Oncogenic KRAS engages an RSK1/NF1 pathway to inhibit wild-type RAS signaling in pancreatic cancer

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is a lethal malignancy with limited treatment options. Although activating mutations of the KRAS GTPase are the predominant dependency present in >90% of PDAC patients, targeting KRAS mutants directly has been challenging in PDAC. Similarly, strategies targeting known KRAS downstream effectors have had limited clinical success due to feedback mechanisms, alternate pathways, and dose-limiting toxicities in normal tissues. Therefore, identifying additional functionally relevant KRAS interactions in PDAC may allow for a better understanding of feedback mechanisms and unveil potential therapeutic targets. Here, we used proximity labeling to identify protein interactors of active KRAS in PDAC cells. We expressed fusions of wild-type (WT) (BirA-KRAS4B), mutant (BirA-KRAS4B^G12D), and nontransforming cytosolic double mutant (BirA-KRAS4B^G12D/C185S) KRAS with the BirA biotin ligase in murine PDAC cells. Mass spectrometry analysis revealed that RSK1 selectively interacts with membrane-bound KRAS^G12D, and we demonstrate that this interaction requires NF1 and SPRED2. We find that membrane RSK1 mediates negative feedback on WT RAS signaling and impedes the proliferation of pancreatic cancer cells upon the ablation of mutant KRAS. Our findings link NF1 to the membrane-localized functions of RSK1 and highlight a role for WT RAS signaling in promoting adaptive resistance to mutant KRAS-specific inhibitors in PDAC.

Keywords: BioID; KRAS; NF1; PDAC; RSK.

Fig. 1.

Properties of BirA-KRAS fusions. (A) Design of BirA-KRAS fusion constructs. Myc-tagged BirA was fused to the N terminus of murine WT KRAS4B or mutant KRAS4BG12D with a short glycine linker (GGSG). Membrane-localized BirA was created by fusing the last 20 amino acids of murine KRAS to myc-tagged BirA. KRAS4BG12D/C185S was created by mutating the cysteine residue in the CAAX prenylation motif to mislocalize the protein. (B) mT42-2D cells expressing B-WT, B-G12D, B-CAAX, and B-G12D/C185S were sequentially fractionated into cytosolic and membrane isolations and probed for myc-tag, as well as EGFR and AKT representing membrane and cytosolic controls, respectively. (C) Immunofluorescence (IF) of B-WT, B-G12D, B-CAAX, and B-G12D/C185S in mT93-2D cells. Staining of the myc-tagged BirA fusions are represented in red while DAPI-stained nuclei are represented in blue in the IF images. The arrows indicate the profiles taken for the relative fluorescent intensities from one edge to the other edge in cells (adjacent panels). (D) mT95-2D FPC cells were infected to express B-WT, B-G12D, B-CAAX, B-G12D/C185S, or empty control. Cells were incubated with 50 μM biotin for 24 h prior to lysis and streptavidin pulldown. Elutions and 1.5% input were probed for RAF1 and myc-tag BirA. (E) Lysate from mT95 expressing B-WT and B-G12D was used for both immunoprecipitation (IP) and streptavidin (SA) pulldown under similar lysis and washing conditions and probed for RAF1, p110-α, and the myc-tagged BirA-Kras fusions.

Fig. 2.

Proximity labeling nominates RSK1 as a mutant- and membrane-specific KRAS interactor. (A) Schematic of the method for determining mutant-specific interaction by comparing B-G12D to B-WT and membrane-specific interactions by comparing B-G12D to B-G12D/C185S. (B) MS relative quantification of internal control KRAS peptides and positive controls (RAF1 and niban-like 1) after normalization to BirA counts for B-WT, B-G12D, B-CAAX, and B-G12D/C185S samples from vehicle-treated FPC cells. Error bars represent SEM (n = 3). (C) Venn diagram of the overlap of candidates that met the criteria for being mutant specific and membrane specific in both the 4-OHT–treated and control-treated conditions for two out of three biological replicates. The 32 candidates met all criteria. (D) Enrichment of peptide abundances for the 32 candidate interactors as well as KRAS and BirA control peptides. Known KRAS interactors include RAF members (purple), NF1/SPRED proteins (blue), mTORC2 (magenta), and RBD-containing proteins (orange) with the internal controls BirA and KRAS (yellow). Enrichment is plotted on a scale of log base 10. (E) BioID was performed on mT42 expressing B-WT, B-G12D, B-CAAX, or B-G12D/C185S followed by Western blot analysis to confirm RICTOR, RSK1, ARAF, BRAF, and RAF1 as B-G12D substrates. (F) mT42 cells expressing B-WT and B-G12D were treated with 50 μM biotin for 24 h and harvested in HEPES-based Nonidet P-40 lysis buffer. Lysates were either incubated for 4 h with streptavidin (SA) beads or 3 h with myc-tag antibody followed by 1 h with protein G Dynabeads to compare the performance of BioID to coimmunoprecipitation. SA pulldowns and immunoprecipitations (IPs) were analyzed by Western blot for the detection of NF1, p110-α, RSK1, ARAF, RAF1, and myc-tag.

Fig. 3.

NF1 recruits RSK1 to the oncogenic KRAS interactome. (A) BioID of mT42 and mT93 expressing B-G12D with CRISPR depletion of MAPKAP1 and RICTOR were analyzed by Western blot to detect the biotinylation of these mTORC2 components by B-G12D. (B) BioID of mT42, mT93, and mT95 expressing B-G12D with CRISPR depletion of NF1. (C) BioID of mT93 expressing B-G12D with CRISPR depletion of NF1, RSK1, SPRED1, and SPRED2. (D) FPC cell lines expressing B-G12D were treated with 500 nM trametinib during the 24 h biotin incubation. The lysates and streptavidin pulldown were analyzed for the presence of RSK1, NF1, ARAF, and myc-tag as well as the phosphorylation status of MAPK. (E) The 5-d cell proliferation curves of mT42 cells expressing B-G12D treated with 5, 50, 200, 500 nM trametinib were compared to negative control DMSO and positive control gemcitabine.

Fig. 4.

Membrane-localized RSK1 exerts negative feedback on WT RAS. (A) FPC cell line mT42 cells ectopically expressing RSK1, myr-RSK1, or empty vector control were plated for cell proliferation and treated with DMSO or 4-OHT, then measured by CellTiter-Glo luminescence assay every day for 4 d. (B) mT42 FPC cells ectopically expressing RSK1, myr-RSK1, or empty vector control were treated with DMSO or 4-OHT and analyzed by Western blot. Gemcitabine was used as a proliferation control. Immunoblotting was performed for RSK1 with RAS pathways represented by phospho-ERK and phospho-AKT, and the RSK1 substrates, phospho-EPHA2 and phospho-S6. (C) MIA PaCa-2 cells expressing RSK1, myr-tag RSK1, and empty vector control were seeded at 2,000 cells per well and treated with either 100 nM AMG 510 or DMSO control, then measured by CellTiter-Glo luminescence assay every day for 4 d. (D) MIA PaCa-2 cells were seeded at 2,000 cells per well and treated 24 h after seeding with 5 μM BI-3406 or DMSO control and AMG 510 (from 5 μM to 0.1 nM). Cell viability was measured after 4 d with CellTiter-Glo. (E) mT42 cells were seeded at 1,000 cells per well and treated with 4-OHT or DMSO control and BI-3406 in a dose–response manner (from 20 μM to 2 nM). Cell viability was measured using CellTiter-Glo Luminescence assay at day 4. (F) FPC cell line mT42 cells were treated with 5 μM BI-3406, 20 nM trametinib, or DMSO vehicle control in the presence or absence of mutant KRAS by treatment with DMSO vehicle control or 4-OHT, respectively. Immunoblotting was performed for RAS pathways by phospho-ERK and phospho-AKT and the RSK1 substrate phospho-S6.

Fig. 5.

Oncogenic KRAS engages an RSK1/NF1 pathway to inhibit WT RAS in pancreatic cancer cells. Schematics illustrating RSK1/NF1 interactions and known mechanisms of RSK1-mediated negative feedback on the RAS pathway. Mutant KRAS activates the RAF/MEK/ERK/RSK and PI3K/AKT pathways. Upon ERK activation, RSK1 transiently localizes to the membrane (59). (A) In KRAS-mutant PDAC cells, RSK1 depends on MEK activity and NF1/SPRED2 expression to be recruited to the mutant-KRAS interactome on the membrane. Membrane-localized RSK1 negatively regulates RAS activation by inhibiting the RasGEF, SOS1, and activating the RasGAP, NF1. Compared to mutant RAS, WT RAS exhibits greater sensitivity to RasGEFs and RasGAPs; therefore, the RSK1-mediated negative feedback mechanism potently inhibits WT RAS but not mutant RAS. (B) Upon mutant Kras ablation, decreased SPRED2 expression and ERK-mediated RSK1 phosphoactivation disengages the negative feedback exerted on WT RAS by NF1/RSK1, thereby enabling RAS-addicted cells to survive.