SMURF2 Facilitates GAP17 Isoform 1 Membrane Displacement to Promote Mutant p53-KRAS Oncogenic Synergy

Abstract

Cooperativity between mutant p53 and mutant KRAS, although recognized, is poorly understood. In pancreatic cancer, mutant p53 induces splicing factor hnRNPK, causing an isoform switch that produces overexpression of GTPase-activating protein 17 isoform 1 (GAP17-1). GAP17-1 is mislocalized in the cytosol instead of the membrane, owing to the insertion of exon 17 encoding a PPLP motif, thus allowing mutant KRAS to remain in the GTP-bound hyperactive state. However, the role of PPLP in influencing GAP17-1 mislocalization remains unclear. We show that Smad ubiquitination regulatory factor 2 (SMURF2), a known stabilizer of mutant KRAS, interacts with GAP17-1 via the PPLP motif and displaces it from the membrane, facilitating mutant p53-mediated mutant KRAS hyperactivation. We used cell lines with known KRAS and TP53 mutations, characterized SMURF2 expression in multiple pancreatic cancer mouse models (iKras*; iKras*, p53*, and p48-Cre; Kras*), and performed single-cell RNA sequencing and tissue microarray on preclinical and clinical samples. We found that SMURF2 silencing profoundly reduces the survival of mutant TP53; KRAS-driven cells. We show that a GAP17-1 AALA mutant does not bind to SMURF2, stays in the membrane, and keeps mutant KRAS in the GDP-bound state to inhibit downstream signaling. In mouse models, mutant KRAS and SMURF2 upregulation are correlated with pancreatic intraepithelial neoplasia and ductal adenocarcinoma lesions. Furthermore, patients with pancreatic ductal adenocarcinoma who received neoadjuvant therapy and express moderate-to-high SMURF2 show decreased overall survival (P = 0.04).

Implications: In TP53 and KRAS double-mutated pancreatic cancer, SMURF2-driven GAP17-1 membrane expulsion facilitates mutant p53-KRAS oncogenic synergy.

Figure 1.. SMURF2 knockdown has more profound impacts on survival of KRAS; p53 double mutant cells.

(A) Different lung cancer cells lines with known KRAS and TP53 mutations (as indicated) were subjected to siRNA transfection using either control (C) or SMURF2 (S) siRNAs, as reported earlier. Forty-eight hours post-transfection, cell lysates were prepared and immunoblotted using indicated antibodies. (B) Above siRNA transfected cells were subjected to clonogenic survival assays and survival fractions were calculated considering control siRNA treated cells as ‘1’. Compared to either wild type, TP53 only, KRAS only mutants, TP53 and KRAS double mutated cells showed significantly (p=0.001, ****) greater loss in survival upon SMURF2 knockdown. Data representing from three independent studies. (C, D) Parental NCI-H460 cells carrying only KRAS^Q61H and another H460 line created that stably expressing TP53^R175H, were treated with either control or SMURF2 siRNAs. Cell lysates were prepared 48 hours post-transfection and subjected to immunoblotting using indicated antibodies (in C), and further plated for clonogenic assays (in D). Survival fraction of control siRNA treated cells were considered as ‘1’. (E, F) We further used U2OS parental (carrying wild type KRAS and p53) or overexpressing an anti-p53 GSE56 element. These cell lines were then transfected with either control or SMURF2 siRNAs. Forty-eight hours post-transfection, cell lysates were prepared and subjected to immunoblotting as indicated (in E), and further plated for clonogenic assays (in F) and analyzed as described above. (G) NCI-H441 cells overexpressing either Myc-tagged KRAS-4A or GFP-tagged KRAS-4B were transfected with different siRNAs (C - Control, H5 - UBCH5, H7 - UBCH7, or S- SMURF2). Cell lysates were subjected to immunoblotting as indicated. (H) Indicated cell lines with known KRAS and TP53 mutations, were transfected with indicated siRNAs and cell lysates prepared 48 h post-transfection were immunoblotted as shown.

Figure 2.. SMURF2 interacts with GAP17–1 in a PPLP motif dependent manner to modulate mutant KRAS activity.

(A) DDK-tagged SMURF2 and tGFP-tagged GAP17–1 (the wild type PPLP or AALA mutant) were co-overexpressed in HEK293 cells as indicated. Twenty-four hours post-transfection, cell lysates were prepared and immunoprecipitated using Affi-FLAG beads and immunoblotted using indicated antibodies. (B) Above cells were also fixed using 10% buffered formalin and subjected to immunofluorescence staining using DDK antibodies and tGFP was used for GAP17–1 localization. Compared to cytosolic localization for the GAP17–1, we noted punctate, membrane anchored localization of the AALA mutation (yellow arrowheads). Scale bar, 5 μm. (C) Cell fractionation (cytosolic, membrane, and nuclear) were performed as detailed in materials and methods. EGFR was used as a membrane, α-Tubulin as both membranous and cytosolic, and Histone H3 was used as a nuclear marker. (D) Overexpression of either wild type (PPLP) or AALA GAP17–1 mutant in H441 cells were subjected to RAS binding domain (RBD) pull-down assay according to manufacturer’s instruction. Either the pulldown fraction or total cellular inputs were subjected to immunoblotting using indicated antibodies. (E) TCGA data were extracted using cBIOPORTAL (for KRAS and TP53 mutation status) and TSVdb (GAP17 isoform expression as RSEM) online interfaces. We contrasted the GAP17–1/GAP17–2 transcript ratio in lung adenocarcinoma with mutant KRAS; mutant TP53 (n=24) or mutant KRAS alone (n=50) against those of normal lung tissue from cancer patients (n=59). Data are expressed as mean +/− SEM with dots representing individual patient samples. We used a non-parametric Kruskal-Wallis test to compare mutant KRAS; mutant TP53 samples to the other two groups, with P values adjusted for the multiple two-group tests.

Figure 3.. Smurf2 is upregulated in PanIN lesions in inducible mutant Kras^G12D and in PDA model of inducible Kras^G12D; Tp53^R172H double transgenic mice.

(A) Different tissues (Lung, pancreas and spleen) were harvested from three C57BL/6 mice and tissue lysates were subjected for immunoblotting using indicated antibodies. (B) Immunohistochemistry (IHC) of Smurf2 was performed on iKras* mice pancreas following induction of Kras* using doxycycline (DOX) after one day (1D), one week (1W) and three week (3W) post-treatment. Scale bar, 100 μm. (C) H&E and Smurf2 IHC staining of iKras* murine pancreas without DOX (top panel), 3W post-DOX treatment inducing Kras* and cerulean injection (middle panel) and 3D after DOX ithdrawal after the initial mutant Kras induction for 3W. Scale bar, 100 μm. (D) PDA lesions isolated from murine iKras*; p53* pancreas and stained with either H&E, Smurf2, or CK19 after 3W of DOX treatment to induce Kras* expression. Scale bar, 100 μm. (E) IHC staining of p48-Cre; Lsl-Kras^G12D mice pancreas collected after 400 days of birth showing Smurf2 staining in PanIN and PDA lesions. Scale bar, 100 μm. (F & G) Single cell RNA sequencing of normal, PanIN and PDA lesions isolated from iKRas*; p53* pancreas showing ductal cell specific expression of Smurf2 gene expression. For analysis, we have included 63, 178, and 2,102 ductal cells isolated from the healthy, PanIN, and PDA groups, respectively and presented either as violin (in panel F) or dot (in panel G) plots. (H) Two (#4292, 9805) mouse PDA cell lines previously established from iKras*; p53* mice were treated either with vehicle control of DOX for indicated time periods and cell lysates were subjected to immunoblotting using indicated antibodies. (I & J) The two above iKras*; p53* PDA cell lines were treated with DOX for 24 h and cells were subjected to either protein (in panel I) or RNA (in panel J) isolation and quantified for Smurf2 and Gapdh protein and transcript levels. (K) The two above cell lines were grown in the presence of DOX, transfected with either control or Smurf2 siRNAs and then subjected to clonogenic survival. Both cell lines showed reduced survival (***, p=0.001) following Smurf2 loss. (L) Cell lysates of indicated cell lines were cultured in the presence of DOX and were subjected to siRNA mediated Smurf2 knockdown as indicated. Forty-eight hours post-transfection, cell lysates were prepared and immunoblotted as indicated.

Figure 4.. Higher SMURF2 expression linked to poor OS of PDA patients.

(A, B) Single cell RNA sequencing analysis of SMURF2 of ductal cells from PDA patients and adjacent normal pancreas. For analysis, we have included 622, 178, and 11,023 ductal cells isolated from the tumor adjacent normal, and PDA groups, respectively and presented either as violin (in panel A) or dot (in panel B) plots. The SMURF2 expression in the human PDA cells were significantly higher (p_adj =7.142954e-05; log2fc= −0.089663640) compared to the adjacent normal. (C) Representative images of PDA patients showing intense SMURF2 staining compared to normal pancreas. (D & E) Kaplan Meier plot and detailed analysis showing poor OS for patients showing moderate to high SMURF2 expression in PDA as assessed using the Cox regression model (p=0.04). (F) First, we established Panc 1 cells stably transduced with SMURF2 sh#3. Cell lysates were prepared from either control or SMURF2 shRNA Panc 1 cells and subjected to either immunoblotting or RBD pulldown assay to quantify GTP-bound KRAS^G12D. (G) Panc 1 cells stably transduced control or SMURF2 shRNA were plated at clonal density (250 cells per 35 mm dish) in triplicate and allowed to adhere. Next day, cells were treated either with DMSO (vehicle control) or different concentrations (1, 2.5 and 5 μM) and allowed to form colonies. Seven days post-treatment, colonies were fixed and stained using crystal violet and counted to determine survival fraction as shown.

Figure 5.. Model showing SMURF2’s role in mutp53-mutKRAS oncogenic synergy in pancreatic ductal adenocarcinoma (PDA).

Mutant p53 is reported to induce hnRNPK spicing factor expression leading to isoform switch of ARHGAP17 (GAP17 in short) to enrich GAP17 Iso1 (GAP17–1) over GAP17–2. GAP17–1 incorporates a unique exon (ex17) encoding PPLP motif responsible for membrane dislocation to cytosol by an unknown mechanism. We previously reported that SMURF2 catalytic activity degrades β-TrCP1 to keep mutant KRAS stable and active (12). Mutant KRAS, in turn, can upregulate SMURF2 level via transcriptional activation to amplify the oncogenic signal. We further show that SMURF2 physically binds to GAP17–1 via PPLP motif and scavenges it from the membrane to cytosol, thus allowing mutant KRAS to remain in GTP-bound hyperactive state and activating downstream effectors and promoting PanIN and PDA progression. In contrast, site-directed mutagenesis of PPLP motif to AALA can restore GAP17–1 membrane localization like GAP17–2 to perform efficient GTP hydrolysis, thus can keep mutant KRAS in inactive state.