Therapeutic modulation of ROCK overcomes metabolic adaptation of cancer cells to OXPHOS inhibition and drives synergistic anti-tumor activity

Abstract

Genomic studies have identified frequent mutations in subunits of the SWI/SNF chromatin remodeling complex including SMARCA4 and ARID1A in non-small cell lung cancer. Previously, we and others have identified that SMARCA4-mutant lung cancers are highly dependent on oxidative phosphorylation (OXPHOS). Despite initial excitements, therapeutics targeting metabolic pathways such as OXPHOS have largely been disappointing due to rapid adaptation of cancer cells to inhibition of single metabolic enzymes or pathways, suggesting novel combination strategies to overcome adaptive responses are urgently needed. Here, we performed a functional genomics screen using CRISPR-Cas9 library targeting genes with available FDA approved therapeutics and identified ROCK1/2 as a top hit that sensitizes cancer cells to OXPHOS inhibition. We validate these results by orthogonal genetic and pharmacologic approaches by demonstrating that KD025 (Belumosudil), an FDA approved ROCK inhibitor, has highly synergistic anti-cancer activity in vitro and in vivo in combination with OXPHOS inhibition. Mechanistically, we showed that this combination induced a rapid, profound energetic stress and cell cycle arrest that was in part due to ROCK inhibition-mediated suppression of the adaptive increase in glycolysis normally seen by OXPHOS inhibition. Furthermore, we applied global phosphoproteomics and kinase-motif enrichment analysis to uncover a dynamic regulatory kinome upon combination of OXPHOS and ROCK inhibition. Importantly, we found converging phosphorylation-dependent regulatory cross-talk by AMPK and ROCK kinases on key RHO GTPase signaling/ROCK-dependent substrates such as PPP1R12A, NUMA1 and PKMYT1 that are known regulators of cell cycle progression. Taken together, our study identified ROCK kinases as critical mediators of metabolic adaptation of cancer cells to OXPHOS inhibition and provides a strong rationale for pursuing ROCK inhibitors as novel combination partners to OXPHOS inhibitors in cancer treatment.

Keywords: AMPK; IACS-10759; ROCK; SMARCA4; functional genomics; oxidative phosphorylation.

Figure 1:. CRISPR and clinical drug screening identifies ROCK1/2 as a synergistic combination agent with OXPHOS inhibition.

(A) Schematic representation of the workflow for CRISPR screens performed in A549, H1299, and H2023 SMARCA4-mutant lung cancer cells. Image was created using BioRender.com. (B) NormZ scores of candidate genes common at both early and late timepoints from FDAome library CRISPR screening results in H1299 cells cultured in presence or absence of 4 nM IACS-10759. The NormZ score was used to define a possible synthetic lethal interaction with IACS-10759. All genes targeted by the FDAome library were scored according to the fold change of levels of their respective sgRNAs. Genes whose loss of function led to IACS-10759 sensitivity appear on the bottom left quadrant, and genes whose loss of function led to IACS-10759 resistance appear on the top right quadrant. High-confidence candidate genes are shaded in orange and those selected for further analysis are indicated in red. (C) List of selected therapeutics and respective gene targets that are FDA approved or in clinical development tested in combination with IACS-10759 for synergistic effects on cell growth. (D) Overall synergy scores determined in A549, H1299, and H2023 exposed to therapeutics against selected CRISPR hits in combination with IACS-10759. An overall synergy score >10: the interaction between two drugs is likely to be synergistic. (E) Dose-response curves of H1299 cells exposed to increasing concentrations of KD025 in the presence or absence of IACS-10759. Cell growth (measured as % Confluency) was assayed 5 days after drug exposure. (F) Representative clonogenic growth assays of H1299 cells cultured in the presence of KD025 and IACS-10759 alone or in combination for 12 days. Surviving cells after the treatment were fixed and visualized by crystal violet staining. (G) Dose-response curves of H1299 cells exposed to increasing concentrations of KD025 in the presence or absence of IM156. Cell growth (measured as % Confluency) was assayed 5 days after drug exposure. (H) Representative clonogenic growth assays of H1299 cells cultured in the presence of KD025 and IM156 alone or in combination for 12 days. Surviving cells after the treatment were fixed and visualized by crystal violet staining. Representative images of three independent experiments are shown. (I)Overall synergy scores determined in H1299 cells exposed to indicated OXPHOS inhibitors in combination with KD025. An overall synergy score >10: the interaction between two drugs is likely to be synergistic. For panels (D, E, G, I) are presented as mean +/− SEM of three biological replicates (n=3).

Figure 2:. Combination of KD025 and IACS-10759 induces cell cycle growth arrest, cell death, and severe energy stress.

(A) FUCCI cell cycle analysis showing percentage of H1299 cells in M-G1, G1, G1/S or S/G2/M following treatment with DMSO, IACS-10759, KD025, or the combination over a 96 hr timecourse. Percentage of cells at each cell cycle phase were quantitatively assessed by cell by cell analysis software from Incucyte S3. (B-C) Quantitation of apoptosis using either Cytotox green or Annexin V showing the percentage of cells over a 96 hr time course. (B) or at increasing doses of KD025 after 96 hr. (C) as assessed by cell-by-cell incucyte analysis following treatment with DMSO, IACS-10759, KD025 or the combination in H1299 cells. (D) Seahorse mitochondrial stress test assay showing a representative trace measuring mitochondrial oxygen consumption rate (OCR) in H1299 cells cultured with DMSO, KD025, IACS-10759, or the combination for 6 hr from which (E) basal respiration, maximum respiratory capacity, ATP-linked respiration, and spare capacity were calculated. Data indicate mean +/− S.E.M from DMSO (n=4), IACS-10759 (n=7), KD025 (n=11), and combination (n=7) independent experiments. (F) Seahorse glycolysis stress test assay showing a representative trace measuring extracellular acidification rate (ECAR) in H1299 cells cultured with DMSO, KD025, IACS-10759, or the combination for 6 hours from which (G), glycolysis, glycolytic capacity, and glycolytic reserve were calculated. Data indicate mean ± SEM from DMSO (n=10), IACS-10759(n=11), KD025(n=10), combination (n=9) independent experiments. (H-J) Quantitation of ATP production by Seahorse XF real-time ATP rate assay following treatment with DMSO, IACS-10759, KD025 or the combination for 6 hours in H1299 cells from which (H), total ATP production, (I) mitochondrial ATP production, and (J) glycolytic ATP production rates were calculated. Data shown are mean ± SEM from DMSO (n=10), IACS-10759(n=11), KD025 (n=10), combination (n=9) independent experiments. (K) Bioenergetic profile map in H1299 cells by plotting basal OCR and ECAR of the indicated treatment groups. For panels (E-K) One-way ANOVA was used corrected for multiple comparisons. ****indicates P-values <0.0001; ***<0.001; **<0.01; *<0.05; ns, not significant.

Figure 3:. KD025 promotes metabolic reprogramming by suppressing adaptive increase in glycolysis due to OXPHOS inhibition.

(A) Heatmap of significantly different metabolite abundances involved in glycolysis, pentose phosphate pathway, and TCA cycle following treatment with DMSO, KD025, IACS-10759, or the combination for 24- and 48- hr in H1299 cells. Log₂ FC > 0 (red) represent an increase of metabolite abundance and Log₂ FC < 0 (green) represent a decrease of metabolite abundance. (B) Relative abundance of select metabolites from glycolysis, pentose phosphate pathway, and TCA cycle pathways following treatment with DMSO, KD025, IACS-10759, or the combination for 24- and 48 hours. Metabolite abundance expressed as relative peak intensity. (C) Glucose uptake in H1299 cells following 6 hours treatment as measured by uptake of 2-deoxyglucose and normalized to cell number. (D) Lactate secretion into extraceullar media from H1299 cells were treated for 6 hours, and lactate levels accumulated in the media were measured and normalized to the endpoint cell number. (E) Heatmap of significantly labeled metabolites after isotope incorporation. Data reflect the relative sum of abundance of all ¹³C isotopologues. (F) Fractional isotopic incorporation of ¹³C₆-glucose into glycolytic and TCA cycle metabolite intermediates as measured by GC/MS in H1299 cells following treatment with DMSO, KD025, IACS-10759 or the combination, and cultured in ¹³C₆-glucose containing medium for 24 hours. m, number of labeled carbons. Data indicate mean ± SEM of three independent experiments. All data indicate mean ± SEM of three independent experiments (n=3). For panels (C, D, F, H) One-way ANOVA was used corrected for multiple comparisons. ****indicates P-values <0.0001; ***<0.001; **<0.01; *<0.05; ns, not significant.

Figure 4:. Phosphoproteomics uncovers RHO GTPase signaling as a top regulated pathway in the KD025 and IACS-10759 combination

(A) Experimental workflow for proteomic and phosphoproteomic analyses. Four replicates were analyzed for each treatment group. Image was created using BioRender.com. (B) Volcano plots of global proteome showing differentially expressed proteins (FDR < 0.05, 1.5 log₂ FC) following treatment with KD025, IACS-10759, or the combination for 6 hours in H1299 cells. (C) Volcano plots of global phosphoproteome showing differentially expressed phosphosites (FDR < 0.05, 1.5 log₂ FC) following treatment with KD025, IACS-10759, or the combination for 6 hours in H1299 cells. PSP annotated and putative AMPK and ROCK regulated phosphosites are indicated. (D) Top Reactome, KEGG, GOBP gene sets significantly enriched (P > 0.05) in down-regulated phosphoproteins in the combination. RHO GTPase signaling-related pathways is highlighted in red (E) Top Reactome, KEGG, GOBP gene sets significantly enriched (P< 0.05) in up -regulated phosphoproteins in the combination. RHO GTPase signaling-related pathways are highlighted in red. (F) Venn diagram comparing overlap of proteins with down-regulated phosphosites, or in proteins with up-regulated phosphosites in the combination. The total number of phosphosites in each group are indicated. (G-I) Top enriched terms of Reactome gene sets significantly enriched (P < 0.05) in phosphoproteins with only down-regulated phosphosites (G), in phosphoproteins with up- and down-regulated phosphosites (H), or in phosphoproteins with only up-regulated phosphosites (I). Terms are sized and colored by significance of enrichment and number of associated genes. Less specific RHO GTPase signaling pathways are highlighted in red. More specific pathways related to RHO GTPase signaling are highlighted in blue.

Figure 5.. AMPK engages RHO GTPase signaling to regulate a dynamic phosphoproteome network at key substrates

(A) Schematic diagram of substrate scoring process and network analysis. Image was created using BioRender.com. (B) Kinase-motif enrichment [log2(frequency factor)] comparing IACS-10759 versus combination (left) and KD025 versus combination (right) phosphoproteomes derived from motifs of significant phosphosites shown in Fig. 4C. Fill color represents the significance of enrichment in either IACS-10759 or KD025 phosphoproteomes, and size represents the significance in the combination phosphoproteome. Select enriched kinases and kinase motif class are indicated. The enrichments in these plots were determined using one-sided exact Fisher’s tests and corrected for multiple hypotheses using the Benjamini–Hochberg method. (C) Network of co-regulated phosphoprotein members in top enriched Reactome gene sets with up- and down-regulated phosphosites. Phosphoproteins are outlined in red if a differentially regulated phosphosite was identified with a AMPK motif. Phosphoprotein size is proportional to the total number of up- and down-regulated phosphosites, and pathways are sized by the number of phosphoprotein members. (D) Top kinase interaction scores for pathways in Reactome gene sets of co-regulated phosphoproteins. Kinase-pathway interactions were constructed using kinase-motif specificities , kinase-substrate specificities from PhosphoSitePlus , and kinase-substrate interactions from BioGRID for all phosphoproteins annotated for each pathway.(E) Annotated Chronos dependency rank plot ranked by DepMap dependency scores for co-regulated phosphoproteins. Chronos dependency scores for each phosphoprotein is represented as a point on the rank plot. Annotations indicate membership in Reactome gene sets. Phosphoproteins with a Chronos scores ≤–0.5 (dotted line) are considered essential, genetic dependent proteins. (F)Kinase-substrate networks and genetic dependency of co-regulated phosphoproteins in top Reactome gene sets after combined IACS-10759 and KD025 treatment. Kinase-substrate interaction scores were computed from kinase-pathway analysis in Figure 5E. Each phosphoprotein was annotated with genetic dependency scores from DepMap. The top 20 kinases are shown with nodes and text sized by their strength in the network. Edges are colored by direction of association with up- or down-regulated phosphosites. Edge width represents the magnitude of interaction score. Co-regulated phosphoprotein nodes and text are colored according to their CRONOS dependency score. (G) Phosphosites on essential, genetic dependent co-regulated proteins that interact with AMPK. Additional top kinases and their associated phosphosites computed by kinase-motif interaction analysis are also indicated. *indicates kinase-phosphosite interaction also annotated by PSP or by manual curation.

Figure 6:. Combination of ROCK and OXPHOS inhibition displays synergistic anti-tumor activity in vivo.

(A) In vivo xenograft in H1299 cells showing anti- tumor efficacy of IACS-10759 and KD025 alone or in combination after daily administration by oral gavage over 21 days. (B) In vivo xenograft in A549 cells showing anti- tumor efficacy of IACS-10759 and KD025 alone or in combination after daily administration by oral gavage over 47 days. (C) Comparison of final tumor volumes of H1299 (left) and A549 (right) tumor xenograft from (A, B). (D-E) Body weight changes of each treatment group over the during of the experiment in H1299 (D) and A549(E) tumor xenografts. For panel (C) Mann Whitney U-test was used. ****indicates P-values <0.0001; ***<0.001; **<0.01; *<0.05; ns, not significant.