NGF-induced TrkA/CD44 association is involved in tumor aggressiveness and resistance to lestaurtinib

Abstract

There is accumulating evidence that TrkA and its ligand Nerve Growth Factor (NGF) are involved in cancer development. Staurosporine derivatives such as K252a and lestaurtinib have been developed to block TrkA kinase signaling, but no clinical trial has fully demonstrated their therapeutic efficacy. Therapeutic failures are likely due to the existence of intrinsic signaling pathways in cancer cells that impede or bypass the effects of TrkA tyrosine kinase inhibitors. To verify this hypothesis, we combined different approaches including mass spectrometry proteomics, co-immunoprecipitation and proximity ligation assays. We found that NGF treatment induced CD44 binding to TrkA at the plasma membrane and subsequent activation of the p115RhoGEF/RhoA/ROCK1 pathway to stimulate breast cancer cell invasion. The NGF-induced CD44 signaling was independent of TrkA kinase activity. Moreover, both TrkA tyrosine kinase inhibition with lestaurtinib and CD44 silencing with siRNA inhibited cell growth in vitro as well as tumor development in mouse xenograft model; combined treatment significantly enhanced the antineoplastic effects of either treatment alone. Altogether, our results demonstrate that NGF-induced tyrosine kinase independent TrkA signaling through CD44 was sufficient to maintain tumor aggressiveness. Our findings provide an alternative mechanism of cancer resistance to lestaurtinib and indicate that dual inhibition of CD44 and TrkA tyrosine kinase activity may represent a novel therapeutic strategy.

Keywords: CD44; NGF; TrkA; lestaurtinib; resistance.

Figure 1. NGF induces binding of CD44 to Trk

AHA-TrkA MDA-MB-231 cells were treated with NGF (5 and 30 min). (A) CD44 and TrkA complexes were detected by IP with anti-HA antibody followed by immunoblotting with anti-CD44 antibody. CD44 binding to HA-TrkA was normalized and indicated as Fold Change. CD44 binding to HA-TrkA at 0 min was considered as 1. (B) Reverse IP with anti-CD44 antibody followed by immunoblotting with anti-HA antibody. HA-TrkA binding to CD44 was normalized and indicated as Fold Change. HA-TrkA binding to CD44 at 0 min was considered as 1. (C) TrkA/CD44 association was visualized by a PLA; interactions between TrkA and CD44 are shown as red spots. Density of red spots was quantified by Image J software. TrkA/CD44 PLA density without stimulation was considered as 1. (D and E) Flow cytometry analyses of membrane levels of TrkA (D) and CD44 (E) Fold Changes were determined using the median value of each histogram.

Figure 2. NGF induces association of TrkA with CD44/p115RhoGEF/RhoA/RhoC/ROCK1 and RhoGTPase activation

(A–D) NGF induces TrkA association with CD44/p115RhoGEF/RhoGTPases/ROCK1. HA-TrkA MDA-MB-231 cells were treated with NGF (5 and 30 min). IPs were done with anti-HA or anti-CD44 antibodies, and immunoblotting was used to detect the presence of p115RhoGEF (A and B) and ROCK1 (C and D) in the eluate. (E) NGF increases RhoGTPase activity. HA-TrkA MDA-MB-231 cells were stimulated with NGF (5 or 30 min) and GTP-bound RhoA and RhoC were determined in cell lysates by an affinity pull-down assay with the GST-Rho binding domain followed by immunoblotting for RhoA and RhoC. Whole cell lysate samples were immunoblotted for total RhoA/C as a control. Figures are representative of three independent pull-down assays. (F) CD44 invalidation inhibits RhoA and RhoC activities. HA-TrkA MDA-MB-231 cells were transfected with siCTRL (scramble siRNA) or siCD44 stimulated with NGF (5 or 30 min) and GTP-bound RhoA and RhoC were determined in cell lysates by an affinity pull-down assay with the GST-Rho binding domain followed by immunoblotting for RhoA and RhoC. Whole cell lysate samples were immunoblotted for total RhoA/C as a control.

Figure 3. CD44, p115RhoGEF and ROCK1 are involved in NGF-stimulated cell invasion

HA-TrkA MDA-MB-231 cells were transfected with si CTRL (scramble siRNA), siCD44 (A) or sip115RhoGEF (B). ROCK1 involvement in cell invasion was assessed using the Y-27632 compound (C). HA-TrkA MDA-MB-231 cells were treated with NGF and invading cells were evaluated using Transwells. Data are mean ± S.D. of three experiments done in triplicate and are presented as a percentage of controls. Statistical analysis was performed with one-way ANOVA followed by Bonferroni's post-test. Error bars represent S.D. *p < 0.001 for NGF stimulation versus no stimulation; § p < 0.001 for experimental versus control under NGF stimulation. siRNA efficiency was assessed by immunoblot with specific antibodies against CD44 or p115RhoGEF. β-actin was used as a loading control.

Figure 4. TrkA binds to CD44 independently of its kinase domain activity

(A–C) HA-TrkA MDA-MB-231 cells were incubated with DMSO (A) or with K252a (B) HA-TrkA kinase-dead MDA-MB-231 cells (C) were treated with NGF (5 or 30 min). The interactions between TrkA and CD44 were visualized by the detection of red spots by PLA. Density of red spots was quantified by Image J software. TrkA/CD44 PLA density without stimulation in DMSO was considered as 1. (D) K252a impedes TrkA internalization as monitored by flow cytometry. Results are representative of three independent experiments. (E) HA-TrkA (wild-type) and HA-TrkA Kinase-dead expressing cells were treated with NGF (5 and 30 min). IPs were done with anti-HA and immunoblotting was used to detect the presence of p115RhoGEF. P115RhoGEF binding to HA-TrkA was normalized and indicated as Fold Change. P115RhoGEF binding to HA-TrkA at 0 min was considered as 1. (F) GTP-bound RhoA and RhoC were determined in cell lysates by an affinity pull-down assay with the GST-Rho binding domain followed by immunoblotting for RhoA and RhoC. Whole cell lysate samples were immunoblotted for total RhoA/C as a control.

Figure 5. Effects of combined targeting of TrkA and CD44 on tumor cell growth

(A and B) Effects of siCD44 and TrkA kinase inhibition by K252a on colony formation. Cells were plated at single cell density and cultured for one week. (A) Representative pictures of forming colonies. (B) Quantification of colony forming units (CFU). (C and E) Impact of TrkA or/and CD44 inhibitions on tumor growth in vivo. Xenograft experiments were conducted using HA-TrkA MDA-MB-231 cells. The tumors were allowed to develop for 14 days and the mice were then submitted to 3 injections (every 3 days; black arrows) of either in vivo Jet PEI + scramble siRNA (7.5 μg/mouse, Control), lestaurtinib (10 mg/kg), Jet PEI + siCD44 (7.5 μg/mouse) or lestaurtinib and Jet PEI + siCD44 (7.5 μg/mouse). (C) Tumor volumes measured at different intervals. A Mann-Whitney test was performed between the control group and lestaurtinib (a), the control group and siCD44 (b), the control group and lestaurtinib + siCD44 (c), and between the siCD44 group and lestaurtinib + siCD44 (d). *P < 0.05; **P < 0.01; ns: not significant. (D) Dot plot representation of tumor volume at the end of the experiment. (E) Brightfield PLA assays of TrkA/CD44 interaction on MDA-MB-231 tumor xenografts. PLA assays were performed on tumor xenografts from control group. Pictures are representative of three independent PLA assays.

Figure 6. Proposed model of lestaurtinib resistance through a phospho-TrkA-independent pathway involving downstream CD44 signaling

Lestaurtinib inhibits the kinase activity of TrkA and phospho-TrkA-dependent downstream signaling including Akt. However, a phospho-TrkA-independent pathway that uses CD44 signaling may serve as an alternative pathway to strengthen tumor aggressiveness and to escape lestaurtinib inhibition.