Databases and literature PPI curation
Information about genes containing mutations in their coding regions was retrieved from the COSMIC database, evaluated; organized and selected genes were submitted for experimental analysis. To establish a catalog of genes and mutations associated with Acute Lymphoblastic Leukemia (ALL), we used the version 71 of COSMIC, previously downloaded to a local server and we extracted data only related to ALL. We developed and implemented a procedure that automatically collects information and check the consistency of changes with the coding sequences and find the corresponding positions on clones from the human ORFeome (http://horfdb.dfci.harvard.edu/).
The retrieved information include details provided at either nucleotide or protein level (mutation syntax), sample id (portion of a tumor being examined for mutations), tissue from which the sample originated, histological classification of the sample and the Pubmed id of the article that published the study. For each gene tested for PPIs using the yeast two-hybrid, we identified mutations positions on the ORFeome clone by sequence alignment (BLAST) then we verified if the protein sequence has undergone modifications as described by mutation syntax.
Human PPIs were collected and verified from the different interactomics databases IntAct [23], HPRD [24] and BioGRID [25]. Only physical PPIs validated at least in two independent references or by two methods were considered as confident and maintained for the analysis.
Network data analyses and visualization
Network analyses and visualization of protein-protein interactions were carried out with Cytoscape software, which is a free software for visualizing, modeling and analyzing molecular and genetic interaction networks. Due to its features, Cytoscape and its plugins provide a powerful tool kit allowing to answer specific biological questions using large amounts of cellular network and molecular profiling information [26]. In our maps, the nodes represent proteins that are connected with edges representing pairwise interactions extracted from interaction databases and from our Y2H experimental assay.
Cell culture and transfection
HEK293, HeLa and HeLa Notch1∆E-eGFP cells were cultured in DMEM supplemented with 10 % fetal bovine serum (FBS), 2 mM glutamine and penicillin/streptomycin. The same medium was used for U2OS Tet-on flp-in cells bearing isogenic transgenes encoding Notch1-Gal4. As for K562-control and K562 expressing Dll4 cells, they were grown and maintained in RPMI 1640 supplemented with 15 % FBS and antibiotics. T-ALL cell lines were grown and maintained in RMPI containing 10–20 % FBS and supplemented with antibiotics.
HEK293 cells were DNA-transfected with polyethylenimine (PEI) purchased from Sigma, reagent was dissolved in water at 1 mg/ml and preserved at −80 °C. Transfection with PEI was performed on HEK cells cultured in DMEM at 80 % confluence. Medium was changed before transfection and cells were collected 24 h post-transfection.
HeLa cells and HeLaN1ΔE-eGFP cells were DNA-transfected with lipofectamine 2000 reagent (Invitrogen) according to manufacturer’s instructions and collected 24 h post-transfection.
SiRNA transfection was performed with Calcium Phosphate using ProFection Mammalian Transfection kit from Promega according to manufacturer’s instructions on cells cultured in DMEM at 40–50 % confluence. Medium was changed 24 h later and cells were collected 48 h post-transfection.
For experiments involving both DNA and siRNA transfections, siRNA-transfection was performed according to manufacturer’s instructions and 24 h later after changing medium cells were transfected with DNA using lipofectamin 2000 reagent (Invitrogen) and cells were collected 24 h post DNA-transfection.
For proteasomal degradation inhibition, cells were treated with 10 μg/ml MG132 for 6 h before being collected.
siRNA sequences:
-
siEXT1: 5′-GGAUUCCAGCGUGCACAUUtt-3′
-
siFBXW7: 5′- GCAUAGAUUUUAUGGUAAtt-3′
-
siCtrl: 5′- GGCUGCUUCUAUGAUUAUGtt-3′
qRT-PCR
Total RNA was extracted using GeneJET RNA Purification Kit (Thermo scientific), DNaseI-treated on the column (Thermo Scientific) and reverse-transcribed with random primers (Thermo scientific). qPCR was performed using SYBER Green detection from Roche and run on Lightcycler 480 (Roche). mRNA quantification was performed relative to GAPDH housekeeping gene. Relative expression levels were calculated for each gene using the ΔΔCt method.
Plasmids
Open reading frames (ORF) encoding Notch1 partners (tested for Protein complementation assay) were obtained from human ORFeome v5.1 (center of cancer systems biology: CCSB) as entry clones. Human NICD plasmid, was obtained from Addgene. FBXW7α expressing vector was kindly provided by Dr. E. Dejardin from the laboratory of molecular immunology and signal transduction—(GIGA-ULg). ORFs that were not available from the hORFeome V5.1 (BRAF, HRAS, ABL1, JAK2 and SMARCB1 and NOTCH1 genes), were purshassed from Genecopea and cloned by Gateway recombination technology (Invitrogen) using specific primers flanked with the following AttB1 and AttB2 Gateway sites: 5′- GGGGACAACTTTGTACAAAAAAGTTGGCATG-3′ (AttB1) and 5′- GGGGACAACTTTGTACAAGAAAGTTGA-3′ (AttB2). These constructions were verified by PCR and sequencing.
Inserts from pDONR223 were transferred by LR cloning (Invitrogen) into different destination vectors: pAD-destCYH and pDB-dest the Y2H expression vectors, and pDEST1899 (flag tag), pDEST491 (YFP-tag) and pDEST-mcherry for mammalian expression studies.
High-throughput yeast Two-hybrid (HT-Y2H)
We used the hORFeome version 5.1, a collection of human ORFs cloned from the Mammalian Gene Collection (MGC) resource, representing a resource of ORFs that can be transferred easily to any Gateway compatible destination vectors. This collection contains 15 483 ORFs representing almost half of the human genome. They are cloned into the pAD-dest-CYH and pDB-dest encoding the yeast Gal4 Activating and DNA-binding domains, respectively. The resulting individual clones were transferred into MATa Y8800 (pAD) and MATα Y8930 (pDB) S. cerevisiae strains. Twenty-one selected ALL-genes were screened for interactions with the hORFeome V5.1 as described in [27].
One pool of 21 AD of selected genes into Y8800 yeast strain was mated to each of the 15483 dB-ORFs Y8930 of the hORFeome v5.1 and each of ALL-genes-DB into Y8930 yeast strain was mated to 165 pools of 94 AD-ORFs of the hORFeome v5. One Y2H screening was performed in the reciprocal orientation, as described in [27]. Positive colonies for the GAL1:: HIS3 and GAL1:: ADE2 selective markers but negative for autoactivation were selected for PCR-amplification (Zymolyase 20 T from Seikagaku Biobusiness, and Platinum® Taq DNA Polymerase from Invitrogen) and identification of interacting proteins by sequencing of the respective AD- and DB-ORFs.
Luciferase reporter assays
Cells were seeded in 24-well plates and transfected with 300 ng of either TP1 luciferase reporter plasmid (TP1-luc) or CBF1 reporter plasmid (CBF1-luc) and 30 ng of renilla Luciferase (R-Luc). Twenty-four hours post-transfection luciferase activity was measured in cell lysates.
U2OS N1-Gal4 cells they were transfected 300 ng of Gal4-firefly luciferase and 30 ng R-Luc reporter plasmid. After 24 h, K562 cells expressing Notch ligands DLL4 or K562 control cells were added to the transfected cells in the presence of tetracycline (2 μg/mL). After 24-h of coculture, luciferase activity was measured in cell lysates.
Cell lysis and luciferase assays were performed in triplicate using Dual-luciferase reporter assay system from Promega. Luciferase measurements were performed in 96 well plates using DLR automated machine. Firefly luciferase values were normalized to R-luc values and calculated ratio represent luciferase activity.
Immunofluorescence and confocal microscopy
HeLa Notch1∆E-eGFP cells were seeded onto coverslips in 24-well plates and transfected with 1 μg of EXT1-mcherry plasmid using lipofectamine2000 (Invotrogen). Twenty-four hours post-transfection, cells were washed in warm PBS, fixed in 3,7 % PBS-paraformaldehyde for 20 min at room temperature, washed 3 times with PBS, and mounted on glass coverslips using ProLong Gold Antifade montant with DAPI (life technologies).
Slides were examined by confocal microscopy using the Nikon A1R confocal system and images processed with the IMAGI software.
Protein complementation assay (PCA)
NICD and FBXW7 were cloned in pN1Gluc vector and pN2Gluc vectors respectively (for Gaussia luciferase 1 and 2) using the Gateway cloning technology. HEK293 cells were seeded in 24 well-plates at a concentration 5.10 4 cell/well, then transfected with GL1 or/and GL2 plasmids and 24 h post-transfection, luciferase activity was measured on lysates transferred into 96-well plate using and automated machine DLR with Renilla luciferase substrate. Normalized luciferase ratio was calculated as follows: NLR = luciferase value GL1 + GL2/(luciferase value GL1 + luciferase value GL2). An interaction is considered positive or validated when NLR ≥ 3.5. Cell lysis and luciferase assays were performed in triplicate for each condition.
EXT1 silencing in zebrafish
Transgenic zebrafish line Tp1bglob:eGFP line [28] were maintained according to EU regulations on laboratory animals. Knockdown experiments were performed by injecting embryos at the one- to two-cell stage with 10 ng of single splice-blocking morpholino designed specifically for both EXT1 a and b orthologs.
RNA sequencing
Total RNA was extracted from HeLaN1ΔE-eGFP cells (siCTRL, siEXT1, siFBXW7), quantified and tested for RNA quality controlled using Agilent 2100 bioanalyzer using the Eukaryote Total RNA Nano assay. Total RNA strands were used to generate libraries and sequenced by HiSeq2000 sequencer.
Statistical analysis
Graph values are presented as mean +/− standard deviation, calculated on at least three independent experiments. Unless stated otherwise, significance was determined using a two-tailed Student’s t-test (comparison of means). P-value thresholds are depicted as follows; *: p < 0.05; **: p < 0.01; ***: p < 0.001 and ****: p < 0.0001.
To prioritize ALL-genes, we combined the ranking from separate results (rank per number of mutation, rank per number of samples, rank per degree) by using order statistics. First, ranks are divided by the total number of ranked genes and we calculated the Q statistic [29], which represents the probability of obtaining the observed ranks r by chance, calculated using joint cumulative distribution of order as:
$$ Q\left({r}_1,{r}_2\kern0.5em ,\dots, \kern0.5em {r}_N\right)\kern0.5em =\kern0.5em N!{V}_N\kern3em {V}_0\kern0.5em =\kern0.5em 1,{V}_k=\kern0.5em {{\displaystyle {\sum}_{i=1}^k\left(-1\right)}}^{i\kern0.5em -\kern0.5em 1}\frac{V_{k\kern0.5em -\kern0.5em i}}{i!}{r}_{N-k+1}^i $$
Where ri is the rank ratio for result i, N is the number of genes used.
