The TP53-activated E3 ligase RNF144B is a tumour suppressor that prevents genomic instability

Human lung adenocarcinoma cell line A549 (ATCC, CRL-3588) and the colon carcinoma cell line HCT116 (ATCC, CCL-247) were obtained from Hospital del Mar Research Institute and authenticated using Short Tandem Repeat profiling (CSIC-UAM, Madrid, Spain). Mouse KRAS^G12V lung cancer cell lines (mKLC) were a gift from D. Santamaria (CIC, Spain) [38] and were grown in DMEM (L0102, Biowest) containing 10% Fetal Bovine Serum (FBS, S181BH, Biowest) and 100 µg/ml penicillin/ streptomycin (15,140,122, Gibco). HBEC3-KT (ATCC, CRL-4051) immortalized bronchial epithelial cells were a gift from Silvestre Vicent (CIMA, Spain) [39, 40]. HBEC3-KT cells were cultured in KSFM media (17,005,042, Gibco) containing 50 µg/mL of Bovine Pituitary Extract (BPE, 13,028,014, Gibco) and 5 ng/mL of human epidermal growth factor (hEGF, E9644, Gibco). HBEC3-KT cells expressing KRAS^G12D and sgRNA knockout populations were also cultured in RPMI-1640 (L0500, Biowest) media supplemented with 10% FBS and 100 µg/ml penicillin/ streptomycin. Mouse embryonic fibroblasts (MEFs) were generated from E13.5 C57BL/6J embryos. MEFs were grown in DMEM containing 10% FBS, 100 µg/ml penicillin/ streptomycin, 100 μM asparagine (A4159, Sigma) and 50 μM 2-mercaptoethanol (63,689, Sigma). Cells were grown in 5% CO₂ at 37ºC. All cell lines were regularly tested for mycoplasma. Only mycoplasma-negative cells were used.

To generate CRISPR knockout bulk populations or clones, cell lines were transduced with a two-construct lentiviral pFUGW-derived system: a constitutive vector with an mCherry-labeled Cas9 [41] and a sgRNA expression vector [23] expressing CFP. sgRNAs sequences were cloned after BsmbI (R0580S, NEB) digestion. sgRNAs targeting human genes were the following: For human Tp53: 5’-GGCAGCTACGGTTTCCGTCT-3’, and for human Rnf144b: 5’-TGACATGGTGTGCCTAAACC-3’. A non-targeting control sgRNA was used (sgCTRL: 5’-CCAGTTGCTCTGGGGGAACA-3’).

shRNAs GFP-labeled targeting mouse RNF144B (shRNF144B: 5’-TGCTGTTGACAGTGAGCGCCAGGTTATTTACATACTTTCATAGTGAAGCCACAGATGTATGAAAGTATGTAAATAACCTGATGCCTACTGCCTCGGA-3’), TRP53 (5’-TGCTGTTGACAGTGAGCGCCCACTACAAGTACATGTGTAATAGTGAAGCCACAGATGTATTACACATGTACTTGTAGTGGATGCCTACTGCCTCGGA-3’) or the shRenilla control (5’TGCTGTTGACAGTGAGCGCAGGAATTATAATGCTTATCTATAGTGAAGCCACAGATGTATAGATAAGCATTATAATTCCTATGCCTACTGCCTCGGA-3’) were generated into LMS (LTR/MCSV/SV40-puro-IRES-GFP) retroviral vector [23]. 3KT cells were infected with a Lenti-CMV-KRAS^G12D construct [39]. To immortalize MEFs cell cultures, retroviral vectors expressing E1a and Hras^G12V were used [23]. For the in vivo competition assay, control MEFs were transduced with a lentiviral plex-Renilla-mCherry (gift from Dr A. Celià-Terrassa, Hospital del Mar Research Institute, Spain). To perform drug response analysis of live cells, a construct expressing a nuclear localization signal (NLS) coupled to GFP was used: pTRIP-SFFV-EGFP-NLS (NLS-GFP) was a gift from Nicolas Manel (Addgene plasmid # 86677). For the overexpression studies, RNF144B cDNA construct (NM_182757.4) was generated in a pcDNA3.1( +)-C-6His vector (Genscript, Netherlands).

Lentiviral supernatant was generated by transient transfection of HEK293T (ATCC, CRL-3216) cells with the following packaging constructs: pMDL (5 μg), pRSV-rev (2.5 μg) and pVSV-G (3 μg)[23]. For retroviral particle production GAG (4.8 µg), and pENV (2.4 µg) constructs were used [23]. 10 μg of vector DNA was transfected using calcium phosphate precipitation. Viral supernatant from HEK293T cells was collected after 48 h, filtered, transferred to cell cultures, and centrifuged at 2200 rpm at 32ºC during 2 h. After 48-72 h, cells were FACS-sorted for the corresponding fluorescence using a BD Influx cell sorter (BD Biosciences). If needed, CRISPR single cell clones were seeded in 96 well plates and expanded to generate isogenic populations. MEFs infected with E1a and Hras^G12V were selected with Puromycin (3 μg/ml, P7255, Sigma) and Hygromycin (200 μg/ml, 400,052, Sigma) for 72 h. MEF immortalized cell lines were used at low passage (passage 6–14) to avoid phenotypes arising from prolonged passaging.

All animal experiments are compliant with ethical regulations regarding animal research and were conducted under the approval of the Ethics Committee for Animal Experiments (CEEA-PRBB, Barcelona, Spain). All animals were euthanized before or at the moment of achieving maximum tumour volume. Subcutaneous tumour models were performed by injection of 1 million cells suspended in 100 μl of PBS in both flanks of 7–10-week-old female Athymic Nude-Foxn1nu mice (Envigo). Tumours were grown for approximately 3 weeks and harvested at the endpoint. For in vivo competition assay, MEF cells were infected with the plex-Renilla-mCherry lentiviral construct or with GFP-labeled shRNA targeting RNF144B or TRP53. Cells were mixed 1:1, evaluated by FACS (LSR Fortessa, BD Biosciences) and 1 million cells were injected subcutaneously into the flanks. After 3 weeks tumours were harvested, minced, and digested in a solution of DMEM, 0,3 mg/ml Collagenase I (C1-BIOC, Sigma) and 10 μg/ml DNAse I (DN25, Sigma) at 37ºC while shaking for 2 h. Digested tumours were filtered through a 45 μM mesh, cleaned of red blood cells with Red Blood Lysis Buffer (11,814,389,001, Roche) and analyzed by cytometry (Fortessa). Subcutaneous tumour growth was followed by caliper measurements and the following formula applied to measure tumour volume: volume = 1/2(length × width²). In the case that tumours did not grow in the flank, measurement was excluded from the comparative analysis.

Intercostal intrapulmonary model was performed by injecting 200.000 A549 cells suspended in 10 μl of PBS through the ribcage into the left lung with a 29G insulin needle and a depth of 4–4.5 mm. 10–12-week-old female Athymic Nude-Foxn1nu mice were used for this study. Weight was monitored biweekly, and animals were euthanized at 6 weeks post-inoculation. Only mice with localized intrapulmonar tumours were considered for tumour burden analysis.

3KT experiments were performed by injecting 1 million cells in 100ul PBS intravenously in the tail vein and after 5 months, animals were euthanized to study lung lesions.

Lungs were inflated with 4% paraformaldehyde (PFA, sc-281692, SCBT) through the trachea and fixed overnight for histological evaluation. Lung sections were performed and scanned with an Aperio ScanScope (Leica) at the Anatomy Department (Hospital del Mar). Tumour area and lung area were measured with ImageJ to calculate tumour burden. Those mice that didn’t present any tumour growth, or that had tumoural growth outside the lung and into the thoracic space were excluded from the analysis. Mice were housed in groups of 5 per cage and irradiated chow and water were provided ad libitum.

50.000 3KT cells were seeded in 6 well plates in triplicates and after 6 days of growth, cells were counted using Trypan Blue staining and a Countess 3 Automatic cell counter (Thermofisher). The experiment was performed in five independent replicas.

3.000 3KT cells were seeded in 6 well plates and after 8 days of growth, cells were fixed using 4% PFA for 10 min and stained with 0.5% crystal violet solution (V5265, Sigma-Aldrich) for 1 h. Plates were scanned with an Amersham Typhoon™ (Cytiva). Crystal violet was dissolved with 10% acetic acid and absorbance was read at 590 nm in a Biotek Synergy HTXmachine (Agilent). The experiment was performed in four independent replicas.

1.000 3KT cells were resuspended in 50 μl of cold Matrigel GFR (354,230, Corning) and seeded as a drop in the wells of a 24 well plate. Soon after seeding the Matrigel domes, the plate was turned upside down and placed in the incubator for 30 min. Afterwards, 1 ml of KSFM media with 20% FBS and 1% penicillin/ streptomycin was added. Spheroids were monitored for 7 days and pictures were taken using a brightfield Olympus CKX53 microscope and an Olympus EP50 camera. Pictures were taken of 4–5 random fields per well with a 10 × objective. Spheroid diameter was analyzed by ImageJ. The number of spheroids quantified was between 180 and 410 depending on the cell line. Experiment was repeated twice and performed in three technical replicates each time.

Cells were lysed in RIPA buffer containing protease inhibitors (cOmplete protease inhibitor cocktail, 11,836,170,001, Roche). Protein extracts were quantified using the Protein Assay Dye Reagent (5,000,006, BioRad) and 20 μg were separated by SDS-PAGE and transferred onto nitrocellulose membranes (Cytiva Amersham). Membranes were blocked for 1 h in 5% milk in PBS-T (PBS with 0.1% Tween 20) and incubated overnight with the corresponding primary antibody in PBS-T 5% milk. For probing antibodies against TRP53 (NCL-L-p53-CM5p, Leica Biosystems), TP53 (sc-126, SCBT), p-γH2AX (9718 T, CST), βACTIN (sc-47778, SCBT), His-Tag (66,005–1-Ig, ThermoFisher), and secondary antibodies anti-rabbit (sc2357, SCBT), and anti-mouse (sc-516102, SCBT) were used. Membranes were developed using the ECL Prime system (RPN2232, Cytiva) and imaged using a ChemiDoc MP (BioRad).

2,5 × 10⁵ A549 or 5 × 10⁵ 3KT cells were seeded in 6 well plates. 24 h after seeding were transfected with 1500 ng of the empty vector (pcDNA3.1 + C-6His) or the OE-RNF144B vector (RNF144B_OHu07981C_pcDNA3.1( +)-C-6His) using Lipofectamine 2000 reagent (11,668,027, ThermoFisher) following manufacturer instructions. Cells were counted after 72 h using Trypan Blue staining and the automatic cell counter Countess 3. A pellet of cells was collected to perform western blot and confirm overexpression. Experiment was repeated thrice and performed in triplicates.

Tissues were collected and fixed in 4% PFA overnight and processed for paraffin-embedding. Slides were stained for Hematoxylin and Eosin (H&E) using standard protocols. Immunohistochemistry was performed with antibodies against Ki67 (12202s, CST) and pH3 (3377 T, CST). Briefly, paraffin sections were re-hydrated and antigen retrieval was performed in a pressure cooker with Sodium Citrate Buffer pH6 for 20 min. 3% H₂O₂ was used to quench the peroxidase for 15 min and blocking was done with PBS / BSA 1% (A9647, Sigma) / 0,3 Triton-X (11,332,481,001, Sigma) for 30 min. Slides were incubated overnight at 4ºC in a humid chamber with primary antibody. The next day, sections were incubated with the 2º antibody (Impress HRP Goat Anti-Rabbit, MP-7451–15, Vector Laboratories) for 1.30 h and afterwards incubated with DAB peroxidase kit (K346711-2, Agilent) and hematoxylin. Slides were mounted with DPX mounting media (06522 Sigma). A Cell Observer (Zeiss) microscope was used for imaging. Images were analyzed and quantified using Qupath [42] (v0.3.2).

A549 (isogenic clones) and 3KT cells (bulk population) carrying Cas9 and sgNT or sgRNF144B were evaluated by amplicon sequencing to detect INDELs in the sgRNA target site. Genomic DNA was extracted from the cells using the DNeasy Blood and Tissue kit (69,504, Qiagen). The sgRNA target sites were PCR amplified using primers flanking the site of interest with recommended overhangs (Fwd:5’-ACACTCTTTCCCTACACGACGCTCTT CCGATCTGTGGCTGAAATGTGTGAGCA-3’ and Rev: 5’-GACTGGAGTTCAGACGTG TGCTCTTCCGATCTCTGTATTTTCTTGCTAGACTCC-3’). PCR was performed to ensure a single band was amplified and PCR products were purified using the QIAquick PCR Purification Kit (28,104, Qiagen) and sent to Genewiz (Leipzig, Germany) using Amplicon-EZ service, able to read from 150-500 bp.

Cells were treated with 10 μM Nutlin-3a for 6 h to stimulate TP53 activation or left untreated, depending on experiment. Total RNA was isolated from cells using TRIzol reagent (15,596,018, ThermoFisher) and reverse transcribed using SuperScript III (18,080,400, ThermoFisher) or SuperScript IV (18,090,050, ThermoFisher), Reverse Transcriptase and Oligo-d(T) primers (18,418,020, ThermoFisher). qRT–PCR was performed using either SYBR green (Roche, 4,707,516,001) or Taqman Gene Expression assays (ThermoFisher). For Taqman: Human TP53 (Hs01034249_m1), mouse TRP53 (Mm01731287_m1), human CDKN1A (Hs00355782_m1), mouse CDKN1A (Mm00432448_m1), human RNF144B (Hs00403456_m1), mouse RNF144B (Mm00461356_m1), housekeeping human HMBS (Hs00609297_m1) and mouse HMBS (Mm01143545_m1). For SYBR green the primers were as follows: Human RNF144B: 5´-TTGTCCTGCCAACAGAGCAC-3´ and human GAPDH: 5´-GCACAGTCAAGGCCGAGAAT-3´. Samples were analyzed in QuantStudio 12 K equipment (Applied Biosystems). The mRNA expression levels of TP53 target genes of interest were standardized with corresponding housekeeping genes and normalized to the untreated control.

1,5 million cells were seeded in 10cm² plates and treated the following day with 0.3 mg/ml of colcemid (10,295,892,001, Roche) for 3 h. Cells were collected and resuspended dropwise with KCL 0.056 M and incubated during 20 min at RT. Cells were then fixed in cold methanol:glacial acetic acid solution (3:1) and washed 3 more times with the fixative solution. Cells were dropped on glass slides from 1,5 m height, dried and stained with 3% Giemsa (GS500, Sigma). After washing, coverslips were mounted and pictures were captured using a brightfield Olympus CKX53 microscope and an Olympus EP50 camera, using a 40 × objective. Chromosomes were counted manually with ImageJ Software. At least 25 cells were analyzed per cell line/genotype.

450.000 cells were seeded in 6 well plates and the next day, cells were trypsinized, washed with PBS and fixed with cold ethanol (70%) in a dropwise manner while vortexing. After 2 h of fixation, cells were pelleted, washed twice with PBS and resuspended in working solution, containing 15 μg/ml of Propidium Iodide (00–6990-50, ThermoFisher) and 300 μg/ml RNAse A (10,109,142,001, Sigma). Cells were incubated for 2 h at room temperature (RT) and cell cycle distribution was analyzed by flow cytometry using a BD LSRII-B cytometer (BD Biosciences). Data was analyzed using FlowJo software.

Edu incorporation was performed using the Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kit (C10424, ThermoFisher). Between 250.000–350.000 cells were seeded in 12 well plates and pulsed with 10 μM EdU for 2 h. Next, cells were harvested, fixed, permeabilized and stained using the Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kit following manufacturer’s instructions. Cells were co-stained with a solution containing Propidium Iodide (15 μg/ml) and RNAse (300 μg/ml) to measure DNA content. Samples were analyzed using BD LSRII-B cytometer and FlowJo Software.

15.000 cells were seeded in Phenoplate (6,055,302, PerkinElmer) black well plates and the following day cells were gamma-irradiated at 5 Gy with an IBL-437C (CIS Biointernational). Control plate was left untreated. Cells were fixed with 4% PFA. Afterwards, blocking and permeabilization was performed with PBS/5% BSA/0,3% Triton-X during 1 h at RT. Staining with the primary antibody p-γH2AX (9718 T, CST) dissolved in PBS/1% BSA/0,3% Triton-X was performed overnight at 4ºC. The following day, cells were washed × 3 with PBS and secondary anti-rabbit Alexa Fluor 647 (A21244, Invitrogen) was added during 2 h at RT in the dark. Cells were washed again × 2 with PBS and incubated with 1 μg/ml DAPI (D9542, Sigma) for 10 min. After washing, cells were imaged with the Opera or Operetta High Content Screening System (Perkin Elmer), using the 40 × objective. Segmentation of the nuclei using the DAPI signal and quantification of the number of p-γH2AX foci per cell was done using Harmony® High-Content Imaging and Analysis Software.

1 × 10⁵ MEFs, A549 and 3KT cells were seeded in a 24-well flat bottom plate in medium containing 10% FCS. 24 h after seeding, the cells were incubated with Doxorubicin (0.05 μg/ml or 0.2 μg/ml), Nutlin-3a (10 μM) or with 0% FBS media, respectively. For 0% FBS experiments, cells were washed 3 × with PBS to remove any residual FBS before addition of medium. Cells were harvested 24 h or 72 h after, stained with APC Annexin V kit (640,920, Biolegend) and 1 μg/ml DAPI and analyzed with an BD LSRII-B cytometer and FlowJo Software.

Between 120.000 and 150.000 cells were grown in 24-well plates. The day after cells were treated or not with 15 µM RO-3306 for 18 h at 37 °C, 5% CO₂. Cells were washed with PBS, fixed with 4% PFA for 10 min at RT and permeabilized with PBS / 0.1% Triton X-100 for 5 min at RT. Blocking (RT, 20 min) and incubations with antibodies (RT, 1 h) were performed with 10% FBS in PBS 0.1% / Triton X-100 and washes were done with PBS 0.1% / Triton X-100 at RT for 3 × 5 min. The antibodies targeted α-tubulin (T9026, Sigma) and γ-tubulin (T6557, Sigma). An Alexa 555 Goat anti-Mouse antibody (A-21424, Invitrogen) was used as a secondary antibody. Nuclei were counterstained with 1 μg/mL DAPI for 2 min at RT and cells were mounted using the ProLong Gold antifade reagent (P10144, Thermofisher). Confocal microscopy pictures were taken with a Leica STELLARIS microscope. For counting lagging chromosomes, DNA bridges, multipolar mitosis or centrosome numbers, at least 200 cells were analyzed by eye for each condition.

150.000 cells were grown in 24-well plates. The day after, cells were washed with PBS and fixed in freshly prepared 4% PFA for 10 min at RT. Nuclei were counterstained with 1 μg/mL DAPI in PBS for 2 min at RT and cells were mounted using the ProLong Gold antifade reagent. Confocal microscopy pictures were acquired in a z-stack mode with a Leica STELLARIS microscope. Micronuclei analysis has been made with Fiji software and for each field (45 random field/sample) the number of micronuclei were divided by the number of nuclei.

100.000 cells were grown on 4-well chambered coverslips (80,426, Ibidi). The day after, cells were treated with 15 µM RO-3306 for 18 h. One hour before imaging, siR-Hoechst (SC007, Spirochrome) was added to the media at 1 µM and cells were incubated at 37 °C and 5% CO₂. Just before imaging, media was replaced by FluoroBrite DMEM (A1896701, ThermoFisher) supplemented with 10% FBS and siR-Hoechst. Time-lapse live-cell imaging was performed using a Leica STELLARIS confocal system with white light laser inverted microscope maintaining temperature at 37 °C and CO₂ at 5%. Images were taken every 4 min with a × 64 objective. Exposure time was optimized so that no phototoxicity or photobleaching was caused to cells. Image processing was performed using FIJI software.

A549 and 3KT cells expressing Cas9 and sgNT, sgRNF144B or sgTP53 were infected with the NLS-GFP construct and sorted for GFP + cells. 5 × 10³ cells were seeded in 96 Phenoplate black well plates. 24 h after seeding were treated with Palbociclib (1–3 μM, 3 μM, Hospital del Mar), Abemaciclib (0,5–3 μM, Hospital del Mar), Paclitaxel (10–20 nM, S1150, Selleckchem), Docetaxel (5–20 nM, Hospital del Mar), Etoposide (10–20 μM, 341,205, Sigma), Doxorubicin (0,05–0,2 μg/ml, N31815, Sigma), Carboplatin (50–100 μM, Hospital del Mar), RO-3306 (5 μM, HY-12529, MedChem) and Nutlin-3a (20 μM). Imaging was performed as described previously [43] with the Operetta High Content Screening System using the × 20 magnification. Cell number represented by the sum of the nuclear GFP intensity/well was quantified with the Harmony Software at day 0 (prior to drug treatment) and after 48 or 72 h, depending on the cell line. Cell confluency was normalized to that of day 0 of the same well.

MEFs were infected with the corresponding shRNAs in three independent biological replicas and sorted for GFP by flow cytometry using a BD Influx cell sorter (BD Biosciences). Afterwards, cells were washed with PBS, scrapped with 6 M Urea and 200 mM Ammonium Bicarbonate and sonicated at 4ºC. 10 μl of each sample at 1 mg/ml was submitted for analysis. The samples were digested with Trypsin and LysC and 2 μg were analyzed by LC–MS/MS using a 90 min gradient in the Orbitrap Eclipse. Raw MS files were processed in Proteome Discoverer version 2.3.0.523 (Thermo Scientific, Waltham, MA,) [44]. Samples have been searched against SP_Mouse database (June 2020), using the search algorithm Mascot v2.6 (http://​www.​matrixscience.​com/​). Peptides have been filtered based on FDR and only peptides showing an FDR lower than 5% have been retained. Normalized protein abundances with “Total Peptide Amount” from Proteome Discoverer were used as input for the analysis with the DEP R package [45].

6513 quantified protein profiles were expressed on 9 samples. We only kept proteins that were based at least in two unique peptides, leading to a final protein quantification data matrix of 5389 proteins. Proteins with missing values showed a lower expression in reference to those without missing values. A full normalized matrix of protein expression values was obtained by imputing missing quantifications with a mixed methodology. Proteins with missing at random (MAR) values were imputed with k-nearest neighbors (knn) algorithm and missing not at random (MNAR) values were imputed with random draws from a Gaussian distribution centered around a minimal value (MinProb). We conducted a protein differential expression analysis based on protein-wise linear models and empirical Bayes statistics using limma [46]. Proteins with p-value < 0.05 and a minimum fold-change of 50% were considered as statistically significant. 5039 proteins had no significant change in expression while 350 proteins were differentially expressed between the control replicates and the RNF144B knockdown cells.

Functional enrichment analysis of the biological processes was conducted with the Gene Ontology (GO) database using the clusterProfiler package [47]. Significant GO terms are shown with an associated p-adjusted value (determined by circle color) and GeneRatio (Number of differentially abundant proteins associated with the GO terms / number of input differentially abundant proteins). The circle size is given by the count of proteins detected that are involved in each GO term.

MEFs were infected with the corresponding shRNAs in three independent biological replicas and sorted for GFP with a BD Influx cell sorter. After, cells were trypsinized and the pellet was snap frozen. RNA from 1 million cells was extracted using the Purelink RNA kit (10,307,963, Invitrogen) and submitted for analysis. Libraries were prepared using the TruSeq stranded mRNA Library Prep (20,020,594, Illumina) according to the manufacturer's protocol. Briefly, 1000–500 ng of total RNA were used for poly(A)-mRNA selection using poly-T oligo attached magnetic beads using two rounds of purification. RNA was fragmented under elevated temperature and primed with random hexamers for cDNA synthesis. Then, cDNA was synthesized using reverse transcriptase (SuperScript II, 18,064–014, Invitrogen) and random primers. The addition of Actinomycin D to the First Strand Synthesis Act D mix (FSA) prevents spurious DNA-dependent synthesis, improving strand specificity. After that, second strand cDNA was synthesized, incorporating dUTP in place of dTTP to generate ds cDNA using DNA Polymerase I and RNase H. A corresponding single T nucleotide on the 3’ end of the adapter provided a complementary overhang for ligating the adapter to the fragments. It was followed by subsequent ligation of the multiple indexing adapter to the ends of the ds cDNA. Finally, PCR was performed with a PCR Primer Cocktail. Final libraries were analyzed using Bioanalyzer DNA 1000 or Fragment Analyzer Standard Sensitivity (DNF-473, Agilent), and were then quantified by qPCR using the KAPA Library Quantification Kit KK4835 (07960204001, Roche) prior to the amplification with Illumina’s cBot. Libraries were sequenced 1 * 50 + 8 bp on Illumina's HiSeq2500.

We performed a quality control on the 9 raw single-end reads samples using the nf-core/rnaseq (v. 3.10.1) [48, 49]. Raw FASTQ files were aligned to the GRCm38.p6 version of the reference mouse genome using STAR (v. 2.7.6a) [50] with default parameters except for –sjdbOverhang 49, producing a set of 9 BAM files. Aligned reads in BAM files were reduced to a table of 55,487 genes by 9 samples. Genes were annotated using the GENCODE vM25 GTF file. Following previously established recommendations [51, 52], we filtered out lowly expressed genes by discarding those that did not show a minimum reliable level of expression of 20 counts per million reads of the smallest library size, in at least all the samples of the smallest group, which was 3. After the filtering, we ended up with a final table of counts of 14,668 genes by 9 samples. The DESeq2 package (v. 1.40.0) [52, 53] was used for the differential expression analysis. Surrogate variables were calculated with SVA [54]. Genes with adjusted p-value < 0.05 (5% FDR) and absolute log2FC > 1 were considered as statistically significant.

Integrative transcriptomics vs proteomics analysis was conducted to show the expression relationship patterns of differentially expressed genes vs differentially abundant proteins. Results are represented with Log₂ expression ratio. The cut-offs are a minimum fold-change of 50% for the proteomics expression profile and minimum fold-change of 100% for the transcriptomics expression profile.

To access comprehensive data on GTEx, GDC and TCGA Pan-Cancer normalized gene expression, phenotypic information, and somatic mutations, we utilized XenaBrowser [55] to extract publicly available data from The Cancer Genome Atlas (TCGA) (https://​www.​cancer.​gov/​tcga). The combined cohort of TCGA, and GTEx [56] samples were employed to investigate gene expression differences between normal and tumour samples. To study RNF144B expression in Tp53 wild type or mutated tumors, the GDC-TCGA Pancancer dataset was utilized. Samples were stratified depending on their classification as Tp53 wild type or mutant and Tp53 was considered wild type in the following conditions: no mutation present, synonymous variants (silent) or located in the intronic, 5' UTR, or 3' UTR regions. Tp53 was considered mutant in the following conditions: splice mutations, frameshift, stop codon gain, missense mutation, coding sequence variants, inframe insertions and loss of start or stop point mutations. The considerations for Tp53 status stratification and the specific mutations present in the patient samples are shown in Supplementary Table 1. When analyzing RNF144B expression data, we focused on cancer types that contained a minimum of 20 samples, with both Tp53 wild type and Tp53 mutated entries present in the gene expression matrix. The cancer types that didn’t reach the minimum 20 samples per group are: Ovarian (OV), uterine (UCEC), testicular (TGCT), papillary kidney (KIRP), cervical (CESC), thymus (THYM), mesothelioma (MESO), skin melanoma (SKCM), bile duct (CHOL), clear cell kidney (KIRC), thyroid (THCA), myeloid leukemia (LAML), rectum (READ), B-cell lymphoma (DLBC), uterine (UCS), adrenocortical (ACC), pheocromocytoma (PCPG) and uveal melanoma (UVM) malignancies. Unpaired two-tailed t-test was conducted to evaluate the statistical differences in log-normalized read counts of RNF144B between tissues or cancer types. In order to facilitate visual comparison across TCGA datasets with wild type or mutant Tp53, the expression of RNF144B was mean centered to zero prior to plotting.

GDC-TCGA Pancan datasets were analyzed with www.​xenabrowser.​net for Kaplan Meyer analysis. Samples were stratified by Tp53 status, and by the gene expression levels of RNF144B (being high expression those samples with normalized expression values equal or above the median value and low expression the lower half). Samples containing null data were excluded. Kaplan Meier plots for 10-year overall survival were plotted for remaining samples. Comparison between groups was evaluated with a log rank test.

Analysis of RNF144B as a Tp53 target gene in different mouse and human databases was performed using the TargetGeneRegulation database [57].

RNF144B dependencies in human lung cancer cell lines were analyzed using the Achilles DepMap dataset (DepMap Public 22Q4 + Score, Chronos) [58]. Cell lines were categorized in Tp53 mutant or Tp53 wild type and the CERES dependency score was plotted for each of them.

To perform the analysis of ChIP-seq data, the FASTQ files were acquired from the Sequence Read Archive (SRA) within the Gene Expression Omnibus (GEO) public repository. The specific accession numbers GSE71175 [59] and GSE55727 [60] were utilized to retrieve the FASTQ files corresponding to mouse and human cells, respectively. We used the nf-core/chipseq pipeline (v. 1.2.2) [48, 49]. FASTQ reads were aligned to the GRCm38.p6 reference genome. MACS2 [61] was used to call peaks in the narrowPeak mode. Peaks with a q-value < 10e-5 were considered statistically significant. The resulting data was visualized using the Gviz R package [62].

For the assessment of aneuploidy scores, we utilized the gene expression dataset from the TCGA Pan-Cancer (PANCAN) cohort. Aneuploidy scores were directly obtained from [63]. RNF144B log-normalized read counts were stratified in high and low expression using the median as cut-off. Samples were also stratified by TP53 status, following the criteria used in TCGA Pan-Cancer dataset. The term "PANCAN" denotes the analysis across all cancer types together.

GSEA [64] was carried out in R using the fgsea package v1.24.0 [65], using as a gene set the list of 70 genes (CIN70) with the highest levels of consistent correlation with ‘total functional aneuploidy’ (tFA) from [66]. The C2 curated collection from the Molecular Signatures Database (MSigB) portal was associated with the CIN70 signature. GSEA was used to test for enrichment of specific gene sets within a ranked list based on p-value and log₂FC to define whether the chromosomal instability profile is enriched among the overexpressed proteins of our analysis.

RNF144B has been described as a potent tumour suppressor in mouse B cell lymphoma models [23]. To test if RNF144B could have a tumour suppressor activity in human cancers, we analyzed its expression in publicly available datasets of cancer and healthy tissues (XenaBrowser [55]). We found that RNF144B expression was both downregulated or upregulated in different tumour tissues when compared to normal tissues, respectively (Fig. 1A), indicating its context-dependent regulation. To evaluate if RNF144B expression is TP53 dependent, we classified tumour tissue samples from GDC and TCGA datasets as TP53-deficient and TP53-proficient, based on their TP53 status (Supplementary Table 1). Pan-Cancer analysis revealed that RNF144B expression was significantly lower in the TP53-deficient versus the TP53-proficient tumours (Fig. 1B). Several TP53-deficient tumors, including colon, head and neck, liver, lung adenocarcinoma and stomach cancers showed lower expression levels of RNF144B in comparison to TP53-proficient tumour samples, however this association only reached statistical significance in colon and stomach cancer datasets (Fig. 1B). These correlative studies suggest that RNF144B could have a tumour suppressive role in certain cancer types, particularly when TP53 is wild-type.

The TP53-dependent expression of RNF144B in various human cancer types prompted us to address if RNF144B expression is associated with disease outcome. Human GDC and TCGA Pan-Cancer analysis revealed that disease-survival of the patients with low RNF144B expression was significantly reduced compared to those with higher RNF144B expression, independently of their TP53 status (Fig. 1C). Interestingly, when analysing this correlation across multiple cancer types (Fig. S1), we observed that decreased expression of RNF144B significantly correlated with a worsened prognosis exclusively in the LUAD patients with wild-type TP53, and not in those with mutant TP53 (Fig. 1D), supporting the role for RNF144B in LUAD, particularly when TP53 wild-type is functional.

To further examine the role of RNF144B in lung cancers we analysed lung cancer cell lines data from Project Achilles [58] and found that RNF144B inactivation enhances proliferation of lung cancer cells, especially those with wild-type TP53 (Fig. 1E).

Together, insights from human cancer analyses suggest potential tumor suppressor role of RNF144B in lung adenocarcinoma.

Given that human cancer data indicated the importance of RNF144B in tumor suppression in lung context, we sought to interrogate the role of RNF144B using lung adenocarcinoma models. TP53 plays a key tumor suppressive role in LUAD, with nearly 50% of tumors carrying TP53 mutations [67]. Moreover, our analysis of RNF144B expression and patient prognosis indicates potential regulation of RNF144B by TP53 in LUAD cancer. To establish controlled cell platforms to assess the expression of RNF144B in TP53-dependent manner and its impact in cellular phenotypes, we used TP53 wild type non-transformed and tumour-derived lung cell lines, and their TP53-deficient derivatives. Of note, the non-transformed (but immortalised) cell models we used maintain naturally near diploid characteristics and have limited and defined genetic alterations.

To investigate the role of RNF144B we used CRISPR-Cas9-mediated gene editing in Kras-driven lung models. We selected the HBEC3-KT cell line derived from human normal lung bronchial epithelia (hereafter 3KT) [40]. 3KT cells were transduced with KRAS^G12D-expressing lentivirus and then infected with Cas9 and sgRNAs targeting TP53 or non-targeting control, to establish polyclonal TP53 wild type (sgCTRL^3KT) and TP53 null (sgTP53^3KT) human normal lung bronchial epithelial cell lines (Fig. 2A). As a tumour derived cell line, we selected human A549, a lung adenocarcinoma cell line derived from Type II alveolar epithelium expressing TP53 wild type and KRAS^G12S. Using CRISPR/Cas9-mediated gene editing we generated TP53 null (sgTP53^A549) and control (sgCTRL^A549) isogenic A549 cell lines (Fig. 2A). In addition to lung cancer cell lines, we used primary mouse embryonic fibroblasts (hereafter MEF) cell lines expressing E1A and HRas^G12V oncogenes. These cells serve as a widely used cellular model with an intact TRP53 signaling pathway, where TRP53 plays a critical role in tumor suppression [68]. For this, three independent early-passaged E1A and HRas^G12V oncogene-expressing MEFs, were transduced with shRNAs targeting Trp53 (shTRP53^MEF) or a control shRNA targeting Renilla luciferase (shCTRL^MEF) (Fig. 2A). Western blot analysis of the engineered cell lines confirmed the efficient removal or knockdown of the wild type TP53 protein levels (Fig. 2B). Quantitative RT-PCR (qRT-PCR) analysis of p21 expression, canonical target of TP53 [69, 70], confirmed the abrogation of TP53 signaling in the isogenic TP53 knockout cellular models (Fig. S2A), as well as their resistance to the MDM2 inhibitor Nutlin-3a, a potent activator of TP53 (Fig. 2C, D).

To test the notion that Rnf114b is a TP53 target gene, we examined the expression of Rnf114b in various cell types. We showed that Rnf144b expression is induced in TP53-dependent fashion in TP53 wild type MEFs, A549, and 3KT cells in response to Nutlin-3a (Fig. 2E). In addition, Rnf144b was also induced in other cell types, such as well characterised mouse TP53 wild type KRAS^G12V-driven lung adenocarcinoma cells (mKLCs) [38] and human HCT116 colorectal cancer cells [71] (Fig. 2E and Fig. S2A and B). These findings demonstrate that TP53 can control the expression of RNF144B in certain mouse and human cellular contexts.

To assess deeper RNF144B regulation by TP53 we analysed chromatin immunoprecipitation sequencing (ChIP-seq) using previously published mouse [59] and human [60] datasets. We confirmed that the RNF144B locus is directly bound by TP53, both in human (Fig. S2C) and mouse cells (Fig. S2D). The observed peak in the human sample dataset corresponded with the p53 target sequence identified as p53BS1, which was already characterized to be present in the promoter region of RNF144B in a human glioblastoma cell line [33].

Finally, we consulted the TargetGeneReg database which contains detailed information on the network of genes that are regulated by TP53 [57]. As expected, p21 achieves a perfect "TP53 Expression Score" of 57 (within a range of -55 to 57), indicating consistent and widespread TP53 regulation. RNF144B obtains a score of 31, with 32 out of 57 human datasets analyzed showing a positive correlation between TP53 activation and RNF144B expression, and only 1 dataset showing a negative correlation. (Fig. S2E, Supplementary Table 2). A similar pattern was obtained for TP53 bona fide targets such as Noxa [14, 16], PTPN14 [72], or SLC43A2 [73] (32, 12, and 34 datasets, respectively), underscoring the context-specific nature of TP53 regulation. Regarding mouse datasets, the correlation was not as robust, as TP53 activation correlated with increased RNF144B expression in just 4 out of 15 datasets (Fig. S2E, Supplementary Table 3). Altogether, these results suggest that RNF144B could be directly regulated by TP53 in different contexts.

To test whether RNF144B itself has growth suppressor capacity in vivo we transduced three independent oncogene-expressing E1A and HRas^G12V MEF lines with retroviruses expressing shRNAs against RNF144B (shRNF144B^MEF) (Fig. 2A). We confirmed the RNF144B knockdown in MEFs by qRT-PCR (Fig. S3A). Analysis of these cells demonstrated that attenuated expression of RNF144B protein increased tumour growth in vivo upon subcutaneous injection into immunocompromised mice (Fig. 2F and Fig. S3B). shRNF144B^MEF subcutaneous tumours showed no differences in the level of phospho-Histone H3 (pH3, a proliferation marker) compared to control shCTRL^MEF, while shTRP53^MEF tumours exhibited a significant increase in pH3 staining (Fig. 2G) indicative of more aggressive tumours. We further quantitatively examined the growth advantage of shRNF144B^MEF relative to control shCTRL^MEF by performing in vivo competition assay. We mixed 1:1 mCherry expressing shCTRL^MEF and GFP expressing shRNF144B^MEF or shTRP53^MEF and injected the cells subcutaneously into recipient immunocompromised mice (Fig. S3C and Supplementary Table 4.1 and 4.2). GFP-labeled shRNF144B^MEF and shTRP53^MEF dominated the resulting subcutaneous tumours, validating the in vivo growth advantage caused by RNF144B knockdown.

To further investigate the role of RNF144B in oncogene driven tumour growth, we selected a A549 LUAD model, in which RNF144B is induced by TP53 (Fig. 2E). By CRISPR/Cas9-mediated gene editing we generated an isogenic A549 cell line lacking full-length RNF144B protein (sgRNF144B^A549) (Fig. 2A). The efficiency of RNF144B knock-out was confirmed by qRT-PCR (Fig. S3D) and amplicon sequencing of the targeted gene region (Fig. S3E). Interestingly, our results showed that sgRNF144B^A549 cells had enhanced tumour growth upon orthotopic intrapulmonary injection into immunocompromised mice, displaying a significant increase in tumour size and tumor burden (Fig. 2H). We observed no significant differences between the RNF144B^A549 and control tumours in the proliferation index, as determined by Ki67 staining (Fig. 2I). In contrast, TP53^A549 displayed increased numbers of Ki67 positive nuclei, indicative of more aggressive adenocarcinoma lesions (Fig. 2I), consistent with the crucial tumor suppressive function of p53 in mutant KRAS-driven LUAD.

To evaluate whether RNF144B plays a role in lung adenocarcinoma growth we knockout RNF144B in non-transformed KRAS^G12D expressing lung bronchial 3KT epithelial cell line by CRISPR/Cas9 (sgRNF144B^3KT) (Fig. 2A). RNF144B knock-out efficiency was confirmed by qRT-PCR (Fig. S3D) and amplicon sequencing (Fig. S3E). Combined overexpression of KRAS^G12D and knockout of RFN144B enhanced growth of 3KT cells in 2D and 3D cultures compared with control sgCTRL^3KT cells (Fig. 2J-L), highlighting its potency in lung cancer growth and transformation. Interestingly, upon intravenous injection into immunocompromised mice, sgRNF144B^3KT cells developed lung adenocarcinoma lesions, while these lesions were not evident in control sgCTRL^3KT mice (Fig. S3F). Next, we tested the impact of enforced RNF144B expression on cell proliferation driven by the loss of TP53 in 3KT cells and A549 cells. TP53-deficient sgTP53^A549 and sgTP53^3KT cells transduced with a vector encoding RNF144B had reduced proliferation capacity compared to control cells (Fig. 2M and Fig. S3G). These findings show that RNF144B can impede the growth of the human LUAD cells.

Taken together, oncogene-expressing cells lacking RNF144B gained proliferation and transformation capacity, in particular MEF and lung cells. We next examined apoptosis in response to two different insults: acute DNA damage and serum starvation. In agreement with previously published data in EμMyc-overexpressing B cells [23], we showed that RNF144B has no impact on cell death in oncogene expressing MEF, 3KT and A549 cellular models (Fig. S3H). We next sought to understand which additional cellular processes might be also regulated by RNF144B during transformation suppression by performing various molecular and cellular assays.

To gain mechanistic insights into how RNF144B suppresses cellular proliferation and transformation, we sought to identify the molecular players involved in RNF144B directed pathways. RNF144B contains a conserved RING-between-RING domain and possesses E3 ubiquitin ligase activity [27‐29] and therefore can participate in the targeting and proteasomal degradation of other proteins by ubiquitin transfer. To identify molecular pathways regulated by RNF144B we performed steady-state proteomics in low passaged oncogene-expressing shRNF144B^MEF and shCTRL^MEF (Fig. 3A). Direct comparison of shRNF144B^MEF with shCTRL^MEF cells revealed differentially abundant proteins associated with cell cycle (p21, TRP53, Top2a, Cdk2), chromatin remodeling (Baz1b, H1-1), DNA damage repair and microtubule organization (TPX2, POLB, BCL7C, RAD21) (Fig. 3B-C, Fig S4A, Table S5). Among 291 differentially upregulated proteins, TRP53 appeared elevated in shRNF144B^MEF (Fig. S4A), prompting us to validate steady-state levels of TRP53 protein levels in various cell lines. While TP53 levels were consistently elevated in a relatively narrow range within shRNF144B^MEFs, the large variations in TP53 levels were observed between individual experiments within sgRNF144B^3KT and no differences in sgRNF144B^A549 LUAD cancer cells (Fig. S4B). This indicates that TP53-regulation is likely independent of RNF144B or could occur at tissue- or stage- specific manner.

In parallel, we performed RNA-seq analysis to examine whether the detected changes in protein levels upon RNF144B knockdown were due to the underlying differences in mRNA levels. As expected, considering the E3 ubiquitin ligase function of RNF144B, the downregulation of RNF144B did not provoke major alterations at the transcriptomic levels (Fig. 3D and Supplementary Table 5). Integrative analysis of our transcriptomic and proteomic data showed that most of the changes occurred at the proteomic level, with a clear trend towards upregulation of protein abundance in shRNF144B^MEF cells (Fig. 3E). Together, these results suggest that RNF144B may control the degradation of proteins related to cell cycle progression, microtubule organisation and DNA damage response during transformation suppression.

Proteomics results prompted us to further investigate the role of RNF144B in controlling cell cycle progression and DNA repair, processes implicated in cellular transformation suppression. Analysis of DNA content showed that RNF144B deficient MEFs and 3KT lung cells displayed a subpopulation (> G2) (Fig. 4A and B and Fig. S5) that has features of aneuploidy (DNA content > 4n), similar to TP53 deficient cells (positive control), but absent in control TP53 proficient cells (Fig. 4A and B and Fig. S5).

To further investigate karyotype abnormalities, we conducted chromosome spread analysis, and revealed that RNF144B deficiency in both oncogene-expressing MEF and lung 3KT cells exhibited a greater proportion of cells with elevated chromosome numbers compared to control cells, similar to TP53 deficient cells (Fig. 4C and D).

Beyond maintaining genomic integrity by regulating ploidy, RNF144B has been shown to promote DNA repair in EμMyc-overexpressing B cells [23]. Interestingly, the capacity of RNF144B to regulate DNA repair has been supported by our proteomics analysis (Fig. 3C and Fig. S4A). To determine whether RNF144B function is linked to DNA repair in oncogene-expressing cells we conducted immunofluorescence analysis with γH2AX (a marker for double strand breaks) both in RNF144B deficient MEFs and 3KT before and after γ-Irradiation (IR) (Fig. 4E). Absence of RNF144B led to significantly higher levels of γH2AX foci in the basal state and late response to DNA damage (0 h and 24 h after IR), in both MEFs and 3KT cells (Fig. 4F-H), similar to TP53 deficient cells. These findings suggest that RNF144B plays an important role in maintaining ploidy and supporting DNA damage response in non-transformed oncogene expressing cells.

Aneuploidy and DNA damage have been proposed to be a product of chromosomal instability (CIN) in cancer cells [74]. We sought to determine whether RNF144B downregulation triggers hallmarks of large-scale CIN, such as lagging chromosomes, anaphase bridges, multipolar mitosis or micronuclei, in unstressed oncogene-expressing MEFs. In mitotic shRNF144B^MEF cells, the number of lagging chromosomes or DNA bridges was significantly increased than in control shCTRL^MEF cells (Fig. 5A). Analyses of the micronuclei content resulted in a significantly increased frequency of cells containing micronuclei in shRNF144B^MEF, that could be attributed to missegregated chromosomes (Fig. 5B). We observed significant differences in the number of centrosomes and multipolar mitosis in non-stressed shTRP53^MEF, but comparatively fewer in shRNF144B^MEF and shCTRL^MEF (Fig. S6A and B). To confirm that aneuploidy in shRNF144B^MEFs cells results from defects in chromosome segregation, we performed time-lapse video and immunofluorescence analysis of mitotic oncogene-expressing MEFs cells released after a temporary G2-arrest by RO-3306 (G2/M inhibitor) and found that shRNF144B^MEF have significantly elevated proportion of cells with DNA bridges/lagging chromosomes (Fig. 5C, Fig. S6C and Movie S1) and multipolar mitosis (Fig. 5D and Movie S2). As expected, the positive control shTRP53^MEF cells displayed significantly higher frequency of cells with lagging chromosomes or DNA bridges, micronuclei, multipolar mitosis and an increased number of centrosomes in comparison to control shCTRL^MEF (Fig. 5A-D, Fig. S6A-C and Movies S3, 4 and 5). Collectively, these findings indicated the occurrence of severe chromosomal aberrations, confirming that downregulation of RNF144B triggers hallmarks of large-scale genomic instability due to mitotic failures.

The analysis of aneuploidy scores in human TCGA datasets indicated that low mRNA expression of RNF144B correlated with higher aneuploidy score in LUAD patients with TP53 wild type while this correlation was not significant when testing for TP53 mutant LUAD tumours (Fig. 5E). The negative correlation of RNF144B expression and aneuploidy showed to be significant too when human TCGA Pan-Cancer data were analyzed (Fig. S5D). In addition, we have observed correlation between the RNF144B regulated protein signatures (Fig. 3B and Fig. S4A) with a previously described chromosomal instability signature [66] (Fig. S6E), supporting the link between the genomic instability and RNF144B molecular axis.

All together, we found that RNF144B is a tumour suppressor involved in maintaining genomic stability and its knockdown triggers the appearance of several mitotic defects that eventually lead to chromosomal aberrations and DNA damage in oncogene expressing cells.

Our results support the notion that RNF144B plays a significant role in the DNA damage response, mitosis progression and chromosomal instability. These could indicate that cancer cells that are deficient for RNF144B have reduced sensitivity to drugs that specifically target the cell cycle, CIN or induce DNA damage. To test this hypothesis, we tested a panel of cytotoxic agents in A549 and 3KT cells lacking RNF144B. We used different families of drugs, including microtubule stabilizers (paclitaxel, docetaxel), CDK4/6 inhibitors that cause prolonged arrest cells in G1/S (palbociclib, abemaciclib), CDK1 inhibitor that promotes G2/M arrest (RO-3306), DNA damaging topoisomerase II inhibitors (etoposide, doxorubicin) and an intercalating DNA agent (carboplatin). Nutlin-3a was used as a positive control for the cell viability assays. Cells were treated with increasing concentrations of drugs and analysed for cell growth. Interestingly, RNF144B deficiency led to significant protection to CIN-inducing [75, 76] cell cycle inhibitors Palbociclib and Abemaciclib, as well as to microtubule stabilizers Paclitaxel and Docetaxel, and the DNA damaging agent Etoposide in A549 and 3KT cell lines. As anticipated, positive control TP53 deficient cells were resistant to the majority of the drugs tested, while A549 and 3KT transduced with control sgRNAs were sensitive to the treatments (Figs. 5F and S7A). We next observed that RNF144B deficient cells had elevated levels of γH2AX following Etoposide, Doxorubicin, Paclitaxel and Docetaxel treatment, which was particularly evident in the A549 LUAD cells, indicating that chemoresistance could be caused due to increased tolerance to the presence of DNA damage. While, treatment with Palbociclib, Abemaciclib, and RO-3306 did not induce DNA lesions, as shown by the absence of γH2AX (Fig. S7B). Many reports indicate that the status of p53 determines the cellular response to cytotoxic agents [77]. However, our results show that p53 is not the sole determinant because RNF144B deficient 3KT and A549 cells, both have different p53 levels yet demonstrated quite similar responses to cytotoxic agents.

Collectively, TP53-activated target RNF144B plays a crucial role in maintaining the genomic stability by controlling the degradation of multiple proteins involved in mitotic progression and DNA damage. Decreased RNF144B levels are associated with heightened aneuploidy in human tumors and can ultimately impact the effectiveness of drugs that target the cell cycle and induce chromosomal instability in human lung adenocarcinoma cancer.

The precise molecular mechanism and target genes involved in RNF144B-mediated maintenance of genomic stability and tumor prevention are yet to be determined. Assessment of the direct causative relationships between RNF144B and TP53 in regulating genomic stability are not defined. An additional constraint of our study was the lack of specificity of available anti-RNF144B antibodies.