2.1 Cell culture

The H2228 cell line was obtained from American Type Culture Collection. The H3122 cell line was a gift from Dr. John Minna. H2228 and H3122 cells were grown in RPMI 1640. Media was supplemented with 10% fetal bovine serum (FBS) and 1× antibiotics (Gibco Life Technologies), maintained at 37 °C in 5% CO₂/95% air, and passaged when cells reached 80–90% confluency.

2.2 Drugs

Alectinib (CH5424802) and ceritinib (LDK378) were purchased from Selleck Chemicals. Crizotinib (PF-2341066) was purchased from ChemieTek. NMS-E628 (NMS; RXDX-101) was obtained from Nerviano Medical Sciences (NCI MTA#31471-11). HSP90 inhibitors 17-AAG (17-allylamino-17demothoxygeldanamycin) and ganetespib were purchased Synta Pharmaceuticals. Drugs were dissolved in DMSO, aliquoted, and stored at −80 °C, and diluted in fresh medium before use. The final DMSO concentration in medium in all experiments was <0.1%.

2.3 Generation of ALK inhibitor resistance in H2228 and H3122 cells

H2228 and H3122 resistant lines were generated by continuous exposure to increasing concentrations of specific ALK inhibitor. H2228 cells and H3122 cells were exposed up to 2 μM and 3 μM NMS, respectively, and designated H2228 NMS_R(2) and H3122 NMS_R. H3122 was exposed up to 3 μM and 4 μM of crizotinib and ceritinib and designated H3122 Criz_R and H3122 LDK_R. Established resistant cell lines were maintained in medium containing maximal dose of ALK inhibitor to maintain selective pressure for ALK inhibitor resistance. In all experiments involving resistant cell lines, cells were seeded without relevant drug to eliminate the effect of drug interaction.

2.4 Western blot analysis

Cells were seeded in 6-well plates at a density of 5 × 10⁵ cells/well in drug-free media. The next morning, cell lysates were extracted using RIPA buffer (Sigma–Aldrich) containing Halt protease and phosphatase inhibitor cocktail (Thermo-Scientific) and clarified by centrifugation. Protein concentrations were quantified using Pierce BCA protein assay (Thermo-Scientific), and equal amounts of protein was separated by 4–20% SDS-PAGE, transferred onto PVDF membrane (Bio-Rad) for protein blot analysis, blocked with 5% non-fat milk in Tris-buffered saline with 0.1% Tween-20, and incubated with primary antibody overnight at 4 °C. Following incubation with appropriate horseradish peroxidase-conjugated secondary antibody (Bio-Rad) in blocking buffer, protein expression was detected using SuperSignal West Pico Chemiluminescent Substrate (Thermo-Scientific). Primary antibodies to phospho-ALK (Y1078), total ALK, E-cadherin, phospho-AXL (Y779), total AXL, phospho-ERK (T202/Y204), total ERK, and vimentin were purchased from Cell Signaling Technology. N-cadherin was purchased from BD Pharmingen. Primary antibody to twist was purchased from Santa Cruz Biotechnologies. β-actin or α-tubulin was used as a loading control.

2.5 Cell viability assays

Cells were seeded overnight at a density of 3000 cells per well in 96-well plates in RPMI 1640 containing 10% FBS. The next day, cells were treated with relevant kinase inhibitors for 72 h. Viable cell numbers were determined using CellTiterGlo (Promega) according to manufacturer's protocol. Each assay consisted of triplicates and was repeated at least twice. Data was expressed as percentages of control cells, calculated from the absorbance corrected for media-only background.

2.6 Cell migration and invasion assay

Migration and invasion of cells were assessed using transwells (12-well inserts; pore size 8uM; Millipore). 2 × 10⁵ cells in serum-free medium were plated in the top chamber of the transwell and the lower chamber was filled with medium supplemented with 10% FBS as a chemoattractant. After 24 h of incubation at 37 °C, cells were fixed with 4% paraformaldehyde and stained with 0.2% crystal violet. Non-migrating cells were removed using cotton swabs, and cells that migrated to the lower surface of the filter were counted using a light microscope. The invasion assay was conducted as described above except transwell inserts were coated with Matrigel and incubated at room temperature for 30 min prior to seeding cells.

2.7 Transfection of siRNA

Small interfering RNA (siRNA) oligonucleotides targeting AXL, Twist1, and vimentin were obtained from Dharmacon. Transfection of siRNA was performed using Lipofectamine RNAiMAX (Life Technologies) in accordance with manufacturer's protocol. Target gene expression was analyzed using western blot after 48 or 72 h of incubation. For cell viability assays, 3000 cells/well were seeded onto 96 well plate on day 0, transfected with siRNA on day 1, and treated with respective drug on day 2 for 72 h.

3.1 Generation of H3122 and H2228 cell lines resistant to ALK inhibitors

To investigate the mechanisms of acquired resistance to ALK inhibitors, H3122 and H2228 cell lines which harbor EML4-ALK rearrangement (variant 1 and 3, respectively) were selected and shown to be sensitive to 1st generation ALK inhibitors, crizotinib and NMS (Figure 1A). The 2nd generation ALK inhibitors, ceritinib and alectinib, were superior to the first 1st generation inhibitors in inducing cell death in H3122 cells. Therefore, these cell lines have clinical relevance for determining the mechanisms of resistance. Of note, NMS (renamed RXDX-101) was initially developed as an ALK inhibitor, but is also highly potent to Trk and ROS1 rearrangements, which make up two other molecularly based subtypes of NSCLC (Gower et al., 2014; De Braud et al., 2014; Vaishnavi et al., 2013). RXDX-101 is currently under clinical investigation in patients with solid tumors with relevant genetic alterations (NCT #02097810).

H3122 and H2228 cells were treated with continuous, increasing concentration of ALK inhibitors until they were growing at a concentration significantly above their IC₅₀. H3122 NMS_R and H2228 NMS_R cell lines both had IC₅₀ values greater than 2 μM (Figure 1B). These resistant cell lines showed dramatic changes to their morphology compared to parental H3122 and H2228 cell lines (Figure 1C). Both resistant cell lines underwent a morphological change consistent with epithelial-to-mesenchymal transition (EMT), a developmental process in which cells change polarity, reorganize their cytoskeleton, and acquire migratory and invasive properties to contribute towards cancer metastasis (Lamouille et al., 2014). H2228 NMS_R cells changed from a polarized shape to a fibroblastic morphology, consistent with EMT. On the other hand, the selective pressure of NMS drug induced a portion, but not all, of H3122 NMS_R cells to change morphologically. Of note, H3122 NMS_R and H2228 NMS_R did not reveal any secondary ALK mutations that can confer resistance to ALK inhibitors (data not shown). Additionally, H3122 NMS_R was cross-resistant to ceritinib and alectinib (Supplementary Figure S1). It is noteworthy that acquired resistance to crizotinib in H3122 (H3122 Criz_R) and to NMS in H2228 (H2228 NMS_R2) showed upregulation of heregulin (Supplementary Figure S2), a ligand that activates the HER family of receptor tyrosine kinase. The resistant lines can acquire resistance through various bypass resistance mechanisms and this data corroborates a previously published paper (Wilson et al., 2015).

3.2 Mechanisms of acquired resistance to 1st generation ALK inhibitors

EMT-associated invasive and migratory potential of H2228 NMS_R and H3122 NMS_R cell lines were assessed. Both resistant lines proved to have significantly increased invasive and migratory capabilities compared to parental control cells (Figure 2A,B). To reveal the potential mechanisms of acquired resistance to 1st generation ALK inhibitors, a series of EMT-related proteins such as E-cadherin, vimentin, twist, and AXL were examined (Figure 2C). Both resistant cell lines had downregulation of E-cadherin compared to parental lines, with H2228 NMS_R completely losing expression of E-cadherin, consistent with the extensive mesenchymal morphology. H2228 NMS_R also exhibited dramatic upregulation of vimentin, an important marker of EMT, whereas H3122 NMS_R has moderate increase of vimentin expression. H3122 NMS_R had significant upregulation of AXL kinase, and pAXL indicating activation of AXL. Additionally, H3122 NMS_R had upregulation of master regulating transcription factor, twist, which likely contributed towards the changes in gene expression that repress the epithelial phenotype and activate the mesenchymal phenotype (Lamouille et al., 2014). It is noteworthy that ZEB, Slug, and Snail transcription factors were examined, however, there was no detectable levels of protein. Collectively, these findings indicated that the mechanisms of resistance to 1st generation ALK inhibitors were associated with EMT, however, the mechanism through which EMT occurs is different and is cell-line dependent.

3.3 Knockdown of mesenchymal markers reduces invasive and migratory capabilities, but does not revert ALK inhibitor resistance

AXL is a receptor tyrosine kinase (RTK) that has previously been implicated in EMT, and activation of AXL can confer resistance to EGFR-targeted therapies in NSCLC (Kim et al., 2013; Vuoriluoto et al., 2011; Zhang et al., 2012). Twist transcription factor has been known to have a key role in EMT progression, specifically its ability to repress E-cadherin in cancer cells (Lamouille et al., 2014). E-cadherin is critical in maintaining the cell-to-cell contact in epithelial cells, and decreased expression of E-cadherin is necessary in the context of EMT. In H3122 NMS_R, AXL kinase was activated and twist was upregulated. To determine if modulation of these mesenchymal markers could restore sensitivity to ALK inhibitors, H3122 NMS_R was treated with siRNA specific for AXL or twist. Genetic knockdown of AXL using two different targeted siRNA sequences at 10 nM caused a loss of N-cadherin, which was acquired during EMT (Figure 3A). In addition, reduction in MAPK p38 was observed, a kinase that plays a role in invasion and migration. Although there were no changes to vimentin or twist following AXL siRNA treatment, the invasive and migratory capabilities of H3122 NMS_R were significantly reduced (Figure 3B,C). Knockdown of twist in H3122 NMS_R restored E-cadherin expression (Figure 3D), and also decreased its invasive and migratory capabilities (Figure 3E,F). Similarly to the knockdown of AXL, twist knockdown did not affect AXL or vimentin protein expression. Knockdown of AXL and twist in H3122 NMS_R (Figure 3G,H; Supplementary Figure S3), and vimentin in H3122 LDK_R (Supplementary Figure S5) did not restore sensitivity to ALK inhibitors. Thus, attempting to revert the resistant cell phenotype back to its epithelial state by genetic knockdown of upregulated mesenchymal markers suggest that EMT was involved in the physiological process of invasion and migration, but was not driving the acquired resistance.

3.4 Interrogation of EMT-associated acquired resistance to ceritinib in H3122

We generated ceritinib resistant H3122 cells by continuous exposure to increasing concentrations of the drug. When the cells acquired resistance to ceritinib, the cells showed a morphological change consistent with EMT (Supplementary Figure S4). H3122 LDK_R had a significantly higher IC₅₀ to ceritinib than parental cells (Supplementary Figure S5) and were resistant to crizotinib as well (Supplementary Figure S6). ALK gene sequencing of H3122 LDK_R detected no mutations that confer resistance to ALK inhibitors (data not shown). H3122 LDK_R had increased invasive and migratory capabilities, and this was confirmed by blotting for EMT markers (Figure 4A,B). H3122 LDK_R expressed high levels of AXL and vimentin. Treatment with vimentin siRNA caused reduction in AXL levels in H3122 LDK_R, whereas knockdown of AXL or twist had no effect on vimentin levels in H3122 NMS_R. This suggests that AXL is functioning downstream of vimentin in mediating EMT. Similar to the genetic knockdown of AXL and twist in H3122 NMS_R cells, vimentin knockdown in H3122 LDK_R did not restore sensitivity to ceritinib, suggesting that EMT is associated with, but does not drive, the resistance to ALK inhibitors (Supplementary Figure S5B). HSP90 inhibitors have previously been shown to be potent against EML4-ALK-positive NSCLC (Normant et al., 2011; Sang et al., 2013). H3122 cell lines that were grown to be resistant to 1st and 2nd generation ALK inhibitors and had morphological and functional changes consistent with EMT were sensitive to both 17-AAG and ganetespib (Figure 4C, Supplementary Figure S7). Ganetespib was shown to be more potent than 17-AAG in inducing cell death. When H3122 LDK_R cells were grown without the presence of ceritinib for 5 weeks, H3122 LDK_R (5wks), sensitivity to ALK inhibition was restored at 100 nM (Figure 4D). It is noteworthy that H3122 LDK_R cells grown without drug for up to 3 weeks did not restore sensitivity (Supplementary Figure S8).