Conserved chamber-specific polyploidy maintains heart function in Drosophila

Polyploidy (whole-genome duplication) is widespread in somatic tissues (Nandakumar et al., 2021; Peterson and Fox, 2021) and is a common mechanism to generate larger cells (Marshall et al., 2012; Øvrebø and Edgar, 2018; Shu et al., 2018). Although functions of tissue polyploidy largely remain mysterious (Fox et al., 2020), one potential clue is the local variation in tissue ploidy levels. For example, although all Drosophila midgut enterocytes and ovarian follicle cells are polyploid, the level of polyploidy varies along the anteroposterior axis of these tissues (Lilly and Spradling, 1996; Viitanen et al., 2021). Similar examples of local polyploidy level variation are found in the fly nervous system, mouse liver and mouse placenta (Nandakumar et al., 2020; Parisi et al., 2003; Tanami et al., 2017), and can be influenced by genetic background (Swift et al., 2023). Given the common relationship between ploidy level and cell size, controlled ploidy variation across a tissue could have profound impacts on tissue form and function.

Perhaps no animal cell type is more frequently polyploid than cardiomyocytes. Developmental polyploidization of cardiomyocytes is highly conserved, occurring across mammals, avians and the fly Drosophila (Brodskiĭ, 1994; Derks and Bergmann, 2020; Hirose et al., 2019; Soonpaa et al., 1996; Swift et al., 2023; Yu et al., 2013). Such developmental cardiomyocyte polyploidy can involve an increase in the number of nuclei per myocyte or in the DNA content of mononucleate cardiomyocytes (Alkass et al., 2015; Bergmann et al., 2015; Brodskiĭ, 1994; Mollova et al., 2013; Peterson and Fox, 2021). Interestingly, developmentally programmed cardiomyocyte ploidy levels may vary by chamber, as suggested by comparisons of ventricular and atrial cardiomyocytes in mice and quail hearts (Anatskaya et al., 2001; Raulf et al., 2015). As summarized recently, functional implications of developmentally acquired cardiomyocyte polyploidy are relatively unclear (Kirillova et al., 2021).

Moreover, a current focus of regenerative medicine research is to prevent developmentally programmed cardiomyocyte polyploidy. This focus comes from intriguing studies in mice and zebrafish that show that developmental polyploidization can block cardiac regeneration (González-Rosa et al., 2018; Han et al., 2020; Patterson et al., 2017). The interest in preventing developmental polyploidy is in line with findings that heart injury can cause cardiomyocytes to exceed polyploidy levels that are set by development, resulting in cardiac hypertrophy (Beltrami et al., 1997; Gan et al., 2020). However, the current focus on blocking developmental cardiomyocyte ploidy may prove problematic given the current lack of understanding of requirements for such naturally occurring whole-genome duplication.

Drosophila melanogaster offers a highly tractable system to uncover conserved developmental regulation and functions of cardiomyocyte biology (Martínez-Morentin et al., 2015; Saha et al., 2022; Wessells et al., 2004; Wolf et al., 2006; Yu et al., 2013, 2015; Zechini et al., 2022). Flies and mammals share evolutionarily conserved transcription factors that specify cardiac fate (Olson, 2006). Mutations in these genes cause heart defects in humans and Drosophila (Akazawa and Komuro, 2005; Reim and Frasch, 2010; Stennard and Harvey, 2005). The Drosophila cardiac organ is known as the dorsal vessel (Demerec, 1950; Rizki, 1978). In embryos and larvae, this tubular organ consists of 104 cardiomyocytes spanning the anteroposterior body axis from segments thoracic 2 (T2) to abdominal 7 (A7) (Molina and Cripps, 2001; Ocorr et al., 2007; Ponzielli et al., 2002; Zaffran et al., 1995; Fig. 1A,A′). A major function of the actively contracting larval cardiac organ is to disperse new immune cells in the lymph (blood) throughout the animal (Johnstone and Cooper, 2006; LaBeau et al., 2009; Rotstein and Paululat, 2016; Rugendorff et al., 1994).

As in mammals, the larval dorsal vessel is composed of molecularly distinct cardiac chambers (Bodmer and Frasch, 1999; Gonzalez et al., 2022; Lo and Frasch, 2001; Lovato et al., 2002; Swift et al., 2023). Embryonic expression of the Hox genes Antennapedia (Antp), abdominal-A (abd-A), Abdominal-B (Abd-B) and Ultrabithorax (Ubx) specifies cardiomyocytes into one of two chambers: either the anterior aorta chamber (T2-A4) or the posterior heart chamber (A5-A7). The aorta is further divided into the anterior aorta (T2-T3) and the posterior aorta (A1-A4) (Bataille et al., 2015; Lo and Frasch, 2001; Lovato et al., 2002; Monier et al., 2005; Perrin et al., 2004; Ponzielli et al., 2002; Schroeder et al., 2022). Each segment from A1 to A7 contains eight cardiomyocytes separated by pairs of ostial cells, except for the posterior A7 segment (hereafter referred to as the apex), which consists of only four cardiomyocytes (Monier et al., 2005). An intracardiac valve between segments A4 and A5 further distinguishes the aorta and heart chambers (Lammers et al., 2017; Molina and Cripps, 2001; Rizki, 1978; Sellin et al., 2006; Zeitouni et al., 2007). In larvae, new lymph cells perfuse into the heart chamber through ostia and then move anteriorly through first the heart chamber and then the aorta before exiting the dorsal vessel (Babcock et al., 2008; Cevik et al., 2019; Curtis et al., 1999).

Drosophila cardiomyocytes are known to be polyploid (Rizki, 1978; Yu et al., 2013), but the developmental regulation and function of such polyploidy is unknown. Intriguingly, the molecularly distinct cardiomyocytes of the aorta and heart chambers are of different sizes (Lovato et al., 2002), hinting at possible ploidy regulation. Here, we determine chamber-specific developmental regulation of cardiomyocyte polyploidy in larvae and identify functional deficits in cardiac ploidy-reduced animals. We reveal that the transition from embryo to larva coincides with cardiomyocytes entering G/S cycles known as endocycles (Schoenfelder and Fox, 2015; Shu et al., 2018; Zielke et al., 2013). This finding mirrors the transition to polyploid cardiomyocytes in many adolescent mammals (Alkass et al., 2015; Bergmann et al., 2015; Walsh et al., 2010). Heart cardiomyocytes endocycle faster and reach a higher final ploidy and cell size than aorta cardiomyocytes. We further show chamber-specific ploidy differences in human donor heart samples, which correlates with higher levels of insulin signaling in the left ventricle compared with the left atrium. In Drosophila, we then demonstrate that differential cell growth and cell cycle signaling sensitivity underlies chamber-specific asymmetry in cardiomyocyte ploidy. Reducing cardiomyocyte ploidy by knocking down the insulin receptor (InR) or the endocycle regulator fizzy-related (fzr) preferentially impacts the heart chamber compared with the aorta chamber. Animals with perturbation of chamber-specific polyploidy exhibit altered heart chamber dimensions, which are associated with diminished cardiac function, including stroke volume, cardiac output and hemocyte velocity. These defects in cardiac ploidy-reduced animals resemble human cardiomyopathies. Overall, our studies highlight a role for precise, chamber-specific polyploidy in heart development and function, and caution against regenerative medicine strategies that block developmentally acquired cardiac polyploidy.

To pinpoint the developmental onset of cardiomyocyte polyploidy in Drosophila, we examined embryonic cardiac mesoderm cell cycle activity (Fig. 1A). Many Drosophila cells become polyploid in late embryogenesis or early larvae by undergoing mitotic-to-endocycle transitions (Edgar and Nijhout, 2004; Smith and Orr-Weaver, 1991). The entire mesoderm undergoes two divisions at embryonic stages 8 and 9, whereas more spatially distinct cardiac mesoderm divisions occur during late stage 10 and early stage 11, the time of cardiac fate commitment (Alvarez et al., 2003; Bate, 1993; Bodmer and Frasch, 1999; Foe, 1989; Han and Bodmer, 2003; Ward and Skeath, 2000). We used the mitotic marker phospho-histone H3 (PH3) along with Gal4-induced UAS-Fly-FUCCI (fluorescent ubiquitin-based cell cycle indicator; Zielke et al., 2014) to investigate cell cycle dynamics (Fig. 1B-E).

As expression strength of commonly used cardiomyocyte lineage drivers differs at specific stages (Fig. S1A), we examined UAS-Fly-FUCCI driven by different drivers at different stages. Embryonic cardiac mesodermal cells are strongly twist (twi)-GAL4⁺ (Greig and Akam, 1993) at stage 8, are twi-GAL4⁺ and strongly pannier (pnr)-GAL4⁺ (Heitzler et al., 1996) at stages 9-11, are strongly Mef2-GAL4⁺ (Ranganayakulu et al., 1996) at stages 14-15, and are strongly Mef2-GAL4⁺ and HandC-GAL4⁺ (Albrecht et al., 2006) at stages 16-17 (Fig. S1A). Consistent with previous studies of embryonic cell cycle dynamics (Foe, 1989), stage 8 twi-GAL4⁺ mesodermal cells are either positive for RFP-CycB_1-266 (S phase), dual positive for RFP-CycB_1-266 and GFP-E2F1_1-230 (G2 phase), or positive for PH3 (M phase). This expression pattern indicates S/G2/M cycles (Fig. 1B-B‴,C,E,F). At stages 10-15, we observe these same cell cycle states in pnr-Gal4⁺ and Mef2-GAL4⁺ cardiac mesoderm, along with the addition of GFP-E2F1_1-230 positive G1 phase. This expression pattern is consistent with a transition to G1/S/G2/M cycles (Fig. 1B-B‴,E,F). Immediately after stage 15, both S and M phase states become undetectable in cardiac mesoderm. Instead, a majority (90%) of HandC-GAL4⁺ cells enter G1, whereas 10% enter G2 for the remainder of embryogenesis (Fig. 1B-B‴,D-F). Given the co-occurrence of S phase and M phase (Fig. 1E) at every embryonic stage examined prior to stage 15, we conclude that embryonic cardiomyocytes do not endocycle, but instead undergo mitotic cycles followed by G1 or G2 arrest.

We next examined larval cardiomyocytes, which are organized into aorta and heart chambers (Fig. 1A′; Lovato et al., 2002; Molina and Cripps, 2001; Monier et al., 2005). Comparing cardiomyocytes between late embryonic and late larval stages (wandering larval third instar, WL3) highlights a clear nuclear size increase (Fig. 1G,H). Over the same time period, cardiomyocyte nuclear number does not change (Fig. 1I). Both L1 and WL3 A1-A7 segments contain exactly 52 cardiomyocytes, consistent with previous counts of late embryo cardiomyocyte nuclei (Ponzielli et al., 2002; Zaffran et al., 1995). These nuclear size and number comparisons support prior conclusions that cardiomyocytes are post-mitotic after embryogenesis (Molina and Cripps, 2001) and likely ‘polytenize’, a term associated with G/S cycles frequently termed endocycles (Stormo and Fox, 2017), during the larval stages (Rizki, 1978). The lack of increased nuclear number from embryo to larva also suggests that cardiomyocytes do not undergo nuclear division followed by incomplete cytokinesis to become multinucleate.

To identify larval cardiomyocytes, we analyzed expression of several Drosophila cardiomyocyte drivers in both the heart and the aorta (Fig. 1A′, Fig. S1). Many commonly used cardiomyocyte drivers for studies of the embryo or adult heart are weakly, regionally or not expressed in the larval cardiac organ (Fig. S1). However, NP5169-GAL4 (hereafter NP>) and Mef2-GAL4 (hereafter Mef2>) (Molina and Cripps, 2001; Monier et al., 2005) are expressed in cardiomyocytes throughout the larval life cycle (Fig. S1D,D′,G,G′). As Mef2> also expresses in non-cardiac muscle (Nguyen et al., 1994; Taylor et al., 1995), we focused our study on NP>, which strongly and specifically expresses in all cardiomyocytes of segments A1-A7 of the larval heart and posterior aorta (hereafter referred to as ‘aorta’), but is not expressed in pericardial cells (Fig. 1K). In our analysis, we excluded thoracic cardiomyocytes (Materials and Methods). To assess whether cardiomyocytes re-enter the cell cycle during larval stages, we continuously fed larvae 5-bromo-2′-deoxy-uridine (BrdU) (Materials and Methods) to identify S-phase activity from embryo hatching through all larval stages. From this feeding regimen, all WL3 cardiomyocytes (4 days post-hatching) were BrdU⁺ (Fig. 1J-J‴). Overall, our results complete the picture of cell cycle dynamics in the embryonic and larval cardiomyocytes, revealing a progression from cell cycle arrest in late embryogenesis to larval endocycles.

To determine the duration and timing of cardiomyocyte endocycles, we performed a BrdU pulse-chase assay, examining S phase over successive 24-h time periods (Materials and Methods). Immediately after hatching and continuing through the first 48 h of larval development at 25°C, nearly every cardiomyocyte of the heart and the aorta underwent DNA replication (Fig. 2A-B′,D). This dramatically decreased after 48 h (Fig. 2C-D). These results suggest that cardiomyocytes initiate endocycles at embryo hatching, during which ploidy increases.

We next examined the level of polyploidy in larval cardiomyocytes (segments A1-A7) using established microscopy methods. Our FUCCI analysis of late embryonic cardiomyocytes predicts that newly hatched larvae would have ploidy values of 2C (G1 arrest) or 4C (G2 arrest). To determine how final larval cardiomyocyte ploidy is impacted by larval endocycles, we measured ploidy (C) of NP>⁺ cardiomyocytes in WL3 animals relative to a haploid 1C standard (Materials and Methods). WL3 cardiomyocytes cluster into two populations with a median ploidy of ∼8.5 C and ∼15.8 C, which is close to a difference of one genome doubling (Fig. 2E). Intriguingly, the two different populations of cardiomyocyte ploidy are mostly separated into distinct cardiac organ chambers (Fig. 2E). The higher ploidy population primarily corresponds to heart chamber cardiomyocytes, whereas the lower ploidy population primarily corresponds to aorta chamber cardiomyocytes. Polyploidy frequently tracks with cell volume (Marshall et al., 2012; Schoenfelder and Fox, 2015). Notably, cell volume and lumen size have been reported to be higher in the heart chamber than the aorta (Lovato et al., 2002). In agreement, our sporadic FLP/FRT-mediated recombination (Xu and Rubin, 1993) labeling of mononucleated WL3 cardiomyocytes (Materials and Methods) revealed a ∼1.6-fold larger area of heart versus aorta cardiomyocytes (Fig. 2F-H). Ploidy can also track with nuclear volume (Cohen et al., 2018; Galitski et al., 1999; Huber and Gerace, 2007; Storchova et al., 2006; Yahya et al., 2022); in agreement, we found that heart cardiomyocyte nuclei have a larger volume that aorta nuclei (Fig. S2A-C). Taken together, cardiomyocyte ploidy, nuclear volume and cell size are higher in the heart chamber compared with the aorta (Fig. 2I). Interestingly, despite having a similar endocycle duration (Fig. 2D), the heart chamber reached a higher final ploidy than the aorta (Fig. 2E), suggesting that endocycles are faster in the heart relative to the aorta (Fig. 2I).

We further analyzed ploidy within each chamber by cardiac organ segment at different temperatures and in different control genotypes. Overall chamber ploidy at 25°C was similar between the two different control genotypes, namely NP> and Mef2> (Fig. S2D). Except for aorta segment A3, all aorta chamber segments (A1-A4) primarily contained cells of the lower ploidy population (Fig. S2E-H). In contrast, heart chamber cells (A5 and A6) were almost exclusively of the higher ploidy population (Fig. S2I,J). We noticed minor, but reproducible, impacts on cardiomyocyte ploidy by temperature. Shifting culturing temperature from 25°C to 29°C slightly altered the distribution of ploidy (Fig. S2D). Upon closer examination, temperature slightly impacted ploidy in most cardiac organ segments except A3 and A4 (Fig. S2E-K). In A1, A2 and A7, ploidy slightly increased at 29°C, but, in contrast, in A5 and A6 ploidy slightly decreased at 29°C (Fig. S2E-K).

We next assessed whether chamber differences in cardiomyocyte ploidy and cell size are reflected in organ-level chamber size differences. Using 3D rendering (Materials and Methods), we noticed an apparent increase in cardiomyocyte layer thickness in the heart relative to the aorta (Fig. 2J,K). To assess these differences further, we used transverse and sagittal optical coherence tomography (OCT) imaging (Choma et al., 2006; Wolf et al., 2006) in living WL3 animals. Indeed, compared with the lower ploidy aorta, the higher ploidy heart chamber exhibited a thicker chamber muscle wall (t; Fig. 2L-N, Fig. S2I, Materials and Methods). Furthermore, the heart chamber exhibits a higher measure of lumen area known as the end diastolic dimension (EDD, a_D; Fig. 2L,M,O). Therefore, the higher ploidy heart chamber exhibits larger and thicker cardiomyocytes and an increased lumen area compared with the lower ploidy aorta. Our findings argue that endocycle regulation aids in sculpting chamber-specific differences in the larval cardiac organ.

The early larval onset of cardiomyocyte polyploidy in Drosophila mirrors the early developmental transition to cardiomyocyte polyploidy in mammals (Bergmann et al., 2015). We next assessed whether chamber-specific polyploidy is conserved in humans. As we report here in Drosophila, the human heart is known to become polyploid during a mid-point in development (adolescence). The average DNA content of the left ventricular cardiomyocytes increases ∼1.7-fold at this time (Bergmann et al., 2015). As in Drosophila, we measured both nuclear ploidy and volume, comparing atrial (A) and ventricular (V) cardiomyocytes in adult human samples (Materials and Methods). Tissue sections of explanted donor hearts from five subjects (three females, two males, ages 41-44, no history or pathology of heart diseases) were procured from the Duke Human Heart Repository (DHHR; Fig. 3A). Because the left (L) side of the heart (LA and LV) pumps oxygenated blood throughout the human body and the LV chamber wall is the thickest (Ho and Nihoyannopoulos, 2006; Walpot et al., 2019), we compared relative DNA content (ploidy) and nuclear volume of LV and LA cardiomyocytes. Ploidy and nuclear volume of LV cardiomyocytes was larger relative to LA cardiomyocytes. Specifically, LV ploidy and nuclear volume were ∼1.6- and ∼1.9-fold higher, respectively (Fig. 3B-D, Fig. S3A,B), and were positively correlated (r=0.6978, Fig. S3C). These findings reveal that, as in Drosophila, the human heart also has chamber-specific asymmetry in ploidy.

To identify potential cell signaling underlying chamber ploidy differences, we analyzed single-cell and nuclear RNA-sequencing datasets from human atrial and ventricular cardiomyocytes (Litviňuková et al., 2020). KEGG pathway analysis of upregulated genes in ventricular cardiomyocytes showed enrichment of biological processes involving insulin signaling pathway genes (Table S1, ‘V_UP’ and ‘KEGG_V_UP’; Fig. 3E, Fig. S3D,E). Furthermore, one of the top upregulated ventricular genes (based on adjusted P-value) was INSR (insulin receptor;  Table S1, ‘DE_V_vs_A’; Fig. 3E). A STRING analysis (Materials and Methods) of genes identified in biological processes enriched by KEGG of upregulated human ventricular genes (Table S1, ‘KEGG_V_UP’; Fig. S3D,E) also revealed insulin signaling downstream targets, such as RPTOR, which regulates mammalian target of rapamycin complex 1 (mTORC1) (Chong and Maiese, 2012). To assess our mRNA findings at the protein level, we immunostained LV and LA sections with a human insulin receptor (INSR) antibody. Indeed, LV cardiomyocytes exhibited a distinct INSR enrichment near membrane-associated intercalated discs (Fig. 3F,F′, Fig. S3F,F′; Estigoy et al., 2009). This pattern was not as prominent in LA sections (Fig. 3G,G′). INSR expression was significantly higher in intercalated discs in LV cardiomyocytes relative to the LA (Fig. 3H-H″). Overall, our results indicate that chamber-specific asymmetry in ploidy is conserved between Drosophila and humans and highlight differential insulin signaling as a likely underlying mechanism in humans (Fig. 3I).

Based on our analysis of single-cell RNA-sequencing data of human ventricular and atrial cardiomyocytes, we examined whether InR also regulates chamber-specific ploidy asymmetry of the Drosophila larval heart. Diet influences PI3 kinase/insulin signaling in Drosophila (Britton et al., 2002; Sudhakar et al., 2020; Zielke et al., 2011), which can regulate polyploidization, cell growth and cell proliferation (Bohni et al., 1999; Britton et al., 2002; Celton-Morizur et al., 2009; O'Brien et al., 2011; Saucedo et al., 2003; Shingleton et al., 2005; Shirakawa et al., 2017; Tamori and Deng, 2013; Zielke et al., 2011). Indeed, cardiomyocyte endocycles are prevented during starvation (Fig. S4A-D). To assess whether Drosophila InR activity regulates larval cardiomyocyte polyploidization, we used larval-onset NP> to drive UAS-InR RNA interference (RNAi; NP>InR RNAi). This RNAi construct has been validated for knockdown of InR function (Texada et al., 2019). Nuclear sizes of WL3-stage cardiomyocytes were visibly smaller in NP>InR RNAi animals compared with NP> controls (Fig. 4A,B; Materials and Methods). Furthermore, cardiomyocyte DNA content decreased in InR RNAi animals (Fig. 4A′-A‴′,B′-B‴′,C, Fig. S4E). We obtained similar results with a second InR RNAi line (Fig. S4E, InR RNAi 2). NP>InR RNAi did not impact cardiomyocyte nuclear number, indicating that our RNAi treatment does not impact embryonic mitotic cycles (Fig. S4F). Furthermore, NP>InR RNAi did not alter NP> expression or noticeably change the pattern of actin-rich cardiac muscle striations, indicating that reduced InR signaling does not impact cardiac cell fate (Fig. 4A,B). Next, to test whether InR activity is sufficient to drive endocycles, we constitutively expressed active InR (NP>InR^CA) in larval cardiomyocytes. This treatment increased overall ploidy of cardiomyocytes without altering cardiomyocyte number (Fig. 4C, D, Fig. S4E,F). These results show that InR activity is necessary and sufficient for WL3 cardiomyocytes to achieve final ploidy through endocycles.

To perturb polyploidy more directly, we additionally assessed whether the well-known endocycle regulator and activator of anaphase-promoting-complex/cyclosome (APC/C) Fizzy-related (fzr)/Cdh1 (Cohen et al., 2018, 2021; Schoenfelder et al., 2014; Sigrist and Lehner, 1997; Zielke et al., 2013) also controls cardiomyocyte ploidy. Indeed, NP>fzr RNAi cardiomyocytes exhibited significantly reduced ploidy (Fig. 4C,D, Fig. S4E). Importantly, NP>fzr RNAi did not impact cardiomyocyte nuclear number (Fig. S4F), confirming that our experimental conditions do not impact the role for fzr in embryonic cardiomyocyte cell division (Drechsler et al., 2018). Furthermore, to determine whether insulin signaling acts through Fzr, we overexpressed constitutively active InR together with fzr RNAi (NP>UAS-InR^CA,UAS- fzr RNAi) in Drosophila larval cardiomyocytes. Increased cardiomyocyte ploidy due to InR gain of function was significantly suppressed by fzr knockdown in both chambers (Fig. 4C,D, Fig. S4E), indicating that InR receptor signaling likely acts through Fzr to increase ploidy. To determine whether other insulin signaling regulators may be involved in cardiac ploidy regulation, we knocked down the kinase mTor. UAS-mTor RNAi targeted to cardiomyocytes significantly reduced Drosophila cardiomyocyte ploidy (Fig. S4C,D). Overall, our results suggest that insulin signaling is important for regulating endocycles and ploidy of cardiomyocytes.

Although NP> is expressed well in both the aorta and heart (Fig. 1K), our genetic manipulations using this driver revealed more profound impacts on ploidy in the heart chamber. The mean ploidy of A5 and A6 heart chamber segments was reduced by 51% in NP>InR RNAi and NP>fzr RNAi WL3 animals, whereas ploidy in A1-A4 aorta segments was reduced only by 41% (Fig. 4C,D, Fig. S4H,I,K). These results suggest that the heart chamber upregulates endocycle machinery to a greater extent than the aorta to achieve faster endocycles and a higher final ploidy. In agreement, experimentally elevating InR in both chambers (NP>InR^CA) increased median ploidy equally (by 0.7-fold from controls in each chamber; Fig. 4C,D, Fig. S4J). Notably, the fold increase in ploidy that we could achieve with NP>InR^CA was much less than the fold decrease in ploidy that we could achieve with NP>InR RNAi, which may reflect an experimental limitation in our system or an upper limit to the tolerated ploidy level in the larval fly heart. Overall, these results highlight differences in cardiomyocyte InR sensitivity that underlie ploidy differences between two cardiac organ chambers (Fig. 4E). Of the two chambers, the heart is the most impacted by loss of InR/Fzr and achieves the higher final ploidy.

We next examined cardiac organ dimensions in animals with altered cardiac ploidy. For ploidy reduction, we used both InR and fzr RNAi animals, as InR can also have ploidy-independent roles (Saltiel and Kahn, 2001; Yoon, 2017). As discussed below, InR and fzr RNAi yielded similar results in essentially every case, suggesting that the effects we are seeing with InR RNAi are primarily due to interference with endocycles/polyploidy and not due to a ploidy-independent InR function.

Our manipulations using fzr RNAi and InR RNAi did not alter body weight (Fig. S4G), suggesting that we are not substantially altering overall body dimensions. Using OCT in live WL3 animals, we next investigated chamber-specific cardiac dimensions in animals with reduced (NP>InR RNAi and NP>fzr RNAi) or increased (NP>InR^CA) ploidy. We measured chamber length (l), wall thickness (t), EDD ( a_D) when the heart is relaxed, and the end-systolic dimension (ESD, a_S) when the heart is contracted (Fig. S2L; Materials and Methods). Previously, wall thickness changes in a model of Raf-induced cardiac hypertrophy were shown to be independent of ploidy (Yu et al., 2013, 2015). Similarly, our larval ploidy manipulations (both increasing and decreasing cardiomyocyte ploidy) did not impact wall thickness in either the heart or aorta (Fig. S5A).

Although our experimental approach to alter cardiomyocyte ploidy had no impact on wall thickness, we found several size alterations in the heart chamber in cardiomyocyte ploidy-reduced animals. In both NP>InR RNAi and NP>fzr RNAi animals, heart chamber length (l_H; Fig. 5A,B,D,E, Fig. S5B,C,E, Movie 1) and EDD (a_D) area (Fig. 5F,G,I,J, Movie 1) were significantly reduced. In contrast, ESD (a_S) of the heart chamber was not affected in cardiac ploidy-reduced NP>InR RNAi and NP>fzr RNAi animals (Fig. S5G,H,J,K). From our length and area measurements, we computed stroke volume, which approximates the amount of hemolymph pumped during each systolic contraction (Fig. S2L; Materials and Methods). Heart chamber stroke volume was significantly decreased for both NP>InR RNAi and NP>fzr RNAi animals (Fig. 5P). In contrast to the heart, the aorta of NP>InR RNAi and NP>fzr RNAi animals remains similar to controls in terms of most functional measurements. We did, however, observe reduced aortic a_D and stroke volume in NP>InR RNAi animals and reduced aortic chamber length in NP>fzr RNAi animals (Fig. 5K-O, Fig. S5F,Q).

In cardiac ploidy-increased NP>InR^CA animals, there was no change in a_D, a_S, length, or stroke volume of the heart or aorta chambers (Fig. 5C,E,H,J,M,O,P, Fig. S5D,F,I,K,N,P,Q). Therefore, our NP>-driven approach to increase larval cardiomyocyte ploidy by InR^CA did not cause noticeable cardiac hypertrophy. These negative results with NP>InR^CA compared with our results with NP>InR RNAi mirror the very mild impact on cardiomyocyte ploidy that we experimentally achieved with NP>InR^CA, again possibly reflecting an upper limit to ploidy in the larval fly heart. Overall, our results indicate that the heart, but not the aorta, of cardiac ploidy-reduced mutants exhibits numerous alterations in heart chamber size, which impact cardiac physiology. The preferential impact on the heart is consistent with this chamber's higher ploidy and dependency on InR signaling.

As the ultimate function of a cardiac organ is to regulate blood or lymph flow, we next measured cardiac output and hemocyte velocity. We compared control and cardiac ploidy-reduced animals (NP>InR RNAi and NP>fzr RNAi), as the mild ploidy increase in NP>InR^CA had no effect on heart chamber dimensions. Using OCT, we determined the heart rate of NP>InR RNAi and NP>fzr RNAi WL3 animals. We measured heart beats per minute for animals of each genotype. Heart rate was not significantly different in either chamber between control, NP>InR RNAi and NP>fzr RNAi WL3 animals (Fig. 6A-D, Fig. S5R). These results differ from findings of heart rate regulation by InR in adult Drosophila during aging (Wessells et al., 2004). We next calculated cardiac output (Fig. S2L), which is the product of stroke volume (Fig. 5P, Fig. S5Q) and heart rate (Fig. 6D, Fig. S5R). Heart chamber cardiac output was significantly reduced in cardiac ploidy-reduced NP>InR RNAi and NP>fzr RNAi animals, but not in the aorta chamber (Fig. 6E,F). Our cardiac output results are consistent with cardiac ploidy-reduced animals exhibiting a compromised measure of heart (not aorta) function.

Given the impact on cardiac output in cardiac ploidy-reduced animals, we next examined the velocity of circulating blood cells (hemocytes) as they move through the cardiac organ before being dispersed throughout the animal (Babcock et al., 2008; Choma et al., 2011). Circulating hemocytes enter the heart chamber at the posterior end and rapidly travel through the aorta with an anterior-directed flow (Babcock et al., 2008) (Fig. 6G). Hemocytes in control animals (NP>) traveled through the heart at an average of 4.6 µm/ms before accelerating to an average of 6.6 µm/ms (Fig. 6H,J,K, Movie 2). However, in NP>InR RNAi and NP>fzr RNAi animals, the average hemocyte speed significantly decreased in both chambers (Fig. 6K). Overall, our results highlight that hemocytes accelerate as they travel from the heart to the aorta (Fig. 6H,K,L), but hemocyte velocity is dysregulated in cardiac ploidy-reduced animals (which have numerous heart chamber defects; Fig. 6I-K,L′).

Here, we identify temporal and molecular regulation of cardiomyocyte polyploidy in the Drosophila larval cardiac organ. Like human cardiomyocytes, which become polyploid during adolescence (Bergmann et al., 2015), polyploidization of Drosophila cardiomyocytes occurs during early juvenile (larval) stages. By combining genetics with light microscopy or optical tomography approaches in both living and fixed animals, we highlight multiple defects in cardiac ploidy-reduced animals. Our data strongly argue for an important role of developmentally programmed polyploidy in Drosophila larvae. Polyploidy in Drosophila cardiomyocytes coincides with larger cell size. Animals with reduced ploidy in the heart chamber exhibit reduced chamber size, stroke volume and cardiac output, and acceleration of circulating hemocytes. Given the high conservation of both transcriptional programming and polyploidy in cardiomyocytes (Akazawa and Komuro, 2005; Olson, 2006; Reim and Frasch, 2010; Stennard and Harvey, 2005), our findings likely apply to any organism with polyploid cardiomyocytes. The idea that programmed polyploidy plays a productive role in cardiac function is not surprising when considering the evolutionary diversity of cardiomyocyte ploidy levels. For example, the giraffe heart is of similar mass as that of other mammals yet generates twice the blood pressure to overcome a severe gravitational challenge. Giraffe left ventricle cardiomyocytes contain, on average, four nuclei and are thus commonly multinucleate polyploid (Ostergaard et al., 2013). Additionally, a large-scale analysis across numerous vertebrate species found that cardiomyocyte polyploidy was likely positively selected for in conjunction with high metabolic demands of mammalian metabolism (Hirose et al., 2019). Based on our findings here, we argue for further study of the role of developmentally acquired cardiac polyploidy in diverse species. One interesting area of future focus is the role of polyploidy by increasing the number of genomes per nucleus versus increasing the number of nuclei per cell, as this can differ between mammalian species (Kirillova et al., 2021; Peterson and Fox, 2021).

Blocking developmental polyploidy in mammals is a current strategy aimed at improving heart regeneration (Derks and Bergmann, 2020). Clearly, manipulation of developmental cardiomyocyte polyploidy impacts regenerative capacity (González-Rosa et al., 2018; Han et al., 2020; Patterson et al., 2017). However, our findings argue that blocking developmental increases in cardiomyocyte ploidy to favor increased recovery from an acute or chronic injury (regeneration) may ultimately weaken cardiac function. Rather, we argue that developmental polyploidy should be disentangled from anti-regenerative gene expression that may arise in conjunction with the transition to the polyploid state. For example, polyploidy and decreased regenerative cell division in mammals is accompanied by higher levels in thyroid hormone (TH) signaling (Hirose et al., 2019). We note that our analysis of chamber-specific human heart expression (Litviňuková et al., 2020) also suggests that TH signaling is upregulated in ventricular cardiomyocytes relative to atrial cardiomyocytes (Fig. S3D). In the future, it may be possible to identify TH targets that are specific to either polyploidy or anti-regeneration properties, such as inflammatory signaling that produces scarring. Encouragingly, in pigs, overexpression of microRNA-199a successfully enabled some cardiac tissue regeneration without compromising polyploidy. Yet, more work is needed, as this microRNA therapy led to uncontrolled cell proliferation (Gabisonia et al., 2019). Polyploidy is not inherently anti-regenerative, as demonstrated by studies in the mouse liver (Liang et al., 2021; Wilkinson and Duncan, 2021) and many other organs (Bailey et al., 2021), and roles for polyploid mitosis in developmental organ remodeling in Drosophila and mosquito (Fox et al., 2010; Schoenfelder et al., 2014; Stormo and Fox, 2016, 2019). Another important distinction for future focus is between productive developmentally programmed cardiac ploidy and maladaptive polyploidy after myocardial injury (Beltrami et al., 1997; Gan et al., 2020). We argue that there may be a sweet spot, set by development, of polyploidy level in cardiac organ chambers that promotes optimal heart function.

Our data argue that specific polyploidy levels in specific locations are important in the larval Drosophila heart. We show here that differential sensitivity to the insulin receptor leads heart chamber cardiomyocytes to undergo speedier endocycles than those of the aorta. Our effects with InR knockdown are phenocopied in fzr knockdown animals, arguing that ploidy itself is a major contributor to the differences in chamber dimension and function. Our results show that heart cardiomyocytes achieve a higher size and ploidy, and this difference is associated with a wider chamber area in the heart versus aorta. It is intriguing that the wall thickness is different between the higher ploidy heart and lower ploidy aorta of control animals, but was unchanged with ploidy manipulations by insulin signaling or Fzr. These results suggest that wall thickness regulation may be regulated independently of ploidy, as previously observed with fzr knockdown in Raf-induced cardiac hypertrophy (Yu et al., 2013, 2015). Notably, we find that hemocytes accelerate when moving from the wider heart to the narrower aorta, in agreement with the Bernoulli principle of fluid dynamics (Badeer, 2001; Bernoulli, 1738). This acceleration is disrupted when heart chamber size is compromised in cardiac ploidy-reduced mutants. We speculate that chamber ploidy asymmetry in fly larvae may function to accelerate and thereby more effectively disperse newly generated clusters of hemocytes throughout the body cavity, to respond to infection more acutely. Future work could reveal whether differential insulin signaling is driven by Hox gene patterning differences between the heart and aorta chambers (Bataille et al., 2015; Lo and Frasch, 2001; Lovato et al., 2002; Monier et al., 2005; Perrin et al., 2004; Ponzielli et al., 2002; Schroeder et al., 2022). Additional future work can explore why the InR^CA larval heart may be refractory to higher ploidy and cardiac hypertrophy, and whether these animals have an uncoupling of ploidy and cell size, as has been observed in some instances in Drosophila skeletal muscle (Windner et al., 2019).

In human donor heart samples, we also find asymmetry in chamber ploidy and insulin signaling between the left ventricle (higher ploidy/insulin signaling) and left atrium (lower ploidy/insulin signaling), suggesting conservation of chamber-specific ploidy regulation by insulin signaling. Deficiency of the insulin receptor in mice has been linked to dilated cardiomyopathy and heart failure (Laustsen et al., 2007; Riehle et al., 2020). Notably, cardiomyocyte-specific insulin receptor ablation in mice reduces cardiomyocyte size and cardiac mass (Belke et al., 2002), and loss of insulin-like growth factor 1 receptor (IGFR1) in cardiomyocytes disrupts sarcomeres and intercalated disks (Riehle et al., 2022). Our RNA-sequencing analysis also shows higher ventricular expression of insulin signaling regulators, such as RPTOR, PRKAA2, FOXO1 and FOXO3 in ventricular cardiomyocytes (Litviňuková et al., 2020). As FOXO can both positively and negatively regulate insulin (Ni et al., 2007) signaling, our data here suggest a potential positive role in ventricular ploidy. The purpose of chamber asymmetry in humans (and mammals in general) may be distinct from that of Drosophila, but likely relates to the greater pressure demand on the left ventricle. Our findings here of a 1.6-fold increase in human ventricular versus atrial nuclear ploidy is consistent with studies in other vertebrates with polyploidy. In quails, ventricular cardiomyocytes are 20% higher in ploidy on average relative to atrial cardiomyocytes (Anatskaya et al., 2001). In mice, where polyploidy is almost exclusively driven by increased nuclear number per myocyte, 77-90% of the ventricular cardiomyocytes are binucleated compared with only 14% in the atrium (Raulf et al., 2015). Connections between insulin and ploidy/cardiomyocyte size may impact our understanding of cardiac disease. Individuals with type 1 diabetes have reduced ventricular mass without affecting the atrium (Hjortkjær et al., 2019), mirroring our finding of cardiac chamber-specific impacts by the insulin receptor in Drosophila larvae.

Beyond cardiac biology, our findings here suggest that ploidy differences along an organ axis may act as a common principle in development to sculpt tissue form and impact tissue function. The unequal growth of body parts is an age-old question in developmental biology (Thompson, 1942). Future work can uncover whether, for example, local ploidy variation accounts for the curvature of polyploid tissues, such as the insect egg, or for specialized performance of distinct skeletal muscle fibers (Cramer et al., 2020; Windner et al., 2019). Notably, whole-genome duplication alters the transcriptome and proteome in non-linear ways (Coate and Doyle, 2010; Maqbool et al., 2010; Yahya et al., 2022; Zhang et al., 2010), highlighting how even a single genome doubling can be transformative in a developmental context. Our work here suggests that future studies should move from a binary (diploid/polyploid) comparisons to take into account the specific level of ploidy. Doing so may accelerate our understanding of the implications of the ubiquitous whole-genome duplications in nature.

Flies were raised on standard fly food provided by Archon Scientific. Fly stocks used in this study are: w¹¹¹⁸, hsFLP;tub<FRT>Gal4,UAS-GFP/CyO (Tub FLP out stock) for clonal analysis, UAS-mCherry-NLS (RRID:BDSC_38424, RRID:BDSC_38425), UAS-GFP-NLS (RRID:BDSC_4775), UAS-GFP-NLS, UAS-Fly-FUCCI (RRID:BDSC_55117, RRID:BDSC_55118) (Zielke et al., 2014), UAS-InR RNAi (VDRC: v992) (Texada et al., 2019), UAS-InR RNAi 2 (RRID:BDSC_35251), UAS-fzr RNAi (VDRC: v25550) (Cohen et al., 2018; Schoenfelder et al., 2014), UAS-InR^CA (RRID:BDSC_8248), UAS-mTor RNAi (RRID:BDSC_34639), UAS-GFP RNAi (gifted by Dr Zhao Zang, Duke University, NC, USA) pnr-Gal4 (RRID:BDSC_3039), Mef2-Gal4 (RRID:BDSC_27390), twi-Gal4 (RRID:BDSC_914) and NP5169-Gal4/cyo (Kyoto: 113612) (Monier et al., 2005), TinC-Gal4 (Lo and Frasch, 2001), HandC-Gal4 (Albrecht et al., 2006), Hand4.2-Gal4 (Han and Olson, 2005) and HandC-GFP (Sellin et al., 2006). HmlΔ-DsRed was a gift from Drs Katja Brükner (Makhijani et al., 2011) and Todd Nystul (UCSF, CA, USA).

Crosses were raised at 25°C, except when heat-shocked at 37°C for 10-30 min to induce FLP-out clones, or when expressing transgenes (29°C). NP> is referred to in three contexts: NP5169-Gal4>UAS-GFP-NLS, NP5169-Gal4>UAS-mCherry-NLS or NP5169-Gal4>UAS-mCherry-NLS X w¹¹¹⁸. Each context is indicated in the corresponding figure legend. Ploidy-based phenotype of the salivary gland and PCR were used to determine the recombination of UAS-InR^CA,UAS-fzr RNAi/CyO flies. For all BrdU feeding experiments, BrdU (100 mg/ml, Sigma-Aldrich) was mixed with standard fly food and fed to larvae of indicated ages for 24 h (BrdU pulse-chase assay) or for 96-120 h (continuous BrdU assay). For the restricted diet, embryos were hatched on agar grape juice plates for 72 h before performing a BrdU pulse-chase assay. To measure larval body weight, WL3 larvae were washed in distilled water and measured individually using a balance.

Embryos were fixed and stained following standard protocols (Rothwell and Sullivan, 2007). Briefly, homozygous UAS-Fly-FUCCI female flies were crossed to male cardiac mesoderm Gal4 lines (Fig. S1A). Embryos were collected on grape plates with yeast paste for 24 h at 25°C. After 24 h, embryos were collected, dechorionated with 50% bleach, and fixed with 4% paraformaldehyde. Fixation buffer was replaced with 1:1 heptane:methanol. Embryos were shaken vigorously, rinsed in methanol, then rinsed in 1× PBS in 0.1% Triton X-100 (PBST). Embryos were then blocked in 1% goat serum in PBST (block solution) followed by primary antibody incubation in block solution, followed by PBST wash, followed by secondary antibody incubation in block solution. Embryos were mounted in VECTASHIELD (Vector Laboratories). Primary antibodies used were mouse anti-PH3 (1:1000; 9706, Cell Signaling Technology) and rabbit anti-RFP (1:1000; PM005, MBL). Secondary antibodies used were goat anti-mouse Alexa Fluor 633 (1:500; A-21052, Invitrogen) and goat anti-rabbit Alexa Fluor 568 (1:500; A-11011, Invitrogen). Images were acquired using a Zeiss AxioImager M.2 microscope (20×/0.5 EC Plan-Neofluar objective).

Larval heart tissues were fixed and stained following standard protocols (Ponzielli et al., 2002). Owing to technical challenges with cleanly recovering the thoracic segments of the dorsal vessel, we used dissection scissors to make an anterior cut at T3. Briefly, tissue was dissected in 1× PBS and fixed in 3.7% paraformaldehyde with 0.3% Triton X-100. Tissues were washed in 1× PBS and blocked in 1× PBS with 0.3% Triton X-100 and 1% normal goat serum. If performing BrdU immunostaining, tissues were incubated with 10 U DNAse (M0303L, New England Biolabs) in 66 mM Tris pH 7.5 and 5 mM MgCl₂ at 37°C for 1 h. Tissues were then rinsed in PBST and blocked in block solution followed by primary antibody incubation in block solution, followed by PBST wash, followed by secondary antibody incubation in block solution. DAPI (5 µg/ul) was added in a final wash step. Laval tissue was mounted in VECTASHIELD (Vector Laboratories). For actin labeling, Alexa Fluor 488 and 555 Phalloidin (1:250; 8878, 8953, Cell Signaling Technology) was added together with the secondary antibody in block solution for 2 h. Primary antibody used was rat anti-BrdU (1:200; ab6326, Abcam). Secondary antibodies used were goat anti-rat Alexa Fluor 488 (1:500; A-11006, Invitrogen) and goat anti-rat Alexa Fluor 568 (1:500; A-11077, Invitrogen). Images were acquired using an upright Zeiss AxioImager M.2 microscope (20×/0.5 EC Plan-Neofluar objective or 63×/1.4 Oil EC Plan-Neofluar objective). For FLP-out clone analysis, z-stack imaging enabled us to determine the full 3D boundaries of cardiomyocytes that wrap around the heart tube (as in Fig. 2F).

Cardiomyocyte ploidy measurements were carried out as previously described (Clay et al., 2023). Briefly, the larval cardiac organ was dissected in 1× PBS and fixed with 4% formaldehyde solution. Dissected hearts were then washed with 1× PBS and then transferred to a siliconized cover slip. A charged slide was gently placed on top of the siliconized cover slip with a drop (∼10 µl) of 1× PBS. Pressure was gently applied to all corners of the cover slip and then further pressure was applied using a vise. The glass slide was submerged into liquid nitrogen and then using a razor blade the siliconized cover slip was quickly removed. The slides were then transferred to a Coplin jar containing cooled 90% ethanol (−20°C). Samples were air-dried and rinsed with 1× PBS before labeling with DAPI (5 µg/ml) for 10 min. Samples were then rinsed twice with 1× PBS and then mounted in VECTASHIELD (Vector Laboratories). For ploidy measurement, testes from adult flies were dissected and processed together with the dissected hearts on the same slide. z-stack images of 1 µm thickness were acquired using an upright Zeiss AxioImager M.2 (63×/1.4 Oil EC Plan-Neofluar objective or 40×/1.3 Oil EC Plan-Neofluar objective). Ploidy analysis was carried out using Fiji. DAPI fluorescence intensity of 10 to 20 individual haploid spermatids was measured and then the median DAPI fluorescence intensity of these measurements was calculated. All the cardiomyocyte nuclei were labeled with mCherry-NLS (NP5169-Gal4>UAS-mCherry-NLS) to distinguish them from other the nonspecific nuclei (such as hemocytes, plasmocytes, pericardial cells). Note that the four cardiomyocytes of the apex (A7 segment) were excluded from our heart chamber ploidy plots (Figs 2E,4C) because of their distinct cell shape.

Cardiac function of Drosophila WL3 cardiac organs was measured using Thorlabs' Telesto-II OCT Systems as previously described (Wolf et al., 2006; Yu et al., 2013, 2015). The resolution of the OCT system is 2 µm. To image live larval hearts, animals were quickly rinsed in distilled water and transferred to an adhesive platform made from transparent tape (3M Heavy Duty Scotch tape) attached to a glass slide. Each larva was placed ventral side down, with the dorsal side exposed for OCT imaging. Videos of the posterior heart and anterior aorta were acquired using the ThorImage OCT 4.2. Each video was acquired at a speed of 100 frames/sec for 5 s duration. Multiple M-Mode transverse and sagittal real-time videos were recorded for the heart and the aorta chamber. As the rhythmic movement of the Drosophila heart is only in one dimension (transverse), chamber length was determined from diastolic sagittal OCT videos. Each segment can be observed during the diastolic and the systolic movement of the heart (Movie 1). A7 and A6 segments were not clearly distinguishable so for diastolic heart chamber length, segments A7-A5 were measured. For diastolic aorta chamber, segments A4-A3 were measured. As the larval cardiac organ lumen was not exactly circular, we directly calculated the area (a) of the lumen using Fiji software. To do so, ESD (a_S) and EDD (a_D) were calculated by tracing the lumen area from the transverse OCT images. Each chamber length (l) was measured from the sagittal M-Mode OCT images directly using ThorImage OCT 4.2 software. For measuring the total chamber wall thickness during diastole, anterior (t_A) and posterior (t_P) wall thicknesses were measured using Fiji software and summed. The number of heart beats for each chamber was recorded from the transverse M-mode OCT videos using orthogonal view of stacks in Fiji. Heart beats per second were calculated by dividing the number of heart beats by the length of the video (i.e. 5 s). Heart rate is defined as number of heart beats per minute. To calculate the heart rate, the number of heart beats per second was multiplied by 60 (Fig. 6D, Fig. S5N).

The stroke volume and cardiac output of each chamber was measured as previously described (Klassen et al., 2017) (Fig. S2L). To calculate the stroke volume of each chamber, the EDD area was subtracted from the ESD area and then multiplied by the chamber length. The cardiac output of each chamber was measured by multiplying the chamber heart rate by the chamber stroke volume.

To visualize the chamber-specific movement of the hemocytes inside the larval cardiac organ, we imaged the fluorescently labeled hemocytes (HmlΔ-DsRed) (Makhijani et al., 2011) in live animals using an Andor Dragonfly 505 unit with a Borealis illumination spinning disk confocal and a Zyla PLUS 4.2 Megapixel sCMOS camera and 20×/0.75 HC PL APO CS2 (Leica 11506517) dry objective, WD 0.62 mm. To restrict the movement of larvae, a sticky glass slide platform was created using transparent strong adhesive tapes. Each larva was placed on the sticky plate with the ventral side attached to the sticky platform and the dorsal side was then exposed for imaging using the Andor platform. Furthermore, to identify the cardiac organ while imaging, it was fluorescently labeled (NP5169>UAS-GFP-NLS). Multiple videos were recorded for each animal to determine the chamber-specific movement of the hemocytes.

Flash-frozen LV and LA samples from explanted hearts of five subjects (Male 1: age 42; Male 2: age 41; Female 1: age 42; Female 2: age 44; and Female 3: age 44) were procured from the DHHR under the review of Duke University Health System (DUHS) institutional review board (Pro00005621). Consent was obtained for all experiments with human tissue. Tissue was mounted in O.C.T. compound (Scigen, #4586) and sectioned at 10 µm by the Duke Substrate Service Core & Research Support (SSCRS). Each section was transferred to a positively charged glass slide and stored at −80°C before immunostaining.

Frozen tissue sections were thawed at room temperature for 10 min and then washed twice with 1× PBS for 5 min each wash to dissolve the O.C.T. medium. Tissue sections were then fixed with 4% paraformaldehyde solution for 15 min at room temperature (RT). Samples were then washed twice with 1× PBS and incubated with Alexa Fluor 633 conjugate of wheat germ agglutinin (WGA) (1:250; W21404, Invitrogen) in 1× PBS overnight at 4°C. Note that WGA staining was performed before permeabilization using Triton X-100. To wash off the WGA solution, samples were then washed thrice with PBST for 5 min each wash and then incubated in blocking solution (5% goat serum in PBST) for 40 min at RT. Samples were then incubated with primary antibody in blocking solution overnight at 4°C. Tissues samples were then washed thrice with PBST for 10 mins each and then incubated with secondary antibody, Hoechst (1:1000, 62249, Thermo Scientific Pierce) and Alexa Fluor 488 Phalloidin (1:250; 8878, Cell Signaling Technology) in block solution at RT for 2 h. Tissues were then washed thrice with PBST for 10 mins each wash and mounted in VECTASHIELD (Vector Laboratories). For nuclear volume analysis, tissue sections were labeled first with WGA in 1× PBS, overnight and then incubated with Phalloidin and Hoechst in 1× PBS for 2 h. The primary antibody used for this study was rabbit anti-insulin receptor (1:200; ab137747, Abcam). The secondary antibody used was goat anti-rabbit Alexa Fluor 546 (1:500; A-11006, Invitrogen).

We used Fiji software to analyze the fluorescence intensity of INSR. Using the ‘Plot profile’ program in the ‘Analysis section’, we measured the intensity of INSR at the intercalated regions of a single z-section that had the highest enrichment of signal. Intensity profiles of 30 cells from multiple images were calculated and data were plotted as previously described (Sawyer et al., 2017).

For nuclear volume analysis, z-stack images of each tissue section were imaged using an Andor Dragonfly 505 unit with a Borealis illumination spinning disk confocal, with a z-step size of 0.5 µm. Samples were imaged using an Andor Zyla PLUS 4.2 Megapixel sCMOS camera together with a 63×/1.47 TIRF HC PL APO CORR (Leica 11506319), oil objective, WD 0.10 mm. For analysis of the nuclear volume of cardiomyocytes, we used the Imaris Version 8.2 module, called ImarisCell. The protocol for analyzing the cardiomyocyte nuclear volume was adapted from Bensley et al. (2016). Briefly, using ImarisCell, the nuclear volume of all nuclei in the z-stack were recorded. With the help of both phalloidin and WGA labeling, cardiomyocyte nuclei were selected manually. We also observed that most human cardiomyocytes showed vesicles nearby the nucleus, which helped to identify the cardiomyocyte nuclei easily. The WGA staining pattern of the vesicles were prominent when Triton X-100 was not used for permeabilization. Only mononucleated cells were imaged, and binucleated cells were excluded. About 50-100 complete nuclei were processed for each group from multiple images. Simultaneously, using Fiji software the relative fluorescence intensity of Hoechst staining was detected for the same cell to determine the ploidy. Nuclear volume of fluorescently labeled (NP5169>UAS-mCherry-NLS) Drosophila cardiomyocytes stained with DAPI and Phalloidin was also determined using Imaris.

‘Heart Global’ data sets [available through the Human Cell Atlas (HCA) Data Coordination Platform (DCP) under accession number ERP123138] containing single-cell and single-nuclei RNA-sequencing data were downloaded from (https://www.heartcellatlas.org) (Litviňuková et al., 2020). Differential expression between ventricular (LV+RV) and atrial cardiomyocytes (LA+RA) was analyzed with Seurat R (Hao et al., 2021). Differentially expressed genes between ventricular and atrial cardiomyocytes are listed in Table S1, ‘DE_V_vs_A’. Upregulated genes in ventricular cardiomyocytes are listed in Table S1, ‘V_UP’. Pathway analysis was performed using DAVID (Huang da et al., 2009; Sherman et al., 2022). A list of KEGG upregulated pathways in ventricular cardiomyocytes can be found in Table S1, ‘KEGG _V_UP’. Protein–protein interaction of the upregulated ventricular genes for biological processes with P<0.05 were performed using STRING (Szklarczyk et al., 2023; von Mering et al., 2003). For STRING analysis, active interaction sources were based on experiments, text mining and databases; K-means clustering was selected to be three clusters.

Statistical methods of analysis, number of biological replicates, values of n, and P-values are detailed in the figure legends. Statistical analysis was performed using GraphPad Prism 9.3.1. Statistical notations used in figures: ns, P>0.05 (not significant); *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

We thank Bloomington Drosophila Stock Center, Vienna Drosophila Resource Center, Dr Zhao Zhang (Duke University) and Dr Sarah Goetz (Duke University) for providing reagents. Fox lab members Ruth Montague, Dr Jessica Sawyer, Dr Matt Andrusiak and Rebeccah Stewart provided technical assistance. Dr Lisa Cameron, Dr Yashen Gao and the Duke Light Microscopy Core provided imaging assistance. The Duke Human Heart Repository (DHHR) provided human heart donor tissue. We thank Paula Newell, Cameron Leonard and the Substrate Service Core & Research Support (SSCRS) from Duke School of Medicine (SOM) OneDukeBio integrated research support, Duke University, for providing assistance with sectioning of human tissue. Dr Sarah Goetz, Dr Zhao Zhang and Fox lab members provided manuscript comments.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201896.reviewer-comments.pdf.