ADARp150 counteracts whole genome duplication

Abstract Impaired control of the G1/S checkpoint allows initiation of DNA replication under non-permissive conditions. Unscheduled S-phase entry is associated with DNA replication stress, demanding for other checkpoints or cellular pathways to maintain proliferation. Here, we uncovered a requirement for ADARp150 to sustain proliferation of G1/S-checkpoint-defective cells under growth-restricting conditions. Besides its well-established mRNA editing function in inversely oriented short interspersed nuclear elements (SINEs), we found ADARp150 to exert a critical function in mitosis. ADARp150 depletion resulted in tetraploidization, impeding cell proliferation in mitogen-deprived conditions. Mechanistically we show that ADAR1 depletion induced aberrant expression of Cyclin B3, which was causative for mitotic failure and whole-genome duplication. Finally, we find that also in vivo ADAR1-depletion-provoked tetraploidization hampers tumor outgrowth.


Introduction
Evading growth suppressors and resisting cell death have been identified as hallmarks of cancer ( 1 ).Many cancers have lost the canonical tumor suppressor pRB or other proteins that activate the G1 / S checkpoint to prevent DNA synthesis in the absence of mitogenic signaling.This condition is often accompanied by loss of the guardian of the genome p53 that halts cell cycle progression or induces apoptosis in response to overwhelming DNA damage.Further suppression of cell death is frequently achieved by amplification of the antiapoptotic protein Bcl2.To mimic these hallmarks of cancer, we previously disrupted all three retinoblastoma family members in Mouse Embryonic Fibroblasts (Triple Knock Out (TKO) MEFs), alleviating the G1 / S checkpoint.Indeed, these cells can enter S-phase under mitogen-deprived conditions but consequently suffer from severe DNA replication stress, leading to cell cycle arrest and apoptosis.Evading these responses was achieved by overexpression of Bcl2 (TB) and disruption of p53 (TBP).The concomitant suppression of apoptosis and cell cycle arrest allowed G1 / S-checkpoint-defective cells to proliferate mitogen independently, thus mimicking in a cell culture model the unrestrained proliferative capacity of cancer cells ( 2 ,3 ).However, mitogen-independentlyproliferating TBP cells still suffered from DNA replication stress and critically relied on the intra-S-phase checkpoint proteins ATR and CHK1.In shRNA and CRISPR / Cas9 drop-out screens, we sought to identify additional pathways essential for TBP cells to mitigate the deleterious consequences of replication stress: the loss of specific shRNAs or gRNAs from libraries points to pathways that are needed to sustain proliferation under growth-restricting conditions ( 4 ).
Here, we demonstrate that TBP cells critically relied on ADAR1 to maintain mitogen-independent proliferative capacity.ADAR1 has recently received much attention as an RNA editing enzyme that catalyzes the conversion of adenosine (A) to inosine (I) in double-stranded RNA (dRNA).Although the A to I conversion by ADAR1 can recode transcripts, editing by ADAR1 predominantly occurs in non-coding regions such as introns and 3 untranslated regions ( 5 ).ADAR1 produces two protein isoforms, ADARp110 and ADARp150, containing three dRNA binding domains (dRBD1-3) and a catalytic deaminase domain.Unique to ADARp150 is the Z α-domain that allows binding to left-handed Z-RNA / Z-DNA ( 6 ).The short isoform ADARp110 is constitutively expressed in the nucleus and has been implicated in resolving R-loops at telomeres ( 7 ).The longer ADARp150 is well known for its role in editing inversely oriented short interspersed nuclear elements (SINEs) found in endogenous mRNAs.Editing of such RNA structures prevents recognition by a family of cytosolic pattern recognition receptors known as retinoic acid-inducible gene I (RIG-I)-like receptors (RLRs), key sensors of virus infection ( 8 ).Recent studies suggested activation of this pathway to be causative for developmental defects in patients suffering from the type 1 interferonopathy Aicardi-Goutières syndrome (AGS), who carry an inherited defect in ADAR1 ( 6 ,9 ).
We uncovered an unexpected role of ADARp150 in proper chromosome segregation during mitosis: upon ADAR1 depletion, TBP cells gradually became tetraploid by occasional whole genome duplication (WGD) events.Tetraploidy turned out to hamper mitogen-independent proliferation of TBP cells, thus explaining the requirement for ADAR1.Mechanistically we show ADARp150 depletion induced overexpression of Cyclin B3 that was critical for WGD.Finally, we show that tetraploidy also hampered tumor outgrowth in vivo suggesting that in vitro growth-restricting rather than growthpromoting culturing conditions reflect the tumor microenvironment in vivo .

Cell culture
Doubling time of cell cultures was measured by growing 0.15 × 10 6 cells in a 10 cm culture dish for 3-4 days.Cells were counted and the doubling time was calculated using the following formula: incubation time * ln(2) / ln(final cell count / seeding cell count) ( https://www.omnicalculator.com/biology/cell-doubling-time ).For serum starvation cell were seeded and allowed to attach for 4 h, washed with PBS, and incubated for the indicated number of days in serum deprived media.To generate growth curves, we grew cells in μClear 96well plate from Greiner and imaged using IncuCyte ZOOM instrument (Essen Bioscience) every 4 h.

Flow cytometry
The fraction diploid and tetraploid G1 cells on FACS were analyzed using the FUCCI cell cycle reporters and DAPI.Asynchronous FUCCI + cultures were harvested using a two-step fixation protocol.First for 10 min in 4% formaldehyde and subsequently in 90% ice-cold methanol.Before flow cytometry analysis using the LSR2 SORP (BD Biosciences) cells were stained with 7 mg / ml DAPI.FACS-sorting of diploid and tetraploid TBP ARPE cells was performed by using livecell fluorescent dye Hoechst 33342 (Invitrogen).To detect cell death among serum starved TBP MEFs we used Zombie NIR TM Fixable Viability Kit (Biolegend).Finally, cell cycle profiling was performed as described previously ( 2 ).In short, TBP ARPE cells were labeled with 10 mM BrdU, ethanolfixed and stained using propidium iodide.All FACS data was analyzed using FloJo TM software version 10.7.1 (Becton Dickson & Company).

Chromosome spreads
Cells were treated for with 0.1 μg / ml KaryoMAX TM Colcemid TM (Gibco) for 3 hours before harvesting.Next, harvested cells were resuspended in 75 mM KCl solution for 10 minutes and fixed using 3:1 methanol / glacial acetic acid.Fixed cells were dropped onto IHC microscopy slides (DAKO) and stained with 1 mg / ml DAPI.Chromosome spreads were imaged using the metafer system (Metasystems)

Neutral comet assay
Neutral comet assay was performed according to Olive et al. ( 12 ) In short, TBP ARPE cells were harvested and embedded in 1% low agarose gel onto CometSlide (R&D Systems), electrophoresed and stained with propidium iodide.Finally, comets were imaged using the inverted Zeiss AcioObserver Z1 microscope (63 × objective) and analyzed using CASP software ( 2 ).
RNA sequencing 300 000 cells were serum starved for 4 days or grown in the presence of FCS.Cells were washed 3 times with PBS and dissociated using trypsin.Collected cells were washed again and 1 × 10 6 -1.5 × 10 6 were directly resuspended in 800 μl RLT buffer (Roche).Library preparation was performed with the TruSeq polyA stranded RNA prep kit (Illumina) according to the manufacturer protocol.The libraries were analyzed for size and quantity of cDNAs on a 2100 Bioanalyzer using a DNA 7500 chip (Agilent), diluted, and pooled in multiplex sequencing pools.The libraries were sequenced as 51 bp pairedend on a Novaseq 6000 (Illumina) with 20 million reads.Differential expression analysis of the RNAseq data was done with DESeq2 in the statistical programming language R (version 4.0.1)( 14 ).Further downstream exploration and analyses was done in QIAGEN IPA ( 15 ).

Time-lapse microscopy
Time-lapse microscopy of the FUCCI fluorophores was performed as described before ( 2 ).Images were prepared using the 4 × 4 binning mode.Data was stitched including tilefusion using 8% overlap and 3% shift.Analysis of individual mitoses was done by manually following single cells using the Zeiss software.
Ex vivo analysis of tumors NMRI mice ( https:// janvier-labs.com/en/ fiche _ produit/ nmri _ mouse/) were used for xenograft experiments.2 × 10 6 FUCCI-expressing TBP ARPE cells, either diploid or tetraploid and either with ADAR1 shRNA or NT shRNA, in 200 μl PBS were injected in a single flank.These experiments were carried out after approval of the animal welfare committee of the Netherlands Cancer Institute.Tumor size was measured bi-weekly with a caliper, tumor volume was calculated (length × width × width / 2).Most mice (12 / 20) were sacrificed when the maximal tolerated volume of 1.5 cm 3 was reached.The remaining (8 / 20) mice were sacrificed earlier because of ulcerating tumors (4 / 20), secondary tumors (2 / 20) or rectal prolapse (2 / 20).All these mice formed visible tumors which were analyzed ex vivo for the presence of tetraploid cells.We manually minced part of the tumor with a razor (Personna™) and enzymatically in DMEM medium with 3 mg / ml collagenase A (Roche), trypsin and 25 μg / ml DNAse I (Sigma) for 30 min at 37 • C ( 16).Digested tumor cells were filtered (100 μm) and cultured in DMEM media supplemented with 10% FCS (Capricorn) and 0.1 mg / ml penicillin-streptomycin (Sigma) for 2-4 days depending on the confluency of individual tumors and subsequently subjected to FUCCI FACS experiments to detect the level of tetraploid cells among mK O2-hCD T1 + mAG1-hGem − cells.

Statistical analysis
GraphPad Prism software was used for statistical analysis and the generation of graphs.For survival analysis of animal studies, we used Log-rank tests (Mantel-Cox).One-and two-way ANOVA were used for comparisons of multiple groups.All statistical tests were performed two-tailed and adjusted for multiple testing when appropriate.Sample sizes and specific statistical test used were presented in the legends.

ADAR1 is critical for mitogen-independent proliferation
To identify pathways that are critical for mitogen-independent proliferation of G1 / S-checkpoint-defective cells, we performed a genome-wide CRISPR-Cas9 dropout screen.One of the hits of this screen (this unpublished CRISPR screen will be published elsewhere by Van Gemert et al.) indicated that TBP MEFs critically relied on ADAR1 for proliferation in non-permissive culturing conditions.We created ADAR1 knockout clones and observed minimal or no effect on the doubling time in unperturbed culturing conditions (Figure 1 A and Supplementary Figure S1 A).In contrast, upon serum starvation, TBP MEFs critically depended on ADAR1 to maintain proliferative capacity (Figure 1 B).Next to the murine TBP MEFs we also created diploid human retinal epithelial cells (ARPE) with the same genetic background.Like TBP MEFs, TBP ARPE cells ( Supplementary Figure S1 B-D) were able to proliferate in the absence of mitogens and proliferation depended on the intra-S-phase-checkpoint kinase CHK1 ( Supplementary Figure S1 E).Importantly, also for TBP ARPE cells, ADAR1 depletion ( Supplementary Figure S1 F) was lethal only in non-permissive culturing conditions (Figure 1 C, D).

ADAR1 depletion does not aggravate DNA replication stress
We previously showed that mitogen-deprived TBP MEFs had reduced replication fork speed ( 2 ,10 ).Since ADARp110 has been implicated in RNA editing of DNA:RNA hybrids at telomeres we wondered whether ADAR1 depletion aggravates DNA replication stress ( 7 ,17 ).Therefore, we studied replication dynamics using the DNA fiber assay.As expected, TBP ARPE cells transduced with a control (non-targeting; NT) shRNA showed reduced replication fork speed and a trend towards reduced levels of origin firing upon serum deprivation, both indicative of replicative stress ( Supplementary Figure S1 G, H).ADAR1 depleted TBP ARPE cells showed similar changes in replication fork dynamics.In addition, DNA double-stranded break (DSB) formation, measured by the comet assay, was similar in ADAR1-proficient and -depleted cells in non-permissive culturing conditions ( Supplementary Figure S1 I).These results suggest that ADAR1 depletion is synthetic lethal in serum-starved conditions for reasons other than aggravation of DNA replication stress.

ADARp150 is essential in mitogen-deprived conditions
The knockout / knockdown of ADAR1 affected both the short (ADARp110) and long isoform (ADARp150) of ADAR1.To study the effects of the two isoforms separately, we performed reconstitution experiments using ADARp110 and ADARp150 cDNA.In both, mouse and human TBP cells, we observed rescue of mitogen-independent proliferation upon reconstitution by ADARp150 but not by ADARp110 (TBP MEFs Figure 1 E-F and TBP ARPE in Figure G-H).Both isoforms were found to have extensive overlapping editing sites, as expected given their shared dRBD and deaminase domain ( 18 ).The Z α domain, however, is unique to the ADAR-p150 isoform and allows binding to a left-handed structure of RNA and DNA called Z-RNA / Z-DNA ( 6 ).We created point mutations in the ADARp150 cDNA expression plasmid to inactivate the Z α-domain (N173A / Y177A) or catalytic deaminase-domain (E912A) and expressed these ADARp150 mutants in ADAR1-depleted cells ( 6 ,7 ).In contrast to wildtype (WT) ADARp150 cDNA, reconstitution with these mutants failed to rescue mitogen-independent proliferation (Figure 1 H), suggesting that both the Z α-and deaminase domain of ADARp150 are critical in supporting mitogen-independent proliferation.

ADAR1 depletion does not affect interferon signaling of TBP cells
The consequences of ADAR1 deficiency have been studied in mice modelling the type 1 interferonopathy Aicardi-Goutières syndrome (AGS).Over 60% of AGS cases are caused by an inherited point mutation (p.P193A) in the Z α domain of ADAR1 combined with a defective deaminase domain or ADAR1 -null allele ( 19 ,20 ).In mice, ADAR1 deficiency triggered the integrated stress response (ISR) causing postnatal mortality.This could be rescued by an ISR inhibitor (ISRIB), or knockout of components of this pathway ( 9 ).
We therefore considered the possibility that the ISR was responsible for the massive death of serum-starved ADAR1-deficient TBP ARPE cells.Surprisingly though, in our hands, knockouts of the type 1 IFN receptor ( IF-NAR1 ), the RIG1-like receptor (RLR) family members MDA5 ( IFIH1 ), LGP2 ( DHX58 ), RIG-1 ( RIG-1 ), PKR ( EIF2AK2 ) or endoribonuclease RNAseL ( RNASEL ) ( Supplementary Figure S2 A-G and I) did not rescue mitogen-independent proliferation of ADAR1-depleted TBP cells, nor did ISRIB ( Supplementary Figure S2 I).Similarly, disruption of the Z-DNA / Z-RNA binding protein 1 (ZBP1), another mediator of postnatal lethality in AGS mouse models (21)(22)(23)(24), did not rescue lethality in serum-starved, ADAR1 depleted TBP ARPE cells ( Supplementary Figure S2 A, H and I).Finally, we used total RNA sequencing to assess transcripts and pathways that were differentially expressed upon ADAR1 depletion and ADARp150 reconstitution ( Supplementary Figure S2 J).Unexpectedly, in mitogen-proficient conditions, ADAR1 depletion reduced interferon signaling.Introduction of ADARp150 rescued interferon signaling and stimulated pathways involving RIG1-like receptors and PKR.Reduced interferon signaling upon ADAR1 depletion and rescue by ADARp150, were also seen in the absence of mitogens, albeit both to a lesser extent and accompanied by a slight dampening of the RIG1-like receptors pathway as well as the NF-κB activation and JAK / ST A T signalling.Collectively, these analyses make it unlikely that suppression of type 1 interferon signalling explains the requirement for ADARp150 for mitogenindependent proliferation.

ADAR1 depletion results in tetraploidy
The absence of increased IFN signaling upon ADAR1 knockdown prompted us to further characterize ADAR1-depleted TBP cells.Pulse labeling of TBP ARPE cells with the thymidine analogue BrdU revealed that ADAR1-depleted TBP ARPE cells in both, perturbed and unperturbed conditions, contained a significant fraction of cells with tetraploid DNA content (Figure 2 A).To distinguish diploid cells in G2-and tetraploid cells in G1-phase (both 4n and BrdU Neg ), we used FUCCI cell cycle reporters ( 25 ) that allowed us to measure the DNA content of G1 cells in asynchronous cultures (Figure 2 B) ( 26 ).In both, ADAR1-depleted TBP ARPE and TBP MEFs cultured in the presence of mitogens, G1 cells contained significantly more cells with a DNA content larger than 2 n (Figure 2 C, D).
As we are not aware of studies describing a role for ADAR1 in preventing whole genome duplication (WGD), we wondered whether tetraploidization was specific for TBP cells.p53 depletion alone already significantly increased the fraction of cells with a DNA content larger than 2 n in ARPE cells and, to a lesser extent, also in HCT116 cells.ADAR1 depletion aggravated tetraploidization in both cell types, but not in the corresponding ADAR1-proficient cells ( Supplementary Figure S3 A-D).On the other hand, we did not observe tetraploidization in p53KO-MCF7 cells with or without ADAR1 depletion ( Supplementary Figure S3 E-F).These experiments indicate that p53 serves as a powerful safeguard against spontaneous and ADAR1-depletion-induced whole genome duplication, although other (cell-type-specific) protection mechanism likely operate as well.

ADARp150 depletion promotes WGD
We next tested whether reconstitution of ADAR1-depleted cells with wild-type or mutant ADAR1 would affect the formation of tetraploid cells.To this end, we prepared chromosome spreads of all ADAR1-reconstituted cell lines and counted the number of chromosomes (Figure 3 A).Diploid TBP MEFs (mouse origin) are expected to have 40 chromosomes while TBP ARPE (human origin) should contain 46 chromosomes per cell.Although for both cell lines we measured considerable heterogeneity in chromosome counts, most TBP cells had a near-diploid chromosome count (Figure 3 B and D; below horizontal dashed line).However, among ADAR1-depleted TBP MEFs and ARPE cells we found a sizable population with approximately twice the expected chromosome content (Figure 3 B and D; vector only).ADARp150 reconstitution clearly reduced the number of cells with an abnormal karyotype in both TBP cell types (Figure 3 B, D; quantified in Figure 3 C, E), while ADARp110 had no effect.This effect of ADARp150 required its catalytic and Z-RNA / Z-DNA-binding activity as ADAR1-depleted TBP ARPE cells reconstituted with catalytic dead or Z α-domain-defective mutant cDNA remained highly tetraploid (Figure 3 F).Finally, in ADAR1-depleted ifnar1 −/ − TBP ARPE cells, which were unable to proliferate in mitogen-independently, we detected high levels tetraploidy, indicating suppression of whole genome duplication by ADARp150 was not mediated by suppression of the ISR ( Supplementary Figure S3 G).These results show that ADARp150 depletion not only restricted mitogenindependent proliferation but also induced tetraploidization of TBP cells.

ADAR1 depletion blocks mitogen-independent proliferation of tetraploid cells
Does genome duplication upon ADARp150 loss underly the synthetic lethal interaction we observed between ADAR1 depletion and serum starvation?To test the effect of WGD on mitogen-independent proliferation, we used the flow cytometry-based method described above (Figure 2 B) to sort diploid and tetraploid G1 TBP cells with or without ADAR1 depletion (Figure 4 A).FACS-sorted populations were maintained in unperturbed culturing conditions and showed similar doubling times (Figure 4 B).However, in serum-starved conditions tetraploidy reduced the proliferative ability of control TBP ARPE cells (Figure 4C; compare NT shRNA diploidand tetraploid-sorted).Furthermore, while during the short 2week culturing period used here ADAR1 knockdown (Figure 4 D) did not grossly affect the proliferation of sorted diploid TBP cells, ADAR1-depleted tetraploid TBP cells were unable to proliferate upon mitogen deprivation (Figure 4 C).These results indicate that mitogen-independent proliferation is negatively affected by tetraploidy and even completely blocked upon continuing polyploidization in the absence of ADAR1.

Mitotic failures in the absence of ADAR1 activity underlies WGD
The tetraploid cells we observed in the absence of ADARp150 likely resulted from WGD events.To better understand the onset of tetraploidy we cultured diploid-sorted TBP ARPE cells, control and ADAR1 knockdown, over time in unperturbed conditions.It appeared that tetraploidy developed gradually, visibly starting to accumulate after 5-6 weeks, and reaching near-completion in approximately 10 weeks (Figure 5 A, B).To visualize WGD events, we followed FUCCI-expressing TBP ARPE cells over time by time-lapse microscopy.The alternation of FUCCI fluorophores in combination with physical separation of individual cells revealed normal mitotic progression and successful cell division in the majority of both control and ADAR1 depleted cells (Figure 5 C, Supplementary Video S1 ).However, upon ADAR1 depletion we observed increased levels of failed cell divisions, in up to 2% of all mitoses.Consistent with the accumulation of tetraploid cells, these TBP ARPE cells generally continued the cell cycle and remained viable until the end of the experiment (Figure 5 D, E and Supplementary Video S2 -S6 ).These data suggest that the accumulation of tetraploid karyotypes upon ADAR1 depletion likely resulted from a mitotic defect.

Tetraploidization upon ADARp150 depletion by induction of Cyclin B3
The editome of ADAR1 includes a wide range of transcripts and pathways whose expression might be affected by A to I editing ( 27 ,28 ).Notably, ADAR1 has been shown to promote expression of transcripts associated with DNA replication and meiotic synapsis ( 29 ).Ingenuity pathway analy-sis (IPA) of our RNA sequencing data set ( Supplementary Figure S2 J) did not reveal notable changes in selected pathways related to cell cycle control, DNA replication and mitosis upon ADAR1 depletion ( Supplementary Figure S4 A).However, within the 'mitotic roles of polo-like kinase' pathway we observed a strong ( ±95-fold) induction of CCNB3 in ADAR1-depleted cells.CCNB3 encodes Cyclin B3, the third member of the B-type cyclin family.B-type cyclins are induced at the end of G2 phase and by activating cyclindependent-kinase-1 (CDK1) drive mitotic initiation and progression (30)(31)(32).Cyclin B3-CDK1 has been shown to activate the anaphase-promoting complex / cyclosome (APC / C) to promote metaphase to anaphase transition (33)(34)(35).Moreover, overexpression of non-degradable Cyclin B3 halts mitotic progression in late anaphase ( 36 ).Aberrant expression of CCNB3 during mitosis could thus underlie the WGD events that we observed upon ADAR1 depletion.First, we con- firmed that in our RNA sequencing dataset, despite some minor variations, CCNB3 was the only cyclin whose expression was induced upon ADAR1 depletion and restored upon ADARp150 reconstitution (Figure 6 A).Next, RT-PCR revealed an 8-fold increase of CCNB3 transcripts upon ADAR1 depletion, which was abolished upon ADARp150 reconstitution, confirming that upregulation of CCNB3 was specific to ADARp150 depletion (Figure 6 B).The CCNB3 levels in diploid-and tetraploid-sorted TBP ARPE cells transduced with a NT shRNA (Figure 6 C) did not differ, indicating that ADAR1 depletion rather than tetraploidization underlies the induction of CCNB3.Next, we sought to functionally test the involvement of Cyclin B3 in tetraploidization upon ADAR1 depletion.To this end, we used ADAR1-depleted TBP ARPE cell cultures that had accumulated approximately 10% tetraploidy ( ±5 weeks after sorting; Figure 5 A, B) and introduced a NT control shRNA and two independent shRNAs targeting CCNB3 (Figure 6 D).Having successfully prevented the induction of CCNB3 upon ADAR1 depletion we used the approach used in Figure 5 A-B to monitor the accumulation of tetraploid TBP cells over time.It is noteworthy that depletion of Cyclin B3 in ADAR1-depleted cells did not affect the population doubling time of these cells (Figure 6  depletion in TBP ARPE cells induced Cyclin B3 expression, resulting in mitotic failures and accumulation of tetraploid karyotypes.

Relevance of ADAR1 depletion and WGD for tumor growth in vivo
During tumor evolution, the (transient) lack of blood supply to provide sufficient nutrients and oxygen may reflect the nonpermissive culturing conditions that we used in vitro .Likewise, the microenvironment of G1 / S-checkpoint-defective incipient tumor cells may cause unscheduled S-phase entry and DNA replication stress.In this scenario, tetraploidy and ADAR1 depletion would hamper tumor growth of TBP cells.diploid cells remained diploid (Figure 7 C, light grey bars).ADAR1-depleted diploid tumors showed variable levels of tetraploid karyotypes, in one case almost reaching 50% (Figure 7 C, left panel).The fraction of tetraploid cells in tumors originating from ADAR1-proficient tetraploid-sorted cultures showed a variable but clear reduction, while tetraploidy was almost lost when ADAR1 was depleted (Figure 7 C, right panel).These results show that tetraploidization not only conferred a proliferative impediment to TBP cells cultured in serum-deprived conditions in vitro but also when grown as tumors in vivo .The absence of ADAR1 may aggravate this proliferative defect likely because of on-going polyploidization leading to chromosome abundance incompatible with proliferation.

Discussion
In this study we identified an unexpected consequence of ADAR1 depletion: whole genome duplication.We demonstrated that both the catalytic-and Z α domain of ADARp150 are essential to prevent tetraploidization.Absence of functional ADARp150 induced expression of Cyclin B3 which was causative for an increased frequency of mitotic catastrophe and the accumulation of tetraploid cells.We identified this mitotic requirement for ADARp150 in G1 / S-checkpoint defective, apoptosis-resistant, p53 defective cells (TBP cells), cultured in the absence of mitogens and suffering from DNA replication stress.However, also in growth-promoting conditions ADAR1-depleted TBP cells accumulated to near-complete tetraploidy, without showing a discernable proliferative defect.In contrast, in growth-restricting conditions, tetraploidy arising from ADARp150 depletion severely hindered proliferation.But why is division of tetraploid cells hindered in mitogen-deprived conditions, i.e. under conditions of perturbed DNA replication?We envisage that mitogen-deprived cells with higher DNA content may suffer more from replication-induced DNA damage, e.g. because of a higher chance of mitotic catastrophe.This may explain the reduced colony formation of mitogen-deprived, ADAR1proficient TBP ARPE cells shown in Figure 4 C.This effect may be exacerbated by high Cyclin B3 activity in the absence of ADAR1 that could prematurely drive cells with damaged DNA into catastrophic M-phase.
In several isogenic cell lines, ADAR1 deficiency only yielded tetraploid karyotypes upon concomitant p53 inactivation.Presumably, p53 activation by WGD itself or certain types of DNA damage that may promote WGD, results in cell cycle arrest or apoptosis, thereby preventing the accumulation of tetraploid cells ( 37 ,38 ).The requirement for p53 deficiency could explain why a mitotic role of ADAR1 has been underappreciated, albeit not fully ignored.
Earlier studies have found a link between ADAR1 and mitotic progression ( 7 , 39 , 40 ).Shiromoto et al. observed an increase in mitotic abnormalities such as the formation of microand multi-nuclei and anaphase bridges upon ADAR1 depletion in HeLa cells, which was associated with an accumulation of R-loops at telomeres due to shortage of ADARp110 activity ( 7 ).In contrast, the mitotic defects in ADAR1-depleted cells we describe here appeared to result from an aberrant induction of Cyclin B3, which was abolished upon re-introduction of ADARp150.None of the other A-, B-, D-or E-type cyclins was induced.B-type cyclins are critical regulators of mitotic entry and are degraded in a timely manner during mitosis ( 41 ).Cyclin B1 is mainly localized in the cytoplasm during interphase and translocates to the nucleus during prophase to promote nuclear envelope breakdown and mitotic entry ( 42 ,43 ).Inactivation of the spindle assembly checkpoint leads to activation of APC / C CDC20 which targets Cyclin B1 for proteolytic destruction at anaphase onset ( 44 ).Cyclin B3 is a late degrading B-type cyclin and is degraded during anaphase ( 35 ).Studies have proposed that Cyclin B3 promotes APC / C CDC20 activity and anaphase initiation ( 33 ,34 ).Aberrant induction of Cyclin B3 could disturb the metaphase to anaphase transition and mitotic exit thus causing the mitotic failures that we observed upon ADARp150 depletion.Indeed, expression of a non-degradable form of Cyclin B3 has been shown to induce mitotic arrest late in anaphase ( 36 ).How ADARp150 affects Cyclin B3 expression and whether this occurs in a particular cell cycle phase remains an open question.
A role of ADARp150 in suppressing IFN signaling in response to endogenous transcripts has been well established (45)(46)(47)(48)(49). Somewhat unexpectedly, we did not observe increased IFN signaling upon ADARp150 depletion.In fact, total RNA sequencing indicated that IFN signaling was reduced upon ADAR1 knockdown.Possibly, TBP ARPE cells have an impairment in IFN signaling, which is supported by the marginal induction of IFNAR1 expression upon treatment with the immunostimulant poly(IC) ( Supplementary Figure S2 B).Alternatively, ADAR1 knockdown may force a new, attenuated level of IFN signaling.Anyway, the lack of IFN signaling upon ADAR1 depletion in the TBP cells we used may have allowed us to find the mitotic defect that underlies ADAR1 dependency of mitogen-starved TBP cells.
ADARp150 depletion and the resulting tetraploidy also significantly delayed tumor outgrowth in vivo .The inability of tetraploid cells to proliferate in vivo as well as in nonpermissive culturing conditions in vitro is suggestive for a nonpermissive microenvironment that impacts tumor outgrowth.While such environment may be transient and differ spatially within tumors depending on the vascularization, several studies have provided evidence for a growth-inhibiting tumor microenvironment ( 50 ,51 ).
ADAR1 has been proposed as a therapeutic target to enhance the efficacy of immunotherapy (52)(53)(54)(55).Collectively, these studies showed that ablation of ADAR1 results in activation of pattern recognition receptors, thereby activating IFN signaling.The whole genome duplications that we observed upon ADAR1 depletion represent another layer of ADAR1 biology.The relevance of our results for AGS biology and ADAR1 as a target for immunotherapy remains to be determined.ADAR1 has been found overexpressed in several tumor types ( 29 , 56 , 57 ).This has been mostly attributed to the canonical function of ADAR1 as a negative regulator of IFN signaling.Our studies provide another perspective, preventing whole genome duplication.

Figure 2 .
Figure 2. ADAR1 depletion induces tetraploidization of TBP cells.( A ) Cell cycle profiling of BrdU pulse-labeled TBP ARPE cells transduced with a NT or ADAR1 shRNA grown in the presence or absence of serum for the indicated days.Propidium iodide was used to detect DNA content.( B ) Gating strategy of FUCCI FACS protocol ( 26 ) used to quantify the fraction of G1 cells with a DNA content larger than 2n.Cells (1 st column) were selected based on side-scatter (SSC) and f orw ard scatter (FSC), single cells were separated by side-scatter-area (SSC-A) and side-scatter-height (SSC-H), cells gated by using mKO2-hCDT1 + / mAG1-hGem − cells represent G1 cells and finally DAPI was used to identify G1 cells with a DNA content larger than 2 n. (C, D) Percentage of G1 cells with a DNA content larger than 2n in TBP MEF ( C ) and TBP ARPE ( D ) cultures transduced with control (NT) (grey) and Adar1 (blue) sgRNA or ADAR1 shRNA, respectively) cultures.Dots represent independent measurements.Error bars indicate the standard deviation.Asterisks represent adjusted P -value of two-sided one way ANO V A test (Dunnett's multiple comparisons test) (** P -value = 0.008, *** P -value = 0.0 0 09, **** P -value < 0.0 0 01).

Figure 3 .
Figure 3. ADARp150 depletion causes whole genome duplication.( A ) Examples of individual chromosome spreads of non-targeting (NT) and ADAR1 shRNA transduced TBP ARPE cells.( B ) Dots represent the number of chromosomes in individual chromosome spreads of NT (grey) and ADAR1 knockout TBP MEF clone reconstituted with linearized pmGFP (vector only; blue), pmGFP-ADAR-p110 (light blue) and pmGFP-ADAR-p150 (purple).Cells w ere gro wn in unpert urbed cult uring conditions (+10% FCS).Dashed horiz ontal line represent a cut-off of n = 49 belo w which cells are considered (near-)diploid.( C ) Quantification of the fraction of cells with a chromosome count larger or equal to 49 in NT (grey) and ADAR1 knockout TBP MEF cult ures reconstit uted with the indicated vectors.Dots represent independent e xperiments which w ere combined in (B).Error bars indicate the standard deviation.Asterisk show statistical adjusted P -value by one-way ANO V A (Kruskal-Wallis test).Comparisons to NT sgRNA: vector only P -value = 0.0295, ADARp110 P -value = 0.0151, ADARp150 P -value = 0.716.( D ) Number of chromosomes in individual chromosome spreads of TBP ARPE cells transduced with a NT (grey) and ADAR1 shRNA, reconstituted with linearized pmGFP (vector only; blue), pmGFP-ADAR-p110 (light blue), pmGFP-ADAR-p150 (purple).Cells were grown in unperturbed culturing conditions (+10% FCS).Dashed horizontal line represent a cut-off of n = 50 below which cells are considered diploid.( E ) Quantification of the fraction of cells with a chromosome count larger or equal to 50 in TBP ARPE cultures transduced with a NT (grey) or ADAR1 shRNA and reconstituted with the indicated vectors.Dots represent independent experiments which were combined in (D).Error bars indicate the standard deviation.Asterisk show adjusted P -value by one-way ANO V A (Kruskal-Wallis test).Comparisons to NT shRNA: vector only P -value = 0.0167, ADARp110 P -value = 0.0498, ADARp150 P -value > 0.999.( F ) Number of chromosomes in individual chromosome spreads of TBP ARPE cells transduced with an ADAR1 shRNA reconstituted with linearized pmGFP (vector only; blue), pmGFP-ADAR-p150 (purple), pmGFP-ADAR-p1 50-E91 2A (dark-green) and pmGFP-ADAR-p150-Z α mut (dark-blue).Cells were grown in unperturbed culturing conditions (+10% FCS).Dots represent the number of counted chromosomes in individual chromosome spreads.

Figure 4 .
Figure 4. P rogressiv e polyploidization upon ADAR1 depletion causes lethality under mitogen-depriv ed conditions.( A ) DNA content (DAPI) of G1 TBP ARPE cells transduced with non-targeting (NT; grey) or ADAR1 shRNA (blue).The 1st row represents the parental population.The 2nd and 3rd row represent the diploid-and tetraploid-sorted fraction, respectively, that were FACS-sorted from the parental cultures.( B ) Population doublings (h) of diploid-(left three bars) and tetraploid-sorted (right three bars) TBP ARPE cells transduced with NT (grey) and ADAR1 shRNA (blue).Cells were grown in unpert urbed cult uring conditions (+1 0% FCS) and passed before reaching confluency .Dots represent six independent measurements.Error bars indicate standard deviation.( C ) Representative images of 25 0 0 0 diploid-and tetraploid-sorted TBP ARPE cells from ( B ) after 14 da y s of serum starved culturing.Plates were stained with crystal violet.( D ) Protein levels of ADARp110 and ADARp150 of cultures used in (B) and (C).γ-Tubulin is used as a loading control.

Figure 5 .Figure 6 .
Figure 5. Mitotic failures in the absence of ADAR1 activity underlies WGD. ( A ) DNA content (DAPI) of diploid sorted G1 TBP ARPE cells transduced with a non-targeting (NT; grey) and ADAR1 shRNA (blue), cultured over time (rows; weeks).Percentages represent the fraction of G1 cells with 2n (left) or > 2n (right) DNA content.( B ) Percentage of TBP ARPE cells in G1 phase transduced with a non-targeting (NT; grey) and ADAR1 shRNA (blue) with a DNA content > 2 n cultured o v er time (weeks) based on the method described in Figure 2 C. Cultures were measured on FACS bi-weekly and passed in unpert urbed cult uring conditions (+10% FCS).( C ) R epresentativ e e xample of a FUCCI e xpressing TBP ARPE cell transduced with a NT shRNA track ed through mitosis using time-lapse microscopy (20 × objective, 10-min intervals).Pictures represent key events (arrows and text) over time (rows; hours:minutes).( D ) R epresentativ e e xample of a FUCCI e xpressing TBP ARPE cell transduced with a shRNA targeting ADAR1 that underw ent a mitotic f ailure, track ed using time-lapse microscop y (20 × objectiv e, 10-min interv als).Pictures represent k e y e v ents (arro ws and te xt) o v er time (ro ws; hours:minutes).( E ) Percentage of successful cell divisions and failed mitotic events as a percentage of all mitosis.Bars indicate TBP ARPE cells transduced with a NT (grey) or ADAR1 shRNA (blue).Sample size: NT shRNA ( n = 1704) and ADAR1 shRNAs ( n = 1 51 9) individual mitosis divided over three independent experiments (dots).Error bars indicate the standard deviation.Asterisks represent adjusted p-value of a two-sided one-way ANO V A test (Šidák multiple comparison test).Significance le v els per category: 'Cell division' *** P -value = 0.0 0 05, 'Failed mitosis' * P -value < 0.0113, 'Cell death during mitosis' ns = non-significant; P -value = 0.2020.

Figure 7 .
Figure 7.In vivo relevance of ADAR1 depletion and whole genome duplication.( A ) Individual growth curves of diploid and tetraploid TBP ARPE cells with normal (NT) or reduced ADAR1 expression ( ADAR1 shRNA) injected under the skin of NMRI mice (5 animals per cell line).Dashed line indicates the maximally tolerated tumor size.( B ) Kaplan-Meier survival curves of mice injected with the indicated cells.Asterisk indicates P -value.Significance levels in comparison to Diploid NT shRNA: Diploid ADAR1 shRNA * P -value = 0.0177, Tetraploid NT shRNA * P -value = 0.0222, Tetraploid ADAR1 shRNA ** P -value 0.0045).( C ) Percentages of diploid-or tetraploid-sorted TBP ARPE cells with a DNA content larger than 2 n before injection (pre-injection) and after tumor growth in vivo and brief culturing ex vivo ( ex vivo tumors).Grey bars indicate cells with normal ADAR1 expression; blue bars cells with ADAR1 knockdown.