NRAS Status Determines Sensitivity to SHP2 Inhibitor Combination Therapies Targeting the RAS–MAPK Pathway in Neuroblastoma

Neuroblastoma, a tumor of the peripheral nervous system and the most common pediatric extracranial solid tumor, is clinically and biologically heterogeneous. Although the majority of patients diagnosed with low- or intermediate-risk neuroblastoma are cured with surgery alone or low doses of chemotherapy, fewer than half of patients with high-risk disease survive despite intensive chemotherapy, radiation, immunotherapies, and stem cell transplant (1–3). Thus, there is significant interest in identifying aberrant signaling pathways that may be targeted therapeutically. However, whole-exome and genome-sequencing studies have demonstrated that recurrent somatic mutations are relatively rare in neuroblastoma at diagnosis, with the most common alterations being MYCN amplification (20%), TERT rearrangements (23%), NF1-loss (6%), and ALK (9%) or PTPN11 (3.5%) mutations (4–9). Analyses focused on relapsed neuroblastoma reveal a higher incidence of coding mutations (10, 11). One recent study comparing paired tumors at diagnosis and relapse reported that 78% of mutations detected in relapse samples were predicted to activate the RAS–MAPK pathway, including mutations in RAS, NF1, ALK, and PTPN11 (10). These results suggest that pharmacologic targeting of this pathway may be beneficial for the treatment of recurrent neuroblastoma.

The RAS–MAPK signaling pathway regulates a variety of cellular processes and is commonly dysregulated in cancer. Rat sarcoma (RAS) oncogenes KRAS, HRAS, and NRAS encode small membrane-bound GTPase proteins that regulate cellular proliferation, differentiation, and survival. Under physiologic conditions, RAS proteins cycle between their GTP-bound active and GDP-bound inactive states to regulate activation of downstream effectors proto-oncogene serine/threonine kinase (RAF), MAPK kinase (MEK), and extracellular-signal-regulated kinase (ERK; refs. 12, 13). RAS-activating mutations predominantly localize to hotspot codons 12, 13, and 61, and are detected in 20%–40% of adult-onset cancers (14). While somatic RAS aberrations are typically considered classic cancer drivers, germline mutations in genes in the RAS–MAPK pathway often give rise to developmental disorders known collectively as “RASopathies.” For example, Noonan and Costello syndromes are associated with mutations in PTPN11 or RAS, respectively, and these patients have an increased risk of developing childhood embryonal cancers, including neuroblastoma (15–19). Furthermore, evidence in zebrafish suggests that PTPN11 cooperates with MYCN to promote neuroblastoma tumorigenesis and RAS pathway activation (20).

Because therapies aimed at directly targeting RAS have not shown clinical efficacy, more recent attempts to target the RAS pathway have focused on inhibiting its upstream or downstream effectors. We previously reported that the PTPN11-encoded tyrosine phosphatase SHP2 is an activator of RAS, which promotes its dephosphorylation to increase RAS binding to RAF and activation of the RAS–MAPK pathway (21, 22). Notably, pharmacologic inhibition of SHP2 suppressed HRAS mutant–driven glioblastoma in mice and decreased tumor growth and burden in pancreatic ductal adenocarcinoma patient-derived xenografts harboring KRAS mutations (22, 23). SHP2 inhibition has also been shown to be effective for other RAS-driven cancers including myeloid leukemia, melanoma, non–small cell lung cancer (NSCLC), and osteosarcoma (24–28). However, the sensitivity of neuroblastoma harboring RAS–MAPK alterations to SHP2 inhibitors is unknown.

Here, we show that RAS mutations in neuroblastoma are associated with decreased sensitivity to SHP2 inhibitors NSC-87877 (29), II-B08 (30), RMC-4550 (31), and SHP099 (27), and that NRAS^Q61K mutation confers resistance to SHP2 inhibition. However, dual inhibition of SHP2 and RAS effector RAF, MEK, or ERK show synergistic effects in neuroblastoma cells harboring RAS-activating mutations, and combination of SHP099 with MEK inhibitor trametinib reduces tumor volume and increases survival in vivo. These results suggest that in certain tumors, based on the genetic status of RAS–MAPK signaling components, combinations of drugs targeting this pathway could be effective strategies for relapsed neuroblastoma.

Human neuroblastoma cell lines, SK-N-AS, CHP-212, IMR-32, SK-N-SH, SK-N-Fl, Kelly, LAN-5, LAN-6, and SH-EP were purchased from the ATCC or obtained from Dr. Patrick Reynolds (COG Childhood Cancer Repository, https://www.cccells.org). NB-EB cell line was provided by Dr. Marielle E. Yohe (NCI, NIH, Bethesda, MD). SK-N-AS-TR cells expressing luciferase were described previously (32). Media and culture conditions for cell lines are listed in Supplementary Methods. Short tandem repeat (STR) cell authentication (TCAG DNA/Sequencing Facility, Toronto, Ontario, Canada) and Mycoplasma testing (InvivoGen) was performed prior to experiments. Whole-exome sequencing was performed to assess KRAS, NRAS, and PTPN11 mutations in all cell lines.

For in vitro assays, inhibitors were diluted in DMSO to a final stock concentration of 50 mmol/L or 100 mmol/L, and added to 10% FBS-containing media. For in vivo experiments, SHP099 was resuspended in final concentrations of 0.6% methylcellulose, 0.5% Tween 80, and 0.9% saline, each added sequentially and in that order. Ulixertinib was dissolved in 0.5% methylcellulose. Trametinib was dissolved in 4% DMSO in corn oil. Drug information is listed in Supplementary Methods.

Cell proliferation was assessed by bromodeoxyuridine (BrdU) incorporation (Cell Signaling Technology). A total of 3–5 × 10³ cells were seeded in 96-well plates, treated with indicated inhibitors for 48 or 72 hours, and analyzed 16 hours after BrdU addition. The absorbance was measured at λ450 using a spectrophotometer plate reader (VersaMax, Molecular Devices). Cell viability was assessed by AlamarBlue assays (Invitrogen). A total of 3–7 × 10³ cells were seeded per well in 96-well plates and treated with inhibitors for 72 hours. AlamarBlue reagent was added 16–18 hours prior to assessing fluorescence intensity on a microplate reader (Spectra MAX Gemini EM, Molecular Devices), with a λ540 excitation/λ590 emission filter. IC₅₀ curves were generated and calculated using GraphPad Prism 6 software (GraphPad Software Inc). All experiments were performed in triplicate or sextuplicate wells for each condition and repeated at least three times.

Endogenous SHP2 in HEK293T and SK-N-AS cells was silenced using GIPZ Lentiviral shRNA plasmids (Horizon Discovery) with the indicated sequences: shPTPN11-#1 (5′–TAGCGTATAGTCATGAGCG); shPTPN11-#3 (5′–ATATTTGTATATTCGTGCC); nontarget shCTL (5′–TAAACATCCATATCAACAC). Cells were selected as described in ref. 23.

Stable cell lines were transfected with NRAS and PTPN11 mutants generated by site-directed mutagenesis (21, 23) and selected as described in Supplementary Methods. RAS-GTP was precipitated using immobilized GTP beads (Jena Bioscience) and analyzed as described in Supplementary Methods.

Animal studies were performed in accordance with University Health Network and SickKids Institutional Animal Utilization Protocol guidelines. Male NOD/SCID mice (6–8 weeks) were injected subcutaneously with 1 or 5 × 10⁶ cells in 0.1 mL suspension containing 50% Matrigel (Corning) in PBS. Bioluminescence imaging and analysis, and tumor volumes were monitored at least twice per week and calculated as described previously (32). Once tumors reached 60–100 mm³ in size and 3.5–4.0 × 10⁸ photons/second in bioluminescence signal mice were randomly cohorted to receive a 0.2 mL suspension containing either vehicle, SHP099 (100 mg/kg, every other day), trametinib (0.25 mg/kg, once daily), or the combination of SHP099 plus trametinib; or vehicle, SHP099 (75 mg/kg, every other day), ulixertinib (75 mg/kg twice daily), or the combination of SHP099 plus ulixertinib by oral gavage 5 days per week. Mice reached endpoint once tumor measurements surpassed 10 mm in two of three dimensions, or earlier depending on their health, weight, or length of the study.

Data reported represents the mean and SD of three independently conducted experiments, each performed in triplicate or sextuplicate. Unpaired two-tailed variance Student t test was used to assess statistical significance between two treatment groups. ANOVA followed by a post-Tukey was used for pair-wise comparisons. For Kaplan–Meier survival curves, a log-rank (Mantel–Cox) test was performed. All statistical analyses were performed using GraphPad Prism 6 (GraphPad Software Inc) or SPSS Statistics (IBM). P value < 0.05 was considered statistically significant. Excess over Bliss (EOB) assessments were calculated as described with EOB values > 0 considered synergistic (33).

To determine whether neuroblastoma cells were sensitive to SHP2 inhibitors NSC-87877, II-B08, RMC-4550 and SHP099, we selected a panel of human neuroblastoma cell lines with differing genetic status of RAS, ALK, NF1, and MYCN (Supplementary Table S1). The basal expression and phosphorylation status of RAS–MAPK components, including SHP2, BRAF, MEK, and ERK, was assessed in all neuroblastoma cell lines (Fig. 1A). Neuroblastoma cells were treated with increasing concentrations of SHP2 inhibitors to determine their half-maximal inhibitory concentration (IC₅₀). ALK aberrations, along with RAS mutants, were considered RAS-associated mutations (RAM) because all ALK single-nucleotide variants (SNV) expressed in the included neuroblastoma cell lines have been shown to promote activation of the RAS–MAPK pathway (34). Interestingly, neuroblastoma cells harboring RAS mutations showed relative resistance to the allosteric SHP2 inhibitors SHP099 and RMC-4550 (Fig. 1B; Supplementary Fig. S1A–S1F), as well as SHP2 catalytic inhibitors II-B08 and NSC-87877 (Fig. 1C; Supplementary Fig. S2A and S2B). The average IC₅₀ values were lowest for cells with wild-type RAS (no-RAM) and ALK missense mutations (ALK^mut), and highest for cells with RAS mutations (RAS^mut; Fig. 1D and E; Supplementary Fig. S1A and S1B). The SHP099 IC₅₀ for the most sensitive neuroblastoma cells was similar to the higher IC₅₀ ranges reported for non-neuroblastoma cells (25, 35–37). Notably, we did not observe any significant correlation between MYCN amplification (MYCN^amp) and sensitivity to SHP2 inhibition (Supplementary Fig. S1C and S1E; Supplementary Fig. S2C). Growth inhibition was, at least in part, due to induction of apoptosis as increased expression of cleaved PARP was detected in NRAS wild-type (NRAS^WT) cells, but minimally in mutant (NRAS^Q61K) cells treated with SHP099 (Fig. 1F) and II-B08 (Supplementary Fig. S2D).

To determine whether pharmacologic inhibition of SHP2 differentially affected downstream signaling, Western immunoblots of cells exposed to vehicle or drugs were performed. Although treatment with SHP2 inhibitor SHP099 or II-B08 was associated with decreased phosphorylated SHP2 (p-SHP2) levels in NRAS^WT and NRAS^Q61K cells, phosphorylation of the downstream effector ERK (p-ERK) decreased in cells with NRAS^WT expression, but not in NRAS^Q61K cells (Fig. 1G and H; Supplementary Fig. S1D and S1F). These results suggest that treatment with SHP2 inhibitors does not lead to RAS-MAPK inactivation in neuroblastoma cells harboring NRAS^Q61K mutation. Furthermore, because mutations in the SHP2-encoding PTPN11 gene are observed in some neuroblastomas and there are conflicting data as to whether PTPN11 mutational status determines SHP2 inhibitor sensitivity in non-neuroblastoma cells, we assessed the efficacy of SHP2 inhibitors in isogenic SH-EP and Kelly cells that ectopically overexpress PTPN11 mutations previously identified in human neuroblastoma tumors (9). In comparison with PTPN11 wild-type cells, those expressing PTPN11-mutant proteins displayed similar sensitivities to SHP099 and II-B08 and showed similar decreased levels of p-ERK following treatment (Supplementary Fig. S3A–S3E). These results suggest that single-agent SHP2 inhibition might be effective for tumors with wild-type RAS, but not beneficial for tumors with NRAS alterations, such as NRAS^Q61K.

To determine whether expression of NRAS^Q61K mutant in neuroblastoma cell lines SK-N-AS and CHP-212 directly mediates the observed differences in sensitivity to SHP2 inhibitors, we generated isogenic cell lines. SH-EP cells, which endogenously express NRAS^WT, were engineered to overexpress either NRAS wild-type (SH-EP-NRAS^WT), mutant (SH-EP-NRAS^Q61K), or an empty vector control (SH-EP-EV). As expected, higher levels of wild-type or mutant NRAS were associated with increased activation of p-ERK (Fig. 2A) and cell proliferation (Supplementary Fig. S4A) compared with control. SH-EP-NRAS^Q61K cells were markedly more resistant to SHP099 than cells expressing either endogenous or exogenous NRAS^WT, as determined by increased viability and higher IC₅₀ (Fig. 2B; Supplementary Fig. S4B–S4E). In contrast to resistant SH-EP-NRAS^Q61K cells, following SHP099 treatment, SH-EP-EV and SHEP-NRAS^WT cells showed lower proliferation rates detected by BrdU incorporation (Fig. 2C) and increased levels of cleaved caspase-3 and PARP (Fig. 2D), suggesting higher rates of apoptosis. The analysis of downstream signaling effectors showed decreased levels of p-ERK in cells expressing NRAS^WT, but negligible changes were observed in SHEP-NRAS^Q61K cells exposed to SHP099 (Fig. 2E). These results suggest that Q61K mutation in NRAS confers resistance to SHP2 inhibitors.

In addition, NRAS^WT cells showed a marked reduction in p-ERK levels following SHP2 (PTPN11) knockdown (Fig. 2F) compared with NRAS^Q61K cells, which exhibited negligible change in ERK activation (Fig. 2G). Moreover, sensitivity to SHP099 was significantly reduced in NRAS^WT cells, but not altered in NRAS^Q61K cells upon SHP2 knockdown. Similar results were observed upon treatment with SHP099 alone or in combination with other inhibitors (Supplementary Fig. S4F and S4G). These findings suggest that SHP2 inhibitors attenuate RAS–MAPK signaling due, in part, to specific SHP2 inhibition.

Inhibition of SHP2 in combination with MEK or ALK inhibition has been reported to reduce growth in vitro, and tumor progression in vivo in RAS-mutant melanoma, gastroesophageal, lung, pancreatic, and triple-negative breast cancer models (35, 38–43). We asked whether relative resistance to SHP099 in NRAS^Q61K cells could be overcome with combination strategies targeting other RAS signaling components. Resistant neuroblastoma cells were treated with SHP2 inhibitors alone or in combination with other MAPK inhibitors including the BRAF inhibitor vemurafenib, MEK inhibitor trametinib, or ERK inhibitor ulixertinib (Fig. 3A). We first assessed sensitivity to SHP2 inhibition in the SHP099-resistant NRAS^Q61K-expressing SK-N-AS cell line, and showed that combination treatment with vemurafenib, trametinib, or ulixertinib sensitized these otherwise resistant cells to SHP099 (Supplementary Fig. S5A). Next, to assess combination regimens, NRAS-mutant cells were treated with increasing doses of MAPK inhibitors to determine concentrations that resulted in ≤30% growth inhibition. We then treated NRAS^Q61K resistant neuroblastoma cell lines with either SHP099 or anti-MAPK drugs alone or in combination. Treatments with single agents showed mild growth inhibition, whereas combination treatments markedly reduced viability in both NRAS-mutant SK-N-AS and, to a larger extent, in CHP-212 cells (Fig. 3B and C). Similar effects were observed with II-B08 in resistant SK-N-AS and more sensitive IMR-32 cells (Supplementary Fig. S5B and S5C). We next assessed activation of downstream MAPK components upon combination treatments in cells harboring NRAS^Q61K. Consistent with our cell viability studies, SHP099 treatment with ulixertinib was associated with some inhibition of downstream MAPK signaling, whereas SHP099 plus trametinib led to stronger inactivation in SK-N-AS and CHP-212 cells (Fig. 3D and E).

To determine whether SHP099 could sensitize NRAS-mutant cells to vemurafenib, trametinib, or ulixertinib, the two relatively resistant cell lines were treated with a wide range of anti-MAPK drugs alone or in combination with SHP099 to determine their IC₅₀. As expected, vemurafenib alone showed negligible sensitivity; however, in combination with SHP099, the IC₅₀ was potently reduced in NRAS^Q61K-expressing cell lines (Fig. 3F; Supplementary Fig. S5D). Single treatments with trametinib and ulixertinib showed mild sensitivity; however, the addition of SHP099 further sensitized both NRAS^Q61K cell lines (Fig. 3F; Supplementary Fig. S5E and S5F). These results suggest that neuroblastoma cells that express NRAS^Q61K are relatively resistant to SHP2 inhibitors but can be markedly sensitized to combination strategies targeting different effectors of the RAS–MAPK signaling cascade.

Acquired resistance to inhibitors targeting MAPK or ALK has been a recurrent challenge in the treatment of patients. Therefore, we investigated whether combination with SHP2 inhibitors could enhance MAPK inhibitor effectiveness in relatively resistant neuroblastoma cells. We first assessed sensitivity to trametinib in a panel of neuroblastoma cell lines with diverse genetic profiles (no-RAM, ALK^mut, or NRAS^Q61K). In agreement with recent publications (10, 44, 45), in comparison with ALK^mut and no-RAM cells, those expressing NRAS-mutant proteins were sensitive to trametinib (Supplementary Fig. S6A). Interestingly, addition of SHP099 at low doses sensitized trametinib-resistant cells with logIC₅₀ values significantly lower in both no-RAM and ALK^mut cells treated with SHP099 and trametinib combinations as compared with trametinib alone (Supplementary Fig. S6B).

To assess drug interactions and determine synergistic combinations, we used the EOB model (33). Compared with other methods, such as Chou–Talalay, EOB optimally controls for high variability of drug responses. We compared the combined effect of SHP099 plus trametinib between NRAS^WT and NRAS^Q61K-expressing cells treated with doses equal or lower to their IC₃₀ concentrations. Synergy was observed in all neuroblastoma models; however, NRAS^Q61K-expressing cells showed the greatest effects, as determined by EOB >25 scores (Fig. 4A). Moreover, in NRAS^Q61K cells treated with SHP099 plus trametinib, activation of RAS–MAPK downstream effectors was particularly reduced as shown by decreased p-MEK and p-ERK levels (Fig. 4B). Similarly, SHP099 and ulixertinib combinations were synergistic and associated with markedly decreased ERK in NRAS^Q61K cells as compared with non-NRAS mutants (Fig. 4C and D). Consistent with other reports (46, 47), treatment with ulixertinib elicited increased p-ERK levels despite decreasing total ERK expression (Fig. 4D) and inhibiting the activation of its downstream targets (Supplementary Fig. S6C). Similar effects were observed with II-B08 in NRAS-mutant SK-N-AS and no-RAM IMR-32 cells (Supplementary Fig. S6D and S6E). Furthermore, anti-MAPK drugs plus SHP099 or II-B08 inhibitors were also synergistic in SK-N-SH and LAN-5 cells harboring ALK mutations (Supplementary Fig. S7A–S7C). Finally, assessment of inhibitor combinations of SHP2 with MEK or ERK inhibitors in our established isogenic SH-EP cell lines also revealed synergy (EOB>25) in NRAS^WT and, to a larger extent in NRAS^Q61K-overexpressing cells (Supplementary Fig. S4D and S4E). These results suggest that NRAS^Q61K-associated resistance to SHP2 inhibition could be successfully overcome with dual targeting of RAS–MAPK pathway components.

To elucidate the mechanisms underlying the synergy observed between SHP099 and trametinib, we assessed RAS activation and signaling. Upon treatment with SHP099, we observed a mild reduction in RAS–GTP association in NRAS^WT, but not NRAS^Q61K cells (Fig. 4E). Notably, trametinib alone or in combination with SHP099 showed markedly reduced RAS-GTP binding in both NRAS^WT and NRAS^Q61K-expressing cells.

To determine whether treatment with SHP099 together with trametinib was effective in vivo, we assessed the MTD of both SHP099 and trametinib alone and in combination in NOD-SCID mice. We observed no significant weight loss or toxicities in SHP099 (100 mg/kg) alone-treated, trametinib (0.25 or 1 mg/kg) alone-treated, or combination-treated groups (100 mg/kg SHP099 plus 0.25 mg/kg trametinib) administered orally daily for up to 19 days (Supplementary Fig. S8A–S8C). Similar to other studies (35, 43), we found that SHP099 long-term tolerability in combination increased when SHP099 was administered every other day.

For combination efficacy studies, SK-N-AS-TR cells containing a luciferase reporter, harboring the SHP099-resistant NRAS^Q61K mutation were generated (32). SK-N-AS-TR cells treated with SHP099 had an equivalent IC₅₀ to SK-N-AS cells in vitro (Supplementary Fig. S8D). Once SK-N-AS-TR xenografts reached approximately 80 mm³ and had similar levels of detectable luciferase activity mice were treated with vehicle control, SHP099 (100 mg/kg, every other day), trametinib (0.25 mg/kg, once daily), or SHP099 (100 mg/kg, every other day) plus trametinib (0.25 mg/kg, once daily) via oral gavage, and mouse weight and tumor size were monitored for up to 37 days. At these doses, SHP099, trametinib, or combination treatments did not elicit significant weight loss (Supplementary Fig. S8E). By day 16 of treatment, in comparison with mice treated with vehicle or single-agent treatment, mice administered SHP099 plus trametinib showed improved tumor response as determined by decreased tumor size (Fig. 5A; Supplementary Fig. S9A and S9B). Treatment with SHP099 or trametinib alone resulted in modest growth inhibition of SK-N-AS-TR xenografts, whereas combination treatment significantly delayed tumor growth as compared with vehicle or single-agent treatments (Fig. 5B). By day 14, imaging of mice receiving combination treatment showed reduced bioluminescence levels compared with mice treated with vehicle or either agent alone (Fig. 5C). Furthermore, survival of mice treated with SHP099 plus trametinib was significantly higher than survival of mice exposed to single agents, as determined by Kaplan–Meier survival analysis (Fig. 5D). The death of three mice treated with vehicle, SHP099 plus trametinib or SHP099 alone on days 5, 14, and 19, respectively, were attributed to causes unrelated to tumor burden. Moreover, immunoblots of representative treated tumor lysates showed increased apoptosis and potent reduction of p-SHP2, p-MEK, and p-ERK levels in mice treated with combination therapy compared with vehicle or drugs alone (Fig. 5E; Supplementary Fig. S9C and S9D), consistent with those detected in vitro. In addition, upon combination treatment, we observed attenuation of RAS-GTP binding (Fig. 5F; Supplementary Fig. S9C), as well as reduced activation of receptor tyrosine kinases (RTK) PDGFR and EGFR (Supplementary Fig. S9E), thus suggesting possible additional mechanisms underlying dual SHP2/MEK inhibition. Furthermore, SHP099 (75 mg/kg, every other day) in combination with ulixertinib (75 mg/kg, twice daily) also resulted in similar tumor growth delay in vivo (Fig. 5G; Supplementary Fig. S10A–S10C). Together, these results support the clinical potential of inhibiting SHP2 in combination with other RAS–MAPK inhibitors such as trametinib or ulixertinib in the treatment of neuroblastomas with RAS mutations, which are more commonly detected at recurrence.

Relapsed neuroblastoma remains a difficult challenge, as more than 50% of patients with high-risk neuroblastoma develop recurrent often chemoresistant tumors. Although activation of the ALK–RAS–MAPK pathways has been shown to be more prevalent in relapse-specific samples (10, 11), there are few pediatric trials targeting the MAPK pathway and no studies evaluating efficacy in patients with neuroblastoma have been reported. Because direct targeting of RAS has historically been unsuccessful, other molecular-based approaches have been explored to indirectly inactivate RAS and the downstream MAPK pathway, including pharmacologic targeting of SHP2, which results in partial-to-complete inactivation of the RAS–MAPK pathway (22, 23, 48). Preclinical reports support SHP2 inhibitors as a promising therapy for RAS-associated cancers, leading to several phase I SHP2 inhibitor trials. Our results support using SHP2 inhibitor combination therapies to target relapsed neuroblastoma, in part, based on RAS status.

Although the connection between ALK and RAS are not fully understood, ALK and NRAS mutations in neuroblastoma are both associated with hyperactive MAPK signaling cascade (10, 34). Here, we show that ALK-mutant neuroblastoma cells are sensitive to SHP2 inhibitors and that cells ectopically expressing neuroblastoma-associated PTPN11 mutations respond to SHP099 at doses that are comparable with cells with wild-type PTPN11, consistent with other reports (24). However, PTPN11 effects may be difficult to detect given the relatively high IC₅₀ of SH-EP and Kelly cells. In contrast, although MYCN is thought to cooperate with SHP2 in neuroblastoma tumorigenesis in zebrafish (20), our results showed no correlation between MYCN status and sensitivity to SHP2 inhibitors. At the time of neuroblastoma recurrence, mutations of RAS are commonly detected and importantly, our results support a correlation between sensitivity and RAS status, as NRAS- and KRAS-mutant neuroblastoma cells showed a relative resistance to SHP2 inhibitors. Furthermore, data from cells overexpressing NRAS^Q61K, a common RAS mutation in neuroblastoma, support a direct consequence of this mutation in mediating decreased cell sensitivity and growth inhibition following SHP099 treatment. This increased resistance was consistent with observations from SHP2-depleted NRAS^Q61K cells. These findings are in line with recent reports that identified Q61 codon mutations in KRAS as critical predictors of resistance to SHP2 inhibitors (40, 42).

While results for clinical studies with MAPK pathway–targeted drugs in relapsed neuroblastoma have not been reported, several MEK inhibitors have shown promising single-agent results in neuroblastoma in vitro and in vivo (10, 45, 49). However, resistance to anti-MAPK drugs has been reported in KRAS-mutant lung, colon, and pancreatic cancers (50, 51), and NRAS-mutant melanomas (52) and neuroblastomas (45, 49). Because mechanisms of acquired resistance often include de novo mutations or amplifications that re-activate the MAPK pathway, strategies to treat relapsed tumors increasingly utilize combinations of drugs targeting multiple pathways or pathway components. Here, we showed that combination inhibition of critical effectors of the RAS–MAPK pathway effectively overcomes SHP2 inhibitor resistance in NRAS-mutant neuroblastoma. Similar to previous reports in KRAS-mutant lung, skin, and breast cancer models (35, 36, 40, 43), we found that in neuroblastoma, SHP2 inhibitors potently synergize with trametinib and decrease RAS–GTP association, RAS–MAPK and RTK signaling activation, and cell survival in vitro, and reduce tumor burden in vivo. Furthermore, we show that SHP099 synergizes with ulixertinib in vitro and in vivo, and that both trametinib and ulixertinib augment SHP2 inhibitor–mediated growth inhibition in cells harboring NRAS mutations commonly detected in relapsed neuroblastoma. Moreover, cells with ALK and non-RAS mutations were also sensitive to both SHP2 inhibitors alone or in combination with trametinib and ulixertinib. This may provide preliminary insights into novel combinations for tumors with specific mutations given that preclinical neuroblastoma studies have shown that monotherapy with trametinib inhibits the growth of RAS-mutant, but not ALK-driven neuroblastomas (10, 53). However, similar to our findings in SHP099-resistant NRAS-mutant cell lines, it is possible that trametinib-resistant ALK-mutant cells and tumors may also be sensitive to combination regimens that target multiple components of the RAS–MAPK pathway. Furthermore, although ALK mutations are common in neuroblastoma at diagnosis and recurrence, many of the hotspot mutations are resistant to current ALK tyrosine kinase inhibitors (TKI; refs. 34, 54). Thus, it will be important to determine whether ALK inhibitors may also be combined with SHP2 inhibitors in neuroblastomas harboring ALK aberrations. A recent publication has demonstrated that this strategy was effective in vitro and in vivo in ALK-TKI–resistant NSCLC harboring ALK fusions (39).

Our results show that in neuroblastoma, treatment with SHP2 inhibitors as a single agent might be an effective approach for certain tumors that do not harbor RAS mutations such as NRAS^Q61K, which is the most commonly reported RAS mutation in relapsed neuroblastoma. We also show that the resistance to SHP2 inhibitors can be overcome with combination inhibition of downstream RAS-MAPK components. Taken together with other studies that have highlighted the clinical potential of SHP2 inhibitors for non-neuroblastoma tumors (13, 20, 26, 35, 39, 40), our results suggest that dual inhibition of SHP2 with MEK or ERK may be effective regimens for neuroblastoma. Importantly, future studies will be required to determine whether other activating mutations of the RAS–MAPK pathway, including ALK, may influence sensitivity.

No potential conflicts of interest were disclosed.

I. Valencia-Sama: Conceptualization, data curation, formal analysis, validation, methodology, writing-original draft, project administration, writing-review and editing. Y. Ladumor: Data curation, formal analysis, validation. L. Kee: Data curation, validation, project administration. T. Adderley: Data curation, validation, project administration. G. Christopher: Data curation, validation. C. Robinson: Methodology. Y. Kano: Methodology. M. Ohh: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, methodology, writing-original draft, project administration, writing-review and editing. M. Irwin: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, methodology, writing-original draft, project administration, writing-review and editing.

This project was funded by the Canadian Institutes of Health Research and James Fund for Neuroblastoma Research. We thank members of the Irwin and Ohh laboratories for critical discussions and reading of this manuscript. This work is supported by grants from the Canadian Institutes of Health Research (PJT-162228 to M.S. Irwin; PJT-166005 to M. Ohh), Sick Kids Neuroblastoma Research (to M.S. Irwin), James Fund (to M.S. Irwin), Lilah's Fund (to M.S. Irwin), and Sebastian's Superheroes (to M.S. Irwin). I. Valencia-Sama is supported by The Hospital for Sick Children RESTRACOMP award, Connaught International Scholarship for Doctoral Students, and CIHR Vanier Canada Graduate Scholarship.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.