A genome-wide in vivo CRISPR activation screen identifies BACE1 as a therapeutic vulnerability of lung cancer brain metastasis

Abstract

Brain metastasis occurs in up to 40% of patients with non-small cell lung cancer (NSCLC). Considerable genomic heterogeneity exists between the primary lung tumor and respective brain metastasis; however, the identity of the genes capable of driving brain metastasis is incompletely understood. Here, we carried out an in vivo genome-wide CRISPR activation screen to identify molecular drivers of brain metastasis from an orthotopic xenograft model derived from a patient with NSCLC. We found that activating expression of the Alzheimer's disease-associated beta-secretase 1 (BACE1) led to a substantial increase in brain metastases. Furthermore, genetic and pharmacological inhibition of BACE1 blocked NSCLC brain metastasis. Mechanistically, we identified that BACE1 acts through epidermal growth factor receptor to drive this metastatic phenotype. Together, our data highlight the power of in vivo CRISPR activation screening to unveil molecular drivers and potential therapeutic targets of NSCLC brain metastasis.

Fig. 1. In vivo CRISPR activation screen identifies BACE1 drives LUAD brain metastasis.

(A) In vivo CRISPR activation screen schematic. NGS – next generation sequencing. (B) Schematic depicting our screen hit prioritization strategy. (C) Distribution and rug plots displaying log₂ fold change of normalized sgRNA read counts (blue ticks – lungs, red ticks – brains) from individual lungs and brains relative to the mean of the control sgRNAs from the respective tissue. The distribution of the control sgRNAs is displayed as gray lines. Dotted line is log₂ fold change of 0. (D) Violin plot depicting the relative enrichment of the sgRNAs of the indicated genes in the brain or lung as log₂ FC compared to the control sgRNAs of the indicated tissue (n=12, one experiment). (E) Experimental scheme for orthotopic (intralung) implantation of CRISPR-activated BACE1 CRUK0748-XCL-GLD cells. (F) Ex vivo bioluminescent images of brains from mice in the indicated groups at endpoint (n=10 per group, one experiment). (G) Quantification of the total flux of the ex vivo brain bioluminescent images in (F). Data in (C) were analyzed by one-sided Wilcoxon rank sum test with Benjamini-Holchberg correction (FDR). Data in (D) were analyzed by two-way ANOVA. Data in (G) were analyzed by Mann-Whitney U test. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001

Fig. 2. BACE1 is expressed in LUAD brain metastasis and is associated with worse prognosis.

(A) (Left) Lung cancer tumor microarray (n=145) including lung adenocarcinomas (n=69) stained for BACE1. Green – BACE1, Blue – DAPI (nuclei). (Right) Pie charts indicating the proportion of tumors staining high or low for BACE1 in all non-small cell lung cancers (left) or in lung adenocarcinomas (right). Scale bar = 250 μm. (B) Immunohistochemical staining of brain metastases from patients with LUAD for BACE1. Colored as in (A). Scale bar = 250 μm. (C) Immunohistochemical staining of BACE1 in primary LUAD and matched brain metastases from Hamilton Health Sciences. Scale bar = 50 μm. (D) Immunohistochemical staining of BACE1 in the primary NSCLC and matched brain metastases TMA from London Health Sciences Centre (LHSC). Scale bar = 300 µm. (E) Violin plot depicting quantitation of BACE1 staining in (D) (n=21). (F) Violin plot depicting mRNA counts from GeoMx analyses of the samples in (D). (G) Kaplan-Meier curve depicting survival proportions for patients from the LHSC cohort according to BACE1 expression. (H) Kaplan-Meier curve depicting survival proportions for patients from the TCGA with LUAD brain metastases stratified according to median BACE1 expression. Hazard ratio determined using multivariate analysis with Cox Proportional Hazards mode (G,H). Data in (E) was analyzed by t test. Data in (H) was analyzed by Log-rank test. *P<0.05

Fig. 3. BACE1 increases the migratory and invasive capacity of primary LUAD cells.

(A) Western blot analysis of BACE1 expression in a panel of non-small cell lung cancer and lung-to-brain metastasis (LBM) lines. (B) Quantification of the migrated cell area of CRUK0748-XCL-GLD cells following BACE1 activation (n=6, N=3). (C) Quantification of the migrated cell area of BACE1^KO CRUK0748-XCL cells (n=4, N=3). (D) Quantification of migrated cell area of MK-8931 (10 μM) treated CRUK0748-XCL-GLD cells (n=6, N=3). (E) Representative micrographs depicting spheroid invasion of BACE1^KO H1299^GFP after 7 days in Matrigel^™. White boundaries indicate extent of invasion for the indicated cell lines. Scale bar = 800 μm. (F) Quantification of the invaded area of the indicated cell lines in (E) quantified using ImageJ (n=7, N=2). (G) Quantification of H1299^GFP invasion in a spheroid invasion model following overexpression of BACE1 (n=4, N=2). (H) Quantification of H1299^GFP invasion in a spheroid invasion model following treatment with MK-8931 (50 μM) for 7 days (n=6, N=2). (I) Representative micrographs depicting spheroid invasion of CRISPR-activated BACE1 CRUK0733-XCL-GLD cells after 80 hours in Matrigel^™. Invasion images illustrate the quantification mask from the Incucyte® spheroid software module utilized to quantify the extent of invasion. Yellow marks the growth of the sphere; orange and white mark the extent of invasion. Scale bar = 1 mm. (J) Quantification of invasion in (I) (n=4, N=2). Bars represent the mean number of migrated cells or invaded area + SEM. Data in (B), (C), (F), (G), (H) and (J) was analyzed by t test. Data in (D) was analyzed by two-way ANOVA with Sidak’s multiple comparisons test. **P<0.01, ***P<0.001****P<0.0001.

Fig. 4. BACE1 is required for the proliferation and self-renewal capacity of LUAD brain metastases.

(A) Western blot analysis of BACE1 expression in control (AAVS1^KO) or BACE1^KO MH1002 cells. (B) Proliferation of BACE1^KO MH1002 (left) and MH1012 (right) cells (n=3, N=3). (C) Micrographs depicting the sphere formation capacity of BACE1^KO MH1002 (Top) and MH1012 (Bottom) cells. Scale bar = 200 μm. (D) Quantification of sphere size for the indicated cell lines in (C) (n=3, N=3). (E) Cell proliferation of the indicated cell lines in response to MK-8931 at the indicated doses after 72 hours (n=3, N=3). (F) Micrographs depicting the sphere formation capacity of the indicated cell lines in response to 50 μM MK-8931 treatment for 96 hours. Spheres were allowed to form for seven days (n=3, N=3). Scale bar = 200 μm. (G) Quantification of sphere size from the images in (F). Bars indicate mean + SEM. Data in (B) were analyzed by two-way ANOVA with Tukey’s multiple comparisons test. Data in (D) and (G) were analyzed by t test. ***p<0.001.

Fig. 5. Perturbing BACE1 expression or activity blocks the formation of LUAD brain metastases.

(A) Experimental scheme for orthotopic injection (intralung) of CRISPR-activated BACE1 CRUK0748-XCL-GLD cells treated daily by oral gavage with 30 mg/kg MK-8931 for 21 days. (B) Representative ex vivo bioluminescent images of brains from the indicated groups. (C) Quantification of the total brain flux for all mice of the indicated groups. Dashed line indicates baseline luminescence (n=9-11 per group, one experiment). BACE1-Act Vehicle vs WT Vehicle P= 0.23. (D) Experimental scheme for intracranial injection of BACE1^KO MH1002 cells. (E) Longitudinal bioluminescent images of representative mice from the indicated groups (n=5 per group, one experiment). (F) Kaplan-Meier curve depicting the survival times of mice across the indicated groups. (G) Experimental scheme for intracranial injection of MH1002 cells treated daily by oral gavage with 30 mg/kg MK-8931 for 21 days. (H) Longitudinal bioluminescent images of representative mice from the indicated groups. (I) Quantification of the total brain bioluminescent flux for the indicated groups over time (n=8 per group, one experiment). (J) Kaplan-Meier curve depicting the survival times of the mice across the indicated groups. Bars indicate mean + SEM. Data in (C) were analyzed by Mann-Whitney U test. Data in (F) and (J) were analyzed by log-rank test. Data in (I) were analyzed by t test. *P<0.05, ***p<0.001, ****P<0.0001.

Fig. 6. BACE1 activity is required for EGFR activation.

(A) Human Proteome Profiler^™ phospho-kinase antibody array in lysates from BACE1^KO cells. (B) Western blot analysis of the indicated proteins in lysates from BACE1^KO cells (N=2). (C) Western blot analysis of the indicated proteins in lysates from cells treated with 30 μM MK-8931 for 72 hours (N=2). (D) Representative micrographs depicting EGFR^Y1068 staining in MH1002 tumors treated with vehicle or 30 mg/kg MK-8931 (n=5 per group, one experiment). Green – EGFR^Y1068; Blue – DAPI. Scale bar = 30 μm. (E) Quantification of proportions of EGFR^Y1068 positivity in MH1002 tumors in response to vehicle or MK-8931 treatment. (F) Western blot analysis of H1299^GFP BACE1^KO cells restored with EGFR^L858R for the indicated proteins (N=2). (G) Representative micrographs depicting spheroid invasion of BACE1^KO H1299^GFP with or without EGFR^L858R expression after 7 days in Matrigel^™. White boundaries indicate extent of invasion for the indicated cell lines. Scale bar = 800 μm. (H) Quantification of the invaded area of the indicated cell lines in (G) (n=5-6, N=2). (I) Schematic for intracardiac injection of H1299^GFP-Luc control, BACE1^KO or BACE1^{KO + EGFRL858R} cells. (J) Representative ex vivo BLI images of brains from the indicated groups (n=9-10 per group). (K) Quantification of total brain flux from individual mice from the indicated groups. Bars represent mean + SEM. Data in (E) and (H) were analyzed by t test. Data in (K) were analyzed by Mann-Whitney U test. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Fig. 7. EGFR is a novel substrate of BACE1.

(A) Western blot analysis of the indicated lysates for BACE1, EGFR (1150-C) and GAPDH following 72 hour expression of the indicated plasmids in HEK293T cells (N=2). (B) Micrographs depicting proximity-dependent amplification of signal following staining in HEK293T cells expressing EGFR and BACE1 for 72 hours for EGFR and BACE1 (N=2). Scale bar = 10 μm. (C) Quantification of PLA foci per cell from images in (B). (D) Micrographs depicting BACE1 and EGFR PLA signal in MH1002 cells (N=2). Scale bar = 10 μm. (E) Quantification of PLA foci per cell from images in (D). (F) Silver stain analysis following co-incubation of rBACE1 and rEGFR overnight at 37 ºC. 0.25 µg of rEGFR was incubated with 0.5, 2 or 4 µg of rBACE1 in 0.1 M sodium acetate buffer pH 4.0 or 7.0 (N=2). (G) Schematic of the ATOMS workflow. (H) Domain schematic of the ectodomain of EGFR (domains I-IV) highlighting ATOMS identified cleavage sites (blue arrows) following 24 hour co-incubation of rBACE1 and rEGFR (ectodomain) at 37 °C. (I) BACE1 FRET activity assay measuring cleavage of EGFR peptide encompassing the ¹¹⁹L↓A¹²⁰ cleavage site following co-incubation in the presence or absence of MK-8931 (1 μM) for 24 hours at 37 °C (N=2). (J) Western blot analysis of the indicated proteins in conditioned media (CM) or whole cell lysates (WCL) from cells transiently expressing the indicated genes for 72 hours (N=2). EGFR (129-160) recognizes an epitope near the N-terminus of EGFR between amino acids 129 and 160. Bars in (C) and (E) indicate mean + SEM. Bars in (I) indicate mean + SD. Data in (C) and (I) were analyzed by one-way ANOVA followed by Tukey’s multiple comparisons test. Data in (E) were analyzed by Mann-Whitney U test. *P<0.05, **P<0.01, ****P<0.0001.