VRK1 as a synthetic lethal target in VRK2 promoter-methylated cancers of the nervous system

Abstract

Collateral lethality occurs when loss of a gene/protein renders cancer cells dependent on its remaining paralog. Combining genome-scale CRISPR/Cas9 loss-of-function screens with RNA sequencing in over 900 cancer cell lines, we found that cancers of nervous system lineage, including adult and pediatric gliomas and neuroblastomas, required the nuclear kinase vaccinia-related kinase 1 (VRK1) for their survival in vivo. VRK1 dependency was inversely correlated with expression of its paralog VRK2. VRK2 knockout sensitized cells to VRK1 loss, and conversely, VRK2 overexpression increased cell fitness in the setting of VRK1 loss. DNA methylation of the VRK2 promoter was associated with low VRK2 expression in human neuroblastomas and adult and pediatric gliomas. Mechanistically, depletion of VRK1 reduced barrier-to-autointegration factor phosphorylation during mitosis, resulting in DNA damage and apoptosis. Together, these studies identify VRK1 as a synthetic lethal target in VRK2 promoter-methylated adult and pediatric gliomas and neuroblastomas.

Keywords: Brain cancer; Cancer; Molecular genetics; Oncology.

Figure 1. VRK1 is a dependency in glioblastoma, NB, and DMG.

(A) Histogram plots showing VRK1 CERES-corrected dependency scores in over 900 cell lines, representing 25 cancer lineages from the DepMap data set (21Q3). Compared with all other lineages, cell lines in the CNS (P = 2 × 10^–12) and PNS (P = 3 × 10^–18) lineages were significantly more dependent on VRK1. (B) Differential dependency of gene KO in CNS and PNS cell lines versus all other lineages. Gene effect size is calculated as the difference in average CERES score between lineage groupings, and q value is determined by limma eBayes methodology. The top enriched dependencies in CNS/PNS lineages are annotated. (C) VRK1 protein expression following expression of 4 different sgRNAs in the NB-1 neuroblastoma cell line. The top 3 guides with greatest VRK1 loss were carried forward in subsequent experiments. (D) Population doubling assay following VRK1 KO with 3 separate guides in NB-1 cells. sgCtrl represents a nontargeting control guide (n = 3; mean ± SD). (E) sgVRK1 KO after 14 days in cell lines representing NB (n = 3), GBM (n = 2), and DMG (n = 2) models. (n ≥ 3; mean ± SD plotted.) (F) Time course of CASP3/7 activity, as measured by cleavage of a peptide reporter, following VRK1 KO in LN443 cells (n = 3; mean ± SD). Total reporter fluorescence signal is normalized by cell confluence. Significance at each time point was determined by 2-way ANOVA (treatment × time). *P < 0.05. (G) Quantification of annexin V–positive cells following VRK1 KO with 2 separate guides in 3 cell lines representing NB and DMG lineages after 7 days. (n = 3; mean ± SD; from 2 separate experiments.) *P < 0.05, **P < 0.001, ***P < 0.0001; significance was determined by 2-tailed Student’s t test (E) and 1-way ANOVA with Tukey’s (D and G).

Figure 2. VRK1 dependency is correlated with VRK2 expression.

(A) Whole-genome Pearson correlations between gene expression from CCLE (21Q3) and VRK1 dependency in the DepMap database (21Q3) and adjusted P values. (B) Scatterplot showing VRK1 dependency versus VRK2 expression. Extent of VRK2 promoter methylation is indicated by dot size. Red dots represent cell lines of CNS lineage, and blue dots represent PNS lineage. (C) VRK2 promoter methylation status stratified by clinical characteristics across the TCGA GBM-LGG cohort. LGG, low grade glioma; GBM, glioblastoma multiforme. Violin plots with mean (solid line) and first and third quartiles (dashed line). (D) VRK2 promoter methylation in pediatric high-grade gliomas and DMGs with wild-type histone H3 and mutant histone H3 (K27M or G34R). Data from Mackay et al., 2017 (39). Violin plots with mean (solid line) and first and third quartiles (dashed line). (E) Cell viability following 14 days’ KO of VRK1 in VRK2^lo LNZ308 and LN443 cell lines and VRK2^hi GAMG and SF172 cell lines. (F) Cell viability analysis 14 days following VRK1 KO in VRK2^hi GBM cell line (SF172), expressing control CRISPR sgRNA or sgRNA targeting VRK2. (n = 3; mean ± SD.) (G) Immunoblot showing the overexpression of exogenous VRK2^WT, VRK1^WT, and kinase-inactive VRK1^K179E in NB-1 NB cells with or without VRK1 KO. RFP, red fluorescent protein. (H) Cell viability analysis for NB-1 cells in G following 14 days of VRK1 KO in cells overexpressing VRK2^WT, VRK1^WT, and kinase-inactive VRK1^K179E. (n = 3; mean ± SD.) (I) Effect of VRK2^WT or VRK2^K168E overexpression on LN443 GBM cell viability following 14 days VRK1 KO. (n = 3; mean ± SD.) *P < 0.05, **P < 0.001, ***P < 0.0001; significance was determined by 2-tailed Student’s t test (E) and 1-way ANOVA with Tukey’s test (C, D, F, H, and I).

Figure 3. Global phospho-proteomics following acute VRK1 degradation.

(A) Schematic of dTAG-VRK1 degrader system. The conjugated FKBP12^F36V binding domain allows small molecule–mediated (dTAG^V-1) recruitment of the VHL ubiquitin ligase complex, targeting exogenous VRK1 for proteasomal degradation. (B) Schematic of VRK1 degrader experiments. Exogenous dTAG-VRK1 is transduced to rescue CRISPR KO of endogenous VRK1. Exogenous dTAG-VRK1 is then under the control of the small molecule degrader (dTAG^V-1) allowing for acute downregulation. (C) Immunoblot validation of the dTAG-VRK1 degrader system in NB-1 neuroblastoma cells. Exogenous dTAG-VRK1 was degraded with dTAG^V-1. Endogenous VRK1 was independently targeted with CRISPR KO. sgLacZ is a nontargeting guide control. (D) Cell viability analysis of dTAG-VRK1-NB-1 cells following addition of either vehicle control or 0.5 μM dTAG^V-1. Significance at each time point was determined by 2-way ANOVA (treatment × time). *P < 0.05, **P < 0.001. (E) Schematic of the quantitative, global phospho-proteomic experiment. Samples were generated in triplicate at 4 hours and 8 hours after dTAG^V-1 (0.5 μM) addition. Following trypsin digestion, peptides were tagged with isobaric tandem mass tags (TMTs), then combined. Phospho-enrichment was performed using IMACs, and then peptides were run on an Orbitrap mass spectrometer. MS2 spectra offer peptide IDs and sample deconvolution through attached mass tags. (F) KSEA of phosphorylation site dynamics following acute degradation of exogenous VRK1. Kinase substrates of CDK1 and AURKA were significantly downregulated following degradation (blue), while substrates of WEE1, BRSK1, and ATM were significantly upregulated (red). (G) Top panel: Venn diagram showing number of unique proteins with a decrease in phosphorylation for at least 1 phosphorylation site in dTAG^V-1–treated samples. Bottom panel: Dot plots showing the overlap of downregulated protein phosphorylation (208 proteins) with select categories of the C5 MSigDB library. All gene sets have FDR ≤ 0.05 as determined by 1-tailed Fisher’s exact test.

Figure 4. VRK1 loss is associated with nuclear envelope malformation.

(A) Nuclear membrane morphology in the LN443 GBM cell line following exogenous VRK1 degradation by dTAG^V-1 after 1 day. White arrows point to nuclear bridges. Blue arrow points to micro-nuclei. (B) Left: Quantitation of irregular nuclei, by LaminB1 staining, following VRK1 degradation as seen in A (n = 3 fields of >50 cells each; mean ± SD). Center: Quantitation of nuclear bridges following VRK1 degradation as seen in A (n = 3 fields of > 50 cells each; mean ± SD). Right: quantitation of irregular nuclei following VRK1 degradation in the NB-1 neuroblastoma cell line expressing GFP-BAF seen in Supplemental Figure 9D (n = 8 fields of >50 cells each; mean ± SD). (C) Quantitation of irregular nuclei, by LaminB1 staining, following KO of both VRK1 and VRK2 in SF172 as seen in Supplemental Figure 9A. (n = 4 fields of >50 cells each; mean ± SD.) (D) Immunoblot of phosphorylated BAF (S4) and total BAF following dTAG^V-1 treatment in dTAG-VRK1-NB-1 cells (left panel) or KO of VRK1 with 2 independent sgRNAs in BT869Luci DMG neurospheres (right panel). Represents 2 independent experiments. (E) Left panel: Nuclear envelope morphology (Emerin-GFP) following doxycycline-induced expression of BAF mutants in LN443 GBM cell line after 3 days: wild-type (WT), S4A (nonphosphorylatable), S4D (phospho-mimetic). Right panel: Quantitation of nuclear bridging phenotype in LN443 cell lines expressing BAF mutants (n = 3; mean ± SD). (F) Live-cell, time-lapse experiment showing nuclear envelope morphology following VRK1 degradation in LN443 (dTAG^V-1 addition at t = 0 hours). White arrows point to cells undergoing mitosis. Blue arrows point to chromatin bridges. Represents 2 independent experiments. Scale bars: 20 μm. *P < 0.05, **P < 0.001, ***P < 0.0001; significance was determined by 2-tailed Student’s t test (B) and 1-way ANOVA with Tukey’s test (C and E).

Figure 5. VRK1 loss results in DNA damage.

(A) Left panel: Nuclear foci of a panel of DNA damage markers — phospho-H2AX (S139), phospho-ATR (S428), and phospho-DNAPK (S2056) — following KO of VRK1 in LN443 GBM cells for 7 days. Right panel: Quantitation of percentage of cells with >2 phospho-H2AX foci following VRK1 KO (n = 3 fields of >50 cells each; mean ± SD). (B) Top panel: Phospho-H2AX foci following 7-day double-KO combinations of sgCtrl/sgCtrl, sgCtrl/sgVRK1, sgCtrl/sgVRK2, and sgVRK1/sgVRK2. Bottom panel: Quantitation of percentage of cells with >2 phospho-H2AX foci following these double-KO combinations (n = 4 fields of >50 cells each; mean ± SD). (C) Top panel: Phospho-H2AX foci following VRK1 degradation with 0.5 μM dTAG^V-1 in both Kelly and NB-1 NB cell lines. Bottom panel: Quantitation of percentage of cells with >2 phospho-H2AX foci following dTAG^V-1 addition (n = 4 fields of >30 cells each; mean ± SD). Scale bars: 20 μm. *P < 0.05, **P < 0.001, ***P < 0.0001; significance was determined by 2-tailed Student’s t test (A and C) and 1-way ANOVA with Tukey’s test (B).

Figure 6. VRK1 is a dependency in vivo.

(A) Left panel: Immunoblot of VRK1 following tamoxifen-induced expression of sgVRK1 in LN443 cells. Right panel: Clonogenic assay in LN443 cells 14 days following tamoxifen-induced KO of VRK1. (B) Schematic of the in vivo xenograft experiment. The SF295 GBM cell line was transduced with Cas9, Cre-ERT2, and Switch-ON guide plasmids and implanted in NSG mouse flanks. When the tumors reached a prespecified size (200 mm³), the mice were treated with tamoxifen. When the tumor size reached approximately 500 mm³ or 40 days following treatment, the mice were euthanized. (C) Tumor volume measurements over time of the flank xenografts. * represents injection of tamoxifen or corn oil vehicle control. (D) Left panel: representative H&E sections of tumors taken from xenografted mice, 7 days following treatment with tamoxifen or vehicle control (scale bar: 50 μm). Sections were stained with an antibody against phospho-H2AX. Right panel: quantitation of number of phospho-H2AX–positive cells per 0.5 mm² in flank xenografts following tamoxifen or vehicle treatment (n = 4 fields; mean ± SD) (*P < 0.05; 2-tailed Student’s t test). (E) Representative bioluminescence imaging of intracranial xenografts of primary DMG neurospheres with doxycycline-inducible control versus VRK1 targeting guides taken 30 days after doxycycline induction. (F) Quantification of bioluminescence images from E (sgCtrl vs. sgVRK1, P = 0.08). (G) Kaplan-Meier survival curves showing overall survival for mice injected with sgCtrl or sgVRK1 DMG neurospheres into the cranium. Significance was determined by log-rank test (sgCtrl vs. sgVRK1, P = 0.10).

Figure 7. Model for mechanism of synthetic lethality between VRK1 and VRK2.

Schematic showing proposed mechanism of synthetic lethality between VRK1 and VRK2. In VRK2-unmethylated tumors (top), VRK2 compensates for VRK1 loss in the phosphorylation of BAF during mitosis. In VRK2^lo tumors (bottom), loss of VRK1 leads to retention of BAF during mitosis and the continued association of the nuclear envelope with chromatin. This leads to impaired chromosomal segregation and DNA damage, including nuclear bridging.