Most anaplastic lymphoma kinase (ALK)-rearranged non-small-cell lung tumors initially respond to small-molecule ALK inhibitors, but drug resistance often develops1,2,3,4. Of tumors that develop resistance to highly potent second-generation ALK inhibitors, approximately half harbor resistance mutations in ALK, while the other half have other mechanisms underlying resistance. Members of the latter group often have activation of at least one of several different tyrosine kinases driving resistance5. Such tumors are not expected to respond to lorlatinib—a third-generation inhibitor targeting ALK that is able to overcome all clinically identified resistant mutations in ALK5,6—and further therapeutic options are limited5. Herein, we deployed a shRNA screen of 1,000 genes in multiple ALK-inhibitor-resistant patient-derived cells (PDCs) to discover those that confer sensitivity to ALK inhibition. This approach identified SHP2, a nonreceptor protein tyrosine phosphatase, as a common targetable resistance node in multiple PDCs. SHP2 provides a parallel survival input downstream of multiple tyrosine kinases that promote resistance to ALK inhibitors. Treatment with SHP099, the recently discovered small-molecule inhibitor of SHP2, in combination with the ALK tyrosine kinase inhibitor (TKI) ceritinib halted the growth of resistant PDCs through preventing compensatory RAS and ERK1 and ERK2 (ERK1/2) reactivation. These findings suggest that combined ALK and SHP2 inhibition may be a promising therapeutic strategy for resistant cancers driven by several different ALK-independent mechanisms underlying resistance.


Multiple molecular mechanisms can lead to acquired resistance to ALK inhibitors. Secondary mutations encoding variants in the ALK kinase domain are observed in 20% and 50% of patients after treatment with first-generation (crizotinib) and second-generation (ceritinib and alectinib) ALK inhibitors, respectively5,7,8,9. Other resistance mechanisms include ALK gene amplification8,10,11 and activation of alternate kinases (epidermal growth factor receptor (EGFR)10,12, KIT proto-oncogene receptor tyrosine kinase (KIT)10, SRC proto-oncogene, nonreceptor tyrosine kinase (SRC)13 and insulin like growth factor 1 receptor (IGF1R)14). These tyrosine kinases cause resistance through maintaining activation of downstream ERK and/or PI3K–AKT signaling despite ALK inhibition10,12,13,14.

The highly potent and selective third-generation ALK inhibitor lorlatinib, which is currently in clinical trials, is active against all known secondary mutations encoding variants in the ALK kinase domain that confer resistance to first- and second-generation ALK inhibitors in preclinical models5,6. However, lorlatinib fails to suppress growth of resistant cells with alternate kinase bypass signaling pathways5, and new therapeutic options are needed for these patients. The heterogeneity of resistance mechanisms makes clinical development of new candidate therapeutic strategies challenging, and this is compounded by the fact that many are not readily identified through genomic analyses13.

We sought to identify a treatment approach that could more broadly address acquired resistance in cases without secondary mutations in ALK. First, we performed a pooled shRNA dropout screen on a panel of seven ALK-rearranged non-small-cell lung cancer (NSCLC) cell lines generated directly from patients who developed resistance to crizotinib or ceritinib in clinic (Supplementary Table 1). Cells were infected with a lentiviral shRNA library (termed tDRIVE) targeting 1,000 cancer-related genes (20 hairpins per gene; see Supplementary Table 2 for the complete gene list) and were treated with the ALK inhibitor ceritinib or with vehicle for 14 d (Fig. 1a). To identify shRNAs that were selectively depleted after ceritinib treatment, we quantified the abundance of barcoded shRNA vectors through next-generation sequencing (NGS) and calculated the relative fold depletion of these vectors after ceritinib treatment as compared to vehicle treatment. In doing this, we identified depletion of shRNAs targeting members of known resistance pathways13 and additional putative resistance drivers, including kinases (EGFR; fibroblast growth factor receptor 1 (FGFR1); Erb-B2 receptor tyrosine kinase 2 (ERBB2); SRC, mitogen-activated protein kinase 1 (MAPK1); and Raf-1 proto-oncogene, serine/threonine kinase (RAF1)), phosphatases (protein tyrosine phosphatase, nonreceptor type 11 (PTPN11)), the adaptor protein fibroblast growth factor receptor substrate 2 (FRS2) and several transcription factors, such as MYC (Fig. 1b). The full list of identified hits is shown in Supplementary Table 3. All cell lines used in the shRNA screen underwent targeted NGS for 1,000 known cancer-related genes (Supplementary Table 4)5,13. Notably, no activating genetic alterations were identified that could readily explain the hits scored in the shRNA screen; there were no alterations in the scored genes themselves or in their signaling pathway neighborhoods. This underscores the challenge of identifying therapeutic alternatives for resistant tumors solely on the basis of results from DNA sequencing5,13.

Figure 1: Complementary shRNA screens and high-throughput drug combination screens identify therapeutic approaches to overcome resistance in ALK-rearranged PDCs.
Figure 1

(a) Schematic of the tDRIVE pooled shRNA screen to identify therapies conferring sensitivity to ceritinib. (b) Top candidates genes that drive resistance to ceritinib as identified by the pooled shRNA screen (raw data in Supplementary Table 3). Each gene is color coded. (cf) EGFR or SRC inhibition restores sensitivity to ALK inhibition in MGH049-1A cells. Primary drug screen data depicting the shift in cellular viability in response to gefitinb (c), erlotinib (d), dasatinib (e) or saracatinib (f) at the dose indicated on the x axis in the presence of 500 nM ceritinib. (g) FGFR inhibition restores sensitivity to ALK inhibition in MGH073-2B cells. Primary drug screen data depicting the shift in cellular viability in response to BGJ398 at the dose indicated on the x axis in the presence of 500 nM ceritinib. (h,i) SRC inhibition restores sensitivity to ALK inhibition in MGH045-2A cells. Primary drug screen data depicting the shift in cellular viability in response to dasatinib (h) or saracatinib (i) at the dose indicated on the x axis in the presence of 500 nM ceritinib. For ci, data are shown as mean ± s.e.m. of n = 3 technical replicates. (j) Results from a long-term proliferation assay of MGH049-1A, MGH073-2B and MGH045-2A cells that were grown in the presence of the indicated drugs (all used at 500 nM) for 14 d followed by crystal violet staining. Images are representative of at least two independent experiments. Combo, combination treatment. (k) Immunoblot of lysates from MGH049-1A cells that were treated with ceritinib (500 nM), erlotinib (1 μM) or cetuximab (100 nM) or the indicated combinations for 4 h using antibodies against the indicated proteins. The blot is representative of at least two independent experiments. p-, phosphorylated. (l,m) Immunoblots of lysates from MGH073-2B (l) and MGH045-2A (m) cells that were treated with the indicated drugs at a dosage of 500 nM for 4 h and for 48 h using antibodies against the indicated proteins. The image is representative of at least two independent experiments. (n) Results from a representative western blot analysis for SHP2 expression in MGH049-1A, MGH073-2B and MGH045-2A cells that were transduced with lentivirus encoding a SHP2-specific or nontargeting (NT) shRNA. SHP2 shRNA-3 was selected from the pooled shRNA library. GAPDH was used as a loading control. Uncropped blots are available in Supplementary Figure 11. (o) Results from a colony-formation assay showing the effect of SHP2 depletion on sensitivity to ceritinib in PDCs. Cells were treated with the indicated drugs for 14 d and stained using crystal violet. The image is representative of two independent experiments.

Second, in a complementary approach, we used a strategy for screening drug combinations that is effective in identifying bypass mechanisms of resistance13. We treated resistant patient-derived cell lines with a panel of 112 targeted agents alone or in combination with ceritinib. Through comparing the ability of these drugs to suppress cell viability in the presence or absence of ceritinib13, we identified that EGFR (in MGH049-1A), FGFR (in MGH073-2B) and SRC (in MGH045-2A and MGH049-1A) were key mechanisms underlying resistance (Fig. 1c–i). These results were confirmed in long-term viability assays, in which erlotinib (EGFR inhibitor) and saracatinib (SRC inhibitor) restored sensitivity to ALK inhibition in MGH049-1A cells, BGJ398 (FGFR inhibitor) did so in MGH073-2B cells, and saracatinib did so in MGH045-2A cells (Fig. 1j). Overall, there was very good agreement between the results of the drug and shRNA screens. However, as expected, not all hits in the drug screen were seen in the shRNA screen, as multiple paralogues can often be targeted with drugs but not through strategies using a gene-specific shRNA. Thus, although small-molecule inhibitors of SRC family members were found to sensitize MGH049-1A to ceritinib (Fig. 1e,f,j)13, SRC and other related kinases were not shRNA screen hits. In support of the high sensitivity and specificity of the shRNA screens, none of the four cell lines with no FGFR or related pathway hits (FRS2) were sensitive to the ceritinib plus BGJ398 combination (Supplementary Fig. 1a).

Analysis of canonical receptor tyrosine kinase downstream signaling events revealed that ERK reactivation rapidly occurred after ceritinib treatment in multiple models harboring ALK mutations, and that effective combinations suppressed this reactivation. In MGH049-1A cells, ceritinib effectively suppressed ERK1/2 phosphorylation 1 h after drug treatment; however, ERK1/2 reactivation was observed 4 h after treatment (Supplementary Fig. 1b). Co-treatment with either erlotinib or cetuximab blocked reactivation of ERK1/2 (Fig. 1k). In the MGH073-2B and MGH045-2A cell lines, ceritinib treatment led to acute suppression of phosphorylated ERK1/2, but ERK1/2 reactivation was seen within 48 h following treatment (Fig. 1l,m). Addition of the appropriate second kinase inhibitor (based on the results of the shRNA and drug screens) overcame this rebound reactivation and led to durable ERK1/2 suppression (Fig. 1l,m). Together, these results confirm that targeting signaling events allowing bypass of receptor tyrosine kinase inhibition may be a useful strategy against resistant cancers. However, these results also demonstrate that different tyrosine kinases drive resistance in distinct cancers, underscoring the challenge of developing a single therapeutic approach that would be effective against a majority of these cancers.

In addition to uncovering the different kinases driving resistance in each cell line, the shRNA screen revealed that depletion of SH2 domain containing tyrosine phosphatase (SHP2; encoded by PTPN11) resensitized multiple PDCs to ceritinib (Fig. 1b). SHP2 mediates GTP loading of RAS downstream of multiple tyrosine kinases, including EGFR, FGFR and SRC15,16,17,18,19,20,21,22,23. Thus, we hypothesized that SHP2 might serve as a common signaling node in cells with different resistance drivers leading to reactivation of downstream RAS–MAPK pathway signaling. To test whether SHP2 is upregulated in ALK-TKI-resistant cells as compared to ALK-TKI-sensitive cells, we compared the level of SHP2 expression in both sensitive and resistant models and did not observe substantial differences (Supplementary Fig. 2a). Further, we overexpressed SHP2 in sensitive cells and found that SHP2 expression also did not confer resistance to ceritinib in these cells (Supplementary Fig. 2b,c). These data suggest that levels of SHP2 expression do not predict sensitivity to ALK inhibition. To validate the results from the initial screen, we used two independent shRNAs to knockdown SHP2 in MGH049-1A cells (EGFR bypass), MGH073-2B and MGH065-1B cells (FGFR bypass), and MGH045-2A cells (SRC bypass) (Fig. 1n and Supplementary Fig. 2d). SHP2 depletion increased sensitivity to ceritinib in all models (Fig. 1o and Supplementary Fig. 2e). To test whether pharmacological inhibition of SHP2 might resensitize resistant cell lines to ceritinib (the original combination drug screen did not include a SHP2 inhibitor), we used SHP099, a highly potent and selective allosteric inhibitor of SHP224,25. As a single agent, SHP099 had a minimal effect on cell proliferation (consistent with the shRNA results); however, the combination of SHP099 and ceritinib led to marked inhibition of colony formation in all examined PDCs (Fig. 2a and Supplementary Fig. 3a). Moreover, in MGH049-1A and MGH045-2A cells, SHP099 enhanced the apoptotic activity of ceritinib. In MGH073-2B and MGH065-1B cells, the combination of SHP099 and ceritinib did not induce apoptosis but rather blocked cell proliferation (Supplementary Fig. 3b).

Figure 2: The SHP2 inhibitor SHP099 attenuates ceritinib-induced ERK reactivation and increases ceritinib efficacy in vitro.
Figure 2

(a) Results from a long-term proliferation assay of MGH049-1A, MGH073-2B, MGH065-1B and MGH045-2A cells treated with ceritinib (500 nM), SHP099 (5 μM) or these in combination for 14 d. Vehicle (DMSO) served as control. Cells were stained using crystal violet. The image is representative of three independent experiments. (be) Immunoblots of lysates from PDCs that were treated with 500 nM ceritinib, 1 or 5 μM of SHP099 or a combination of these for the indicated time points using antibodies against the indicated proteins. GAPDH was assessed as a loading control. The image is representative of three independent experiments. (fi) Immunoblots of whole-cell lysates (WCL) or GST-RBD-immunoprecipitated (IP) lysates from PDCs treated with ceritinib (500 nM), SHP099 (5 μM) or a combination of these for 1 h and 48 h using antibodies against the indicated proteins. The image is representative of at least two independent experiments. Uncropped blots are available in Supplementary Figure 11.

We further tested the effect of SHP2 inhibition in the ALK-TKI-sensitive cell lines H3122 and MGH026-1. Similar to what we observed in ALK-TKI-resistant cells, SHP099 alone did not have antiproliferative activity in these ceritinib-sensitive cells. Furthermore, the addition of SHP099 did not substantially sensitize these cells to ceritinib (Supplementary Fig. 4a,b). Analysis of downstream signaling in H3122 cells revealed that ceritinib alone effectively suppressed phosphorylation of both ERK1/2 and AKT, and SHP099 in addition to ceritinib did not further downregulate the pathway. Furthermore, SHP099 alone did not affect the activation status of ERK1/2 or AKT (Supplementary Fig. 4c).

In accordance with the previous observation that SHP2 mediates RAS–MAPK pathway activation19,25,26, cotreatment of resistant cell lines with SHP099 and ceritinib resulted in durable suppression of phosphorylation of ERK1/2 and the ERK substrate p90 ribosomal S6 kinase (p90RSK; Fig. 2b–e). In accordance with the viability data, SHP099 alone (without ALK inhibition) did not suppress ERK activity at 48 h. However, SHP2 inhibition blocked ERK1/2 reactivation following continued suppression of ALK by ceritinib. In accordance with the hypothesis that SHP099 blocks reactivation of ERK driven by receptor tyrosine kinases (RTKs), in a short and late time window of exposure (the last 4 h of 48 h), SHP099 was able to suppress the ERK phosphorylation observed in the presence of continuous ALK inhibition (Supplementary Fig. 5a–d). We next examined the ERK1/2-dependent gene expression of dual specificity phosphatase 6 (DUSP6; a phosphatase and negative regulator of ERK), ETS variant 5 (ETV5), and sprout homolog 2 (SPRY2) as an additional readout of flux through the ERK signaling pathway, as this allows for a more sensitive measurement the status of the ERK signaling pathway than western blotting for phosphorylated ERK27. Compared to treatment with ceritinib alone, the combination treatment led to greater suppression of expression of ERK-dependent transcripts (Supplementary Fig. 6). To further test the specificity of action of SHP099 on tyrosine kinase–driven bypass mechanisms conferring resistance, we used a PDC (MGH034-2A) harboring an activating mutation in MAP2K1 (encoding MAP2K1-K57N; this gene is also called MEK) as the mechanism of resistance13. In this model, ERK activation should be maintained in the presence of ceritinib without the need for upstream RAS reactivation and thus should be impervious to the addition of SHP099. Consistent with this hypothesis, levels of phosphorylated ERK and phosphorylated RSK were maintained in MGH034-2A cells after treatment with the ceritinib plus SHP099 combination (Supplementary Fig. 7a).

There were some other notable findings in the screens. FGFR1 was a shRNA screen hit in MGH065-1B cells. However, FRS2, an adaptor downstream of FGFR, was a hit in MGH065-1C cells (generated from MGH065-1B cells cultured in croztinib), but FGFR1 was not. Notably, the FGFR inhibitor BGJ398, which acts on FGFR1, FGFR2 and FGFR3, sensitized this model to ceritinib, suggesting that other FGF receptors in addition to FGFR1 could mediate resistance in this model (Supplementary Fig. 8a–c). Consistent with other cases of FGFR-driven resistance, SHP099 could sensitize MGH065-1C cells to ceritinib and suppress ERK reactivation (Supplementary Fig. 8d,e). It is not clear why the shRNA screen failed to detect PTPN11 as a hit in this cell line. In the MGH075-2E line, PTPN11 was not identified as a sensitizer in the shRNA screen, which is consistent with the lack of RTK hits, and this was the only line with a MYC hit. We confirmed that SHP099 did not affect viability or inhibit ERK activity rebound in this model at 48 h following treatment (Supplementary Fig. 7b,c). Interestingly, previous work has shown that in some triple-negative breast cancer models, suppression of ERK yielded to MYC degradation, RTK upregulation and adaptive resistance28. However, MYC downregulation did not appear to support ERK reactivation in this context (Supplementary Fig. 7d). Taken together, our results demonstrate that targeted SHP2 inhibition can overcome resistance in multiple contexts of tyrosine kinase–mediated bypass signaling, leading to survival in the presence of ALK inhibition, most likely through the suppression of ERK signaling.

Echinoderm microtubule-associated protein-like 4-anaplastic lymphoma kinase (EML4-ALK) was recently shown to promote cell survival in NSCLC through activation of MAPK pathway signaling via all three RAS isoforms (KRAS, HRAS and NRAS)29. To better understand the relationship between the RAS isoforms, ALK and SHP2 in the resistant models, we performed pull-down assays for GTP-bound RAS (GTP-RAS) using the RAS-binding domain of RAF. In accordance with a previous report29, ALK inhibition led to initial loss of GTP loading of all three RAS isoforms in all models (Fig. 2f–i). Neither short nor long exposure to single-agent treatment with SHP099 alone decreased RAS activity in any of the PDCs. In contrast, short-term treatment with the combination of ceritinib and SHP099 further decreased the level of GTP-RAS in all models as compared to ALK inhibition alone. After 48 h of single-agent ceritinib treatment, all three RAS isoforms were reactivated to varying degrees in MGH049-1A, MGH073-2B, MGH065-1B and MGH045-2A cells (consistent with the reactivation of ERK signaling observed with single-agent ceritinib). Notably, addition of SHP099 invariably led to greater suppression of all RAS isoforms in all PDCs (Fig. 2f–i). It is clear that SHP2 can modulate ERK1/2 activity downstream of several tyrosine kinases19. Interestingly, short-term treatment with SHP099 did substantially impact ERK activity in the PDCs. In addition, SHP099 did not suppress ERK activity in ceritinib-sensitive models (Supplementary Fig. 4c). Taken together, our results suggest that SHP2 is not a key modulator of RAS and ERK activity downstream of EML4-ALK but rather is a mediator of RAS and ERK reactivation downstream of a diverse set of tyrosine kinases activated upon EML4-ALK inhibition.

To evaluate the in vivo efficacy of combined treatment with SHP2 and ALK inhibitors, we treated nu/nu nude mice bearing subcutaneous xenografts derived from PDCs with SHP099, ceritinib or the two in combination. Treatment of MGH049-1A and MGH073-2B xenografts with ceritinib resulted in modest and short-lived responses, whereas MGH045-2A xenografts were completely resistant (Fig. 3a–c). The combination of SHP099 and ceritinib resulted in profound regression of MGH049-1A and MGH073-2B xenograft tumors and modestly reduced growth of MGH045-2A tumors (Fig. 3a–c) and coincided with significantly decreased DUSP6 mRNA levels (Fig. 3d). Similar results were observed when combining ceritinib with the anti-EGFR antibody cetuximab or the FGFR inhibitor BGJ398, targeting the specific bypass mechanisms in MGH049-1A and MGH073-2B cells, respectively (Supplementary Fig. 9a,b). We also evaluated tumor regrowth after withdrawal of the SHP099 and ceritinib combination. Tumors that regressed did regrow after treatment cessation, although growth was slow at least during the 22 d of follow up in the MGH-073-B model. We tested this in another model (MGH049-1A) and found that tumors also grew back upon drug withdrawal but remained sensitive to drug retreatment (Supplementary Fig. 9c,d). As reported25, single-agent treatment with SHP099 was well tolerated with minor or no body weight loss over the course of treatment. The SHP099 plus ceritinib treatment exhibited minor toxicity in models following the initiation of the treatment, with no more than 7% body weight loss. However, over the course of treatment, toxicity decreased (Supplementary Fig. 10a–c). The same pattern of toxicity was observed in mice treated with BGJ398 and ceritinib (Supplementary Fig. 10b). Thus, SHP2 inhibition suppresses reactivation of ERK1/2 downstream of different tyrosine kinases, which leads to tumor regression in vivo, suggesting that this combination approach might be broadly useful against acquired resistance to ALK inhibitors in cancer.

Figure 3: Combined inhibition of ALK and SHP2 enhances response and overcomes resistance in drug-resistant ALK-rearranged PDCs in vivo.
Figure 3

(a) MGH049-1A subcutaneous xenografts in mice following treatment with ceritinib (25 mg per kg body weight, daily), SHP099 (75 mg per kg body weight, daily) or a combination of both drugs. Data are shown as mean + s.e.m. *P < 0.05 versus vehicle; one-way ANOVA followed by Tukey test. n = 5 mice per group. (b) MGH073-2B subcutaneous xenografts in mice following treatment with ceritinib (25 mg per kg body weight, daily), SHP099 (75 mg per kg body weight, daily) or a combination of both drugs. Data are shown as mean + s.e.m. *P < 0.05 versus vehicle; one-way ANOVA followed by Dunnett's test. n = 5 mice per group, except for the combination group, for which n = 8 mice. (c) MGH045-2A xenografts in mice that were treated as in b. Data are shown as mean + s.e.m. *P < 0.05 versus vehicle; one-way ANOVA followed by Tukey test. n = 4 mice per group. In ac, the Tukey test was used where there was an equal number of mice in each group; Dunnett's test was used where the number of mice differed between groups. (d) DUSP6 levels, assessed through qRT-PCR, in mice bearing MGH049-1A,MGH073-2B or MGH045-2A subcutaneous tumors that were administered vehicle, ceritinib (25 mg per kg body weight), SHP099 (75 mg per kg body weight) or a combination of both drugs for 3 d. Tumors were collected 3 h after the last treatment. Data are shown as mean ± s.e.m. and are expressed as the fold change percentage. *P < 0.01 versus vehicle, ceritinib or SHP099; two-tailed Student's t-test. n = 3 mice per condition were used.

In conclusion, our study suggests that inhibition of SHP2 may abrogate RAS and ERK1/2 reactivation following treatment with an ALK inhibitor in NSCLC. Combined ALK and SHP2 inhibition may provide a broad-reaching therapeutic strategy for overcoming or preventing heterogeneous ALK-independent mechanisms of acquired drug resistance in patients with NSCLC positive for ALK rearrangement.


Cell culture and compounds.

Patient-derived cell lines were established as previously described13. MGH034-2A, MGH049-1A, MGH026-1A and MGH075-2E patient-derived cell lines were previously described5,13. MGH073-2B was developed from crizotinib-resistant pleural effusion. MGH065-1B was developed from crizotinib-resistant lymph node biopsy. MGH065-1C was developed from MGH065-1B in medium containing 300 nM crizotinib. MGH051-2B was established from ceritinib-resistant liver biopsy. MGH045-2A was developed from crizotinib-resistant pleural effusion. All patients signed informed consent to participate in a protocol approved by the Dana-Farber/Harvard Cancer Center Institutional Review Board giving permission for research to be performed on their samples. Cell lines were sequenced to confirm the presence of ALK rearrangements identified through clinical testing of biopsy specimens from the same patients and tested negative for mycoplasma contamination. Additional authentication was performed for MGH049-1A, MGH045-2A, MGH051-2B, MGH034-2A and MGH065-1B through single-nucleotide polymorphism (SNP) fingerprinting. Cells were grown in either DMEM or RPMI-1640 (Corning) supplemented with 10% FBS and 1× Antibiotic-Antimycotic. Ceritinib, BGJ398 and SHP099 were synthesized in the Global Discovery Chemistry Department at Novartis Institutes for Biomedical Research. Saracatinib, erlotinib and gefitinib were purchased from Selleckchem. Each compound was dissolved in DMSO for cell culture experiments. Cetuximab (eribitux) was obtained from Imclone LLC, an Eli Lilly subsidiary.

Drug screen.

A screen of drug combinations consisting of 112 agents was performed as previously described13. Cells were treated with vehicle or varying concentrations of drugs to be screened in the absence or presence of 300 nM ceritinib. After incubation with drugs for 5 d, cell counts were determined using CellTiter-Glo (Promega) per the manufacturer's instructions. GraphPad Prism version 5.0 was used to graphically display data fit to a nonlinear regression model using a four-parameter analytic method.

tDRIVE shRNA screen and next-generation sequencing.

The synthetic lethal pooled shRNA screen was conducted using a custom-made shRNA library (tDRIVE) containing 20,000 hairpins that target 1,000 known cancer-related genes with 20 shRNAs per gene. Cells were transduced with tDRIVE lentivirus at a multiplicity of infection (MOI) of 0.3 using the spin infection protocol (1 h at 2,100 r.p.m.), selected with puromycin and expanded to 60 million cells. These cells were then divided into three equivalent pools of 20 million cells (1,000 cells per shRNA): day 0, DMSO (vehicle)- and ceritinib-treated groups. Day 0 cells were harvested at the time of treatment start and were snap-frozen using liquid nitrogen. The other two pools were treated with vehicle or 500 nM ceritinib for 14 d, and then NGS was performed to determine hairpin distribution. For each cell line, genes presenting with the largest fold change in shRNA abundance depletion in the ceritinib-treated sample as compared with the DMSO-treated sample were selected to undergo further functional studies. P values based on the redundant siRNA activity (RSA) statistic comparing ceritinib-treated cell lines and DMSO-treated cell lines were calculated through pooling the counts of barcodes in each cell line. In addition, a z-score was calculated using the mean and s.d. of the fold change between DMSO- and ceritinib-treated samples in the counts of the shRNAs in the library. A z-score for each gene was calculated as the mean of the z-scores of the shRNAs targeting the gene.

Infection and transfection.

Custom packaged lentiviral particles carrying sh_hPTPN11 no. 165861 (shRNA-3, Hairpin sequence: 5′-ACCGGCGGTTTGATTCTTTGATAGATGTTAATATTCATAGCATCTGTCAAAGAATCAAACCGTTTT-3′), sh_hPTPN11 no. 165860 (shRNA-4, Hairpin sequence: 5′- ACCGGGCAATGATGGCAAGTTTAAAGGTTAATATTCATAGCCTTTAGACTTGCCGTCATTGCTTTT-3′) or shNT (nontargeting control in pRSI16-U6-sh-ubic-TagRFP-2A-Puro) were obtained from Cellecta. Cells were seeded into six-well plates at a density of 3 × 105 cells per well. 24 h later, cells were infected with lentiviral particles. Cells were selected with puromycin 48 h after viral transduction before use in experiments.

SMARTpool ON-TARGETplus MYC siRNA was obtained from Dharmacon. Cells were seeded into six-well plates at a density of 2 × 105 cells per well. 24 h later, cells were transfected with siRNA according to the manufacturer's instructions. Cells were used for experiments 48 h post-transfection. Production of doxycycline (dox)-inducible wild-type streptavidin-binding peptide–tagged (SBP) SHP2 viral particles was performed as previously described25. Cells were seeded into six-well plates at a density of 3 × 105 cells per well. 24 h later, cells were infected with SBP-SHP2 wild-type viral particles. Cells were selected with G418 at 48 h after viral transduction before use in experiments.

Antibodies and immunoblotting.

Cells were seeded in six-well plates and treated with drug or siRNA for specified time points. Lysates were prepared as previously described, and equal volumes of total cell lysate were processed for immunoblotting10. Antibodies against phosphorylated ALK (phospho-ALK) Y1282/1283 (9687), ALK (3633), phospho-AKT S473 (4060), AKT (4691), phospho-ERK T202/Y204 (9101), ERK (9102), phospho-paxillin Y118 (2541), SHP2 (3752) and EGFR (4267) were obtained from Cell Signaling Technology and used at a 1:1,000 dilution. Antibodies against phospho-EGFR Tyr1068 (AB5644) and phospho-RSK 359/S363 (AB32413) were purchased from Abcam and used at a 1:1,000 dilution. Antibody against GAPDH (MAB374) was purchased from Millipore and used at a 1:5,000 dilution. Antibodies against K-RAS (F234), H-RAS (F235) and N-RAS (F155) were purchased from Santa Cruz Biotechnology and used at a 1:500 dilution. All secondary antibodies (anti–mouse IgG HRP-linked (7076S) and anti–rabbit IgG HRP-linked (7074S)) were purchased from Cell Signaling and used at 1:50,000 dilution.

RAS activation assay.

Cells were seeded in 10-cm dishes and treated with the drug for specified time points. The glutathione S-transferase–Ras-like Raf-binding domain fusion (GST–RBD) assay for RAS activation was performed using RAS activation assay kit (Cytoskeleton, catalog no. BK008) according to the manufacturer's instructions.

Crystal violet assay.

Cells were seeded at a density of 5,000–10,000 per well in 12-well plates and were drugged the following day. Medium and drug were replaced every 72 h for 12 d. Cells were fixed with glutaraldehyde for 10 min, washed 2× with H2O and stained with 0.1% crystal violet (Sigma) for 30 min. Cells were then washed 3× with H2O, and plates were dried overnight.


Cells were seeded 24 h before attaining 50% confluence. Cells were treated with drugs in the panel for 4 h and 48 h, and RNA was extracted using the RNeasy Kit (Qiagen). Likewise, RNA was extracted from tumor samples for pharmacodynamic analyses. cDNA was prepared from 500 ng total RNA with the First-Strand Synthesis Kit (Invitrogen) using oligo-dT primers. qPCR was performed on a Light Cycler 480 system (Roche) using FastStart SYBR green master mix (Roche). mRNA expression relative to actin mRNA levels was calculated using the delta–delta threshold cycle (ΔΔCT) method. Primers that were used are as follows: SPRY2, F 5′-TTGCACATCGCAGAAAGAAG-3′, R 5′-GGTCACTCCAGCAGGCTTAG-3′; ETV5, F 5′-CCTACATGAGAGGGGGTTATTTC-3′, R 5′-CGTCAAAGTATAATCGGGGATCT-3′; DUSP6, F 5′- CGACTGGAACGAGAATACGG-3′, R 5′-TTGGAACTTACTGAAGCCACCT-3′; HPRT1 (HPRT), F 5′-TCAGGCAGTATAATCCAAAGATGGT-3′, R 5′- AGTCTGGCTTATATCCAACACTTCG-3′; SDHA, F 5′-TGGGAACAAGAGGGCATCTG-3′, R 5′-CCACCACTGCATCAAATTGATG-3′; TBP, F 5′-CACGAACCACGGCACTGATT-3′, R 5′-TTTTCTTGCTGCCAGTCTGGAC-3′.

Xenograft studies.

All mouse studies were conducted in accordance with the guidelines as published in the Guide for the Care and Use of Laboratory Animals, with Novartis International Animal Care and Use Committee (IACUC) regulations and guidelines and with the Institutional Animal Care and Use Committee (IACUC) of Massachusetts General Hospital. Female nu/nu mice aged 6–8 weeks were obtained from Charles River laboratories Inc., Wilmington, MA. Mice were maintained in laminar flow units in sterile filter-top cages with ALPHA-dri bedding. The MGH045-2A and MGH073-2B cells were harvested while in the exponential proliferation phase. Each mouse was inoculated subcutaneously in the upper right flank with 5 × 106 cells suspended either in 0.2 ml cold PBS (MGH045-2A) or in 20% BD Matrigel Basement Membrane Matrix in PBS (MGH073-2B). The development of MGH049-1A xenograft tumors comprised two steps. In the first step, 1 × 107 MGH049-1A cells harvested during exponential growth were suspended in a 1:1 mixture of cold PBS and Matrigel in a total volume of 0.2 ml and were injected subcutaneously into the upper right flank of mice. Tumors established in this step were collected and fragmented. Tumor fragments were then implanted into the upper right flank of mice. Tumor volumes and mice weights were monitored twice per week. Tumor size was measured using a caliper twice weekly, and weights were determined at the same time. Tumor volume was calculated as follows: length × width2 × 0.51. When tumor volume reached approximately 200 mm3, mice were randomized and orally administered vehicle, 25 mg ceritinib per kg body weight (0.5% methyl cellulose and 0.5% Tween 80); 75 mg SHP099 per kg body weight (0.5% methyl cellulose and 0.5% Tween 80); 30 mg BGJ398 per kg body weight (50% acetic acid/acetate buffer (pH 4.68) and 50% PEG300); 25 mg ceritinib per kg body weight plus 75 mg SHP099 per kg body weight daily; or 25 mg ceritinib per kg body weight plus 30 mg BGJ398 per kg body weight daily respectively. 20 mg cetuximab per kg body weight was injected i.p. twice a week as monotherapy or in combination with 25 mg ceritinib per kg body weight. For pharmacodynamic analyses, tumor-bearing mice were administered drug or vehicle daily for 3 d. Tumor tissue was excised 3 h after the last treatment and was snap-frozen in liquid nitrogen for measurement of DUSP6 levels through qRT-PCR. Ceritinib, SHP099 and BGJ398 were provided by Novartis. Cetuximab (eribitux) was obtained from Imclone LLC., an Eli Lilly subsidiary. When combinations were administered, 30 min elapsed between drug doses.

Next-generation sequencing.

NGS was performed as previously described13. Briefly, hybridization capture was performed using RNA baits to capture 1,000 known cancer genes (RightOn Cancer Sequencing Kit, developed in collaboration with Elim BioPharma). NGS data for MGH075-2E, MGH051-2B, MGH034-2A and MGH049-1A was previously reported5,13.

Statistical analyses.

All analyses for in vivo experiments were performed with SigmaPlot version 13.0. Data sets generated in vivo were analyzed for statistical significance using either a one-way ANOVA followed by Tukey test or a one-way ANOVA followed by Dunnett's test. GraphPad Prism version 6.0 was used to analyze the in vitro data. Data sets generated in vitro were analyzed for statistical significance using an unpaired two-tailed Student's t-test.

Life Sciences Reporting Summary.

Further information on experimental design and reagents is available in the Life Sciences Reporting Summary.

Data availability.

The authors declare that the main data supporting the findings of this study are available within the article and its Supplementary Information files.

Additional information

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


  1. 1.

    et al. First-line crizotinib versus chemotherapy in ALK-positive lung cancer. N. Engl. J. Med. 371, 2167–2177 (2014).

  2. 2.

    et al. Crizotinib versus chemotherapy in advanced ALK-positive lung cancer. N. Engl. J. Med. 368, 2385–2394 (2013).

  3. 3.

    et al. Ceritinib in ALK-rearranged non-small-cell lung cancer. N. Engl. J. Med. 370, 1189–1197 (2014).

  4. 4.

    & Treating ALK-positive lung cancer—early successes and future challenges. Nat. Rev. Clin. Oncol. 9, 268–277 (2012).

  5. 5.

    et al. Molecular mechanisms of resistance to first- and second-generation ALK inhibitors in ALK-rearranged lung cancer. Cancer Discov. 6, 1118–1133 (2016).

  6. 6.

    et al. PF-06463922, an ALK/ROS1 inhibitor, overcomes resistance to first and second generation ALK inhibitors in preclinical models. Cancer Cell 28, 70–81 (2015).

  7. 7.

    et al. EML4-ALK mutations in lung cancer that confer resistance to ALK inhibitors. N. Engl. J. Med. 363, 1734–1739 (2010).

  8. 8.

    et al. Mechanisms of resistance to crizotinib in patients with ALK gene rearranged non-small cell lung cancer. Clin. Cancer Res. 18, 1472–1482 (2012).

  9. 9.

    et al. Therapeutic strategies to overcome crizotinib resistance in non-small cell lung cancers harboring the fusion oncogene EML4-ALK. Proc. Natl. Acad. Sci. USA 108, 7535–7540 (2011).

  10. 10.

    et al. Mechanisms of acquired crizotinib resistance in ALK-rearranged lung cancers. Sci. Transl. Med. 4, 120ra17 (2012).

  11. 11.

    et al. Heterogeneity of genetic changes associated with acquired crizotinib resistance in ALK-rearranged lung cancer. J. Thorac. Oncol. 8, 415–422 (2013).

  12. 12.

    et al. A novel ALK secondary mutation and EGFR signaling cause resistance to ALK kinase inhibitors. Cancer Res. 71, 6051–6060 (2011).

  13. 13.

    et al. Patient-derived models of acquired resistance can identify effective drug combinations for cancer. Science 346, 1480–1486 (2014).

  14. 14.

    et al. Rationale for co-targeting IGF-1R and ALK in ALK fusion-positive lung cancer. Nat. Med. 20, 1027–1034 (2014).

  15. 15.

    , , , & Multiple requirements for SHPTP2 in epidermal growth factor-mediated cell cycle progression. Mol. Cell. Biol. 16, 1189–1202 (1996).

  16. 16.

    et al. A lipid-anchored Grb2-binding protein that links FGF-receptor activation to the Ras/MAPK signaling pathway. Cell 89, 693–702 (1997).

  17. 17.

    , , , & Molecular mechanism for the Shp-2 tyrosine phosphatase function in promoting growth factor stimulation of Erk activity. Mol. Cell. Biol. 20, 1526–1536 (2000).

  18. 18.

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

  19. 19.

    , & The 'Shp'ing news: SH2 domain–containing tyrosine phosphatases in cell signaling. Trends Biochem. Sci. 28, 284–293 (2003).

  20. 20.

    , , , & Protein-tyrosine-phosphatase SHPTP2 couples platelet-derived growth factor receptor beta to Ras. Proc. Natl. Acad. Sci. USA 91, 7335–7339 (1994).

  21. 21.

    , & SH2-containing phosphotyrosine phosphatase as a target of protein-tyrosine kinases. Science 259, 1607–1611 (1993).

  22. 22.

    , , , & Role of SH-PTP2, a protein-tyrosine phosphatase with Src homology 2 domains, in insulin-stimulated Ras activation. Mol. Cell. Biol. 14, 6674–6682 (1994).

  23. 23.

    , & Genetic analysis of protein tyrosine phosphatases. Curr. Opin. Genet. Dev. 8, 112–126 (1998).

  24. 24.

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

  25. 25.

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

  26. 26.

    et al. Regulation of the mitogen-activated protein kinase signaling pathway by SHP2. J. Biol. Chem. 277, 9498–9504 (2002).

  27. 27.

    et al. (V600E)BRAF is associated with disabled feedback inhibition of RAF–MEK signaling and elevated transcriptional output of the pathway. Proc. Natl. Acad. Sci. USA 106, 4519–4524 (2009).

  28. 28.

    et al. Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell 149, 307–321 (2012).

  29. 29.

    et al. RAS–MAPK dependence underlies a rational polytherapy strategy in EML4-ALK-positive lung cancer. Nat. Med. 21, 1038–1047 (2015).

Download references


We thank D. Rakiec for help with NGS and C. Liu for help with in vivo experiments. This study was supported by grants from Novartis Institutes for BioMedical Research, American Association for Cancer Research (AACR)–AstraZeneca Fellowship in Lung Cancer Research (17-40-12-DARD to L.D.), National Cancer Institute (R01CA164273 to A.T.S.), the Wellcome Trust (102696 to C.H.B.) and the National Foundation for Cancer Research (to A.T.S.), by Be a Piece of the Solution and by LungStrong.

Author information

Author notes

    • Luc Friboulet
    •  & Fang Li

    Present addresses: Gustave Roussy Cancer Campus, Université Paris Saclay, INSERM U981, Paris, France (L.F.) and Tango Therapeutics, Cambridge, Massachusetts, USA (F.L.).

    • Cyril H Benes
    • , Fang Li
    •  & Jeffrey A Engelman

    These authors contributed equally to this work.


  1. Cancer Center, Massachusetts General Hospital, Charlestown, Massachusetts, USA.

    • Leila Dardaei
    • , Manrose Singh
    • , Katherine X Shaw
    • , Satoshi Yoda
    • , Michael S Lawrence
    • , Adam Langenbucher
    • , Justin F Gainor
    • , Luc Friboulet
    • , Ibiayi Dagogo-Jack
    • , David T Myers
    • , Melissa Parks
    • , Dana Lee
    • , Richard H DiCecca
    • , Alice T Shaw
    • , Aaron N Hata
    •  & Cyril H Benes
  2. Department of Medicine, Harvard Medical School, Boston, Massachusetts, USA.

    • Leila Dardaei
    • , Satoshi Yoda
    • , Alice T Shaw
    • , Aaron N Hata
    •  & Cyril H Benes
  3. Novartis Institutes for BioMedical Research, Cambridge, Massachusetts, USA.

    • Hui Qin Wang
    • , Paul Fordjour
    • , Kristine Yu
    • , Jinsheng Liang
    • , Yichen Cao
    • , Yan Chen
    • , Emma Labrot
    • , David Ruddy
    • , Susan Moody
    • , Huaixiang Hao
    • , Morvarid Mohseni
    • , Matthew LaMarche
    • , Juliet Williams
    • , Keith Hoffmaster
    • , Giordano Caponigro
    • , Fang Li
    •  & Jeffrey A Engelman
  4. Novartis Institutes for BioMedical Research, Basal, Switzerland.

    • Grainne Kerr
  5. Broad Institute of the Massachusetts Institute of Technology and Harvard, Cambridge, Massachusetts, USA.

    • Michael S Lawrence


  1. Search for Leila Dardaei in:

  2. Search for Hui Qin Wang in:

  3. Search for Manrose Singh in:

  4. Search for Paul Fordjour in:

  5. Search for Katherine X Shaw in:

  6. Search for Satoshi Yoda in:

  7. Search for Grainne Kerr in:

  8. Search for Kristine Yu in:

  9. Search for Jinsheng Liang in:

  10. Search for Yichen Cao in:

  11. Search for Yan Chen in:

  12. Search for Michael S Lawrence in:

  13. Search for Adam Langenbucher in:

  14. Search for Justin F Gainor in:

  15. Search for Luc Friboulet in:

  16. Search for Ibiayi Dagogo-Jack in:

  17. Search for David T Myers in:

  18. Search for Emma Labrot in:

  19. Search for David Ruddy in:

  20. Search for Melissa Parks in:

  21. Search for Dana Lee in:

  22. Search for Richard H DiCecca in:

  23. Search for Susan Moody in:

  24. Search for Huaixiang Hao in:

  25. Search for Morvarid Mohseni in:

  26. Search for Matthew LaMarche in:

  27. Search for Juliet Williams in:

  28. Search for Keith Hoffmaster in:

  29. Search for Giordano Caponigro in:

  30. Search for Alice T Shaw in:

  31. Search for Aaron N Hata in:

  32. Search for Cyril H Benes in:

  33. Search for Fang Li in:

  34. Search for Jeffrey A Engelman in:


L.D., F.L., A.T.S., A.N.H., C.H.B. and J.A.E. designed the study and analyzed the data. L.D., C.H.B. and J.A.E. wrote the paper. G.C., F.L., J.A.E. and C.H.B. supervised the studies. L.D., F.L., M.S., P.F., S.Y., K.Y., Y. Cao, Y. Chen, K.X.S. and E.L. performed cell line and biochemical studies. L.D., D.T.M., J.L. and H.Q.W. performed tumor xenograft studies. M.P., R.H.D. and D.L. developed or maintained patient-derived cell lines. G.K., F.L. and D.R. performed shRNA screen analyses. M.S.L. and A.L. performed NGS analysis. J.F.G., I.D.-J., L.F., E.L., S.M., H.H., M.M., M.L., J.W., K.H. and G.C. were involved with the study design. All authors discussed the results and commented on the manuscript.

Competing interests

H.Q.W., P.F., G.K., K.Y., J.L., Y. Cao, Y. Chen, E.L., D.R., S.M., H.H., M.M., M.L., J.W., K.H., G.C., F.L. and J.A.E. are employees of Novartis. J.F.G. has served as a compensated consultant or received honoraria from Bristol-Myers Squibb, Genentech, Ariad, Loxo, Incyte, Novartis, Merck and Clovis. A.T.S. has served as a compensated consultant or received honoraria from Pfizer, Novartis, Genentech, Roche, Ignyta, Blueprint Medicine, Daiichi-Sankyo, Ariad, Chugai, Taiho Pharmaceuticals and EMD Serono. C.H.B. has received research funding from Novartis and Amgen. A.N.H. has received research funding from Novartis, Relay Therapeutics and Amgen. The remaining authors have no financial interests to declare.

Corresponding authors

Correspondence to Cyril H Benes or Fang Li or Jeffrey A Engelman.

Supplementary information

PDF files

  1. 1.

    Supplementary Tables & Figures

    Supplementary Tables 1–4 & Supplementary Figures 1–11

  2. 2.

    Life Sciences Reporting Summary

About this article

Publication history






Further reading