Targeting wild-type KRAS-amplified gastroesophageal cancer through combined MEK and SHP2 inhibition

An Author Correction to this article was published on 09 August 2018

This article has been updated

Abstract

The role of KRAS, when activated through canonical mutations, has been well established in cancer1. Here we explore a secondary means of KRAS activation in cancer: focal high-level amplification of the KRAS gene in the absence of coding mutations. These amplifications occur most commonly in esophageal, gastric and ovarian adenocarcinomas2,3,4. KRAS-amplified gastric cancer models show marked overexpression of the KRAS protein and are insensitive to MAPK blockade owing to their capacity to adaptively respond by rapidly increasing KRAS–GTP levels. Here we demonstrate that inhibition of the guanine-exchange factors SOS1 and SOS2 or the protein tyrosine phosphatase SHP2 can attenuate this adaptive process and that targeting these factors, both genetically and pharmacologically, can enhance the sensitivity of KRAS-amplified models to MEK inhibition in both in vitro and in vivo settings. These data demonstrate the relevance of copy-number amplification as a mechanism of KRAS activation, and uncover the therapeutic potential for targeting of these tumors through combined SHP2 and MEK inhibition.

Main

KRAS is justifiably renowned as the most commonly mutated oncogene across human cancers1. Research on RAS-driven cancers has focused almost exclusively on RAS-coding mutations. However, recent studies have found genomic amplification of KRAS without canonical mutations5,6. For example, our characterization of somatic copy-number alterations (SCNAs) across gastric, esophageal and colorectal adenocarcinomas found KRAS to lie at the peak of the most significant amplification, events that were almost exclusive to gastroesophageal tumors, with exome sequencing demonstrating that the majority of these tumors lack detectable KRAS mutations3,4. KRAS amplification was similarly noted in ovarian and endometrial cancers, in which their presence was associated with enhanced rates of metastasis or poor survival2,7. In retrospect, early KRAS studies noted KRAS amplification. In 1985, a study reported the occurrence of KRAS amplification in cancer and that overexpression of wild-type KRAS transformed NIH-3T3 cells8. Although several recent studies have identified KRAS amplification as a mechanism of acquired resistance to targeted therapies, the amplifications that were observed in gastroesophageal and ovarian cancers were detected in de novo, untreated cancers9,10,11,12. These results suggest that amplification of wild-type KRAS is an alternative means of activating this oncoprotein in cancer and raise key questions about its activity and the potential vulnerabilities of these tumors relative to those with mutant KRAS.

We first systematically evaluated the frequency of wild-type KRAS amplification across cancer using existing data from The Cancer Genome Atlas (TCGA) and other large-scale genome studies. Using data with matched SCNA profiling and exome analysis, we identified KRAS amplifications without coding mutations in esophageal adenocarcinoma (17%), the chromosomal instability (CIN) variant of gastric cancer (13%) and serous ovarian cancer (10%) with a lower frequency of events in endometrial and lung cancer (Fig. 1a and Supplementary Table 1). These data are consistent with the genomic characterization of these diseases, which identified significant focal amplification peaks at the KRAS locus4,5. Indeed, visual inspection of the SCNA profiles on chromosome 12 demonstrates the focality of amplification at the KRAS locus in gastric cancer, events that were notably absent in colorectal cancer in which KRAS is recurrently activated by canonical mutation (Fig. 1b). We next evaluated the relationship between KRAS copy number and expression. Analyses from the gastric cancer TCGA demonstrated that KRAS amplification commonly occurs at high levels, exceeding 25 estimated copies of the gene (saturating the discriminant capacity of the array-based copy-number platform) and is accompanied by marked elevation of KRAS mRNA (Fig. 1c and Supplementary Fig. 1a).

Fig. 1: Amplification of wild-type KRAS associates with elevated KRAS expression and poor survival in gastric cancer.
figure1

a, Frequencies of KRAS amplification in the absence of mutations across cancer types. b, Representation of SCNAs across chromosome 12p in 615 colorectal compared to 298 CIN gastric tumors. Samples are ordered from left to right by decreasing copy-number level at the KRAS locus. The insets expand the region from 22 to 28 Mb on Chr12 from the 15% of each sample set with the highest KRAS copy number. The horizontal black line indicates the KRAS locus. Color bar represents the scale of copy-number gain (red) and copy-number loss (blue). c, KRAS mRNA expression compared to KRAS copy number in gastric adenocarcinoma. The x axis represents the SNP-array-inferred KRAS copy number and the y axis represents KRAS mRNA expression based on quantification of RNA-sequencing and was set to log2 scale. Estimated copy numbers greater than 25 could not be discriminated based on saturation of array-based profiling. d, Histological analysis of patient-derived primary gastric tumor samples. Panels from left to right represent representative images of H&E staining, KRAS immunohistochemistry (IHC), FISH for KRAS. The right-most panel is a high-power magnification (HPF) image of the inset in FISH analysis that displays the magnitude of KRAS locus amplification in tumors. Q36, Q39 are Q56 are tumors with KRAS amplification. Q20 is a tumor without KRAS amplification and is used as a negative control. e, Kaplan–Meier survival analysis comparing cause-specific survival of patients with gastric cancer with KRAS amplification status (red line; n = 30) to patients with gastric cancer without KRAS amplification (blue line, n = 97) in a Japanese cohort. Log-rank (Mantel–Cox) two-sided P = 0.048. f, KRAS quantification using mass spectrometric analysis of FFPE-extracted tissue from patient-derived gastric cancer samples. Red dotted line represents cut-off between low and high KRAS expression.

We confirmed high KRAS expression and amplification in primary gastroesophageal tumor samples using fluorescence in situ hybridization (FISH) and immunohistochemistry (Fig. 1d). To examine the relationship between KRAS amplification and impact on clinical outcomes, we first compared cause-specific survival of patients with KRAS amplification to patients lacking the amplification in a series of gastroesophageal junction adenocarcinomas from a Japanese cohort (see Methods). In a Kaplan–Meier five-year survival analysis from this Japanese cohort, patients with KRAS amplification–positive status (n = 30, 38.4%) had significantly poorer survival compared to patients with non-amplified KRAS status (n = 97, 69.4%; Fig. 1e). An independent analysis from the TCGA cohort of gastric adenocarcinoma with a mixture of patients from Eastern and Western descent, found that patients with KRAS-amplified tumors had significantly decreased disease-free survival and a trend towards decreased overall survival compared to patients without KRAS amplification (Supplementary Fig. 1b). We also evaluated a Western cohort of gastroesophageal adenocarcinoma (see Methods) for KRAS protein expression via mass spectrometric analysis of paraffin-embedded samples, which confirmed marked KRAS overexpression (Fig. 1f). This observation was correlated with a significantly decreased five-year survival in patients that had tumors with KRAS protein overexpression (Supplementary Fig. 1c).

To evaluate the function of the amplification of wild-type KRAS in gastric cancer (GC), we used the availability of cellular models with amplification (YCC1, KE39, HUG1N) and mutation (GSU, SNU-1; both G12D) and a control cell line that is neither amplified nor mutant for KRAS (IM95). Amplified KRAS GC cell lines displayed strikingly elevated total KRAS levels, quantified by immunoblot densitometry to be 45–300-fold the levels present in IM95 cells (Supplementary Fig. 1d). We performed siRNA-mediated KRAS silencing and found a significant reduction in cell proliferation in all cell lines, except IM95, indicating that mutant and amplified cell lines are dependent on KRAS (Supplementary Fig. 2).

We next evaluated potential therapeutic vulnerabilities in KRAS-amplified and -mutant GC models. Directly targeting KRAS has been challenging owing to the difficulty in disrupting the nucleotide-binding pocket and the marked affinity of KRAS for GTP13. Because targeting downstream effectors in the PI3K and MAPK pathways have been proposed for RAS-driven cancers, we examined the effect of MEK and PI3K inhibitors in GC lines. Whereas AKT or PI3K inhibitors MK2206 and GDC0941 showed equivalently modest effects on viability in KRAS-mutant and amplified models (Supplementary Fig. 3), we found discrepant responses to MEK inhibition. Compared to KRAS-mutant cell lines, KRAS-amplified models demonstrated intrinsic resistance to MEK blockade using either AZD6244 or GSK1120212 (Fig. 2a). Following 72-h treatment with GSK1120212, the more potent of the MEK inhibitors, we observed less apoptosis in amplified KRAS as measured by annexin-V staining (Fig. 2b). We confirmed that the relative failure of MEK inhibition was not attributable to inadequate target engagement, finding stable inhibition of phosphorylated (p)ERK after treatment with GSK1120212 (Fig. 2c).

Fig. 2: Amplified wild-type KRAS GC cell lines and organoids display differential sensitivity to MEK inhibition compared to KRAS-mutant models.
figure2

a, Cell viability (n = 3 independent experiments) in GC lines (blue, KRAS-amplified lines; red, KRAS mutant lines and black, the IM95 wild-type line) after 72 h of MEK inhibition AZD6244 (1 μM) or GSK1120212 (100 nM) with DMSO used as a vehicle control. Cell viabilities are shown as mean ± s.d normalized to the DMSO control group and are expressed as a percentage of maximum proliferation. Statistical comparisons were made between KRAS-amplified and KRAS-mutant cells treated with GSK1120212 (***P < 0.001) or AZD6244 (**P = 0.0062) using a non-parametric Kruskal–Wallis test. Each triangle in this figure represents a biological replicate/individual experiment except in d, where each triangle represents a technical replicate. b, Percentage apoptosis (n = 3 independent experiments) in GC cell lines (blue, KRAS-amplified cell lines; red, KRAS-mutant cell lines; black, the IM95 wild-type cell line) after treating with 100 nM GSK1120212 for 72 h based on cells stained with annexin V–FITC and propidium iodide (PI) and analyzed by flow cytometry to quantify percentage of cells undergoing apoptosis. DMSO was used as a vehicle control. Data are mean ± s.d and are expressed as percentage annexin V–FITC–positive cells. Comparisons between DMSO and treatment groups were made using two-tailed Student’s t-test; *P = 0.048; **P = 0.003; NS, not significant. c, Representative western blot analysis (n = 3 independent experiments) of GC cell lines after treatment with 100 nM GSK1120212 at 1, 6 and 24 h. Lysates were probed with pAKT (Ser473), total AKT, pERK1/2 and total ERK1/2 and β-actin was used as a loading control. d, Relative RAS–GTP expression. RAS–GTP levels were compared between the GC KRAS-amplified cell line (KE39) and GC KRAS-mutant cell line (SNU1) after treatment with 100 nM GSK1120212 at 1, 6 and 24 h. GTP bound RAS was quantified by measuring absorbance in 96-well plates using the RAS G-LISA assay. Fold change in RAS–GTP levels is relative to DMSO control group at each time point. Each point represents the mean fold change relative absorbance from two independent experiments conducted for each cell line. e, Representative western blot analysis (n = 3 independent experiments) of the GC KRAS-amplified line (KE39) and GC KRAS-mutant line (GSU). Cells were transfected with two independent siRNAs against KRAS and non-targeting control siRNA at 10 nM for 48 h and treated with 100 nM GSK1120212 for 24 h. Protein lysates were collected and probed with antibodies against pAKT (Ser473), total AKT, pERK1/2 and total ERK1/2. β-Actin was used as loading control for all western blots. f, Schematic of establishment of primary isogenic gastric organoids from gastric epithelium from adult mice with Trp53fl/fl or Trp53fl/flKrasG12D/+ genotypes and subsequently infected with control lentivirus or lentivirus encoding there tandem copies of codon-optimized KRAS (3×KRAS). Immunoblot analysis of KRAS expression in gastric organoids (Trp53−/−3×KRAS and control Trp53−/− empty vector; Trp53−/−KrasG12D/+3×KRAS and control Trp53−/−KrasG12D/+ empty vector) compared to GC cell line (KE39) with endogenous amplified KRAS expression g, Representative western blot analysis (n = 3 independent experiments) of isogenic KRAS mouse gastric organoids treated with GSK1120212 (50 nM) at 1, 6 and 24 h. Lysates were collected and probed with antibodies against KRAS, pAKT (Ser473), AKT, pERK1/2 and ERK1/2. c,e,g, Uncropped images of gels can be found in Supplementary Fig. 12. h, Representative phase contrast images (n = 3 independent experiments) of in vitro cultures of primary gastric organoids (Trp53−/−3×KRAS, control Trp53−/− empty vector, Trp53−/−KrasG12D/+3×KRAS, control Trp53−/−KrasG12D/+ empty vector) after five days of treatment with 50 nM GSK1120212. DMSO was used as vehicle control. Scale bar, 50 μm.

We next investigated whether KRAS-amplified cells showed adaptive compensation to MEK inhibition. We found that MEK inhibition in cells with KRAS amplification led to a strong induction of pAKT after 1, 6 and 24 h, an effect that was not observed in control or mutant KRAS cells (Fig. 2c). We confirmed that these KRAS-amplified cell lines lack somatic PTEN inactivation or activating PIK3CA mutations14. To evaluate whether these effects were specific to MEK inhibition or more generalizable to blockade of the MAPK pathway, we tested the ERK inhibitor SCH772984, again finding adaptive increases in pAKT in amplified but not in mutant KRAS cell lines (Supplementary Fig. 4a). These results suggested an adaptive response to loss of ERK-mediated negative regulation of upstream cell signaling. Indeed, we confirmed the transcriptional repression of SPRY4, FOS, DUSP4 and DUSP6, indicative of effective blockade of the MAPK pathway and attenuation of expression of mediators of ERK-mediated negative feedback (Supplementary Fig. 4b). Additionally, we observed that ERK signaling was still inhibited at 72 h, suggesting that the adaptive effects of MEK blockade in KRAS-amplified cells do not include evident pERK reactivation (Supplementary Fig. 4c). Collectively, these data suggest that differential adaptive capabilities of KRAS-amplified tumors may serve as a barrier to effective therapy with MAPK inhibition, a focus of many therapeutic strategies for KRAS-driven cancers.

To evaluate adaptation to MEK inhibition, we first confirmed that MEK inhibitor-mediated activation of pAKT in KRAS-amplified cells was dependent upon PI3K by demonstrating that co-inhibition of a PI3K inhibitor (GDC-0941) and GSK1120212 blocked pAKT induction (Supplementary Fig. 5a). Furthermore, PI3K blockade sensitized KRAS-amplified cells to GSK1120212 (Supplementary Fig. 5b), suggesting that blockade of adaptive physiology could facilitate the development of therapeutic combinations for these cancers. Because co-administration of effective doses of PI3K and MEK inhibitors in patients has been greatly limited by the toxicity of these combinations15, we next sought to define the etiology of this adaptive resistance and identify methods to inhibit this adaptation in order to potentiate MAPK inhibition in KRAS-amplified cancers.

We first investigated whether the adaptation to MEK inhibition was stimulated by enhanced activation of other signaling pathways, specifically receptor tyrosine kinases (RTKs). We examined changes in cell signaling using a phosphorylated-tyrosine RTK signaling antibody array that detects phosphorylation of 28 RTKs and 11 key signaling nodes. After 6 h of exposure to GSK1120212, all of the KRAS-amplified cell lines but none of the KRAS-mutant cell lines showed increased AKT phosphorylation (Supplementary Fig. 5c). ERBB3 was the only other RTK for which phosphorylation was upregulated, which was shown to be present in one of the three KRAS-amplified cell lines, HUG1N, and validated using immunoblotting analysis (Supplementary Fig. 5d). These data suggested that ERBB3 could be a direct mediator of resistance, as has been reported with KRAS-mutant lung and colon cancer models following MEK blockade16. Notably, in lung and colorectal cancer models, adding the pan-ERBB tyrosine kinase inhibitor afatinib to MEK inhibition led to the inhibition of pAKT activation and augmented anti-proliferative responses16. However, when we evaluated afatinib with GSK1120212 in KRAS-amplified GC models, we did not observe pAKT attenuation nor any substantial effects on cell proliferation (Supplementary Fig. 5e,f), suggesting that activated ERBB3 was not the primary cause of adaptation. Additionally, we silenced ERBB3 in HUG1N cells using siRNA before treatment with GSK1120212 and this again failed to inhibit pAKT activation (Supplementary Fig. 5g), suggesting that ERBB3 is not the primary mediator of MEK resistance.

We next evaluated mechanisms by which KRAS amplification could promote adaptive resistance. Specifically, we hypothesized that loss of pERK-mediated negative feedback following MEK blockade could enable cells with aberrantly elevated KRAS expression to adapt by enhancing KRAS activation. Notably, although KRAS-amplified models were found to have higher KRAS–GTP levels compared to KRAS-mutant models (Supplementary Fig. 5h), we found that KRAS-amplified models had a greater ability to enhance KRAS activation following MEK inhibition. Specifically, we found that MEK inhibition induced increases of RAS–GTP of up to 500% in KRAS-amplified KE39 cells at 6 and 24 h after GSK1120212 treatment, an effect not similarly observed KRAS-mutant SNU1 cells (Fig. 2d). We investigated whether NRAS or HRAS contribute to RAS-mediated adaptation to MEK inhibition in KRAS-amplified KE39 and KRAS-mutant GSU cells. Although we did not detect HRAS–GTP (i.e., GTP-bound proteins of the RAS superfamily) following MEK inhibition in KE39 cells, we observed a modest increase in NRAS–GTP (Supplementary Fig. 5i). However, when we silenced NRAS expression using pooled siRNAs, we failed to inhibit the adaptive pAKT response to MEK inhibition and did not attenuate the increased RAS–GTP levels (Supplementary Fig. 5j,k), suggesting that KRAS is the primary RAS member to be activated in response to MEK inhibition in KRAS-amplified cells. Indeed, silencing KRAS with siRNAs before GSK1120212 treatment prevented pAKT rebound in KRAS-amplified but not KRAS-mutant GC lines and significantly decreased MEK inhibitor-induced RAS–GTP activation (Fig. 2e and Supplementary Fig. 5l,m).

The data in GC cell lines suggested that activation of amplified wild-type KRAS itself could mediate adaptive responses and resistance to MEK and/or MAPK inhibitor therapy. We first evaluated this hypothesis by testing whether KRAS-amplified models had a different ability to activate KRAS in the setting of a mitogenic stimulus. We confirmed that with EGF stimulation of serum-starved cell lines, the amplified KRAS cell lines had a greater increase in KRAS–GTP levels compared to KRAS mutants (Supplementary Fig. 6a). To further validate this hypothesis, we generated a novel isogenic system using organoids from mouse gastric epithelia (Fig. 2f). Primary gastric organoids were isolated in parallel from glandular stomachs of mice with floxed alleles of p53 (Trp53−/−) or a combination of loxP-stop-loxP (LSL) KrasG12D (KrasLSL-G12D/+) and floxed p53 alleles (Trp53−/−KrasG12D/+), in which the KRAS mutant is expressed from the endogenous promoter following Cre-mediated excision of a floxed stop cassette. Organoids were infected with an adenovirus expressing CMV-Cre (AdCMVCre) and selected with nutlin3 and/or erlotinib to remove cells that had not recombined. These organoids were then infected with a lentivirus expressing either triple tandem codon-optimized KRAS to mimic KRAS amplification (pLX324-3×KRAS) or a control vector (pLX324-Cre). We confirmed that KRAS protein expression in 3×KRAS organoids (Trp53−/−3×KRAS and Trp53−/−KrasG12D/+3×KRAS) was comparable to endogenous KRAS expression of a KRAS-amplified GC cell line (KE39) (Fig. 2f). We then tested this isogenic model with GSK1120212 at 1, 6 and 24 h and confirmed that 3×KRAS organoids consistently induced compensatory pAKT activation after MEK inhibition (Fig. 2g) and compensatory KRAS–GTP augmentation (Supplementary Fig. 6b). Moreover, ectopic 3×KRAS organoids treated with GSK1120212 in vitro displayed decreased sensitivity to MEK inhibition compared to control organoids after five days of treatment (Fig. 2h). These data are consistent with the differential responses of KRAS-amplified and -mutant GC cell lines and highlight that overexpression of wild-type KRAS itself serves as a mediator of adaptive resistance to MAPK inhibitor therapy.

Because our data indicated that KRAS mediates adaptive resistance to MAPK blockade, we next evaluated potential therapeutic approaches to target KRAS-amplified cancers. Although we did not observe increased RTK activity following MEK inhibition in KRAS-amplified cell lines other than the aforementioned increased pERBB3 in HUG1N cells (Supplementary Fig. 5c), we reasoned that basal RTK activity could contribute to KRAS-driven adaptation. We used the induction of pAKT after MEK inhibitor therapy as a marker to identify how blocking individual kinases that are known to be active in gastric cancer could impact adaptive responses. We found that inhibitors of MET (crizotinib) or FGFR (BGJ-398) blocked pAKT induction in combination with GSK1120212 in YCC1, but not in KE39 or HUG1N, cell lines (Supplementary Fig. 7b,c). By contrast, combinatorial MEK and IGF1R (OSI-906) inhibition consistently abrogated MEK-mediated pAKT induction in KRAS-amplified GC cell lines (Supplementary Fig. 7a), suggesting that IGF1R signaling may be important in circumventing MEK resistance. However, the combination of OSI-906 (a selective IGF1R inhibitor) with GSK1120212 showed a moderate, additive decrease in cell viability after treatment for five days (Supplementary Fig. 7d). Longer-term clonogenic assays showed that IGF1R inhibition in KRAS-amplified cells enhanced responses to MEK inhibition; however, the effects of this combination were not as pronounced as the effects of single-therapy MEK inhibition in KRAS-mutant cells (Supplementary Fig. 7e).

Although these data suggest that RTK signaling contributes to KRAS-mediated adaptive resistance, these findings raised concerns that targeting single RTKs alone may not be sufficient to block KRAS reactivation in KRAS-amplified models following MAPK inhibition. We therefore analysed whether there may be alternative strategies to block the mobilization of KRAS–GTP following MAPK inhibition in KRAS-amplified tumors. Specifically, we investigated whether it may be feasible to target the physiological process that integrates the inputs from multiple RTKs to activate RAS proteins. We first evaluated the son of sevenless proteins guanine exchange factors (GEFs) SOS1 and SOS2, which form a complex with RAS to catalyze the GDP-to-GTP nucleotide exchange17. We found that, using pooled siRNAs targeting both SOS1 and SOS2, we could block the pAKT rebound following MEK inhibition in KRAS-amplified models (Fig. 3a). Moreover, we observed that silencing of SOS1 and SOS2 also reduces pERK, indicating that SOS blockade attenuates the basal activity of amplified KRAS. These data bolster our hypothesis that KRAS activation itself mediates adaptive response and demonstrate that GEFs could serve as targets for KRAS-amplified tumors.

Fig. 3: Genetic targeting of SOS enhances efficacy of MEK inhibition in KRAS-amplified GC models in vitro and in vivo.
figure3

a, Representative western blot analysis (n = 3 independent experiments) of GC KRAS-amplified (HUG1N and KE39) and KRAS-mutant (GSU) cell lines was performed on cells that were transfected with a combination of 20 nM pooled siRNAs specific to SOS1 and SOS2 (siSOS1/2) or 20 nM non-targeting control siRNA (NTsiRNA) for 48 h and then treated with 100 nM GSK1120212 for 24 h. Protein lysates were collected and probed with antibodies against SOS1, SOS2, pAKT (Ser473), total AKT, pERK1/2 and total ERK1/2. β-Actin was used as loading control for all western blots. b, Percentage cell viability (n = 3 independent experiments) in KRAS-amplified (KE39 and HUG1N) and KRAS-mutant (GSU) GC lines after transfection with SOS1 and SOS2 siRNAs for 48 h followed by treatment with 100 nM GSK1120212 for 72 h. Cell viability was determined using a CellTiterGlo cell viability assay (Promega, Madison, WI, USA). Data are mean ± s.d. Comparisons were made using a two-tailed Student’s t-test; *P < 0.05; ***P < 0.001. Each triangle in this figure represents a biological replicate/individual experiment except in d, where each triangle represents a technical replicate. c, Representative western blot analysis (n = 2 independent experiments) of KRAS-amplified (KE39) and KRAS-mutant (GSU) cell lines engineered to stably express doxycycline-inducible pTRIPz shSOS or NTshRNA constructs. Silencing of SOS was induced with 1 μM doxycycline for 48 h and cells were treated with 100 nM GSK1120212 for 24 h. Protein lysates were collected and probed with antibodies against SOS1, pAKT(Ser473), AKT, pERK1/2 and ERK1/2. d, Relative RAS–GTP levels of KE39 expressing indicated constructs with or without 48-h induction with doxycycline and a 6-h treatment with 100 nM GSK1120212. DMSO was used as a treatment control. Relative active RAS–GTP levels were quantified using the RAS G-LISA activation assay. Each data point represents mean relative absorbance (n = 2 technical replicates). Data are representative of two independent experiments. e, Top, tumor volumes of NOD/SCID mice injected subcutaneously with KE39 or GSU cells expressing inducible shSOS or NTshRNA control constructs. When tumors were approximately 150 mm3, mice were initiated on a doxycycline-containing diet for 48 h before starting GSK1120212 (2 mg kg−1) or vehicle treatment for five or three weeks, respectively. Data are mean ± s.e.m. (n = 10 mice in vehicle group, n = 8 and n = 5 mice in KE39 NTshRNA and GSU NTshRNA tumors treated with GSK1120212, respectively; n = 8 mice in GSK1120212 treatment group). Statistical comparisons between vehicle and treatment groups were made using a two-tailed Student’s t-test with Welch’s correction; *P < 0.05. Bottom, waterfall plots showing change in percentage tumor volume (compared to initial tumor volume) for individual tumors (each represented by a bar) following doxycycline induction and GSK1120212 treatment for either five weeks (the KE39-derived cell lines) or three weeks (the GSU-derived cells). For tumors with growth exceeding the scale, the raw number is shown above the bar. f, Representative western blot analysis (n = 2 independent experiments) in KRAS-amplified patient-derived GC (CAT12) cells after treatment with 100 nM GSK1120212 at 1, 6, 24 and 72 h. Lysates were probed with antibodies against pAKT (Ser473), total AKT, pERK1/2 and total ERK1/2. g, Percentage cell viability (n = 3 independent experiments) of CAT12 cells after transfection with SOS1 and SOS2 siRNAs for 48 h followed by treatment with 100 nM GSK1120212 for 72 h. Cell viability was determined using CellTiterGlo cell viability assay. Data are mean ± s.d. Statistical analysis was performed using a two-tailed Student’s t-test; **P < 0.01. h, Representative western blot analysis (n = 2 independent experiments) of CAT12 cells engineered to express doxycycline-inducible pTRIPz shSOS or NTshRNA constructs. Knockdown of SOS was induced with 1 μM doxycycline for 48 h and treated with 100 nM GSK1120212 for 24 h. Protein lysates were collected and probed with antibodies against pAKT(Ser473), AKT, pERK1/2 and ERK1/2. a,c,f,h, Uncropped images of gels can be found in Supplementary Fig. 12. i, Left, tumor volumes of NOD/SCID mice injected subcutaneously with CAT12 cells expressing inducible shSOS or NTshRNA control constructs. When tumors were approximately 150 mm3, mice were initiated on a doxycycline-containing diet for 48 h before starting GSK1120212 (2 mg kg−1 day−1) or vehicle treatment for five weeks. Data are mean ± s.e.m. (n = 10 per treatment group). Statistical comparisons were made between vehicle and treatment groups using a two-tailed Student’s t-test with Welch’s correction; ***P < 0.001. Right, waterfall plot showing change in percentage tumor volume (compared to initial tumor volume) for individual tumors (each represented by a bar) following doxycycline induction and GSK1120212 or vehicle treatment for five weeks.

In addition, we investigated whether SOS inhibition sensitizes KRAS-amplified models to MEK inhibitor therapy. We observed that whereas SOS silencing had modest anti-proliferative effects in the KRAS-amplified models, silencing of SOS1 and SOS2 significantly sensitized KRAS-amplified cells to MEK inhibition (Fig. 3b). By contrast, the augmentation of the anti-proliferative effects of MEK blockade with SOS silencing in KRAS-mutant models was modest. Furthermore, we observed that the combination of MEK inhibition and SOS silencing led to significantly augmented induction of apoptosis in KRAS-amplified models (Supplementary Fig. 8a). Notably, in HUG1N cells, the KRAS-amplified GC line with the highest KRAS expression, loss of SOS1 and SOS2 alone led to decreased cell viability and increased apoptosis.

To enable long-term in vivo studies of SOS inhibition, we evaluated a previously validated doxycycline-inducible short hairpin RNA (shRNA) system that targets both SOS1 and SOS218 (Fig. 3c). We first tested the shRNA system in vitro, showing that inducible silencing of SOS in amplified KE39 cells decreased the pAKT adaptive response to MEK inhibition compared to non-targeting shRNA controls (Fig. 3c). In addition, inducible suppression of SOS significantly abrogated RAS–GTP induction after GSK1120212 treatment in the amplified cells compared to mutant cells (Fig. 3d and Supplementary Fig. 8b) and increased the anti-proliferative effects of GSK1120212 treatment in vitro (Supplementary Fig. 8c). We next evaluated MEK inhibition with genetic silencing of SOS in vivo. KE39 and GSU cells expressing the doxycycline-inducible shSOS or control shRNAs were injected subcutaneously into the flanks of immunocompromised mice and the mice were given a doxycycline-containing diet and treated with GSK1120212 (or vehicle control) once the tumors reached a size of approximately 150 mm3. In KRAS-amplified xenografts, SOS silencing significantly enhanced the efficacy of GSK1120212, consistent with in vitro results (Fig. 3e). By contrast, SOS did not impact the effects of GSK1120212 in KRAS-mutant tumors (Fig. 3e). Moreover, immunohistochemical analysis of KRAS-amplified xenografts showed upregulated pAKT expression in the GSK1120212-treated group, an effect that aws abrogated by SOS silencing (Supplementary Fig. 8d). Combination therapy also reduced Ki-67 staining in KRAS-amplified but not -mutant xenografts (Supplementary Fig. 8d).

We next validated our results in CAT12, a new GC cell line that is derived from a pleural effusion and has confirmed KRAS amplification (Supplementary Fig. 9a). Using in vitro cultures, we first validated MEK inhibitor insensitivity and adaptive pAKT induction after MEK treatment (Fig. 3f and Supplementary Fig. 9b). Silencing of SOS1 and SOS2 using pooled siRNAs sensitized CAT12 cells to GSK1120212 treatment in vitro (Fig. 3g). We then confirmed these results with shRNA, showing that silencing of SOS reduced adaptive increases in pAKT and RAS–GTP after 24 h following GSK1120212 treatment (Fig. 3h and Supplementary Fig. 9c). We then evaluated the impact of SOS silencing in vivo. Suppression of SOS in combination with GSK1120212 treatment in CAT12 xenografts led to tumor regression as well as decreased pAKT expression, thus validating our in vitro findings in two different in vivo KRAS-amplified models (Fig. 3i and Supplementary Fig. 9d). These positive results with SOS targeting are consistent with the role of KRAS activation as a mediator of resistance and suggest that targeting physiological mediators of RAS activation to be a candidate therapeutic approach for KRAS-amplified tumors.

Given the success of genetically targeting SOS to enhance MEK inhibition, we sought additional targetable mediators of physiological RAS activation. Most notable among the candidates was the protein tyrosine phosphatase SHP2, which promotes activation of RAS and MAPK signaling and its genetic silencing has been demonstrated to reverse intrinsic resistance to targeted therapies19. SHP099, a potent and selective allosteric inhibitor of the catalytic site of SHP2, was recently developed20,21. We investigated whether SHP099 may be able to block adaptive KRAS activation in KRAS-amplified models and whether the combination of GSK1120212 and SHP099 would be effective in KRAS-amplified models. In cell line models, SHP099 showed a strong effect in combination with GSK1120212 in inhibition of cell proliferation, preventing RAS–GTP activation and preventing adaptive pAKT induction in KRAS-amplified GC cells (Fig. 4a–c). Long-term clonogenic assays also showed synergistic effects of SHP099 with GSK1120212 in KRAS-amplified GC cells (Fig. 4d). We then evaluated SHP099 and GSK1120212 combinations in our isogenic amplified KRAS gastric organoid model system. Addition of SHP099 reversed resistance to GSK1120212 treatment in organoids with ectopic 3×KRAS expression (Fig. 4e). Because of the potency of MEK and SHP2 inhibition in vitro, we pursued in vivo testing. KRAS-amplified KE39 and CAT12 cells and KRAS-mutant GSU cells were subcutaneously injected into immunocompromised mice and the treatment was initiated when tumors were around 150 mm3, after which tumor progression was monitored. We observed no difference in tumor growth with the addition of SHP099 to GSK1120212 treatment in GSU xenografts (Fig. 4f). By contrast, the combination of SHP099 and GSK1120212 displayed a striking reduction in tumor growth and induced regression in both KE39 and CAT12 tumors (Fig. 4f). Moreover, histological analysis of KRAS-amplified xenografts confirmed blockade of GSK1120212-mediated pAKT induction with SHP099 therapy (Supplementary Fig. 10).

Fig. 4: Combination of SHP2 and MEK inhibition displays anti-tumor activity in wild-type KRAS-amplified gastric adenocarcinoma in vitro and in vivo.
figure4

a, Percentage cell viability (n = 3 independent experiments) of GC KRAS-amplified (KE39, YCC1 and HUG1N) and KRAS-mutant (GSU) lines after five days of treatment with GSK1120212 (100 nM), SHP099 (3, 5 or 10 μM), SHP099 (3, 5 or 10 μM) in combination with GSK1120212 (100 nM). DMSO was used as vehicle control. Cell viabilities (mean ± s.d) were normalized to the DMSO control group. Statistical comparisons between DMSO control and treatment groups were made using a two-tailed Student’s t-test with Welch’s correction; **P < 0.01. Each triangle in this figure represents a biological replicate/individual experiment except in b, where each triangle represents a technical replicate. b, Relative RAS–GTP levels (n = 2 independent experiments) of KRAS-amplified (KE39) and KRAS-mutant (GSU) GC cells after treatment with GSK1120212 (100 nM) or SHP099 (3, 5 or 10 μM) alone or in combination for 6 h. DMSO was used as a vehicle control. GTP-bound RAS is quantified by measuring absorbance in 96-well plates using the RAS G-LISA assay. Fold change in RAS–GTP levels is relative to the DMSO control group at each time point. Each data point represents mean fold change relative absorbance (n = 2 technical replicates). c, Representative western blot analysis (n = 2 independent experiments) of KRAS-amplified GC (KE39, YCC1 and HUG1N) lines after treatment with GSK1120212 (100 nM), SHP099 (3 or 5 μM), SHP099 (3 or 5 μM) in combination with GSK1120212 (100 nM) for 24 h. DMSO was used as vehicle control. Protein lysates were collected and probed with antibodies against pAKT (Ser473), AKT, pERK1/2, ERK1/2. β-Actin was used as loading control. Uncropped gel images can be found in Supplementary Fig. 12. d, Clonogenic assay of the KRAS-amplified GC line (KE39) treated with GSK1120212, SHP099 or both drugs combined as indicated after 14 days. Images are representative of two independent experiments. e, Representative phase contrast images (n = 2 independent experiments) of in vitro cultures of primary isogenic KRAS gastric organoids (Trp53−/−3×KRAS, control Trp53−/− empty vector, Trp53−/−KrasG12D/+3×KRAS, control Trp53−/−KrasG12D/+ empty vector) after five days of treatment with GSK1120212 (50 nM), SHP099 (5 μM) or both drugs combined. DMSO was used as vehicle control. Scale bar, 50 μM. f, Top, tumor volumes of NOD/SCID mice injected subcutaneously with KRAS-amplified GC (KE39/CAT12) or KRAS-mutant (GSU) cells. When tumors were around 150 mm3, mice were treated with vehicle, GSK1120212 (1 mg kg−1), SHP099 (50 mg kg−1) or both drugs in combination for either five weeks (the KE39 and CAT12 cell lines) or three weeks (the GSU cell line). Data are mean ± s.e.m. (n = 8 per treatment group). Statistical comparisons between vehicle and treatment groups were made using an unpaired, two-tailed t-test with Welch’s correction; *P < 0.05; **P < 0.01. Bottom, waterfall plots showing change in percentage tumor volume (compared to initial tumor volume) for individual tumors (each represented by a bar) treated with vehicle, GSK1120212, SHP099 or both drugs combined for either five or three weeks. g, Schematic of wild-type KRAS-amplified tumors at baseline (left) and after MEK inhibition (right). MEK inhibition leads to increased activation of KRAS–GTP and adaptive resistance that is mediated in part by upstream signaling. Adaptive responses can be abrogated through inhibition of SOS or SHP2, which can lead to inhibition of tumor growth when combined with MEK blockade.

The specific mechanisms of SHP2 in RAS and MAPK signaling have been an area of active study in the RAS field. SHP2 has been proposed to be a mediator of RAS and MAPK pathway activation by diverse mechanisms and at distinct points of RAS regulation22. Candidate mechanisms of SHP2 include promoting recruitment of the GAB2–GRB2–SOS complex to the plasma membrane, direct dephosphorylation of tyrosyl-phosphorylated RAS for RAS activation as well as indirect endomembrane RAS activation via SYK22,23,24. Furthermore, SHP2 has also been implicated in regulation of KRAS activation through modulation of activity of RAS GTPase proteins25. Because we were able to phenocopy our results with genetic SOS targeting with SHP2 inhibition, we next investigated whether we could rescue the inhibitory effect of SHP099 on GSK1120212-mediated adaptive responses in KRAS-amplified GC through SOS. We engineered cells expressing ectopic wild-type SOS1 or an active variant expressing only the catalytic domain of SOS (SOS-cat) but lacking the C-terminal GRB2-interacting domain or N-terminal Dbl-homology and pleckstrin homology domains26. We found that SOS-cat but not wild-type SOS could rescue the inhibition of cell viability, adaptive RAS–GTP increases as well as the adaptive pAKT induction that we observed with addition of SHP099 to GSK1120212 treatment (Supplementary Fig. 11a–c). Although these data cannot exclude the potential for SHP2 to also mediate RAS and MAPK activation via a GRB2–SOS adaptor complex-independent manner25,27, they are consistent with the hypothesized ability of SHP2 to facilitate SOS-mediated KRAS activation. We further analysed the effects of SHP2 inhibition on SOS function using confocal fluorescence microscopy to evaluate the localization of endogenous SOS1 in response to inhibitor treatment in KRAS-amplified and mutant GC cell lines. With MEK inhibition in the amplified KE39 cell line but not the mutant GSU model, we observed a shift in SOS1 localization that was abrogated with co-administration of SHP099 (Supplementary Fig. 11d), which is again consistent with a role of SOS in mediating adaptive KRAS activation with KRAS amplification. These data support our overarching hypothesis that SHP2 inhibition potentiates MEK inhibition by inhibiting adaptive KRAS activation and provide additional support that SHP2 can facilitate the ability of SOS to activate KRAS.

In summary, we demonstrate that KRAS also commonly acts as an oncogene after amplification of the wild-type gene and there are differences between the physiology of KRAS amplification and mutant KRAS that impact optimal targeted therapy (Fig. 4f, schematic). KRAS-amplified cell line models possess markedly increased basal levels of KRAS protein. This increased protein expression, likely aided by the lack of canonical somatic mutations that alter the KRAS equilibrium, creates a dynamic state with greater potential to mobilize KRAS–GTP, allowing adaptation to pharmacologic MAPK blockade. MAPK signaling induces negative feedback regulation on multiple signaling nodes, including RTKs and SHP2/SOS28,29,30. Loss of negative feedback, coupled with increased KRAS levels, thus enables rapid adaptation after MAPK inhibition. These results are consistent with earlier studies that have suggested that MEK inhibition is less effective in cancers lacking activating missense mutations of RAS (or RAF), because the loss of ERK-mediated negative feedback would enhance RAS activity31,32. Notably, a recent study in KRAS-mutant myeloid leukemia suggested that while the loss of wild-type KRAS increased clonal fitness, the gain of wild-type KRAS provided resistance to GSK112021233. This allelic imbalance of KRAS was found as a means of mediating resistance after tumor relapse, corroborating our idea that KRAS expression can affect therapeutic responses.

Further studies will need to refine our understanding of how the activation of overexpressed KRAS protein is regulated in cancers with KRAS amplification. Our results suggest that RTK signaling contributes to KRAS-driven adaptive resistance to MEK inhibition in KRAS-amplified cells. However, in lieu of upregulation of a single dominant RTK, our data also highlights the difficulty in successfully identifying prospective kinase combination therapies for KRAS-amplified cancers. Indeed, with the loss of ERK-mediated negative feedback, massive KRAS overexpression may serve as an ‘amplifier’, magnifying any upstream RTK signaling that is present. Recent studies that have demonstrated how MEK inhibition could induce transcriptional upregulation of receptor kinases or kinome remodeling to promote resistance34,35, add another layer of complexity to the potential of utilizing individual kinase inhibitors to overcome adaptive resistance to MEK therapy for KRAS-amplified tumors therapies.

We reasoned that targeting mechanisms of physiological RAS activation instead might provide a novel approach for the treatment of KRAS-amplified cancers. Recent studies have illuminated the dynamism of RAS activation, even among RAS mutants, which were traditionally thought to be predominantly in an active state36,37. GEFs, mostly SOS1 and SOS2, have already been under active evaluation as possible targets in RAS-mutant cancers, with candidate small molecules and peptide mimetics showing efficacy in pilot in vitro studies in RAS-mutant cancers38,39. One stated concern diminishing the enthusiasm for development and investment in these targets has been that these agents may have greater efficacy against wild-type RAS proteins than the mutant oncoprotein40. Accordingly, our recognition of the role of wild-type KRAS amplification presents an unrecognized and potential optimal context for inhibition of this target in cancer therapy. Furthermore, the new availability of allosteric inhibitors greatly enhances the feasibility of targeting SHP2 to circumvent adaptive resistance bypass to MEK inhibition in tumors with KRAS amplification. Although the mechanisms of SHP2-mediated regulation of KRAS activity have not been completely elucidated, our findings support the hypothesis that SHP2 and SOS do coordinate to promote RAS activation and strongly suggest the potential for combination therapy with SHP2 and MEK inhibition to be effective in KRAS-amplified cancers. Given the aggressive nature of these cancers and paucity of current therapeutic strategies, development of strategies to couple MAPK inhibition to SOS or SHP2 inhibition and further focused studies of the unique pathophysiology and intrinsic and adaptive signaling of wild-type KRAS amplification are clearly warranted.

Methods

Cell lines and culture conditions

IM95, YCC1, KE39, HUG1N, GSU and SNU1 cell lines were gifts from the Dana-Farber Cancer Institute Belfer Institute, which had obtained them directly from commercial sources and authenticated the lines using standard STR analysis. YCC1 cells were grown in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FBS. IM95 cells were grown in DMEM with 10% FBS and 10 mg l−1 recombinant insulin. KE39, HUG1N, SNU1 and GSU cells were grown in RPMI with 10% FBS. CAT12, a patient-derived xenograft, was established at the University of Chicago, following written patient consent, from a 72-year-old male presenting with iron-deficiency anemia and found to have a poorly differentiated cT3N3M1 esophagogastric adenocarcinoma with metastases to retroperitoneal, supraclavicular and mediastinal lymph nodes, as well as bilateral malignant pleural effusions. CAT12 was simultaneously established in vitro in RPMI with 10% FBS as a mixed suspended and adherent cell line. All cell lines were supplemented with 1 mM penicillin–streptomycin, 2 mM l-glutamine and maintained at 37 °C in a 5% CO2 incubator. All cells were routinely tested for mycoplasma and found to be free of contamination.

Genomic analysis

Summaries for KRAS copy-number alterations and mutation status in each tumor type were obtained from the Memorial Sloan-Kettering Cancer Center cBioPortal for Cancer Genomics (http://www.cbioportal.org/). The amplification frequency was defined as the ratio of patients with only KRAS amplification, excluding those with KRAS amplification and co-occurring mutations over the total number of patients with matched copy-number and/or sequencing data. For the correlation analysis of KRAS expression level and copy-number data, mRNA expression data were downloaded from the TCGA data portal (https://portal.gdc.cancer.gov), and the copy-number data were extracted from the Broad TCGA stomach adenocarcinoma copy-number dataset.

Survival analysis

A Japanese cohort of subjects with gastric adenocarcinoma with outcome analyses that were associated with wild-type KRAS amplification and non-amplified KRAS amplification was tested. Subject data for this Japanese cohort have been published previously41. The Kaplan–Meier method was used to analyze survival probabilities. Survival probabilities were defined as the subject entry into the study until five years after surgery. Comparisons between KRAS amplification and no KRAS amplification and survival probabilities were made using the log-rank (Mantel–Cox) test. These results were validated on cohorts of subjects with gastroesophageal adenocarcinoma from the University of Chicago and from the TCGA stomach cancer cohort, both of which had outcome analyses that were associated with high and low KRAS expression. Subject data for the Chicago cohort and TCGA stomach adenocarcinoma have been published previously42,43. The Kaplan–Meier method was used to analyze the percentage of disease-free and overall survival and these analyses were defined as subject entry into the study until 120 months after surgery. Comparisons between cases with KRAS alteration and without KRAS alteration and the outcome analyses were made using the log-rank (Mantel–Cox) test.

KRAS immunohistochemistry

After deparaffinizing tissue blocks, antigen retrieval was achieved using a wet autoclave (121 °C, 15 min) in antigen retrieval solution, pH 6 (DAKO, S2031, Glostrup, Denmark). In order to block endogenous peroxide enzyme, tissue sections were incubated for 30 min using peroxidase-blocking solution (DAKO, S2023). A primary antibody specific to KRAS (1:100, 415700, Life Technologies, Carlsbad, CA, USA) was applied, and slides were incubated overnight at 4 °C. Visualization was achieved using EnVision+/HRP, (for mouse, DAKO, K4001) and hematoxylin counterstaining.

FISH

Two sets of FISH analysis were performed as follows.

Samples from the Japanese cohort

Dual-color FISH was performed on formalin-fixed paraffin-embedded (FFPE) tissue. After deparaffinization and dehydration, the sections of FFPE tissue were digested in 0.1 N HCl for 20–30 min and washed in phosphate-buffered saline (PBS) for 5 min at room temperature. The FISH KRAS probe was labeled with bacterial artificial chromosomes (BACs) RP11-636P12, RP11-62I19 and RP11-65C2, which were labeled with Cy3 (Chromosomescience laboratory, Sappro, Japan). FISH chromosome 12 centromere (CEN12) was labeled with BACs RP11-267D19, RP11-792O21 and RP11-8P13, which were labeled with Cy5 (Chromosomescience laboratory, Sappro, Japan). After dehydration and drying, each FISH probe was applied to each targeted area, and the slides were sealed with coverslips. The section was denatured at 90 °C for 10 min, followed by overnight hybridization at 37 °C in a wet chamber. Hybridized slides were washed in 2× saline–sodium citrate buffer (SSC) for 5 min and coverslips were removed gently. The slides were stringently washed in 50% formamide and 2× SSC for 20 min at 37 °C, and the kept in 1× SSC for 15 min at room temperature. After post-hybridization washing, the slides were counterstained with DAPI. The FISH images were captured with a fluorescence microscope (BZ-X710, Keyence, Japan).

The University of Chicago sample cohort and CAT12 cell line

Dual-color FISH assays using KRAS (BAC clone CTD-2060B1; 12p12.1), with corresponding chromosome enumeration probe CEP12 (pBR12 alpha satellite 12 control clone), were performed as previously described44. Interpretation of FISH was performed as previously described44, with amplification defined as a KRAS:CEP12 ratio ≥2.

RAS–GTP pull-down assay

The RAS–GTP assay was carried out using the RAS Activation Assay kit (Millipore Sigma, Burlington, MA, USA; #17-218) according to the manufacturer’s instructions. Cells were washed twice with cold PBS and lysed with Mg2+ lysis buffer supplemented with a protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktails (Calbiochem). Activated RAS was precipitated with purified GST–RAF–RBD agarose beads by pre-incubating 1 mg of whole-cell lysates with GST–RAF–RBD pre-bound to gluthiathione–sepharose. Bound RAS was subjected to SDS–PAGE electrophoresis and immunoblotting analysis using anti-KRAS antibody (Millipore).

RAS G-LISA assay

The RAS G-LISA assay was performed using the RAS G-LISA activation (absorbance-based) kit (Cytoskeleton, Inc, Denver, CO, USA; #BK131) according to the manufacturer’s instructions. 12.5 μg of whole-cell lysates was added in duplicate in a 96-well plate and activated RAS was bound to a RAS–GTP-binding protein linked to each well. Bound, active RAS was detected with a RAS-specific antibody and quantified by measuring the relative absorbance at 490 nM using a Molecular Devices SpectralMax M5 plate reader.

Lentiviral infection

Lentivirus was generated using standard protocols. In brief, HEK-293T cells were plated in 6-cm2 plates with fresh medium without antibiotics. Subsequently, 1 μg of the lentiviral vector, 100 ng of the envelope plasmid and 900 ng of the packaging plasmid were diluted in Opti-MEM (Thermo Fisher Scientific, Cambridge, MA, USA; #31985070) and 6 μl of X-tremeGene 9 DNA transfection reagent (Millipore Sigma; #036335779001) was added dropwise and this mixture was incubated for 20 min. The DNA complexes were added dropwise to the cells and incubated for 12 h before aspirating and addition of 6 ml of fresh medium. After 24 h, the virus-containing medium was collected and filtered through a 0.45-μm syringe and the lentivirus was stored at −80 °C. The non-targeting doxycycline-inducible pTRIPz shRNA control vector was purchased from Dharmacon (Lafayette, CO, USA). Constructs used for RNAi were obtained from pLenti CMV/TO vector (gift from D. Bar-Sagi, New York University) and cloned into the doxycycline-inducible pTRIPz vector according to the manufacturer’s instructions (Dharmacon). The oligonucleotide sequences used for cloning are available in the Supplementary Methods. For shSOS knockdown experiments, 2 × 105 cells were plated in a 6-well plate and infected with pTRIPz NTshRNA or shSOS constructs and selected with puromycin (1 μg ml−1) for seven days.

RNAi

siRNA experiments

1 × 105 cells were plated in 6-well plates in duplicate and knockdown of KRAS was performed with 10 nM or 20 nM of two independent KRAS siRNAs (Ambion, Life Technologies); knockdown of SOS1 and SOS2 was performed using 20 nM ON-TARGETplus SMARTpool human siRNAs (Thermo Fisher Scientific). A full list of oligos is available in the Supplementary Methods. ON-TARGETplus SmartPool Non-Targeting human siRNAs were used as negative controls. siRNA transfections were performed using Lipofectamine RNAiMAX (Invitrogen).

shRNA experiments

1 × 105 cells were plated in 6-well plates in duplicate and knockdown of SOS was induced using 1 μM of doxycycline for 48 h before treatment with 100 nM of GSK1120212 or DMSO for 24 h.

Mouse colonies

Mice with Trp53flox/flox and KrasLSL-G12D/+ alleles were provided by K.-K.W., and crossed to generate Trp53flox/flox;KrasLSL-G12D/+ mice. All animal experiments were performed in accordance with the Dana-Farber Cancer Institute’s Institutional Animal Care and Use Committee–approved animal protocols.

Culture of mouse gastric organoids

The procedures for establishing and maintaining mouse gastric organoids were based on previously reported protocols45. In brief, the stomach from an adult mouse of the appropriate genotype was collected, opened lengthwise, and washed in cold PBS. A 0.5-cm segment was minced extensively on ice and digested in 1 ml of collagenase solution (2 mg ml−1 collagenase type I (Millipore Sigma) and 50 μg ml−1 gentamicin in washing medium (1× PBS)) for 30 min with pipetting every 10 min. Gastric glands were filtered through a 70-μm cell strainer, mixed with 9 ml cold washing medium and pelleted by centrifugation. The supernatant was carefully removed and the pellet was resuspended with Matrigel (Corning, Corning, NY, USA) and seeded in a 24-well plate (30 μl per well) in 500 μl 50% L-WRN-conditioned medium (a 50/50 mix of L-WRN conditioned medium and Advanced DMEM/F-12 with 20% FBS, supplemented with 1 mM penicillin–streptomycin, 2 mM l-glutamine) and maintained at 37 °C in a 5% CO2 incubator. The L-WRN cell line was a gift from T. Stappenbeck, Washington University in St. Louis. Organoids were passaged every 4–5 days by dissociation into single cells with TryPLE Express (Millipore Sigma) at 37 °C for 15 min while vortexing every 5 min, followed by vigorous pipetting. Single cells were pelleted, suspended in Matrigel and replated at a 1:6 split into a new 24-well plate for maintenance, or counted for assays.

Viral infection of mouse organoids

For adenoviral infection of gastric organoids, Trp53flox/flox or Trp53flox/floxKrasLSL-G12D/+ organoids were plated in a 24-well plate, incubated directly with 500μl 50% L-WRN conditioned medium containing 108 plaque-forming units. AdCMVCre (University of Iowa), and selected with 10 μM nutlin-3 (Cayman Chemical, Ann Arbor, MI, USA) and 1 μM erlotinib (Selleckchem, Houston, TX, USA) for 7–14 days to generate Trp53−/− or Trp53−/−KrasG12D/+ organoids. The Gateway-compatible pLX324 lentiviral vector was derived from the pLX304 vector (a gift from D. Root, Broad Institute) in which the CMV promoter was changed to the EF1a promoter to avoid its potential silencing in mammalian cells and ensure stable expression. The fragment containing three different human codon-optimized KRAS cDNAs linked by IRES2 was synthesized (GENEWIZ) and cloned into the pLX324 vector. The fragment sequences are available in the Supplementary Methods. For lentiviral infection of Trp53−/− or Trp53−/−KrasG12D/+ gastric organoids, organoids at 3–5 days of growth were collected and incubated with TrypLE Express to dissociate into single cells. Cells were plated in a 48-well plate and centrifuged with 200 μl 50% L-WRN conditioned medium containing pLX324-Cre or pLX324–3×KRAS constructs at 32 °C, 600g, for 1 h in the presence of 10 μM Y-27632 (Enzo Life Sciences, Farmingdale, NY, USA). The plate was then incubated at 37 °C for 6 h before seeding in a 24-well plate. 48 h after infection, the infected organoids were selected with blasticidin (1 μg ml) for seven days.

Cell viability assay

KRAS siRNA proliferation studies

2,500 cells were plated in 96-well plates in triplicate for each siRNA condition and transfected with 10 nM or 20 nM KRAS siRNA or non-targeting (NT)siRNA. Cell viability was measured at 24-h intervals after 72 h transfection as shown. For each siRNA condition, proliferation curves were expressed as fold change difference to 72-h time point.

siRNA transfections and inhibitor studies

2,500 cells were plated in 96-well plates in triplicate for each siRNA condition for 48 h and subsequently treated with either DMSO or 100 nM GSK1120212 for additional 72 h.

For inhibitor assays

5,000 cells were plated in 96-well plates in triplicate and treated with inhibitor for 72 or 120 h. Cell viability was quantified by measuring cellular ATP content using the CellTiterGlo Cell Viability assay (Promega) according to the manufacturer’s instructions.

Organoid drug inhibition studies

Western blot analysis of organoids after inhibitor treatment

Gastric organoids at 3–5 days of growth were collected and dissociated into single cells using TrypLE Express. Cells were plated at a density of 20,000 cells per 30 μl (6 drops per well) in a 6-well plate in 2 ml 50% L-WRN conditioned medium and allowed to grow for three days before treatment with DMSO or 50 nM GSK1120212 for 1, 6, and 24 h. The Matrigel that was surrounding the organoids was removed using Cell Recovery Solution (Corning). The released organoids were then pelleted and lysed as previous described2.

Growth of organoids with inhibitor treatment

Gastric organoids were dissociated and plated at a density of 3,000 cells per 30 μl (1 drop per well) in duplicate in a 24-well plate in 500 μl 50% L-WRN conditioned medium and treated with DMSO, 50 nM GSK1120212, and/or 5 μM SHP099 the following day. Bright-field pictures showing organoid growth were taken five days after treatment using a Nikon Eclipse TE2000 inverted microscope.

Antibodies and inhibitors

Western blot analysis was performed as previously described4. Primary antibodies against phosphorylated ERK1 and ERK2 (pERK1/2) T202/Y204 (#4370), total ERK1/2 (#4695), pAKT Ser473 (#4060), total AKT (#9272), pERBB3 Y1222 (#4784), total HER3 (#4754), pIGF1Rβ Y1135 (#2918), total IGF1Rβ (#3027) and SOS1 (#5890) were purchased from Cell Signaling Technologies (Danvers, MA, USA). The primary SOS2 (#PAS-35070) antibody was purchased from Pierce Protein Biology, Thermo Fisher Scientific. Primary KRAS (Ras10 clone; #17-218) and KRAS (#F234-sc30) antibodies were purchased from Millipore Sigma and Santa Cruz Biotechnology (Santa Cruz, CA, USA), respectively. Primary antibodies against HRAS (C20: sc-520) and NRAS (F155: sc-31) were purchased from Santa Cruz Biotechnology. The anti-β-actin antibody was purchased from Sigma-Aldrich. Horseradish peroxidase-conjugated secondary antibodies (anti-rabbit and anti-mouse) were purchased from Pierce, and Amersham ECL Prime chemiluminescent detection reagent (GE Healthcare Life Sciences) was used to visualize protein expression. Fluorescently conjugated DyLight Fluor (680 and 800) secondary antibodies (anti-rabbit and anti-mouse) were purchased from Thermo Fisher Scientific and a Licor Odyssey Imaging scanner was used to visualize protein expression. AZD6244, GSK1120212, afatinib, GDC0941, OSI-906, crizotinib, BGJ-398 and MK2206 were purchased from Selleckchem, SHP099 was a gift from Novartis Institutes of Biomedical Research and dissolved in DMSO before use.

RNA isolation and qPCR

Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Germantown, MD, USA) and cDNA was synthesized using the Taqman Reverse Transcription Reagents kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Gene-specific primers for SYBR Green real-time PCR were designed by SNAPGENE software and synthesized by Integrated DNA Technologies. Real-time PCR was performed and analyzed using ABI PRISM 7000 sequence detection system software (PE Applied Biosystems) and using Power SYBR Green PCR Master Mix (Thermo Fisher Scientific) according to the manufacturer’s instructions. Relative mRNA expression was determined by normalizing to PPIA expression, which served as an internal control.

Apoptosis assays

Inhibitor assays

2 × 105 cells were plated in duplicate in 6-well plates for each treatment group and treated with either DMSO or inhibitor for 72 h.

siRNA transfections followed by treatment with inhibitors

1 × 105 cells were plated in 6-well plates in duplicate and transfected with 10 nM KRAS siRNA or 20 nM SOS1 and SOS2 siRNA for 48 h. Subsequently, transfected cells were treated with either DMSO or 100 nM GSK1120212 for an additional 72 h. After 72 h, both floating and adherent cells were collected, washed twice with ice-cold PBS and stained using the BD Pharmingen FITC annexin-V Apoptosis Detection kit (BD Biosciences, Billerica, MA, USA) according to the manufacturer’s instructions. Cells undergoing early and late apoptosis were analyzed using a BD LSRFortessa (BD Biosciences) and 10,000 live events per treatment group were collected for analysis.

Mouse xenograft studies

All animal experiments were conducted in accordance with Institutional Animal Care and Use Committee–approved animal protocols at the Dana-Farber Cancer Institute in compliance with NIH guidelines. For xenograft studies, cells were prepared in 1:1 Matrigel:medium ratio and 1–5 × 106 cells were injected subcutaneously into the flanks of NOD/SCID (NOD.CB17/PrkdcSCID/J) female mice (6–8 weeks old) acquired from the Jackson Laboratory. Tumors were palpable in approximately 1–3 weeks. Tumors were measured using electronic calipers and tumor volumes were calculated using the formula, volume = length × width2 × 0.5.

SOS shRNA xenograft studies

The doxycycline-containing diet was initiated for 48 h before drug treatment. GSK112012 (2 mg kg−1) or vehicle (0.5% hydroxypropylmethylcellulose, 0.1% Tween-80) was delivered daily using oral gavage and tumor sizes were monitored.

SHP099/GSK1120212 xenograft studies

GSK1120212 (1 mg kg−1) was delivered daily for one week followed by treatment every other day; SHP099 (50 mg/kg) or vehicle was delivered daily by oral gavage and tumor sizes were monitored. Mice with palpable tumors were randomized into drug treatment groups using covariate-adaptive randomization in order to reduce baseline differences in tumor volumes. Investigators performing the study were not blinded to treatment groups. Sample size (n = 10 per treatment group) was chosen for satisfactory interanimal reproducibility. Tumor volumes were measured twice weekly. When tumors reached a maximum of 1,500 mm3, mice were euthanized and tumors were excised and snap-frozen.

KRAS selected reaction monitoring (SRM) assay development

In silico trypsin digestion mapping of the KRAS protein sequence (UniProtKB accession number P01116) was used to identify unique peptides for KRAS SRM assay development. Peptides containing methionine or cysteine residues were excluded due to their propensity to undergo unpredictable oxidation. The peptide SFEDIHHYR comprising residues 89–97 was found to be the only tryptic peptide that was unique to KRAS by comparing this sequence to the entire human proteome using the BLASTp function within the BLAST search engine (http://blast.ncbi.nlm.nih.gov/Blast.cgi), and sequence analysis using the phosphosite database (https://www.phosphosite.org/homeAction.action). Consequently, unlabeled (SFEDIHHYR) and isotopically labeled (SFEDIHHYR (13C6,15N4)) versions of this peptide were synthesized to develop and perform the assay (Thermo Scientific, San Jose, CA, USA). SRM transitions used for the quantification of the unlabeled KRAS peptide were 401.86/475.24 (y3+1), 485.23 (y7+2), and 558.76 (y8+2) (Q1/Q3) and the transitions used for the isotopically labeled internal standard were 405.19/485.25 (y3+1), 490.23 (y7+2) and 563.77 (y8+2) (Q1/Q3). Instrumental analyses were performed on TSQ Vantage or TSQ Quantiva triple quadrupole mass spectrometers (Thermo Scientific, San Jose, CA, USA) equipped with a nanoAcquityLC system (Waters, Milford, MA, USA), as previously described46.

Mass spectrometry quantification of KRAS expression in tumors

Retrospective gastroesophageal adenocarcinoma patient samples, with annotated clinical and pathological information, were obtained from the University of Chicago (Chicago, IL, USA) under Institutional Review Board–approved tissue bank protocols, as previously described42. KRAS protein was quantified by SRM–mass spectometry (MS) as previously described46,47. In brief, tissue sections (10 μM) from FFPE blocks were placed onto DIRECTOR microdissection slides followed by deparaffinization and hematoxylin staining. Tumor areas were marked by a board-certified pathologist and were microdissected and solubilized to tryptic peptides using liquid tissue technology. The solution was subjected to SRM–MS analysis using stable isotope-labeled internal standard peptides for KRAS quantification. The assay was monitored using actin and tubulin quantification as internal control to verify sample quality and efficiency of microdissection. On-column injection resulted in 5 fmol of isotopically labeled internal standard peptides and 1 μg of total tumor protein as measured by microBCA (ThermoFisher Scientific, San Jose, CA, USA).

siRNA oligonucleotides

The following siRNA oligonucleotides were purchased from GE Healthcare Biosciences/Dharmacon or Ambion/Life Technologies (Supplementary Table 2).

shRNA oligonucleotide sequences

The following oligonucleotide sequence using for cloning into the tetracycline-inducible pTRIPz vector (shSOS) targets human SOS1 and SOS2 and was previously published18.

shSOS: 5′-GACAGTGTTGTAATGAATT-3′.

3×KRAS fragment sequence

Note, IRES (internal ribosome entry site) sequences were placed between each of the codon optimized KRAS sequences. KRAS human codon-optimized sequence #1: 5′-ATGACCGAGTATAAACTGGTGGTCGT GGGCGCTGGCGGAGTGGGCAAATCCG CTCTGACCATCCAGCTGATCCAGA ACCACTTCGTCGATGAGTACGATC CCACCATCGAGGACTCCTATA GGAAACAAGTGGTGATCGA TGGCGAGACCTGTCTGCTCGAC ATCCTGGATACAGCCGGACAGG AGGAGTACTCCGCCATGAGGGAC CAGTATATGAGAACCGGAGAGGG CTTCCTCTGCGTGTTCGCCAT CAACAACACCAAAAGCTTTGAGG ACATCCACCACTACAGGGAACAG ATCAAGAGGGTGAAAGATAGCG AAGATGTGCCCATGGTCCTGGT CGGCAACAAGTGCGATCTGCCC AGCAGAACCGTGGACACCAAGC AGGCTCAGGACCTGGCCAGAAG CTATGGCATCCCCTTCATCGAA ACCAGCGCCAAGACCAGGCAGG GAGTGGACGACGCCTTCTACAC ACTGGTTCGAGAAATTCGA AAACATAAAGAAAAGATG AGCAAAGATGGTAAAAAGA AGAAAAAGAAGTCAAAGACAA AGTGTGTAATTATGTAA-3′. KRAS human codon-optimized sequence 2: 5′-ATGACAGAGTACAAGCTCGTGGT GGTGGGCGCTGGAGGAGTGGGCAAG AGCGCCCTGACCATCCAACTGATC CAAAACCACTTCGTGGACGAATACG ACCCCACCATCGAAGACTCCTACA GGAAGCAGGTGGTGATCGACGGAGA AACCTGTCTGCTGGACATCCTGGAC ACAGCCGGCCAGGAAGAGTACAGC GCCATGAGGGATCAGTACATGAGG ACCGGCGAGGGCTTCCTGTGCGTG TTCGCTATCAATAACACAAAGAGC TTCGAGGACATTCACCACTATAGG GAGCAGATCAAAAGGGTGAAGGAC AGCGAGGACGTGCCCATGGTGCT GGTGGGCAATAAGTGTGACCTGC CCAGCAGGACCGTGGACACAAAGC AGGCCCAGGATCTGGCCAGGTCCT ACGGCATCCCCTTTATCGAGACAT CCGCCAAGACAAGGCAAGGCGTGG ACGATGCCTTTTACACACTGGTTC GAGAAATTCGAAAACATAAAGAAA AGATGAGCAAAGATGGTAAAAAGAA GAAAAAGAAGTCAAAGACAAAGTG TGTAATTATGTAA-3′. KRAS human codon-optimized sequence #3: 5′ATGACGGAATATAAGCTTGTGGTG GTGGGCGCTGGTGGCGTGGGAAAG AGTGCCCTGACCATCCAGCTGATC CAGAACCACTTTGTGGACGAATACG ACCCCACTATAGAGGATTCCTACCG GAAGCAGGTGGTCATTGATGGGGAG ACGTGCCTGTTGGACATCCTGGATA CCGCCGGCCAGGAGGAGTACAGCGC CATGCGGGACCAGTACATGCGCACC GGGGAGGGCTTCCTGTGTGTGTTTG CCATCAACAACACCAAGTCTTTTGA GGACATCCACCATTACAGGGAGCAG ATCAAACGGGTGAAGGACTCGGAGG ACGTGCCCATGGTGCTGGTGGGGAA CAAGTGTGACCTGCCTTCACGCACT GTGGACACTAAGCAGGCTCAGGACC TCGCCCGAAGCTACGGCATCCCCTTC ATCGAGACCTCGGCCAAGACCCGGC AGGGAGTGGATGATGCCTTCTACACA CTAGTTCGAGAAATTCGAAAACATAA AGAAAAGATGAGCAAAGATGGTAAAA AGAAGAAAAAGAAGTCAAAG ACAAAGTGTGTAATTATGTAA-3′.

RTK antibody arrays

To identify relative expression of phosphorylation of RTKs, a PathScan RTK antibody (Fluorescent) array (Cell Signaling Technologies; #7949) was carried out according to the manufacturer’s instructions.

EGF stimulation studies

5 × 106 cells were plated in 10-cm2 plates and serum-starved overnight. Cells were treated with 50 ng ml−1 human recombinant EGF (R&D Systems, Minneapolis, MN, USA). Protein lysates were collected 0, 5, 15, 30 and 60 min after EGF stimulation. Cells were washed twice with cold PBS and lysed with Mg2+ lysis buffer supplemented with a protease inhibitor cocktail (Millipore Sigma) and phosphatase inhibitor cocktail (Millipore Sigma). The RAS–GTP assay was carried out using the RAS Activation Assay kit (Millipore) according to the manufacturer’s instructions. Activated RAS was precipitated with purified GST–RAF–RBD agarose beads by pre-incubating 1 mg of whole-cell lysates with GST–RAF–RBD pre-bound to gluthiathione–sepharose. Bound RAS was subjected to SDS–PAGE electrophoresis and immunoblotting analysis using anti-KRAS antibody (Millipore).

Immunohistochemistry and antibodies

Xenograft tumors were excised and fixed with 10% formalin overnight and embedded in paraffin (FFPE). Unstained sections were stained with the following antibodies purchased from Cell Signaling Technology: SOS1 (#5890), pAKT Ser473 (#4060), pERK1/2 (#4370). The staining kit for Ki-67 (Vector #VP-K451) was used following the manufacturer’s instructions. Representative images were taken using a Leica DM1000 LED light microscope camera.

Generation of SOS1 and SOS1-cat stable cell lines

Human SOS1 and SOS1-cat open reading frames were PCR-amplified from pCGN-SOS1 and pCGN HA-SOS1-cat (gift from D. Bar-Sagi, New York University; Addgene plasmids #32920 and #23917) and cloned into the Gateway-compatible lentiviral vector pLX304 (Addgene plasmid #25890) according to the manufacturer’s instructions (Invitrogen). Lentivirus was generated using standard protocols. For SOS1 and SOS1-cat rescue experiments, 2 × 105 cells were plated in a 6-well plate and infected with pLX304 empty vector, SOS1 or hemagglutinin (HA)-tagged SOS1-cat constructs and selected with blasticidin (5 μg ml−1) for seven days.

Confocal immunofluorescence microscopy

For SHP099 and GSK1120212 inhibition experiments, 2 × 104 cells were plated in 4-well chamber glass slides and grown overnight before treatment with DMSO or inhibitors for 6 h. Cells were then fixed in 4% PFA in PBS for 15 min and permeabilized with 0.1% Triton X-100 in PBS for 5 min at room temperature and subsequently incubated with anti-SOS1 antibody (CST#5890; 1:100) overnight at 4 °C. The next day, cells were washed three times with PBS and incubated with goat anti-rabbit Alexa Fluor 555 antibody (Invitrogen, Carlsbad, CA, USA; #A-21428; 1:500) for 1 h at room temperature. The cells were then washed three times with PBS and mounted in Vectashield Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA, USA). Images were acquired on a Zeiss LSM510 confocal microscope with a 63× oil-immersion objective.

Statistical analysis and reproducibility

Experiments were performed in triplicate. Data are represented as mean ± s.d unless indicated otherwise. For each experiment, either independent biological experiments or technical replicates are as noted in the figure legends and were repeated with similar results. Statistical analysis was performed using Microsoft Office statistical tools or in Prism 7.0 (GraphPad). Pairwise comparisons between groups (that is, experimental versus control) were performed using an unpaired two-tailed Student’s t-test or Kruskal–Wallis test as appropriate. P < 0.05 was considered to be statistically significant. For all experiments, the variance between comparison groups was found to be equivalent. For xenograft experiments, data are displayed as mean ± s.e.m. and statistical comparisons were performed using unpaired, two-tailed Student’s t-tests with Welch’s correction. Sample sizes and animal numbers were determined from pilot laboratory experiments and previously published literature. Animals were excluded from analysis if they were euthanized due to health reasons unrelated to tumor volume end point. For in vivo experiments, all mice were randomized before studies.

Reporting Summary

Further information on experimental design is available in the Nature Research Reporting Summary linked to this article.

Data availability

All data generated or analyzed during this study are included in this published article (and its Supplementary Information).

Methods

Methods, including statements of data availability and any associated accession codes and references, are available at https://doi.org/10.1038/s41591-018-0022-x.

Change history

  • 09 August 2018

    In the Supplementary Information originally published with this article, a lane was missing in the β-actin blot in Supplementary Fig. 2. The lane has been added. The error has been corrected in the Supplementary Information associated with this article.

References

  1. 1.

    Cox, A. D., Fesik, S. W., Kimmelman, A. C., Luo, J. & Der, C. J. Drugging the undruggable RAS: mission possible? Nat. Rev. Drug Discov. 13, 828–851 (2014).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  2. 2.

    Ross, J. S. et al. Comprehensive genomic profiling of epithelial ovarian cancer by next generation sequencing-based diagnostic assay reveals new routes to targeted therapies. Gynecol. Oncol. 130, 554–559 (2013).

    Article  PubMed  CAS  Google Scholar 

  3. 3.

    Dulak, A. M. et al. Exome and whole-genome sequencing of esophageal adenocarcinoma identifies recurrent driver events and mutational complexity. Nat. Genet. 45, 478–486 (2013).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  4. 4.

    Dulak, A. M. et al. Gastrointestinal adenocarcinomas of the esophagus, stomach, and colon exhibit distinct patterns of genome instability and oncogenesis. Cancer Res. 72, 4383–4393 (2012).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  5. 5.

    Chen, Y. et al. Identification of druggable cancer driver genes amplified across TCGA datasets. PloS ONE 9, e98293 (2014).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  6. 6.

    Das, K. et al. Mutually exclusive FGFR2, HER2, and KRAS gene amplifications in gastric cancer revealed by multicolour FISH. Cancer Lett. 353, 167–175 (2014).

    Article  PubMed  CAS  Google Scholar 

  7. 7.

    Birkeland, E. et al. KRAS gene amplification and overexpression but not mutation associates with aggressive and metastatic endometrial cancer. Br. J. Cancer 107, 1997–2004 (2012).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  8. 8.

    Pulciani, S., Santos, E., Long, L. K., Sorrentino, V. & Barbacid, M. ras gene amplification and malignant transformation. Mol. Cell. Biol. 5, 2836–2841 (1985).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  9. 9.

    Ahronian, L. G. et al. Clinical acquired resistance to RAF inhibitor combinations in BRAF-mutant colorectal cancer through MAPK pathway alterations. Cancer Discov. 5, 358–367 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  10. 10.

    Cargnelutti, M. et al. Activation of RAS family members confers resistance to ROS1 targeting drugs. Oncotarget 6, 5182–5194 (2015).

    Article  PubMed  Google Scholar 

  11. 11.

    Oddo, D. et al. Molecular landscape of acquired resistance to targeted therapy combinations in BRAF-mutant colorectal cancer. Cancer Res. 76, 4504–4515 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  12. 12.

    Valtorta, E. et al. KRAS gene amplification in colorectal cancer and impact on response to EGFR-targeted therapy. Int. J. Cancer 133, 1259–1265 (2013).

    Article  PubMed  CAS  Google Scholar 

  13. 13.

    Cox, A. D. & Der, C. J. Ras history: The saga continues. Small GTPases 1, 2–27 (2010).

    Article  PubMed  PubMed Central  Google Scholar 

  14. 14.

    Barretina, J. et al. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483, 603–607 (2012).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  15. 15.

    Jokinen, E. & Koivunen, J. P. MEK and PI3K inhibition in solid tumors: rationale and evidence to date. Ther. Adv. Med. Oncol. 7, 170–180 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  16. 16.

    Sun, C. et al. Intrinsic resistance to MEK inhibition in KRAS mutant lung and colon cancer through transcriptional induction of ERBB3. Cell Rep. 7, 86–93 (2014).

    Article  PubMed  CAS  Google Scholar 

  17. 17.

    Vigil, D., Cherfils, J., Rossman, K. L. & Der, C. J. Ras superfamily GEFs and GAPs: validated and tractable targets for cancer therapy? Nat. Rev. Cancer 10, 842–857 (2010).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  18. 18.

    Jeng, H. H., Taylor, L. J. & Bar-Sagi, D. Sos-mediated cross-activation of wild-type Ras by oncogenic Ras is essential for tumorigenesis. Nat. Commun. 3, 1168 (2012).

    Article  PubMed  PubMed Central  Google Scholar 

  19. 19.

    Prahallad, A. et al. PTPN11 is a central node in intrinsic and acquired resistance to targeted cancer drugs. Cell Rep. 12, 1978–1985 (2015).

    Article  PubMed  CAS  Google Scholar 

  20. 20.

    Chen, Y. N. et al. Allosteric inhibition of SHP2 phosphatase inhibits cancers driven by receptor tyrosine kinases. Nature 535, 148–152 (2016).

    Article  PubMed  CAS  Google Scholar 

  21. 21.

    Garcia Fortanet, J. et al. Allosteric inhibition of SHP2: identification of a potent, selective, and orally efficacious phosphatase inhibitor. J. Med. Chem. 59, 7773–7782 (2016).

    Article  PubMed  CAS  Google Scholar 

  22. 22.

    Dance, M., Montagner, A., Salles, J. P., Yart, A. & Raynal, P. The molecular functions of Shp2 in the Ras/Mitogen-activated protein kinase (ERK1/2) pathway. Cell. Signal. 20, 453–459 (2008).

    Article  PubMed  CAS  Google Scholar 

  23. 23.

    Bunda, S. et al. Inhibition of SHP2-mediated dephosphorylation of Ras suppresses oncogenesis. Nature Commun. 6, 8859 (2015).

    Article  CAS  Google Scholar 

  24. 24.

    Zhang, S. Q. et al. Shp2 regulates SRC family kinase activity and Ras/Erk activation by controlling Csk recruitment. Mol. Cell 13, 341–355 (2004).

    Article  PubMed  Google Scholar 

  25. 25.

    Agazie, Y. M. & Hayman, M. J. Molecular mechanism for a role of SHP2 in epidermal growth factor receptor signaling. Mol. Cell. Biol. 23, 7875–7886 (2003).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  26. 26.

    Boykevisch, S. et al. Regulation of ras signaling dynamics by Sos-mediated positive feedback. Curr. Biol. 16, 2173–2179 (2006).

    Article  PubMed  CAS  Google Scholar 

  27. 27.

    Araki, T., Nawa, H. & Neel, B. G. Tyrosyl phosphorylation of Shp2 is required for normal ERK activation in response to some, but not all, growth factors. J. Biol. Chem. 278, 41677–41684 (2003).

    Article  PubMed  CAS  Google Scholar 

  28. 28.

    Buday, L., Warne, P. H. & Downward, J. Downregulation of the Ras activation pathway by MAP kinase phosphorylation of Sos. Oncogene 11, 1327–1331 (1995).

    PubMed  CAS  Google Scholar 

  29. 29.

    Kamioka, Y., Yasuda, S., Fujita, Y., Aoki, K. & Matsuda, M. Multiple decisive phosphorylation sites for the negative feedback regulation of SOS1 via ERK. J. Biol. Chem. 285, 33540–33548 (2010).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  30. 30.

    Porfiri, E. & McCormick, F. Regulation of epidermal growth factor receptor signaling by phosphorylation of the ras exchange factor hSOS1. J. Biol. Chem. 271, 5871–5877 (1996).

    Article  PubMed  CAS  Google Scholar 

  31. 31.

    Manchado, E. et al. A combinatorial strategy for treating KRAS-mutant lung cancer. Nature 534, 647–651 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  32. 32.

    Lito, P. et al. Disruption of CRAF-mediated MEK activation is required for effective MEK inhibition in KRAS mutant tumors. Cancer Cell 25, 697–710 (2014).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  33. 33.

    Burgess, M. R. et al. KRAS allelic imbalance enhances fitness and modulates MAP kinase dependence in cancer. Cell 168, 817–829 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  34. 34.

    McNeill, R. S. et al. Combination therapy with potent PI3K and MAPK inhibitors overcomes adaptive kinome resistance to single agents in preclinical models of glioblastoma. Neuro-oncol. 19, 1469–1480 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  35. 35.

    Zawistowski, J. S. et al. Enhancer remodeling during adaptive bypass to MEK inhibition is attenuated by pharmacologic targeting of the P-TEFb complex. Cancer Discov. 7, 302–321 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  36. 36.

    Hunter, J. C. et al. Biochemical and structural analysis of common cancer-associated KRAS mutations. Mol. Cancer Res. 13, 1325–1335 (2015).

    Article  PubMed  CAS  Google Scholar 

  37. 37.

    Lito, P., Solomon, M., Li, L. S., Hansen, R. & Rosen, N. Allele-specific inhibitors inactivate mutant KRAS G12C by a trapping mechanism. Science 351, 604–608 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  38. 38.

    Winter, J. J. et al. Small molecule binding sites on the Ras:SOS complex can be exploited for inhibition of Ras activation. J. Med. Chem. 58, 2265–2274 (2015).

    Article  PubMed  CAS  Google Scholar 

  39. 39.

    Evelyn, C. R. et al. Rational design of small molecule inhibitors targeting the Ras GEF, SOS1. Chem. Biol. 21, 1618–1628 (2014).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  40. 40.

    Wang, W., Fang, G. & Rudolph, J. Ras inhibition via direct Ras binding—is there a path forward? Bioorg. Med. Chem. Lett. 22, 5766–5776 (2012).

    Article  PubMed  CAS  Google Scholar 

  41. 41.

    Tokunaga, R. et al. Fibroblast growth factor receptor 2 expression, but not its genetic amplification, is associated with tumor growth and worse survival in esophagogastric junction adenocarcinoma. Oncotarget 7, 19748–19761 (2016).

    Article  PubMed  PubMed Central  Google Scholar 

  42. 42.

    Maron, S. B. et al. Targeted therapies for targeted populations: anti-EGFR treatment for EGFR amplified gastroesophageal adenocarcinoma. Cancer Discov. https://doi.org/10.1158/2159-8290.CD-17-1260 (2018).

  43. 43.

    The Cancer Genome Atlas Research Network. Comprehensive molecular characterization of gastric adenocarcinoma. Nature 513, 202–209 (2014).

    Article  CAS  Google Scholar 

  44. 44.

    Catenacci, D. V. et al. Durable complete response of metastatic gastric cancer with anti-Met therapy followed by resistance at recurrence. Cancer Discov. 1, 573–579 (2011).

    Article  PubMed  PubMed Central  Google Scholar 

  45. 45.

    Miyoshi, H. & Stappenbeck, T. S. In vitro expansion and genetic modification of gastrointestinal stem cells in spheroid culture. Nat. Protoc. 8, 2471–2482 (2013).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  46. 46.

    Catenacci, D. V. et al. Absolute quantitation of Met using mass spectrometry for clinical application: assay precision, stability, and correlation with MET gene amplification in FFPE tumor tissue. PloS ONE 9, e100586 (2014).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  47. 47.

    Hembrough, T. et al. Application of selected reaction monitoring for multiplex quantification of clinically validated biomarkers in formalin-fixed, paraffin-embedded tumor tissue. J. Mol. Diagn. 15, 454–465 (2013).

    Article  PubMed  CAS  Google Scholar 

Download references

Acknowledgements

This research was supported by funding from Target Cancer Foundation, Sanofi Oncology (A.J.B., G.S.W. and K.J.), Twomey Family Fellowship in Esophageal Cancer Research (G.S.W. and J.Z.), a Research Scholar Grant from the American Cancer Society to A.J.B. and NIH grants P50 CA127003 (A.J.B.). A.J.B., K.-K.W., J.A.D. and A.K.R. were supported by NIH grant P01 CA098101. JSPS Kakenhi grant JP16H06259 and Kobayashi Foundation for Cancer Research supported Y.I. D.C. was supported by the Live Like Katie (LLK) Fund, Sal Ferrara II Fund for PANGEA, NIH K23 CA178203-01A1, University of Chicago Comprehensive Cancer Center (UCCCC) Precision Oncology-Cancer Center Support Grant P30 CA014599.

Author information

Affiliations

Authors

Contributions

G.S.W., A.J.B. and D.C. conceived the study and wrote and edited the manuscript. G.S.W., J.Z., J.B.L., Z.W., T.L., X.X., J.P., C.Z., A.D. and K.J. participated in the planning, data generation and analysis of in vitro and biochemical experiments. G.S.W., J.Z., J.B.L. and Z.W. performed tumor xenograft experiments. S.E.S., J.M., S.F., P.M., S.A.C. and R.B. performed genomic analysis. D.X., L.H., P.X., E.O’D., R.R., W.-l.L., F.C., T.H., S.S. and C.S. developed and maintained patient-derived cell lines, performed histochemical and mass spectrometric analysis. F.G., A.R., K.N., E.O., M.W., H.B. and Y.I. performed immunohistochemical and retrospective clinical outcomes analysis. A.K.R., K.-K.W. and J.A.D. provided critical input. All authors read and edited the manuscript.

Corresponding authors

Correspondence to Daniel Catenacci or Adam J. Bass.

Ethics declarations

Competing interests

G.S.W. is now an employee of Novartis Institutes for Biomedical Research, Inc.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wong, G.S., Zhou, J., Liu, J.B. et al. Targeting wild-type KRAS-amplified gastroesophageal cancer through combined MEK and SHP2 inhibition. Nat Med 24, 968–977 (2018). https://doi.org/10.1038/s41591-018-0022-x

Download citation

Further reading

Search

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing