAT3, CUL5, and α-tubulin in KK1 and ST1 ATLL cells treated with the indicated amount of pevonedistat for 24 hours. The arrowhead indicates the neddylated form of CUL5, which was decreased by using pevonedistat. (D) ST1 cells transduced with sgSTAT3 together with a GFP reporter or with a control sgAAVS1 were treated with pevonedistat for 24 hours. Cell surface expression of PD-L1 in the GFP-expressing cell populations was measured by flow cytometry. (E) Cell surface expression of PD-L1 in KK1, ST1, and Su9T01 ATLL cell lines treated with the indicated concentrations of pevonedistat and ruxolitinib for 24 hours was measured by flow cytometry. (F) Immunoblot analysis of PD-L1, pSTAT3, STAT3, CUL5, and α-tubulin in KK1, ST1, and Su9T01 ATLL cells treated with the indicated amounts of pevonedistat and ruxolitinib for 24 hours. The arrowhead indicates the neddylated form of CUL5. (G-H) Cell surface expression of PD-L1 in unmanipulated KK1, KK1 PD-L1 SV#7, and KK1 PD-L1 SV#15-12 cells treated with the indicated amount of pevonedistat (G) or with a combination of pevonedistat and ruxolitinib (H) for 24 hours was measured by flow cytometry. (I) Cell surface expression of PD-L1 in the mutated STAT3-transduced KK1 ATLL cells treated with the indicated amount of pevonedistat for 24 hours was measured by flow cytometry. (J) Immunoblot analysis of PD-L1, pSTAT3, STAT3, and GAPDH in the mutated STAT3-transduced KK1 ATLL cells. Error bars represent the mean with SEM of replicates. ∗P < .05; ∗∗P < .01; Welch 2-sample t test. All experiments were repeated at least twice. GFP, green fluorescent protein.

Figure 5.

Pevonedistat induces apoptosis and cell cycle arrest in ATLL cell lines. (A) Viable cell numbers measured by the MTS assay for ATLL cell lines treated with the indicated amount of pevonedistat for 96 hours. (B) The percentages of apoptotic cells in KK1 cells treated with or without pevonedistat at 1250 nM for 48 hours were detected by analyzing annexin V and propidium iodide (PI) on flow cytometry. (C) The percentage of PI(−) annexin V(+)/PI(−) KK1 cells treated with or without pevonedistat at 1250 nM was monitored at 48 and 96 hours. The black and red bars represent DMSO- or pevonedistat-treated cells, respectively. (D) DNA content was analyzed in KK1 cells with or without pevonedistat for 48 hours. Error bars represent the mean with SEM of replicates. ∗∗∗P < .001, Welch 2-sample t test. All experiments were repeated at least twice.

Figure 6.

Pevonedistat can enhance the cytotoxicities of anti–PD-L1 monoclonal antibody avelumab in ATLL cells. (A) Values of specific lysis of the target KK1 cells with or without ectopic STAT3^Y640F cDNA expression under cocultivation with the effector PBMNC. The target cells were stained with CellBriteTM Red Cytoplasmic Membrane dye for the discrimination from effector cells and treated with 1 μg/mL avelumab for 1 hour before coculture. After the cocultivation of the target KK1 cells and the effector PBMNC for the indicated time, the dye-stained target cell numbers were analyzed by flow cytometry. The numbers were normalized by cell counting beads. (B) The target KK1 wild type or KK1 PD-L1-SV#7 cells were analyzed as shown in panel A. (C-D) The target KK1 wild-type cells (C) or KK1 PD-L1-SV#7 (D) were pretreated with or without pevonedistat for 24 hours. After washing, the target cells were analyzed as shown in panel A. Error bars represent the mean with SEM of replicates. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; Welch 2-sample t test. All experiments were repeated at least 2 times. cDNA, complementary DNA.

Figure 7.

Pevonedistat can enhance the cytotoxicities of PD-L1 CAR T cells in ATLL cells. (A) Schematic design of the durvalumab-based anti–PD-L1 CAR construct used in the study. (B) Representative plots showing the transduction efficiency of mock-control T cells and PD-L1 CAR T cells analyzed by flow cytometry. (C) Cell surface expression of PD-L1 in KK1 cells transduced with sgAAVS1 or sgPD-L1 was measured via flow cytometry. PD-L1 knockout KK1 cells were established by single-cell cloning from bulk sgPD-L1–transduced cells. (D) Values of specific lysis of KK1 cells transduced with sgAAVS1 or sgPD-L1 under cocultivation with PD-L1 CAR T cells. Data were obtained as shown in Figure 6A. (E) Values of specific lysis of unmanipulated KK1 (wild-type), KK1 PD-L1-SV#7, and KK1 PD-L1-SV#15-12 cells under cocultivation with mock T cells and PD-L1 CAR T cells. Data were obtained as shown in Figure 6A. (F-H) Values of specific lysis of unmanipulated KK1 (wild-type) (F), KK1 PD-L1 SV#7 (F), KK1 PD-L1 SV#15-12 (F), unmanipulated ST1 (G), and unmanipulated TL-Om1 cells (H) under cocultivation with the effector PD-L1 CAR T cells. The target cells were pretreated with pevonedistat (1250 nM) or DMSO for 24 hours, washed twice, stained with CellBrite red cytoplasmic membrane dye, and cocultured with the effector PD-L1 CAR T cells. Data were obtained as shown in Figure 6A. Error bars represent the mean with SEM of replicates. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; Welch 2-sample t test. All experiments were repeated at least 2 times.