Anne-Florence Blandin

Le glioblastome multiforme (GBM) est la tumeur cérébrale primaire la plus fréquente. Une dérégulation des voies de signalisation de l’EGFR et un fort potentiel invasif sont les caractéristiques principales du GBM. Malheureusement, les essais cliniques impliquant des thérapies anti-EGFR dans le traitement des GBM demeurent inefficaces. Nous avons précédemment montré que le récepteur de la fibronectine, l’intégrine α5β1, est associé avec un mauvais pronostic et une résistance des patients au temodal. Les intégrines peuvent coopérer avec les récepteurs aux facteurs de croissance et ainsi amplifier leur potentiel oncogénique. Ici, nous avons cherché à déterminer le rôle de l’intégrine α5 dans la résistance aux thérapies anti-EGFR. Utilisant la lignée U87 de GBM, on a dans un premier temps confirmé que l’activation de l’intégrine sous l’influence de la fibronectine, potentialisait la signalisation de l’EGFR. La perte d’expression d’α5 sensibilise les cellules U87 aux anti-EGFR (cetuximab, gefitinib) dans des essais de clonogénicité en soft agar. L’expression d’ α5 favorise la résistance aux 2 drogues lors de la migration cellulaire. Pour aller plus loin, nous avons développé un nouveau test basé sur la quantification de l’évasion cellulaire à partir d’une sphère tumorale. La perte d’ α5 augmente la sensibilité des cellules U87 à 2 TKI réversibles spécifiques de l’EGFR, gefitinib et erlotinib, mais n’a pas d’effet sur l’efficacité du lapatinib, un TKI irréversible ciblant EGFR, ErbB2, ErbB3 et ErbB4. Grâce à la microscopie confocale, nous avons montré l’effet important du gefitinib sur l’endocytose de l’intégrine et de l’EGFR. Ces résultats suggèrent que l’expression d’ α5 favorise la résistance aux TKI par l’activation des voies de signalisation des récepteurs ErbB ou en contrôlant le trafic membranaire de l’EGFR. On a aussi montré que pour favoriser l’adhésion cellulaire, l’intégrine α5 stimulait la fibrillogénèse. Dans les cellules migrant à distance de la sphère, l’intégrine α5 est strictement engagée dans des adhésions cellule-substrat contenant la protéine FAK activée. Nos résultats soulignent le rôle central du couple fibronectine/ intégrine α5 dans l’invasivité du GBM et la résistance aux thérapies anti-EGFR.

Supplementary Figure 4: Characterization of the lung cancer recurrent DCTN1-ALK fusion identified in cGBM (case ALK.232). A. Integrative Genomic Viewer (IGV) screenshot of the DCTN1-ALK variant.B. Predicted sequence of DCTN1-ALK fusion (DCTN1 exon 1-26, ALK exon 20-29 conserved).C. PDX DCTN1-ALK sub-renal capsule engraftment confimed by IHC.C. Relative viability of the ALK wild-type GBM line (BT164) and BT1857 ceritinib-treated cells versus DMSO control.D. Relative viability of BT164 and BT1857 lorlatinib-treated cells versus DMSO control.E. Relative viability of BT164 and BT1857 cells treated with an EGFR tyrosine kinase inhibitor neratinib.F. PDX DCTN1-ALK sub-renal capsule engraftment confirmed by IHC.

Supplementary Figure 5: Single-cell RNA Sequencing and IHC examination of ALK in human neural development and immature neural cells. A. 2D representation of human brain cell subtypes (48 samples) during cortex development using a single cell RNA-seq atlas (cortex-dev.cells.ucsc.edu) shows that ALK transcript appears to be enriched in a certain cell type. Color-coded by ALK gene expression (beige to red) and age in weeks (rainbow).B. ALK-stained sagittal section of normal human cerebellum at 15 postconceptional weeks (pcw).C. ALK-stained sagittal section of human fetal neural stem cell pellets at 13 postconceptional weeks (pcw).D. ALK-stained sagittal section of human fetal neural stem cell pellets at 15 postconceptional weeks (pcw).

Supplementary Figure 3: PPP1CB-ALK fusion is oncogenic and sensitive to targeted therapy. A. Immunoblot analysis of ALK downstream signaling proteins in cortical mNSC (CTX#6) and brainstem mNSC (BS#3) expressing PPP1CB-ALK. Cells were treated for 4 hours with ceritinib and lorlatinib at the indicated concentrations.B. Left panel: Soft agar colony-forming assay using stable mNSC PPP1CB-ALK demonstrated colony formation with dose-dependent inhibition by ceritinib. mNSC KRAS did not respond to ceritinib validating the drug specificity. Right panel: Quantification of colony formation under targeted drug inhibition using CellProfiler. Values represent colony counts relative to the untreated group ± s.d. The mean of three independent replicates is shown. Significance between treatments determined by the Mann-Whitney test. *P < 0.05, **P < 0.01.C. Cell viability of ceritinib-treated PPP1CB-ALK NIH-3T3 relative to eGFP-positive cells by CellTiterGlo.D. Relative viability of the ALK wild-type GBM line (BT164) and BT1857 lorlatinib-treated cells versus DMSO control.E. Relative viability of BT164 and BT1857 ceritinib-treated cells versus DMSO control.F. Relative viability of BT164 and BT1857 cells treated with an EGFR tyrosine kinase inhibitor neratinib.G. Pharmacodynamic analysis of NIH 3T3-PPP1CB-ALK s.c. tumors treated with vehicle or ceritinib at 30mg/kg for 5 days. Representative H&E stain, ALK, pSTAT3, pAKT S473 and Ki-67 IHC.H. Quantification of Ki-67-positive cells in vehicle and ceritinib-treated tumors using CellProfiler software. Error bars show standard error of the mean. *P < 0.05.I. Quantification of pSTAT3-positive cells in vehicle and ceritinib-treated tumors using CellProfiler software. Error bars show standard error of the mean. *P < 0.05.J. Tumor growth following subcutaneous implantation of PPP1CB-ALK and eGFP-expressing fibroblasts.K. Kaplan-Meier survival curves of SCID mice injected subcutaneously with mNSC CTX-PPP1CB-ALK (red, n = 5) or mNSC CTX-eGFP (black, n = 4) and treated with ceritinib at 30mg/kg/d for 15 days. Mice were euthanized at endpoint (tumor volume of 2000mm3). The grey area represents the treatment period.L. Vehicle or ceritinib-treated mouse weights at start and end of treatment.

Supplementary Figure 2: T1 post-contrast axial and sagittal images of ALK-fused and non ALK-altered GBMs and ALK staining in ALK-fused congenital GBMs. A. Example images from patients with ALK aberrations (76, 89, 116, 197)B. In contrast to those without ALK aberrations (right columns). Contrast is gadolinium for those where applicable and show enhancement within both the tumor categories.C. Representative H&E and ALK IHCs of congenital GBM harboring an ALK fusion.

Additional information tables 1 to 8.

Supplementary Figure 1: Characterization of the novel LRRFIP1-ALK fusion (case ALK.223) A. Visualization of the read coverage at the fusion sites in ALK and LRRFIP1 genes using Integrative Genomic Viewer.B. Pile-up plot generated by SvABA using whole-genome sequencing data shows the breakpoint coordinates.C. Copy number ratio profile by WGS showing no somatic copy number alterations anywhere in the genome. The x-axis represents the chromosomes.D. Copy number ratio profile on chromosome 2 showing no copy number change of ALK and LRRFIP1 genes.E. Two RT-PCR products of 186bp (variant 1) and 189bp (variant 3) were detected by agarose gel electrophoresis with total RNA from tumor specimen ALK.223.F. Primer sequences of ALK-LRRFIP1 (variants 1 and 2) and LRRFIP1-ALK (variant 3) and the predicted DNA product size.G. Western Blot analysis of ALK signaling pathway in subcutaneous tumors treated with vehicle or lorlatinib. Normal brain was used as control.H. Representative 40x images and CellProfiler quantification of Cleaved Caspase 3- and Ki-67- stained subcutaneous tumors.

Supplementary Figure 6: Single cell RNA sequencing comparison of ALK expression and other infant/pediatric glioma kinase drivers in developing human and mouse brain. A. 2D representation of human brain cell subtypes (48 samples) during cortex development using a single cell RNA-seq atlas (http://cells.ucsc.edu/?ds=cortex-dev) shows that ALK positive cells make up less than 1% of all cells in the dataset and aren't localized within a specific cluster.B. Cluster-level summary of the human scRNA dataset shows that ALK expression (TPM) is decreased compared to EGFR, NTRK2, and NTRK3, and ALK positive cells make up a much lower percentage of each cluster compared to those 3 genes.C. 2D representation of developmental mouse forebrain (E12.5 – P6) in a single cell RNA-seq atlas (http://cc-shiny-01.functionalgenomics.ca/braindex/clusters) shows similar results as seen in the human scRNA Nowakowski dataset.D. Cluster-level summary of the mouse scRNA dataset corroborates results seen in the Nowkowski scRNA dataset; that Alk expression (log-normalized) is decreased, and less prevalent than EGFR, NTRK2, and NTRK3.E. 2D representation of RNA sequencing data for NTRK1-3 genes in the Nowakowski developing human cortex dataset. Compared to ALK expression, the NTRK genes have increased expression levels, increased prevalence in the dataset, and show increased localization to specific cluster.F. Similar to (E), the 2D representation of Ntrk genes in the developing mouse scRNA dataset shows increased expression and prevalence compared to Alk.

Additional information tables 1 to 8.