Glucocorticoid mediated inhibition of LKB1 mutant non-small cell lung cancers

Abstract

The glucocorticoid receptor (GR) is an important anti-cancer target in lymphoid cancers but has been understudied in solid tumors like lung cancer, although glucocorticoids are often given with chemotherapy regimens to mitigate side effects. Here, we identify a dexamethasone-GR mediated anti-cancer response in a subset of aggressive non-small cell lung cancers (NSCLCs) that harbor Serine/Threonine Kinase 11 (STK11/LKB1) mutations. High tumor expression of carbamoyl phosphate synthase 1 (CPS1) was strongly linked to the presence of LKB1 mutations, was the best predictor of NSCLC dexamethasone (DEX) sensitivity (p < 10^-16) but was not mechanistically involved in DEX sensitivity. Subcutaneous, orthotopic and metastatic NSCLC xenografts, biomarker-selected, STK11/LKB1 mutant patient derived xenografts, and genetically engineered mouse models with KRAS/LKB1 mutant lung adenocarcinomas all showed marked in vivo anti-tumor responses with the glucocorticoid dexamethasone as a single agent or in combination with cisplatin. Mechanistically, GR activation triggers G1/S cell cycle arrest in LKB1 mutant NSCLCs by inducing the expression of the cyclin-dependent kinase inhibitor, CDKN1C/p57(Kip2). All findings were confirmed with functional genomic experiments including CRISPR knockouts and exogenous expression. Importantly, DEX-GR mediated cell cycle arrest did not interfere with NSCLC radiotherapy, or platinum response in vitro or with platinum response in vivo. While DEX induced LKB1 mutant NSCLCs in vitro exhibit markers of cellular senescence and demonstrate impaired migration, in vivo DEX treatment of a patient derived xenograft (PDX) STK11/LKB1 mutant model resulted in expression of apoptosis markers. These findings identify a previously unknown GR mediated therapeutic vulnerability in STK11/LKB1 mutant NSCLCs caused by induction of p57(Kip2) expression with both STK11 mutation and high expression of CPS1 as precision medicine biomarkers of this vulnerability.

Keywords: LKB1; glucocorticoid; lung cancer; nuclear receptor; targeted therapy.

Figure 1

Dexamethasone (DEX) causes growth inhibition and significant metabolic alterations in some lung cancer cell lines. (A) Cell counting assays show responder cell lines are significantly growth inhibited by dexamethasone (100nM) over a 5-day assay. Data shown from three independent assays with three replicates for each assay with standard deviation on each bar in graph. Triple asterisk (***) represent p-values < 1x10^-5 calculated by Student’s t-test. (B) After 48 hours of 100nM DEX exposure, DEX sensitive cell line (NCI-H1993) undergoes cell-cycle arrest at the G1/S phase transition as demonstrated by FACS analysis. DEX resistant cell line (NCI-H2347) shows no abnormality in cell cycle distribution in response to drug. FACS assays were repeated three times. (C) DEX sensitive cell lines (NCI-H1993) show significantly altered morphology 48 hr. after 100nM DEX exposure with a significant proportion of senescent cells (H1993-DEX (X-gal)). (D) DEX sensitive cell lines show significant ATP accumulation in response to DEX exposure while DEX resistant NCI-H2347 does not. Assays done in triplicate with three technical replicates for each assay. Triple asterisk (***) represent p-values < 1x10^-5 calculated by Student’s t-test. (E) DEX sensitive cell lines (NCI-H1993, A549) take up less glucose and excrete less lactate in response to DEX exposure. Resistant cell line NCI-H2347 exhibits no significant reduction in glucose uptake of lactate excretion. Assay done in triplicate. Double asterisk represents p-value < 1x10^-4 and single asterisk represents p-value < 1x10^-3 calculated by Student’s t-test. * P < 0.05, ** P < 0.01, ns, not significant.

Figure 2

High CPS1 expression is the biomarker for DEX sensitivity and LKB1 mutation in lung cancer cell lines. (A) RNA expression heatmap of cell lines from NR ligand screen. CPS1 (red) was the most highly expressed gene in the DEX sensitive cell lines when compared to DEX resistant cell lines (p < 9.0 x 10^-16). (B) Exemplar western blot demonstrating the inverse correlation between the presence of LKB1 WT protein (50KDa) and the expression of CPS1 in lung cancer cell line panel. (C) Box and whisker plot of RNAseq analysis demonstrating strong association (p< 1x 10^-31) of high CPS1 expression with LKB1 mutation in large panel of NSCLC cell lines (N=164) (D) Analysis of TCGA transcriptome and mutation data demonstrate LKB1 loss is strongly associated with high CPS1 expression. (E) Box and whisker plot of RNAseq data from 164 NSCLC cell lines showing strong correlation between LKB1 loss and increased GR expression (p < 1x10^-8) and no correlation between GR expression and either KRAS or EGFR mutation. (F) Exemplar western blot confirms GR protein is more highly expressed in LKB1 mutants than wild type. Blue stars mark LKB1 mutant cell lines and red font indicates DEX sensitive cell lines.

Figure 3

In vivo xenograft and orthotopic model studies show DEX mediated inhibition of tumor growth. (A) NSCLC cell line xenograft studies using DEX sensitive cell line (NCI-H1993 in red) and resistant cell line (NCI-H2009). Once tumors for each cell line reach an average size of 150 mm3, they are randomized and split into four treatment groups (1); vehicle control (2), DEX only (3), cisplatin only and (4) DEX + cisplatin. Treatments are stopped after four weeks and tumors allowed to progress until they reach ~2000 mm3 at which point mice are sacrificed. DEX and platinum as single agents show similar anti-tumor responses and survival benefit (red, blue line respectively). The combination DEX + cisplatin treated groups performed best in sensitive NCI-H1993 (purple line). No significant response was observed in DEX resistant NCI-H2009. (B) Mouse orthotopic model created by tail vein injection of luciferase labeled A549 cells into 20 mice. After three weeks, 7 mice showed similarly sized tumors as judged by BLI. Mice were randomized and split into two groups (1); vehicle control and (2) DEX treated. The first BLI images shown were taken before first day of DEX or vehicle treatment. The second set of images was taken 35 days after the start of treatment. (C) Body mass of DEX treated (red line) and vehicle control (black line) during progression of experiment. (D) Example of lungs removed after sacrifice for DEX treated (right) and control (left). Black arrows point to visibly significant tumor nodules in vehicle group.

Figure 4

Biomarker driven choice of patient derived xenograft (PDX) and KL tumor GEMMs show response to DEX. (A) RNAseq data confirming high expression of CPS1. DNA seq data confirms and characterizes LKB1 mutation. (B) High CPS1 expressing PDX tumors were grown to approximately 250mm3 before DEX (red), platinum alone (blue), combination (purple) or vehicle (black) treatment started. Tumor growth curve for PDX tumors treated for five weeks with DEX are shown. (C) Average body weight measurement for members of each cohort is plotted over the duration of the experiment. (D) Survival curves for each treatment group are plotted. (E) MRI estimated tumor volumes for mouse KL GEMM at baseline before DEX treatment. (F) MRI estimated tumor volumes plotted as percentage change in tumor volume compared to baseline at two weeks and four weeks. (G) Representative MRI of one control treated mouse and one DEX treated mouse over the course of the four week treatment. * means p-value < 0.05; ** means p-value < 0.01.

Figure 5

The top 50 up- and down-regulated genes in response to DEX in DEX-sensitive NCI-H1993 in vitro and in vivo (xenograft). CDKN1C is highlighted.

Figure 6

GR mediated up-regulation of p57(Kip2) expression is responsible for the cell cycle arrest in DEX-sensitive cells. (A) Western blot showing time course of p57(Kip2) protein appearance after DEX treatment in sensitive (A549, EKVX) and resistant (NCI-H2009, NCI-H2347) cells. (B) Western blot of LKB1 mutant PDX tumor exhibiting p57(Kip2) expression and the appearance of cleaved PARP in the DEX treatment group. (C) Western blot of NCI-H1993 CRISPR clones and parental (H1993-P) after DEX treatment (100nM, 24 hrs.). NCI-H1993 CRISPr clones 2-6 (red) and 3-7 (blue) show no p57(Kip2) after DEX treatment and were used in subsequent cell proliferation assays. (D) Cell counts after DEX treatment with parental NCI-H1993 and CRISPR knockout clones NCI-H1993 cln2-6 and cln3-7. (E) Colony formation assay confirming loss of cell cycle arrest after CRISPR knockout of CDKN1C gene in DEX-sensitive NCI-H1993. Inset shows exemplar brightfield images of colonies from the colony formation assay plates shown.