The PIDDosome controls cardiomyocyte polyploidization during postnatal heart development

The mammalian adult heart is a post-mitotic organ characterized by mainly polyploid cardiomyocytes (CMs). The post-mitotic status of polyploid adult CMs poses a clear limit to heart regeneration. Thus, understanding how CMs acquire this polyploid status is relevant to heart regeneration therapies. Here, we aim to unveil that the PIDDosome, a multi-protein complex known to limit scheduled and accidental polyploidization events by activating p53, implements a CM-specific genetic program that controls polyploidization during postnatal heart development in mice. Flow cytometric DNA content analysis coupled with image-based 3D volumetric measurements of CM nuclei showed that PIDDosome knockout animals harbor significantly more polyploid CM nuclei compared to WT controls. Remarkably, the increased CM ploidy levels do not interfere with cardiac structure or functional output and is implemented in a cell-autonomous manner. During early heart development, ploidy analyses revealed that the PIDDosome starts to shape CM ploidy at postnatal day 7 (P7), reaching a plateau of activity on postnatal day P14. In line with observations made in the hepatocytes, PIDDosome activation in CMs is dependent on PIDD1 localization to extra centrosomes in cells that fail cytokinesis via the distal appendage protein ANKRD26. Interestingly, but in contrast to prior observations in hepatocytes, the PIDDosome limits CM polyploidization in a p53-independent manner but still involves the CDK-inhibitor, p21/Cdkn1a, a notion supported by nuclear RNA sequencing and genetic analyses. Together these results provide a new mechanism how proliferation of polyploid CMs is restricted during development, adding new mechanistic insight into the tightly regulated program of terminal differentiation events occurring in the postnatal heart.


Introduction
Ischemic heart disease (IHD) is one of the leading causes of mortality worldwide with nearly 9 million deaths documented in 2017 1 .The primary cause of a reduced heart function is the initial loss of cardiomyocytes (CMs) and the inability of humans to compensate it 2,3 .In order to reverse this loss, induction of proliferation in pre-existing CMs appears to be a promising approach but is still not applicable due to a clear gap of knowledge of the terminal differentiation process.This terminal differentiation program is activated postnatally and is characterized by several tightly regulated steps in order to generate mature and functional CMs.
After birth, CMs undergo a decrease in cell cycle activity, which is coupled with a change of metabolism from glycolysis to fatty acid oxidation, and an increased maturation status of the sarcomere apparatus 4,5 .In rodents, within the first postnatal week, the majority of CMs fail cytokinesis during mitosis resulting in binucleated CMs 2,6 .At the same time, CMs lose centrosome integrity leading to centriole disengagement, fragmentation of the pericentriolar matrix (PCM), and the translocation of the major microtubule organizing center (MTOC) function from the centrosome to the nuclear envelope 7 .In the second to third postnatal week, a last wave of DNA synthesis occurs in a small subpopulation of binucleated CMs resulting in increased nuclear ploidy in a fraction of cells 8 .Consequently, the adult mouse and rat heart consist of ~80 to 90% binucleated CMs, while the adult human heart contains an estimated ~25% binucleated and ~25 to 40% polyploid mononucleated CMs 2,[9][10][11] Recent studies showed that the number of polyploid CMs limits cardiac regeneration by reducing the proliferation capacity of CMs 12,13 .However, why polyploid CMs are unable to reenter the cell cycle is still unknown.Therefore, a better understanding of mechanisms controlling CM polyploidization is of great interest to uncover new avenues for the treatment of heart disease based on the induction of CM proliferation.
A common feature of mononucleated and binucleated polyploid cells is the presence of supernumerary centrosomes 14 .Centrosomes act as the major microtubule organizing center (MTOC) in most animal cells and form the mitotic spindle poles 15 .Thus, tight control of centrosome copy number is critical for faithful chromosome segregation.A diploid cell contains exactly one centrosome that is duplicated once per cell cycle.In contrast, polyploid cells accumulate centrosomes with each round of incomplete cell division.The centrosome is formed by two centrioles, the mother and the daughter centriole, which are surrounded by the pericentriolar matrix.The mother centriole differs from the daughter centriole as it bears distal and subdistal appendages, including the distal appendage proteins ODF2, CEP83, SCLT1 and ANKRD26 16 .The presence of supernumerary centrosomes, as found in polyploid cells, can alter bipolar spindle organization and metaphase plate formation by acquiring ectopic MTOC activity and thus, creating extra spindle poles 17,18 .This can result in mitotic delays and even chromosome missegregation and thus aneuploidy, which may lead to cell death or senescence 19,20 .To avoid multipolar spindle formation, cells can cluster the extra centrosomes to form a pseudo-bipolar spindle, which can improve but not fully rescue mitotic fidelity 20 21 .
Postnatal binucleated CMs show extra centrosomes.In fact, ~ 65% of P3 binucleated CMs contain 4 centrioles, while ~ 27% have 3 centrioles, with 2 mother centrioles 22 .Moreover, once stimulated with pro-proliferative factors in vitro, P3 binucleated CM can re-enter the cell cycle resulting in either a binucleated cell or a mononucleated cell 22 .In both situations, the nuclear ploidy of the resulting daughter CM is believed to increase.This shows that polyploid CMs have the intrinsic capability to cycle but efforts to induce this process in adults have largely failed.
Polyploid cells with extra centrosomes undergo a p53-dependent cell cycle arrest or cell death response 23,24 .This is initially mediated by the "PIDDosome" 25 .The PIDDosome is a multiprotein complex formed by the auto-processed form of PIDD1 (P53-Induced death domain protein 1), termed PIDD1-CC, which recruits the bipartite adapter RAIDD/CRADD (RIP-Associated ICH1/CED3-homologous protein with Death Domain), and the pro-form of Caspase-2, a member of the family of cysteine-driven proteases known to control cell death and inflammation 26 .PIDD1 localizes to the mother centrioles via the distal appendage protein ANKRD26.The presence and clustering of supernumerary mother centrioles result in the activation of this signaling complex, leading to Caspase-2 activation 27,28 .Active Caspase-2 proteolytically inactivates MDM2, the master regulator of cellular p53 levels, and thereby stabilizes p53 protein to promotes a p21-dependent cell cycle arrest 25 .
Although first described in cancer cells forced to fail cytokinesis 25 , absence of either Caspase-2, RAIDD, or PIDD1, similar to p53 loss itself, increases ploidy in murine primary hepatocytes as part of a physiological polyploidization program during liver organogenesis.Interestingly, the same study has shown that this function of the PIDDosome is preserved also during liver regeneration, and likely also in humans 29 .Altogether, this supports the notion that the PIDDosome is the key player in limiting scheduled as well as accidental polyploidization events via p53 in (non)-cancerous cultured epithelial cells as well as in hepatocytes during liver organogenesis and regeneration 29 .These findings raise the question of whether the PIDDosome is also involved in controlling the scheduled polyploidization of CMs during postnatal heart development, opening new perspectives on heart regeneration therapies based on the induction of CM proliferation by targeting the PIDDosome.

Animals
The generation and genotyping of Casp2 -/-, Casp2 fl/fl , Raidd -/-, Pidd1 -/-mT/mG, p53 -/-and p21 -/- mice was previously described [30][31][32][33][34][35][36] .XMLC2-Cre (XMLC2 + ) mice were a kind gift from Prof. Felix Engel 37 .In order to generate XMLC2 + Casp2 fl/fl mice, XMLC2 + mice were crossed with the Casp2 fl/fl mouse line and bred for more than 10 generations.XMLC2 + mice were crossed with a switchable Tomato and GFP reporter allele (mT/mG) in the Rosa26 locus, allowing expression of membrane bound versions of both fluorescent proteins.The mice used were backcrossed onto C57BL/6N backgrounds for at least 12 generations and handled at the Medical University of Innsbruck Animal Facility.Ankrd26 -/-mice were generated, maintained and genotyped as previously described 38 and heart tissues were kindly provided by Prof. Andrew J. Holland.In all experiments, age-matched animals were used indiscriminate of sex.

Adult and postnatal cardiomyocyte isolation
Adult hearts from 3-month-old animals of the indicated genotypes were sacrificed by CO2 asphyxiation and cervical dislocation according the governmental and international guidelines on animal experimentation.Adult cardiomyocytes were isolated by following the protocol of a Langendorff-free method published by Ackers-Johnson and colleagues 39 with minor changes.
To the Collagenase buffer, 30 µM CaCl2 was freshly added to optimize the digestion step.In order to have a final concentration of 0.5 mg/ml, Collagenase II (255 units/mg, 17101-015, Life Technologies, USA) and Collagenase IV (310 units/mg, 17104-019, Life Technologies) were used at 16.52 mg and 16.81 mg, respectively, as the unit concentrations of these enzymes were different from the original protocol.Moreover, the freshly made collagenase buffer, was prewarmed on a heating magnetic stirrer at 37°C just before being injected in the left ventricle (LV).As first step, 7 ml of EDTA buffer was injected into the apex of the right ventricle with 1ml/min speed as described in the original protocol.Then, 20 ml EDTA buffer was injected into the LV over 10 minutes, and subsequently 3 ml of Perfusion buffer was injected into the LV with 1 ml/min speed.40 ml of Collagenase buffer was finally injected into the LV over 13 minutes.However, after 8 minutes the heart was examined to see whether it has lost its color and structure and if it appeared over-digested.If this was the case, the digestion step was stopped at 8 minutes.After heart digestion and creating ~1 mm 3 pieces, the tissue trituration was done by gently pipetting for 4 minutes, followed by the addition of Stop buffer and further gentle pipetting for another 4 minutes.Adult cardiomyocyte vitality was checked after each isolation via Trypan blue staining.In order to assess cellular ploidy, cells were then centrifuged for 2 min at 100 x g, fixed in 4%PFA diluted in PBS for 20 mins at RT and then washed 3 times with PBS.
Postnatal cardiomyocytes were washed once with PBS (14190169, Thermo Fisher Scientific, USA) and the cell pellet was snap-frozen in liquid nitrogen to preserve RNAs.

Cardiomyocyte nuclei isolation and flow cytometric analyses
The isolation of CM nuclei was performed by following the previously described protocol 40,41 with minor changes.Snap-frozen ventricles from selected genotypes were cut in 1 mm 3 pieces and transferred into a falcon containing 15 ml of lysis buffer and then, homogenized with a TP 18/10 Ultra-Turrax probe homogenizer (IKA, Germany) at 20 000 rpm for 20s.In order to isolate the CM nuclei, the 15 ml lysis buffer solution with the homogenized tissues was passed 8 times up-and -down through a 20 ml syringe with a 20 G needle.The crude nuclei isolate was filtered using at first, a 100 µm cell strainer, and subsequently, a 70 µm cell strainer.After centrifugation for 10 min at 700 x g at 4°C, the nuclei pellet was dissolved in 5 ml sucrose buffer and then, topped up to 20 ml total solution.10 ml sucrose buffer was added into a 1% BSA/PBS pre-coated ultra-centrifuge tube and was overlayed by the 20 ml nuclei suspension before spinning for 60 min at 13 000 x g at 4°C.After centrifugation over sucrose, buffer was quickly removed and the nuclei were resuspended in 700 µl of nuclei storage buffer.Next, 600 µl of nuclei solution was incubated with a rabbit anti-PCM1 antibody (1:300, HPA023370, lot.number: 000007967, Sigma-Aldrich,) overnight at 4°C with constant shaking.100 µl of the nuclei solution was used for the negative control of the staining.On the next day, both nuclei samples were washed by adding 2 ml of PBS, centrifuged for 10 min at 700 x g at 4°C and then, the nuclei pellets were incubated with goat anti-rabbit Alexa 647-conjugated antibodies (1:500, Life Technologies) for 60 min at 4°C.After 1 hour, the samples were washed with PBS and incubated with Propidium Iodide (1:25 of a 1 mg/ml solution).The gating strategy for ploidy analysis is shown in Suppl.Figure 1A, nuclear DNA content was measured on a flow cytometer (LSR-Fortessa, BD Biosystems, USA) and data were analyzed quantitatively, excluding doublets, using FlowJo (FlowJo X, LLC).

Immunohistology
In order to have cardiomyocytes in a relax status (diastole), hearts were injected with a cardioplegic solution (25 mmol/L KCl) in the LV.Once the hearts stopped beating, they were cut from the aorta branch and placed in 4% PFA diluted in PBS overnight at RT. On the next day the fixed hearts were dehydrated and paraffinized.For hematoxylin and eosin staining, 4 µm frontal heart sections were deparaffinized and rehydrated by repeated xylol and ethanol incubation steps and stained.At least 6 different hearts per genotype were analyzed.After the staining, sections were mounted and images acquired on NanoZoomer S210 Digital slide scanner (C13239-01, Hamamatsu photonics, Japan).

Cryosections
Hearts from mentioned genotypes were pre-fixed in 4% PFA overnight at 4°C and on the next day, washed 3 times with PBS and incubated in 15% sucrose diluted in MilliQ water for 4 hat 4°.After the 15% Sucrose solution, hearts were submerged in 30% sucrose diluted in MilliQ water overnight at 4°C.Afterwards, hearts were embedded in Tissue-Tek O.C.T. compound tissue-freezing medium (4583, Sekura, Germany), frozen in liquid nitrogen, and sectioned with a Microm HM550 (Thermo Fischer Scientific) (10 µm).Once the tissues were cut, the sections were left to better adhere on the slide for 30 min at RT.
Immunostainings of pre-fixed heart cryosections were permeabilized with 0.5% Triton in PBS for 10 min at RT. Subsequently, the above-mentioned immunostaining protocol 42 was used with a minor modification of the time of secondary antibody incubation, set to 1h at RT. High resolution images were captured on a LSM980 confocal laser scanning microscope (ZEISS, Germany).

RNA isolation and qRT-PCR
Postnatal CM pellets were processed in order to extract total RNA using TRIzol TM Reagent (15596018, Invitrogen, USA) according to the manufacturer.1µg of total RNA was used for generation of cDNA (iScript cDNA synthesis kit, 170-8891, BioRad, USA).For quantitative real-time PCR (RT-qPCR) experiments 140 ng of cDNA per each reaction was used.RT-qPCR assays were performed in experimental triplicates for each biological replicate using Luna® Universal Probe One-Step RT-qPCR Kit (E3006, New England biolabs, USA) in a StepOne Plus real time PCR system (Applied Biosystems, USA).Relative gene expression was calculated based on ΔCt values using mGapdh as housekeeping gene.

Bulk nuclear RNA sequencing (RNAseq) analysis
For extraction of nuclear CM RNA, postnatal cardiomyocyte nuclei were isolated and stained using anti-PCM1, as described above.Subsequently, cardiomyocyte nuclei were sorted based on their PCM1 staining by using a BD FACSAria™ III Cell Sorter (648282, BD Bioscience).
Importantly, in order to obtain enough RNA material from the isolated cardiomyocyte nuclei, several hearts from the same postnatal days were pooled together per biological replicate (P1: 15 hearts, P7: 6 hearts, P14: 2 hearts).Afterwards, the low-binding Eppendorf tubes containing the sorted cardiomyocyte nuclei were centrifuged for 10 min at 700 x g at 4°C.The resulting nuclei pellets were processed for RNA extraction by utilizing the Quick-RNA MicroPrep Kit (R1050, Zymo Research, USA) following the manufacturer's protocol.RNA-sequencing library were prepared by Lexogen NGS Services (Vienna, Austria) using the QuantSeq 3′ mRNA-Seq Library Prep Kit FWD for Illumina and following the low-input protocol.
Sequencing was preformed on an Illumina NextSeq 2000 at Lexogen NGS Services to produce 100bp single-end reads for each sample.Raw RNA sequencing reads were quality-controlled with FastQC (v0.11.8) 43 and preprocessed with cutadapt (v4.0) 44 to trim poly-G stretches resembling sequencing artefacts, trim low-quality bases from the 3'end, trim adapter and poly-A sequences introduced by the sequencing strategy, remove low-quality reads (more than 1 expected error and/or more than 30% N-bases) and remove short reads (less than 20 nucleotides).Processed reads were aligned against the mouse reference genome GRCm39 from Ensembl (v108) 45 using STAR (v2.6.1e) 46.All following analyses were performed within R v. 4.2.1.The number of reads per gene (considering the full gene, i.e. both exonic and intronic sequences) was counted with HTSeq (v2.0.3) 47 .Gene count normalization and differential gene expression analysis were performed with the R package limma (v.3.52.4) 48.Genes were considered as significantly differentially expressed with an adjusted p-value < 0.05 and an absolute log2 fold change > 1. Gene set enrichment analysis was carried out using the R package gage (v.2.46.1) 49 , based on a selection of gene sets retrieved with the R package msigdbr (v.7.5.1) 50(included gene set collections: hallmark, canonical pathways excluding WikiPathways, transcription factor targets, Gene Ontology).Gene sets were considered as differentially expressed when presenting with a plain p-value < 0.05.

Bulk ncRNA sequencing data pre-processing and analysis
Before gene count normalization, differential gene expression analysis, and all subsequent analysis steps were performed as described above.Three samples (P1 XMLC2 -Casp2 fl/fl replicate 1, P14 XMLC2 -Casp2 fl/fl replicate 3, P14 XMLC + Casp2 fl/fl replicate 1) were excluded due to low quality.The number of detected genes (at least one read count) was particularly low for P14 XMLC2 -Casp2 fl/fl replicate 3 (1676 genes) and P14 XMLC + Casp2 fl/fl replicate 1 (3893 genes) compared to the other samples (between 7223 and 17057 genes) (data not shown).P1 XMLC2 -Casp2 fl/fl replicate 1 clustered particularly far away in a PCA of all samples (PCA calculated with R package labdsv v2.1-0 51 (Suppl.Figure 3A).A fuzzy clustering of all samples showed clusters of genes (clusters 6 and 7) with high expression level in P1 XMLC2 -Casp2 fl/fl replicate 1 compared to replicates 2, 3, and 4 (fuzzy clustering performed with R package Mfuzz v2.56.0 52 (Suppl.Figure 4).The clustering was calculated on the gene log2 fold changes with respect to the average gene counts of the four P1 XMLC2 -Casp2 fl/fl replicates, after each sample's raw counts were normalized with the average count of its interquartile range count values.Subsequent functional enrichment analysis of gene cluster 6 and cluster 7 returned significantly enriched gene sets (adjusted p-value < 0.05) associated with mitochondria, ribosomes and translation (Suppl.Figure 3B).Functional enrichment was calculated with Fisher's exact test, based on the same gene set collection as described above.As these results are indicative of a cytoplasmic contamination of P1 XMLC2 -Casp2 fl/fl replicate 1, this sample was not considered for further analysis.

Imaris image analysis
Nuclear and cellular ploidy measurements were performed by following the indications published by Bensley and colleagues 11 .More in detail, firstly, reconstituting the individual nuclei by the 3D volume module ("3D View" in Imaris) was accomplished and then, analyzing the mean intensity of the Sytox Green was achieved by using the "surface" Imaris software package.Importantly, since signal intensity is fundamental for accurate ploidy measurement, the same setting of laser power, voltage, offset, and pinhole across the board was kept constant for all the experiments.Single nuclei from binucleated CMs were taken as reference for diploid nuclei in this analysis.For nuclear and cellular ploidy analysis, images were captured on a spinning disk confocal microscope, CV1000 Cell voyager (Yokogawa, Japan).

Statistical analysis
Data are expressed as the mean ± standard deviation (SD) or standard error of the mean (SEM) of at least three independent experiments if not stated otherwise.Statistical significance of differences was evaluated by either Student's t test, or 1way or 2way ANOVA followed by Bonferroni's post-hoc test, Sidak's or Tukey's multiple comparisons test (GraphPad Prism 9.0).p < 0.05 was considered as statistically significant.

The PIDDosome controls cardiomyocyte ploidy
In order to decipher the role of the PIDDosome during CM polyploidization, we quantified the nuclear ploidy of freshly-isolated CMs from 3-month-old mice lacking individual PIDDosome components (Figure 1A).The ploidy of single CM nuclei was assessed by 3D volumetric analysis of Sytox Green fluorescence intensity and correlated with the cellular ploidy status.
Moreover, the nuclear staining intensity of single nuclei from binucleated CMs, representing the majority of cells isolated from WT hearts, was used to define 2N ploidy 11 .This analysis revealed that CM nuclei of binucleated cells from all three different PIDDosome knockout strains contain a higher percentage of nuclei with increased DNA staining intensity, compared to nuclei from binucleated CMs from WT hearts.This suggests that binucleated CMs in Pidd1, Raidd and Casp2 deficient animals harbor an increased number of nuclei with > 2N DNA content (Figure 1B).To determine the exact nuclear ploidy of single CM nuclei from PIDDosome-mutant animals, we used a flow cytometry-based strategy.To discriminate CM nuclei from non-myocyte nuclei, we exploited the translocation of the pericentriolar matrix protein PCM1 from the centrosome to the nuclear envelope during terminal CM differentiation 7 (Suppl.Figure 1A).Intriguingly, PIDDosome-mutant animals showed a more than two-fold increase in the percentage of tetraploid (4n) CM nuclei, compared to those isolated from WT animals (Figure 1C).Quantification of the cellular ploidy status of CMs showed that absence of the PIDDosome also causes a significant increase in multinucleated CMs (Figure 1D).
Together, these results indicate that abrogation of the PIDDosome induces a pronounced increase in nuclear as well as cellular polyploidization of CM.This suggests that in the absence of the PIDDosome, tetraploid binucleated CMs enter another round of cell cycle, characterized by cytokinesis failure generating either octoploid binucleated CMs or multinucleated CMs with diploid single CM nuclei.
To test if this phenotype is independent of external parameters, e.g., non-myocytes influencing postnatal CM polyploidization, we generated a mouse mutant harboring a cardiac specific deletion of Casp2 exploiting a floxed Caspase-2 allele and a XLMC2-Cre transgene (XMLC2 + Casp2 fl/fl ) (Suppl. Figure 1B).To test if cardiac CRE expression alters CM polyploidization and to ensure its cardiac-specificity, XMLC2-Cre mice were first crossed with a fluorescence reporter line, carrying a switchable Tomato and GFP reporter allele (mT/mG) in the Rosa26 locus allowing expression of membrane bound versions of both fluorescent proteins.
In the mT/mG model, all cells initially express the Tomato reporter prior to CRE recombination, but not GFP.CRE protein expression leads to excision of both the Tomato cassette and the STOP signal, allowing expression of membrane-bound GFP (Suppl.Figure 1C).Consistently, upon XMLC2-driven CRE activation, CM membranes in heart cryo-sections were GFPpositive, while non-CM membranes retained dTomato positivity.Staining with α-sarcomeric actinin and DAPI was performed in addition to define cell identity and nuclear morphology (Figure 1E).CRE expression under control of the XMLC2 promoter did not affect CM nuclear ploidy (Suppl.Figure 1D).Similar to full body Casp2 -/-mice, 3D volumetric analysis of Sytox Green staining intensity, as well as flow cytometric analysis using DAPI revealed that cardiacspecific Caspase-2 deletion in XMLC2 + Casp2 fl/fl mice significantly increased the percentage of polyploid CM nuclei, when compared to Casp2 fl/fl control mice lacking CRE expression (XMLC2 -Casp2 fl/fl ) (Figure 1F,G).In addition, the cellular ploidy of XMLC2 + Casp2 fl/fl CMs increased to a similar degree as for the full body Casp2 -/-animals (Figure 1H).
Taken together, these data indicate that the PIDDosome controls CM polyploidization, limiting not only their nuclear but also cellular ploidy in a cell autonomous manner.

Increased CM ploidy does not affect cardiac structure or function
Since PIDDosome loss causes an increase of polyploid CMs in adulthood, we wondered whether this might affect the heart structure and thus, cardiac function.The heart ultra-structure of the different genotypes was analyzed by hematoxylin and eosin (H&E) staining on frontal sections of formalin-fixed and paraffin-embedded (FFPE) hearts.No obvious differences were observed in Casp2 -/-, Raidd -/-or Pidd1 -/-hearts compared to WT hearts (Figure 2A), as well as in XMLC2 + Casp2 fl/fl vs. XMLC2 -Casp2 fl/fl (Suppl.Figure 1E).Likewise, echocardiography of 7-10-week old mice lacking Pidd1 confirmed that the observed increases in the polyploid CM population did not affect the cardiac function.In fact, there were no significant differences in ejection fraction (EF) and fractional shortening (FS) between Pidd1 -/-and WT mice (Figure 2B-C).
Polyploid cells harbor multiple copies of the DNA content, and they are usually larger than their diploid counterparts 53 .Since ablation of PIDDosome components causes an increase in polyploid CM, we wanted to investigate whether this ploidy increase was associated with an increase of CM area.For this reason, frontal sections of (FFPE) hearts from Casp2 -/-, Raidd -/-, Pidd1 -/-and WT mice were stained for a membrane marker, Wheat germ agglutinin (WGA), a cardiac marker, Troponin I and DAPI.Quantification of the cross-sectional area of CMs revealed no differences across genotypes, (Figure 2 D), suggesting that increases of ploidy are not associated with an increase in CM size, even though contrasting findings were made in the liver 54 .
Collectively, these data show that PIDDosome deletion and the subsequent increases in the number of polyploid CM do not alter the tissue architecture of the heart and more importantly, cardiac function in steady state.

The PIDDosome restricts cardiomyocyte ploidy in early postnatal development
Cardiomyocytes fail to undergo cytokinesis, thus becoming binucleated during the first week after birth 3,10 .As described by Alkass and colleagues 8 , a second wave of polyploidization occurs between the second and the third postnatal week.To define when the PIDDosome may be engaged during cardiomyocyte development, we first evaluated postnatal mRNA expression of the individual PIDDosome components by qRT-PCR.All PIDDosome components showed low expression levels at P1, while at P7 mRNA levels of all components were found strongly increased.Interestingly, decreasing expression of these genes already at P10 suggest that the PIDDosome components are under tight transcriptional control during heart development (Figure 3A).Next, we examined the exact time-point of PIDDosome-mediated CM-specific ploidy control, hearts from WT, Casp2 -/-, Raidd -/-and Pidd1 -/-mice were isolated on P1, P7, P14 and P21, and CM nuclear ploidy was assessed by flow cytometry.As expected, on P1 no significant difference in the fraction of tetraploid CM nuclei was observed across genotypes (Figure 3B), consistent with diploid mononucleated CMs dominating at this developmental stage 3 .However, at P7 the population of tetraploid CM nuclei in PIDDosome mutant mice increased compared to the same population in WT mice (Figure 3C).This coincides with the formation of binucleated CM at this time point during heart development, as previously reported 3 .Interestingly, the percentage of tetraploid CM nuclei reached a plateau at P14 as no further increases were noted on P21 (Figure 3D-E), matching numbers were seen in adult mice (Fig. 1C).Flow cytometric ploidy analysis of XMLC2 + Casp2 fl/fl mice confirmed that this is a cell-autnonomous effect (Figure 3F).
Collectively, these data show that the PIDDosome controls CM polyploidization as early as P7.

Ploidy control in CM depends on extra centrosomes
PIDDosome activation depends on PIDD1 binding to extra mother centrioles via the centriolar distal appendage protein ANKRD26 in cancer cells 55 , as well as in primary hepatocytes 38 .
Notably, previous work has shown that ~ 65% of binucleated CMs on postnatal day 3 (P3) contain 4 centrioles with extra mother centrioles, while ~ 27% have 3 centrioles 22 .To evaluate the relationship of PIDDosome activation and centrosomes, we quantified the CM nuclear ploidy in 3-month-old Ankrd26 knockout hearts.Indeed, the percentage of tetraploid nuclei in Ankrd26 -/-mice was increased by two-fold, similar to what was observed in the PIDDosome knockout mice (Figure 4A).This strongly suggests that the PIDDosome regulates CM polyploidization based on centrosome number via PIDD1 localization to the centriolar distal appendage protein ANKRD26.

Caspase-2-deficient CMs show enhanced expression signatures related to proliferation
In order to elucidate the biological processes through which the PIDDosome executes its ploidy-regulating function in CMs, CM nuclei of P1, P7 and P14 from XMLC2 -Casp2 fl/fl (control mice) and XMLC2 + Casp2 fl/fl mice were sorted based on PCM1 staining to perform bulk RNAseq (Suppl.Figure 2A).Gene set enrichment analysis results of control CMs supported the previously described developmental changes between P1, P7 and P14 56,57 .In fact, different ion channels were upregulated in P14 compared to P7 in gene-set enrichment analysis, which is a known feature of cardiomyocyte maturation 58,59 .In addition, gene sets associated with cell adhesion were upregulated in P1 compared to P7 CMs, as previously published 57 .Gene sets of RNA splicing, aerobic respiration and oxidative phosphorylation were abundant in P7 CMs in both comparisons, P14 vs. P7 and P7 vs. P1, as a sign of cardiomyocyte maturation 60 .
Interestingly, when comparing nuclear transcriptomes of CMs isolated from P7 vs. P1, gene sets such as mitochondrial protein-containing complexes, as well as mitochondrion organization_ were upregulated, indicating that P7 CMs have a more mature mitochondria structure or organization compared to P1 CMs, as previously shown 59,61 (Suppl.Figure 2B-C).
In addition, a tendency towards the expression of genes, which play a role in sarcomere structure and regulation, previously associated with cardiomyocyte developmental maturation 57,59 , was also noted between P1 and P7 CMs (Suppl.Figure 2D).Since the PIDDosome starts to control CM ploidy status during the first week postnatally (Figure 3), we investigated the differentially expressed genes of Casp2-depleted CMs (XMLC2 + Casp2 fl/fl mice) compared to CMs from control animals (XMLC2 -Casp2 fl/fl mice) at P7 (Figure 4B).Interestingly, genes related to CM proliferation (Foxm1 [62][63][64] , Gas5 65,66 , Sqle 67 , Dusp8 68 , cell cycle regulation (Cdkn1a 69 , Hes1 70 , Trim39 71 , Lin37 72 , Kif18a 73,74 ), ploidy control (E2F8 75 , MAD2L1 76 ) and T-Tubule maturation (Bin1 77 ) were found statistically and significantly altered between genotypes (Figure 4C).Further, gene set enrichment analysis highlighted that GO terms related to chromosome separation, mitotic spindle checkpoint, PLK1 pathway, resolution of sister-chromatid cohesion and microtubule-based processes, were upregulated in Casp2-depleted CMs vs. CMs from control mice on P7, suggesting that Caspase-2-depleted CMs, which are more polyploid compared to their P7 control counterparts (Figure 1), might experience mitotic errors and delays due to their increased DNA content.Importantly, downregulated GO terms in P7 Casp2depleted CMs vs. P7 control CMs were associated with a reduction in fatty acid metabolism and muscle differentiation as well as sarcomere organization (Figure 4C), proposing that CMs lacking Caspase-2 might delay the naturally occurring terminal differentiation program, containing more immature and cycling CMs compared to control mice at that developmental stage.

CM ploidy control does not require p53 pathway activity
Mechanistically, the PIDDosome regulates ploidy by the activation of the protease activity of Caspase-2 and subsequent MDM2 cleavage and its key-components are negatively regulated by E2F8, consistent with hits in our bulk nuclear (nc)RNAseq analysis.As this axis is conserved across several cell types 25,29 , we investigated whether it is also at play in CM polyploidization.
To this end, we analyzed CM ploidy in 3-month-old p53 -/-mice.Surprisingly, the percentage of tetraploid CM nuclei in mice lacking this key ploidy regulator was not significantly different compared to WT animals.In the liver, p53 promotes a p21 induced cell cycle arrest in polyploid hepatocytes 29 .Intriguingly, p21 -/-mice showed a significant increase in ploidy, compared over WT animals.Curiously, the percentage of 4N nuclei in p21 -/-mice was still lower compared to that seen the PIDDosome knockout mice (~16 % vs. ~28%, respectively), suggesting that PIDDosome controls CM ploidy through another route than the canonical MDM2-p53-p21 axis or that p21 and the PIDDosome may act in separate pathways (Figure 4D).However, comparing CM nuclei from Caspase-2 mutant mice with those from Caspase-2/p21 double-knockout animals did not reveal an additional ploidy increase, suggesting that p21 acts downstream of the PIDDosome.
Taken together, these data indicate that the PIDDosome exerts its CM ploidy controlling function depending on extra centrosomes but independently of p53 stabilization, yet engaging p21 in a yet to be defined manner.The transcriptomic profiles of Casp2-depleted CMs suggest that the PIDDosome influences not only the CM cell cycle directly by modulating CM-specific and general cell cycle/ploidy genes, but also indirectly by altering other biological processes, such as lipid metabolism and CM differentiation, which are known to regulate CM proliferation 57,59,78 .As such, we propose that the PIDDosome represents an attractive target to modulate CM differentiation and possibly regeneration.

Discussion
During early postnatal development, cardiomyocytes lose their proliferative capacity and become polyploid, which is an established barrier to heart regeneration.To date, little is known about why polyploid cardiomyocytes cannot re-enter the cell cycle.Here, we show that the PIDDosome multiprotein complex is critical for regulating cardiomyocyte polyploidization.In fact, mice lacking individual PIDDosome components exhibit an increased nuclear and cellular ploidy status of cardiomyocytes, which does not alter cardiac functionality and structure in steady-state.This PIDDosome-mediated cardiomyocyte-specific ploidy control is restricted to the first weeks after birth.Importantly, our data suggest that this pathway is activated in polyploid CMs by "counting" supernumerary centrioles via the localization of PIDD1 with ANKRD26.Interestingly, in CMs the PIDDosome limits polyploidy independently of p53 stabilization, but still involves p21.As Casp2/p21 double-mutant animals show no additional ploidy increase, we conclude that p21 is epistatic downstream of Caspase-2.How Caspase-2 can engage p21 in the absence of p53 is unclear.P73 has been implicated in regulating CM proliferation and is also regulated by MDM2 and may compensate for p53 during postnatal heart development 79,80 , Moreover, stabilizing effects of Caspase-2 on p21 mRNA translation have also been reported 81 ).Regardless of mechanism, these data together with the increased level of 4N cardiomyocyte nuclei in p21 knockout animals suggest that the PIDDosome might exert its function not exclusively via p21, as the increase in 4N cardiomyocyte nuclei in p21 knockout mice is consistently slightly lower than in PIDDosome-deficient CMs.These findings are in line with previous studies which have shown that postnatal cardiomyocyte proliferation depends on p21 82 but not on p53, as a p53 knockout is insufficient to induce cardiomyocyte proliferation 83 .
E2F8 is known to be a fundamental regulator of hepatocyte ploidy during liver development 75 where it transcriptionally targets Casp2 and Pidd1 29 .Recently, it was reported that cardiomyocyte-specific double knockout of E2F8 and E2F7 mice have an increased percentage of mononucleated diploid cardiomyocytes, which had no effect on cardiac function in steady state but also failed to improve heart regeneration after infarction 84 .Our bulk RNAseq data has shown that cardiac-specific deletion of Caspase-2 correlates with an upregulation of E2f8, which could explain the increased level of polyploid cardiomyocyte in these mice, as E2F8 mediated suppression of cytokinesis genes may be more effective.In addition, it seems that the limited change of cardiomyocyte ploidy observed both in E2f7/8 and PIDDosome knockout mice, does not cause alteration in cardiac function suggests a rather broad range of CM ploidy is compatible with the maintenance of heart function and regeneration.The reported association of lower CM ploidy and increased regenerative capacity across a large range of genetic backgrounds in mice 13 needs to be explored further.
Finally, heart regeneration therapies based on cardiomyocyte proliferation face many obstacles in clinical studies and hence it seems worth-while to explore if induction of polyploid cardiomyocyte proliferation is a possible strategy, and if this can be achieved by PIDDosome inhibition.Of notice, small molecules targeting Caspase-2 have been reported [85][86][87] and thus, their validation in preclinical studies is of great interest.

Funding
ML acknowledges support by the Austrian Science Fund (FWF) (Lisa Meitner, M 3115-B), AV acknowledges support by the FWF (P36658), as well as the ERC, AdG 787171 (POLICE).FE acknowledges support from the DOC fellowship program of the Austrian Academy of Sciences (ÖAW).intensity in CM nuclei [arbitrary units] SYTOX Green fluorescence intensity in CM nuclei [arbitrary units] A Leone M et al.

A
Leone M et al.