PrPC controls epithelial-to-mesenchymal transition in EGFR-mutated NSCLC: implications for TKI resistance and patient follow-up

Abstract

Patients with EGFR-mutated non-small cell lung cancer (NSCLC) benefit from treatment with tyrosine kinase inhibitors (TKI) targeting EGFR. Despite improvements in patient care, especially with the 3rd generation TKI osimertinib, disease relapse is observed in all patients. Among the various processes involved in TKI resistance, epithelial-to-mesenchymal transition (EMT) is far from being fully characterized. We hypothesized that the cellular prion protein PrP^C could be involved in EMT and EGFR-TKI resistance in NSCLC. Using 5 independent lung adenocarcinoma datasets, including our own cohort, we document that the expression of the PRNP gene encoding PrP^C is associated with EMT. By manipulating the levels of PrP^C in different EGFR-mutated NSCLC cell lines, we firmly establish that the expression of PrP^C is mandatory for cells to maintain or acquire a mesenchymal phenotype. Mechanistically, we show that PrP^C operates through an ILK-RBPJ cascade, which also controls the expression of EGFR. Our data further demonstrate that PrP^C levels are elevated in EGFR-mutated versus wild-type tumours or upon EGFR activation in vitro. In addition, we provide evidence that PRNP levels increase with TKI resistance and that reducing PRNP expression sensitizes cells to osimertinib. Finally, we found that plasma PrP^C levels are increased in EGFR-mutated NSCLC patients from 2 independent cohorts and that their longitudinal evolution mirrors that of disease. Altogether, these findings define PrP^C as a candidate driver of EMT-dependent resistance to EGFR-TKI in NSCLC. They further suggest that monitoring plasma PrP^C levels may represent a valuable non-invasive strategy for patient follow-up and warrant considering PrP^C-targeted therapies for EGFR-mutated NSCLC patients with TKI failure.

Fig. 1. PRNP gene expression correlates with EMT in LUAD.

GSEA analysis showing enrichment of the EMT signature in the genes most correlated to PRNP in cell lines of the CCLE LUAD (A) and in patients from the Onco-HEGP LUAD cohort (B). C Heatmap summarizing the correlation indexes between the expression of PRNP and that of EMT TF in multiple datasets. D Heatmap summarizing the correlation indexes between the expression of PRNP and that of the pan-cancer epithelial (EMT-epi), mesenchymal (Epi-mes) and EMT scores from [22] and the mesenchymal score derived from [23] in multiple datasets.

Fig. 2. PrP^C is necessary for EMT in LUAD cell lines.

A qRT-PCR analysis of the expression of the EMT TF SNAI1, SNAI2, TWIST, ZEB1, ZEB2 in PRNP-silenced versus control H1975 cells. Western blot analysis of the expression of ZEB1 (B) and SLUG (C) in PRNP-silenced versus control H1975 cells. D GSEA analysis highlighting the EMT signature as one of the most affected pathway in H1975 cells in response to PRNP silencing. E Immunofluorescence images showing PrP^C and F-actin staining in PRNP-silenced versus control H1975 cells. Analysis with the CASY cell counter revealing a decrease in proliferation (F) and an increase in cell volume (G) in PRNP-silenced versus control H1975 cells. Cell index measurements of PRNP-silenced versus control H1975 cells in xCELLigence migration (H) or invasion (I) assay. qRT-PCR analysis of the expression of the EMT TF SNAI1, SNAI2, ZEB1, ZEB2 (J) and EMT genes VIM, CDH1 and CDH2 (K) in HCC827 cells exposed to siRNA against PRNP and recombinant TGFβ1. L Western blot analysis of the expression of SNAIL, ZEB1, ZEB2, Vimentin, E-cadherin and N-cadherin in HCC827 cells exposed to siRNA against PRNP and recombinant TGFβ1. Results are expressed as means ± s.e.m of n = 2 independent triplicates (A–C, J) or n = 1 triplicate (F–I, L) of cell preparations. *p < 0.05, **p < 0.01 ***p < 0.001 vs. control (siScramble), ^#p < 0.05, ^###p < 0.001 vs. TGFβ1 treated, siScramble cells, ^§p < 0.05 vs. TGFβ1 untreated, PRNP-silenced cells. Protein levels in western blots were normalized to α-tubulin (α-tub) with quantifications summarized in Supplementary Fig. S2F.

Fig. 3. PrP^C controls EMT in LUAD cell lines via an ILK-RBPJ axis.

A GSEA analysis showing that the NOTCH signalling pathway is affected in response to PRNP silencing in H1975 cells. B qRT-PCR analysis of the expression of JAG1, JAG2, NOTCH1, NOTCH2 and RBPJ in PRNP-silenced versus control H1975 cells. Western blot analysis of the expression of JAGGED1 (C) and RBPJ (D) in PRNP-silenced versus control H1975 cells. qRT-PCR analysis of the expression of JAG1, JAG2, NOTCH1, NOTCH2 and RBPJ (E) and Western blot analysis of the expression of JAGGED1 (F) and RBPJ (G) in HCC827 cells exposed to siRNA against PRNP and recombinant TGFβ1. H Venn diagram highlighting ILK as a common protein most correlated with PrP^C levels in the Chen and Lehtiö proteogenomic studies and a PrP^C partner in melanoma. Scatter plots illustrating the correlation between ILK and PrP^C levels in the Lehtiö proteogenomic study I. Western blot analysis of the expression of ILK in HCC827 cells exposed to siRNA against PRNP and recombinant TGFβ1 (J). qRT-PCR analysis of the expression of SNAI1, ZEB1 and ZEB2 EMT TF (K) and JAG1, JAG2, NOTCH1, NOTCH2 and RBPJ (L) in QLT0267-treated versus control H1975 cells. qRT-PCR (M) and Western blot (N) analysis of the expression of RBPJ in QLT0267-treated versus control MDST8 cells. O Schematic diagram illustrating the proposed regulation of EMT TF downstream from PrP^C. Results from qRT-PCR and western blots are expressed as means ± s.e.m of n = 2 independent triplicates of cell preparations. *p < 0.05, **p < 0.01, ***p < 0.001 vs. control (siScramble or vehicle), ^#p < 0.05, ^###p < 0.001 vs. TGFβ1 treated, siScramble cells, ^§p < 0.05, ^§§p < 0.01 vs. TGFβ1 untreated, PRNP-silenced cells. Protein levels in western blots were normalized to α-tubulin (α-tub).

Fig. 4. PRNP levels are associated with EGFR activation and are elevated in EGFR-mutated LUAD.

A GSEA analyses showing enrichment of the activated EGFR pathway (EGFR_up.V1_up) signature in the genes most correlated to PRNP in cell lines of the CCLE LUAD and in patients from the TCGA LUAD, Onco-HEGP LUAD, Chen LUAD or Lehtiö LUAD studies. B Boxplots showing the distribution of PRNP mRNA levels according to the EGFR mutational status in the CCLE and Onco-HEGP LUAD studies. C Boxplots showing the distribution of the PrP^C protein levels according to the EGFR mutational status in the Chen and Lehtiö LUAD studies. D Boxplot showing the distribution of mouse Prnp mRNA in lung tissue of several genetically modified mouse models of LUAD bearing the EGFR L858R or T790M mutation or combining the two mutations.

Fig. 5. PrP^C and EGFR are linked by reciprocal regulation of expression, physical and functional interaction.

qRT-PCR (A) and Western blot (B) analysis of the expression of PRNP / PrP^C in EGF-treated versus control H1650 cells. C qRT-PCR analysis of the expression of EGFR mRNA in H1650, HCC827 and H1975 cells after PRNP silencing as compared to control cells. D Western blot analysis of the expression of EGFR protein levels in H1975 cells after PRNP silencing as compared to control cells. qRT-PCR (E) and western blot (F) analysis of the expression of EGFR in HCC827 cells exposed to siRNA against PRNP and recombinant TGFβ1. G qRT-PCR analysis of the expression of EGFR mRNA in QLT0267-treated versus control H1975 cells. H Schematic diagram illustrating the proposed regulation of EGFR downstream from the PrP^C-ILK-RBPJ axis. I Confocal microscopy images showing the staining of PrP^C and EGFR in H1975 cells. White arrows indicate co-localization regions. Scale bar = 20 µm. J Heatmap showing the most differentially expressed genes in H1650 cells exposed to siRNA against PRNP and recombinant EGF. qRT-PCR analysis of the expression of SNAI2 (K), ILK (L), JAG1 and RBPJ (M) mRNA in H1650 cells exposed to siRNA against PRNP and recombinant EGF. N Western blot analysis of the expression of RBPJ in H1650 cells exposed to siRNA against PRNP and recombinant EGF. Results from qRT-PCR and Western blots are expressed as means ± s.e.m of n = 2 independent triplicates of cell preparations. *p < 0.05, **p < 0.01, ***p < 0.001 vs. control (siScramble or vehicle), # p < 0.05, ^###p < 0.001 vs. TGFβ1 or EGF treated, siScramble cells, ^§p < 0.05 vs. EGF untreated, PRNP-silenced cells. Protein levels in Western blots were normalized to α-tubulin (α-tub).

Fig. 6. PRNP levels are associated with resistance to EGFR-TKI.

Boxplots showing the distribution of PRNP mRNA in (A) erlotinib-resistant HCC827 clones versus parental cells (B) the minimal residual cell population obtained after combined treatment of HCC827 cells with osimertinib and trametinib (C) HCC827 cells acutely exposed to osimertinib (left) or HCC827 persister cells after long-term exposure osimertinib (right). D Quantification of cell numbers in H1975 cells pre-treated or not with PRNP siRNA and exposed to different doses of osimertinib for 72 h, according to the schematic workflow (top). E Differential expression of PRNP is shown in violin plots for single cancer cells collected from a patient treated with erlotinib (left) or a patient treated with osimertinib (right). TN treatment naive, RD residual disease, PD progressive disease.

Fig. 7. Plasma PrP^C levels are elevated in EGFR-mutant NSCLC patients and evolve according to disease history.

A Boxplot showing the mean levels of circulating PrP^C in the plasma of healthy subjects or patients with EGFR-mutated NSCLC at the time of pre-treatment with first line EGFR-TKI in the metastatic disease. B Summary of plasmatic PrP^C values according to demographic information in healthy subjects or patients. C Evolution of plasma PrP^C levels in paired samples from EGFR-TKI treated patients from cohort 1 between pre-treatment (T0) and evaluation (T1) with T1 corresponding to progression. D Table summarizing clinical data of patients from (C) at T0 and T1. E Evolution of plasma PrP^C levels in paired samples from EGFR-TKI treated patients from cohort 1 between pre-treatment (T0) and evaluation (T1) with T1 corresponding to a clinical event other than progression. F Table summarizing clinical data of patients from (E) at T0 and T1. G Evolution of plasma PrP^C levels in paired samples from EGFR-TKI patients from cohort 2 between pre-treatment (T0) and evaluation (T1). H Table summarizing clinical data of patients from (G) at T0 and T1. I Kinetics of evolution of plasma PrP^C levels in samples from EGFR-TKI patients from cohort 2 according to disease history. J Table summarizing clinical data of patients from (I) at T0, T1 and T2.