CDK4/6 upregulation promotes osimertinib resistance and replication stress in EGFRmt LUAD CDXs.

(A) Biochemical analysis of CDK4/6 overexpressing and empty vector (E.V.) H1975 EGFR-mutant LUAD CDXs treated with vehicle (0 mg/kg) or osimertinib (2.5-5 mg/kg). Protein expression detected by immunoblotting as indicated. (B) Quantification of p-EGFR immunoblotting shown in (A). (C,D) Growth curves of osimertinib (5 mg/kg) treated CDK4 (C) or CDK6 (D), overexpressing H1975 CDXs compared to empty vector controls (n = 4-6 xenografts per group; error bars representing SEM). (E) Top significantly up-(red) regulated GSEA pathways (n = 3 xenografts per group; p-value calculated as described in methods; NES: Normalized Enrichment Score). (F-H) Representative IHC stained images (F) and quantification (G, H) of Ki-67 from E.V. and CDK4/6 overexpressing H1975 CDXs treated with osimertinib (5 mg/Kg) (n = 3-6 xenografts per group; p-value assessed by one-way ANOVA and Tukey’s multiple comparisons test; error bars representing SEM). (I, J) Immunoblotting of replication stress biomarkers p-Rb, TPX2, p-RPA, and p-ATM in CDK4/6 overexpressing H1975 CDXs (I) and quantification (J) versus empty vector (E.V.) controls (error bars representing SEM; following a significant ANOVA result (p < 0.05), post-hoc analysis was conducted using Tukey’s Honest Significant Difference (HSD) test for pairwise comparisons. Statistical significance is indicated by brackets and p-values, with a threshold of p < 0.05). (E.V.: empty vector; CDK4 and CDK6: CDK4 or CDK6 overexpressing CDXs; Veh.: vehicle; Osi.: osimertinib).

CDK4/6 upregulation promotes DNA damage accumulation in EGFRmt LUAD CDXs treated with osimertinib.

(A, B) Gene expression profiling of empty vector (E.V.), CDK4, and CDK6 overexpression H1975 CDXs demonstrating statistically significant enrichments in the ATM pathway genes RAD51, CHEK1, and CHEK2, across CDK4/6 overexpressing CDXs compared to E.V. controls (n = 3 xenografts per group; p-value assessed by one-way ANOVA and Tukey’s multiple comparisons test). (C, D) Double immunofluorescence stain for γ-H2AX (red) and EdU (green) (C) and quantification of γ-H2AX foci number (D) in EdU positive cells for vehicle-treated H1975 CDXs (white arrows point to double-positive cells; n= 50-100 EdU positive cells per sample, one xenograft per group; p-value calculated with one-way ANOVA and Tukey’s multiple comparisons test). (E, F) Biochemical analysis of γ-H2AX expression in CDK4/6 overexpressing H1975 CDXs treated with osimertinib for one week compared to E.V. overexpressing control CDXs (E) and blot’s quantification (F) (n = 3 xenografts per group; fold change compared to E.V. controls; error bars representing SEM). (E.V.: empty vector; CDK4 and CDK6: CDK4 or CDK6 overexpressing CDXs; Veh.: vehicle; Osi.: osimertinib).

Overlap of CDK4/6 upregulated genes with abnormal CN alterations in CDK4/6 overexpressing EGFRmt LUAD CDXs.

(A) Fraction of Genome Alterations (FGA) analysis from whole exome sequencing data from osimertinib (5 mg/kg) or vehicle-treated H1975 CDK4/6 overexpressing CDXs compared to empty vector (E.V.) controls (n = 2 xenografts per group; p-value assessed by one-way ANOVA and Tukey’s multiple comparisons test; error bars representing SEM). (B, C) Venn diagram illustrating the overlap between genes upregulated in CDK4 (B) or CDK6 (C)-overexpressing CDXs (RNA-seq) and genes with copy number alterations (CNAs) in the vehicle-treated arm (normalized to E.V. CDXs). RNA-seq upregulated genes are shown in sky blue, and CNA-associated genes are shown in green. The intersection indicates genes both transcriptionally upregulated and genomically altered. (D, E) Volcano plot illustrating differential gene expression between CDK4 (D) or CDK6 (E)-overexpressing CDXs vs. E.V. controls. Genes were classified as upregulated (red), downregulated (blue), or not significant (grey) based on log2 fold change (FC) > 0.3 and FDR < 0.3. Selected genes of interest (AGR2, CBX3, ANLN, etc.) are annotated. Vertical and horizontal dashed lines represent fold-change and FDR thresholds, respectively. (F) Venn diagram showing the overlap of genes between osimertinib-treated samples (Osi.) and vehicle-treated samples (Veh.), based on upregulated RNA expression (RNA up) and/or copy number alterations (CN = 3; normalized to E.V. CDXs). (CNAs detected in CDK4 and CDK6 overexpressing CDXs normalized to empty vector control E.V. CDXs in the same treatment group; CDK4: CDK4 overexpressing CDXs; CDK6: CDK6 overexpressing CDXs; for the CNAs analysis, n = 2 xenografts for each group; for the RNAseq experiments, n = 3 xenografts for each group).

Combination therapy with EGFR and CDK4/6 inhibitors suppresses tumor growth and decreases replication stress and genomic instability in CDK4/6 overexpressing EGFRmt LUAD CDXs.

(A, B) Representative Ki-67 IHC stained tumor sections (A) and quantification (B) from E.V. and CDK4/6 overexpressing H1975 CDXs treated with osimertinib (5 mg/kg) or combinatorial treatment with palbociclib (100 mg/kg) (n = 3-4 xenografts per group; p-value assessed by one-way ANOVA and Tukey’s multiple comparisons test; error bars representing SEM). (C-H) Representative IHC stained tumor sections (C) and quantification of p-Rb (D), p-RPA (E), γ-H2AX (F), and CDK4 (G, H) from CDK4 overexpressing H1975 CDXs treated with osimertinib (5 mg/kg) and combinatorial treatments with palbociclib (100 mg/kg) (n = 3-4 xenografts per group for Ki-67, p-Rb, p-RPA, γ-H2AX, p-RPA IHC stains; n = 1 for p-RPA in the combinatorial treatment group due to high necrosis in the tissues; p-value assessed by one-way ANOVA and Tukey’s multiple comparisons tests and Student’s T-test; error bars representing SEM). (E.V.: empty vector; CDK4: CDK4 overexpressing CDXs; CDK6: CDK6 overexpressing CDXs; Veh.: vehicle; Palbo.: palbociclib; Osi.: osimertinib; p-value assessed by one-way ANOVA and Tukey’s multiple comparisons test; error bars representing SEM).

Combination therapy with EGFR and CDK4/6 inhibitors decreases replication stress and genomic instability in a CDK6amp, EGFRmt primary LUAD PDXs.

(A) Immunoblotting with quantification for EGFR pathway, CDK6-Rb pathway, and replication stress biomarkers CCNE1, TPX2, and p-AURKA from TH116 (EGFR L8585, CDK6amp) primary patient-derived xenografts (PDXs), treated with vehicle, palbociclib (150 mg/kg) or osimertinib (5 mg/kg); normalized quantification of the level of a panel of biomarkers provided in the lower panel. (B-E) Representative hematoxylin/eosin, γ-H2AX, and CDK6 IHC stained tumor sections (B) and quantification (C-E) using tissues from TH116 PDXs (n = 4 xenografts per group; Student T-test). (F) FGA analysis in TH116 PDXs, treated with vehicle, osimertinib (5 mg/kg), palbociclib (100 mg/kg), or a combination of osimertinib and palbociclib (Osi. + Palbo.) (n = 2 xenografts per group; error bars representing SEM). (Veh.: vehicle; Palbo.: palbociclib; Osi.: osimertinib; p-value assessed by one-way ANOVA and Tukey’s multiple comparisons tests; error bars representing SEM).

Concurrent EGFRmt and CDK4/6 CNAs are associated with an enhanced percentage of genome alterations in LUAD patient tumors.

Fraction of Genome Alterations (FGA) analysis in EGFRmt LUAD using Foundation Medicine (FM) (A) or MSK-Impact GENIE (B) targeted exome sequencing datasets. (A) FGA was compared in tumors harboring concurrent cell cycle CNAs (CC pos.) to tumors negative for such alterations (CC neg.) in the FM (A) and MSK-Impact GENIE datasets (B). (C, D) FGA analysis in EGFRmt tumors stratified by CDK4/6 CNA status in the FM dataset (C) and MSK-Impact GENIE dataset (D). (E, F) FGA analysis in EGFRmt tumors stratified by CDK4 CNA status in the FM (E) or MSK-Impact GENIE (F) datasets. (G, H) FGA analysis in EGFRmt tumors stratified by CDK6 CNA status in the FM (G) or MSK-Impact GENIE (H) datasets. (I) Positive correlation plot showing degrees of CDK4/6 amplification (y-axis) and corresponding FGA (p-value calculated with F-statistics) (FM Dataset; each dot representing one CDK4/6 pos. tumor). (J) FGA analysis with FM dataset comparing RB1 alteration positive versus RB1 alteration negative EGFRmt LUAD cases. (K, L) Mutual exclusivity analysis (K) and percentage of cases (L) carrying concurrent CDK4/6 amp and RB1 gene alterations (SNVs and CNAs) in EGFRmt tumors from Foundation Medicine (FM) or MSK-Impact GENIE targeted exome sequencing (p-value calculated with Fisher’s test). (FGA boxplots included in violin plots represent the interquartile range, lower quartile 25%, median, and upper quartile 75%; p-value calculated with Wilcoxon test).

Transcriptional changes associated with CDK4amp in EGFRmt LUAD cancer cells from patient biopsies.

(A) UMAP visualization of single-cell RNA-seq data from CDK4amp, EGFRmt LUAD clinical clinical samples (n = 3), with cancer cells annotated in red based on inferCNV inference. (B) UMAP showing high CDK4 expression in CDK4amp tumor cells, demonstrating localization of CDK4 elevated expression to cancer cells. (C) UMAP visualization of single-cell RNA-seq data from CDK4/6wt EGFRmt LUAD clinical samples (n = 11, left scatterplot) and UMAP showing a low degree of CDK4 expression across CDK4/6wt tumor cells (right scatterplot). (D) Violin plots comparing CDK4 expression in cancer cells from CDK4amp (n =3) versus CDK4/6wt (n = 11) tumors, showing significantly higher CDK4 expression in CDK4amp cases. (E) Violin plots pointing to AGR2, STEAP1, GGCT, and ASNS expression in cancer cells as significantly upregulated in CDK4amp versus CDK4/6wt tumors. (F) Difference in Spearman correlation of gene expression with CDK4 gene expression between single cancer cells from CDK4amp (n = 483) and CDK4/6wt (n = 3188) EGFR-mutant LUAD patients (Δρ). P-values were calculated by comparing Fisher z-transformed correlations using a z-test, then adjusted for multiple testing using the False Discovery Rate (FDR). Bars in red indicate genes with significant differences after FDR correction (FDR < 0.05); gray bars are not significant. (G) Violin plot comparing epithelial-mesenchymal transition (EMT) scores in single cancer cells from CDK4amp (n = 483) and CDK4/6wt (n = 3188) EGFR-mutant LUAD patients. EMT scores were calculated as the mean expression of canonical EMT markers (VIM, ZEB1, SNAI1, TWIST1, CDH2, and FN1). Medians are indicated above each violin. Statistical comparison was performed using a two-sided Mann–Whitney U test.

Model of the effects of CDK4 or CDK6 amplification in EGFRmt LUAD.

(Left) Schematic model illustrating the impact of CDK4 or CDK6 amplification in EGFRmt LUAD. In CDK4amp or CDK6amp tumors, Rb phosphorylation persists despite EGFR TKI treatment, leading to continued proliferation, accumulation of replication stress (RS) and genomic instability (GIN), and activation of TKI resistance mechanisms (e.g., TPX2/AURKA, ATM). This promotes an increased fraction of genome altered (FGA) and genomic and transcriptional upregulation of tumor promoting genes (e.g., AGR2, ASNS, STEAP1), and epithelial to mesenchymal transition (EMT) contributing to EGFR TKI resistance. (Right) Co-treatment with a CDK4/6 inhibitor (e.g. palbociclib or abemaciclib restores G1/S checkpoint control, reduces proliferation, and induces apoptosis, leading to tumor regression.