Neuropilin-2b facilitates resistance to tyrosine kinase inhibitors in non-small cell lung cancer

Abstract

Objective: Innate and acquired resistance is the principle factor limiting the efficacy of tyrosine kinase inhibitors in lung cancer. We have observed a dramatic upregulation of the cell surface co-receptor neuropilin-2b in lung cancers clinically treated with tyrosine kinase inhibitors correlating with acquired resistance. We hypothesize that neuropilin-2b plays a functional role in acquired tyrosine kinase inhibitor resistance.

Methods: Non-small cell lung cancer proliferation and survival were determined during chronic tyrosine kinase inhibitor exposure in the presence or absence of neuropilin-2b knock-down. Interactions of neuropilin-2a and neuropilin-2b isoforms with PTEN and GSK3β were assessed by immunoprecipitation. Neuropilin-2a and neuropilin-2b mutants deleted for their cytoplasmic domains were used to identify regions responsible for neuropilin-2b-GSK3β interaction. Because GSK3β is known to phosphorylate and degrade PTEN, phospho-PTEN and total PTEN levels were assessed after transfection of neuropilin-2a and neuropilin-2b wild-type and mutant constructs.

Results: Non-small cell lung cancer cells chronically treated with gefitinib or osimertinib developed drug resistance and exhibited logarithmic growth in the presence of endothelial growth factor receptor tyrosine kinase inhibitors. However, neuropilin-2b knockdown cells remained sensitive to gefitinib. Likewise, neuropilin-2b knockdown suppressed and neuropilin-2a knockdown enhanced cellular migration. Acquired drug resistance and cell migration correlated with neuropilin-2b-dependent AKT activation with the intermediate step of GSK3β-dependent PTEN degradation. A specific binding site for GSK3β on the cytoplasmic domain of neuropilin-2b was identified with truncated protein constructs and computer modeling.

Conclusions: Neuropilin-2b facilitates non-small cell lung cancer resistance to tyrosine kinase inhibitors, and this biological effect relates to AKT activation. Neuropilin-2b GSK3β interactions appear to be essential for PTEN degradation and AKT activation in lung cancer cells. Disruption of the neuropilin-2b GSK3β interaction may represent a novel treatment strategy to preserve sensitivity to tyrosine kinase inhibitors in non-small cell lung cancer.

Keywords: lung cancer; migration; resistance; survival; tyrosine kinase inhibitor.

FIGURE 1.

Schematic molecular model of the relationship between cellular growth of EGFR-mutant lung tumors and the link to phosphatidylinositol-3 kinase (PI3K) and activation of AKT, a kinase known to promote cell survival and migration. TKIs, such as gefitinib or osimertinib, block this pathway at the level of the EGFR receptor, stagnating cell growth and inducing apoptosis. When NRP2b is present, GSK3β is recruited to phosphorylate and degrade NRP2a-recruited PTEN, leading to enhanced AKT activation with promotion of survival and metastasis. This effect is predicted to enhance the effect of residual EGFR activity remaining in the presence of EGFR inhibitors. When NRP2b is absent, NRP2a-recruited PTEN is stabilized, opposing PI3,4,5P3 production by PI3K and thereby inhibiting AKT, survival, and metastatic spread, thus enhancing the effect of EGFR TKI.

FIGURE 2.

Drug-tolerant persister cells require NRP2b. A, EGFR mutant PC-9 NSCLC cells (1 × 10⁵/dish), knocked down for NRP2b, NRP2a, or with scrambled control shRNA (as indicated), were treated with gefitinib (2.5 μM) for 4 weeks. Surviving drug-tolerant persister cells were then cultured in drug-free medium to permit colony formation and quantitation. B, HCC827 NSCLC cells were treated as in (A) except the EGFR TKI osimertinib was used (2.5 μM) for 9 days before expansion into colonies. Significance was determined by 2-tailed t tests. ***P < .001.

FIGURE 3.

Sensitization to EGFR-TKI by NRP2b knockdown. A, NCI-H322 NSCLC cells (EGFR-wild type), knocked down with shNRP2b or nontargeting shCtl, were cultured with 125 nM gefitinib for 4 weeks. Weekly cell counts for the 2 genotypes are plotted on a linear scale versus time. B, Weekly cell counts for dimethyl sulfoxide–treated NCI-H322 (shNRP2b or shCtl) are plotted on a log₁₀ scale versus time. C, EGFR-wt NCI-H1703 cells, knocked down with shNRP2b or shCtl, were cultured in gefitinib or dimethyl sulfoxide, as indicated. Weekly cell counts for shNRP2b or shCtl cells are plotted on a log₁₀ scale versus time. Significance determined by 2-tailed t tests; *P < .05. ***P < .001. n.s., Not significant.

FIGURE 4.

NRP2b inhibits PTEN expression. A, HEK293 cells were induced for 48 hours with doxycycline to express NRP2a or NRP2b, as indicated, and serum starved for the final 24 hours. Cultures were then exposed briefly (10 minutes) to fetal bovine serum at concentrations ranging from 0% to 10%, harvested and analyzed by western blot for PTEN, phosphorylated PTEN (p-PTEN; on Thr366), phosphorylated-AKT (p-AKT; on Ser473), total AKT (pan-AKT), NRP2, and NRP2b. A, β-actin control verified equal loading. B, Densitometric quantitation of mean phospho-AKT and total-AKT signals from (A); control (v), NRP2a (a), and NRP2b (b) expressing samples are plotted as ratios (pAKT:AKT). C, Quantitation of mean PTEN levels from (A). D, Quantitative real-time reverse transcription polymerase chain reaction analysis of PTEN mRNA levels from (A). Significance determined by 2-tailed t tests; **P < .01. ***P < .001. n.s., Not significant.

FIGURE 5.

GSK3β interacts with NRP2b to regulate PTEN. A, HEK293 cells expressing wild-type NRP2a or NRP2b, or with deletion of the cytoplasmic domains (△C), were immunoprecipitated with antibodies against NRP2. Precipitates were probed for co-precipitated GSK3β by Western blot. Lysates before immunoprecipitation were analyzed simultaneously for input levels. B, PTEN levels were assessed by Western blot in lysates from HEK293 cells expressing NRP2a, NPR2b, or vector control, +/− treatment with the GSK3β inhibitor, TZDZ-8. Blots were probed for total PTEN, phosphorylated PTEN [p-PTEN (T366-P)], total NRP2, NRP2b, and mTOR (control for equal loading). C, PTEN levels were assessed as in (B) in lysates of HEK293 cells expressing NRP2a, NPR2b, or vector control treated or not with the proteasome inhibitor MG132. D, Normalized PTEN levels from (C) determined by quantitation.

FIGURE 6.

GSK3β interacts with a 15 amino acid motif within NRP2b. A, Diagram of C-terminal deletion mutants of NRP2b used for GSK3β co-immunoprecipitation. From top; full-length wild-type (2b-wt), cytoplasmic-domain deletion (2b-△C), deletion of the last 21 amino acids (2b-△21), and deletion of the last 6 amino acids (2b-△6). B, Lysates of HEK293 cells transfected with GSK3β and the NRP2b constructs diagrammed in (A) were immunoprecipitated with antibodies against NRP2. Precipitates were analyzed for GSK3β by Western blot. C, In silico modeling of the GSK3β-interacting motif (A881-Q895) of NPR2b suggested an amphipathic α-helical structure with hydrophilic (teal) and hydrophobic (tan) side-chains clustered on opposite faces. D, In silico modeling identified a putative high-affinity docking site (−45 kcal) for the NRP2b aliphatic helix on the 3-dimensional x-ray structure of GSK3β (pdb:1PYX).

FIGURE 7.

NRP2b-dependent migration requires GSK3β activity. A, A549 lung cancer cells, modified by shRNAs targeting NRP2, NRP2b, or NRP2a, were assayed for migration towards 20% fetal bovine serum in the lower chamber (see “Materials and Methods”). Cells in the upper chamber were treated with HGF (50 ng/mL), TDZD-8 (10 μM GSK3β inhibitor), or both, as indicated. Migrated cells were stained, photographed, and counted; results are reported using box and whisker plots; solid line is the median. B, Equivalent migration assays of PC-9 cells. Significance determined by 2-tailed t tests. ***P < .001. n.s., Not significant.

FIGURE 8.

Emergence of NRP2b-dependent, drug-tolerant persister cells requires GSK3β activity. A, PC-9 cells knocked down for NRP2a, NRP2b, or nontargeting controls were treated with gefitinib (2.5 μM) for 4 weeks with or without the GSK3β inhibitor, TZDZ-8 (5 μM). For quantitation, surviving persister cells were grown into colonies in nonselective medium for an additional 3 to 4 weeks. Box and whisker plots indicate colony counts. B, Same as (A) except PC-9 cells were treated with osimertinib (2.5 μM) with or without TZDZ-8 (5 μM) for 9 days; surviving cells were grown into colonies for counting, as in (A). C, Same as (B) except HCC827 cells with knockdown of NPR2a, NRP2b, or nontargeted controls were treated with osimertinib (2.5 μM) with or without TZDZ-8 (5 μM) for 9 days. Significance determined by 2-tailed t tests. **P < .01. ***P < .001.