CRISPR-Cas9 Knockout Screens Identify DNA Damage Response Pathways and BTK as Essential for Cisplatin Response in Diffuse Large B-Cell Lymphoma

Abstract

The recurrence of diffuse large B-cell lymphoma (DLBCL) has been observed in 40% of cases. The standard of care for refractory/relapsed DLBCL (RR-DLBCL) is platinum-based treatment prior to autologous stem cell transplantation; however, the prognosis for RR-DLBCL patients remains poor. Thus, to identify genes affecting the cisplatin response in DLBCL, cisplatin-based whole-genome CRISPR-Cas9 knockout screens were performed in this study. We discovered DNA damage response (DDR) pathways as enriched among identified sensitizing CRISPR-mediated gene knockouts. In line, the knockout of the nucleotide excision repair genes XPA and ERCC6 sensitized DLBCL cells to platinum drugs irrespective of proliferation rate, thus documenting DDR as essential for cisplatin sensitivity in DLBCL. Functional analysis revealed that the loss of XPA and ERCC6 increased DNA damage levels and altered cell cycle distribution. Interestingly, we also identified BTK, which is involved in B-cell receptor signaling, to affect cisplatin response. The knockout of BTK increased cisplatin sensitivity in DLBCL cells, and combinatory drug screens revealed a synergistic effect of the BTK inhibitor, ibrutinib, with platinum drugs at low concentrations. Applying local and external DLBCL cohorts, we addressed the clinical relevance of the genes identified in the CRISPR screens. BTK was among the most frequently mutated genes with a frequency of 3-5%, and XPA and ERCC6 were also mutated, albeit at lower frequencies. Furthermore, 27-54% of diagnostic DLBCL samples had mutations in pathways that can sensitize cells to cisplatin. In conclusion, this study shows that XPA and ERCC6, in addition to BTK, are essential for the response to platinum-based drugs in DLBCL.

Keywords: CRISPR; DNA damage response; cisplatin; diffuse large B-cell lymphoma; ibrutinib; relapse.

Figure 1

Genomewide CRISPR screens using cisplatin in DLBCL cells. (A) Experimental setup for the CRISPR screens using two doses of cisplatin for 10 days (created with biorender.com). (B) From the screens, a total of 326 genes conferring resistance to cisplatin upon knockout and 425 genes conferring sensitivity to cisplatin upon knockout were identified between the two doses. (C) Comparison of gene scores (β-scores) calculated using MAGeCK-MLE between the two doses of cisplatin. Pearson correlation between the two doses was assessed considering all genes (faint dots, and in center). Genes in the enriched category (green if observed in both doses or else black) and depleted category (red if observed in both doses or else black) are illustrated. (D) Mutational frequencies (nonsynonymous variants) of the cisplatin-sensitizing gene knockouts identified from the CRISPR screen were examined for mutation frequencies in three clinical DLBCL cohorts consisting of pre-treatment diagnostic samples. Green genes are from the enriched category, red genes are from the depleted category. (E) gRNA counts from the CRISPR screen of genes chosen for validation experiments. Lines illustrate gRNA-wise comparisons between the saline controls and cisplatin treated knockout populations after 10 days, in which negative slopes indicate a depletion in the cisplatin treated population.

Figure 2

XPA gene knockout. (A) Polyclonal populations of XPA^KO cells were generated using targeting ribonucleoprotein (RNP) complexes alongside scramble control gRNAs and validated using western blot. Two gRNAs were used to generate knockout in each cell line, and the effect was compared to nontargeting scramble (SCR) controls. Gray bars show the ICE scores, i.e., the percentage of indel mutations in the population, whereas the blue bars show the knockout scores, i.e., gene edits leading to frameshifts and ≥21 bp deletion. (B) Proliferation rate was not affected in most XPA^KO cells. RIVA XPA^KO were used for functional validation. (C) MTS-based dose–response screens showed increased sensitivity to platinum compounds in XPA^KO cells across four DLBCL cell lines representing different molecular subtypes. If both gRNA display similar significance when compared to SCR, they are marked with black symbols and placed above SCR curves, and gRNA-specific significance symbols are placed under the respective curves. (D) Cell cycle analysis performed with propidium iodide in RIVA. (E,F) DNA damage measured using phosphorylation of H2AX (γH2AX) as a marker in RIVA XPA^KO cells without (E) and with (F) cisplatin. Values are displayed as mean ± SEM (one-way or two-way ANOVA: n = 9–18 for (B), n = 18–24 for (C), and n = 6–8 for (D–F); ns p >  0.05; * p  ≤  0.05; ** p  ≤  0.01; *** p  ≤  0.001; **** p  ≤  0.0001 compared with SCR controls) and are summarized based on three independent experiments for (B,C) and two independent experiments for (D,F). The uncropped bolts are shown in Supplementary Materials File S1.

Figure 3

ERCC6 gene knockout. (A) Polyclonal populations of ERCC6^KO cells were generated using targeting ribonucleoprotein (RNP) complexes alongside scramble control gRNAs and validated using western blot. Two gRNAs were used to generate knockout in each cell line, and the effect was compared to nontargeting scramble (SCR) controls. Gray bars show the ICE scores, i.e., the percentage of indel mutations in the population, whereas the blue bars show the knockout scores, i.e., gene edits leading to frameshifts and ≥21 bp deletion. (B) Proliferation rate was not affected in most ERCC6^KO cells. RIVA ERCC6^KO were used for functional validation. (C) MTS-based dose–response screens showed increased sensitivity to platinum compounds in ERCC6^KO cells across three DLBCL cell lines representing different molecular subtypes. If both gRNA display similar significance when compared to SCR, they are marked with black symbols and placed above SCR curves, and gRNA-specific significance symbols are placed under the respective curves. (D) Cell cycle analysis performed with propidium iodide in RIVA. (E,F) DNA damage measured using phosphorylation of H2AX (γH2AX) as a marker in RIVA ERCC6^KO cells without (E) and with (F) cisplatin. Values are displayed as mean ± SEM (one-way or two-way ANOVA: n = 12–18 for (B), n = 18–24 for (C), and n = 6–8 for (D–F); ns p >  0.05; * p  ≤  0.05; ** p  ≤  0.01; *** p  ≤  0.001; **** p  ≤  0.0001 compared with SCR controls) and are summarized based on three independent experiments for (B,C) and two independent experiments for (D–F). A representative example is shown in (D,F). The uncropped bolts are shown in Supplementary Materials File S1.

Figure 4

BTK gene knockout. (A) Polyclonal populations of BTK^KO cells were generated using targeting ribonucleoprotein (RNP) complexes alongside scramble control gRNAs and validated using western blot. A total of 1–2 gRNAs was used to generate knockout in each cell line, and the effect was compared to nontargeting scramble (SCR) controls. Gray bars show the ICE scores, i.e., the percentage of indel mutations in the population, whereas the blue bars show the knockout scores, i.e., gene edits leading to frameshifts and ≥21 bp deletion. (B,C) MTS-based proliferation and dose–response screens across three BTK^KO DLBCL cell lines. If both gRNA display similar significance when compared to SCR, they are marked with black symbols and placed above SCR curves, and gRNA-specific significance symbols are placed under the respective curves. (D) Drug combination screening in OCILY7 with mono and combination treatments of platinum drugs + ibrutinib leading to a total of 25 distinct combinations for cisplatin + ibrutinib (top) and carboplatin + ibrutinib (bottom). Synergistic drug interactions (negative Bliss scores, blue) and antagonistic drug interactions (positive Bliss scores, red) are displayed in each matrix figure. On the right side, bar plots display the lowest ibrutinib dose in combination with all five platinum drug doses. Values are displayed as mean ± SEM (one-way or two-way ANOVA: n = 18–24 for (C) and n = 12–16 for (D); ns p >  0.05; * p  ≤  0.05; ** p  ≤  0.01; *** p  ≤  0.001; **** p  ≤  0.0001 compared to SCR controls in (C) and to the combination treatment in (D)). Data are summarized based on 3–4 independent experiments for (B–D). The uncropped bolts are shown in Supplementary Materials File S1.