Epithelial mechanics are maintained by inhibiting cell fusion with age in Drosophila

Aging is often accompanied by permanent changes to the cellular architecture of an organ or tissue, as aging cells turnover or undergo senescence. The lack of resident stem cell populations limits the regenerative capacity of many animal tissues. As a result, cell growth via polyploidy might be the only mechanism to compensate for cell loss over the lifetime of an animal (Gjelsvik et al., 2019; Lazzeri et al., 2019). Polyploidy, therefore, describes a cell that has more than the diploid copy of its chromosomes, hence referred to as 3C or greater chromatin content. Somatic polyploid cells are generated by cell–cell fusion or an incomplete cell cycle, known as endoreplication. As a result, polyploidy occurs in a wide variety of cell types and organisms across the kingdoms of life, allowing cells to grow orders of magnitude larger, as cell size scales with DNA content (Frawley and Orr-Weaver, 2015).

Studies in vertebrate and invertebrate models have demonstrated that polyploidy often arises under conditions of stress, including injury, aging and disease. Polyploid cells have thus been found to result in both beneficial and detrimental effects on tissue physiology. In the short-term, polyploidy has been found to be beneficial as it speeds wound closure, restores tissue mass when cell division is limited, protects against genotoxic stress and promotes resilience to environmental insults (Cao et al., 2017; Cohen et al., 2018; Grendler et al., 2019; Hassel et al., 2014; Losick et al., 2013; Storchova, 2014). However, polyploidy is also associated with detrimental consequences in the long-term, including increased fibrosis, altered metabolism and cell senescence (De Chiara et al., 2022; Dewhurst et al., 2020). With aging, though, the role of polyploidy is even more poorly understood, as it has only recently been observed that polyploid cells arise in neurons in the brain and hepatocytes in the liver by endoreplication (Lopez-Sanchez et al., 2017; Nandakumar et al., 2020; Dewhurst et al., 2020). Multinucleated cells have also been observed in the human and mouse eye (cornea endothelium and retinal pigment epithelium) as well as the fruit fly epithelium with age (Losick et al., 2016; Chen et al., 2016; Saksens et al., 2016; Zhang et al., 2019; Scherfer et al., 2013), but the mechanism of polyploid generation and its impact on aging (the healthspan of an organism) has remained unexplored.

Healthspan is the period of life spent free from chronic disease and the disabilities of aging, including human mobility (DiLoreto and Murphy, 2015; Rowe and Kahn, 1987). Consequently, the velocity of organismal movement is used as an indicator of healthspan in model organisms (Hahm et al., 2015). We recently discovered that the adult Drosophila abdominal epithelium is under tension, which is required for efficient bending in male fruit flies mimicking their copulation behavior (Losick and Duhaime, 2021). Multinucleated polyploid cells have been found to arise during both wound healing and aging in Drosophila (Galko and Krasnow, 2004; Losick et al., 2013; Scherfer et al., 2013). In the adult fruit fly, the epithelial cells both fuse and endocycle to close a wound, with the endocycle being critical to upregulate myosin, a motor protein necessary to restore epithelial tension post wound repair. Wound-healing studies in the larval stage of Drosophila development have revealed that cell fusion is dependent on autophagy-dependent remodeling of epithelial junctions (Kakanj et al., 2022). Similarly, autophagy was first observed to maintain epithelial junctions with age in adult Drosophila; however, the mechanism of epithelial multinucleation with age as well as its impact on tissue and organismal physiology remains unknown (Scherfer et al., 2013).

Here, we utilize the genetic and biophysical tractability of Drosophila to determine the mechanism of age-induced polyploidy and its impact on the biomechanical fitness of the fly. In summary, we have discovered that the adult Drosophila epithelium becomes polyploid with age by cell fusion, not failed cytokinesis. In addition, the maintenance of the ‘youthful’ diploid, mononucleated epithelium is dependent on conserved mechanosensory genes including α-Catenin (αCat), E-cadherin [also known as Shotgun (Shg) in flies], Rac GTPase (Rac; also known as Rac1), and Rho kinase (Rok). Interestingly, αCat is also a regulator of multinucleation in mouse retina pigment epithelium and associated with the human macular disease ‘butterfly shaped-pigment dystrophy’. Thus, genetic regulators of epithelial multinucleation with age appear to be conserved. Using the Drosophila model, we further show that the inhibition of epithelial multinucleation with age does not improve lifespan but does restore the biomechanical fitness of the fly, as epithelial tension and abdominal movements are maintained when cell fusion is inhibited by overexpression of a dominant-negative allele of Rac GTPase (Rac^DN). Thus, our Drosophila model of age-induced polyploidy provides a useful genetic and biophysical system to elucidate the links between epithelial ploidy and biomechanical fitness over the lifespan of an animal.

Multinucleation with age in the ventral epithelium of Drosophila was first reported to be a sign of tissue deterioration as there was found to be significant loss of labeling with the septate junction protein FasIII on cell junctions (Scherfer et al., 2013). Therefore, to start, we aimed to more rigorously quantify the extent of cell junction remodeling with age. To do so, we quantified epithelial cell size based on cell borders labeled with FasIII and the number of nuclei per cell stained with the epithelial specific transcription factor, Grainyhead (Grh). This analysis was conducted over a time course with at least ten flies from each time point: 5 days (d) to 50 d post eclosion (Fig. 1A–G). We used mated epi-Gal4 female flies for this study, which have a standard mean lifespan of 57±4 d (mean±s.e.m.) (Koliada et al., 2020). Thus, epithelial cells were defined as a syncytium (either binucleated or multinucleated) if there was either no FasIII-labeled borders between nuclei, a significant reduction in FasIII labeling (less than twice the background) or a gap of 3 µm or more in the cell border (Fig. 1A,H).

In young flies (5–10 days old), we found the epithelium was composed on average of 90% mononucleated cells with 5.6% binucleated and 2.28% multinucleated cells (Fig. 1B,C,I). At 20 days old, flies showed a noticeable change in epithelial organization with large regions containing multiple epithelial nuclei with significant reduction in FasIII labeling on the epithelial cell junctions (Fig. 1D). There was also a significant increase in polyploidy as we observed a doubling of both binucleated and multinucleated cells. Polyploidy peaked in 40-day-old flies with 18% binucleated and 15% multinucleated epithelial cells (Fig. 1F,I). This was in part because the multinucleated cells enlarge beyond the maximum image size and thus could not be accurately quantified past 40 days of age. The multinucleated cells in young flies contained 3–5 nuclei, whereas older flies had double the number of small multinucleated cells as well as frequent large multinucleated cells containing 6–31+ nuclei (Fig. 1E–G,J).

The above findings are in stark contrast to original report of epithelial multinucleation with age in Drosophila, which reported that it occurred as early as 3 days post eclosion (Scherfer et al., 2013). However, here we did not observe a significant increase in syncytia until flies were at least 20 days old, which is consistent with when other Drosophila tissues exhibit signs of aging, including stem cell turnover in germline (Kai and Spradling, 2004). To assess whether epithelial multinucleation with age was strain dependent, we assessed the extent of polyploidy in the same Drosophila melanogaster w¹¹¹⁸ strain used in the prior study. In young female w¹¹¹⁸ flies, we found that the epithelium resembled that of the epi-Gal4 strain and was composed of 89% mononucleated cells (Fig. S1A,E). In the old w¹¹¹⁸ strain, there was a significant increase in polyploidy and their epithelium was made up of 15% binucleated and 15% multinucleated cells (Fig. S1B,E). Another wild-type Drosophila strain, Canton S, had elevated numbers of binucleated epithelial cells in young flies, but not multinucleated epithelial cells (Fig. S1C,E). However, at 40 days old, flies from the Canton S strain also showed a significant increase in its multinucleated cell population (Fig. S1D,E).

The multinucleated cells, although few in number, were noted to make up the majority of abdominal epithelial tissue area with age (Fig. 1F,G). We calculated the percentage of the epithelial area composed of syncytia, and found the largest increase at 40 days and 50 days, when multinucleated cells made up 48% and 60% of the tissue area, respectively (Fig. 1K), whereas only 11% of the epithelial tissue area was made of multinucleated cells in 10-day-old flies. Similar to what was seen for the epi-Gal4 fly strain, both the 40-day-old w¹¹¹⁸ and Canton S strains had at least 40% of the epithelium made up of multinucleated cells, confirming that multinucleation is an aging-dependent, strain-independent phenomenon in Drosophila (Fig. S1F).

As animals age, cells can be lost via apoptosis and not replaced or repaired, thus leading to tissue deterioration. In Drosophila, we expected syncytia to be generated with age, similar to what occurs in a wound healing response, to compensate for cell loss (Losick et al., 2013, 2016). To test this hypothesis, we measured changes in the cell and nuclear number in the epithelium from the aging time course study (Figs 1A–G and 2A–D). As expected, the total number of epithelial cells declined with age, with a significant reduction in flies that were 20 days old, when there was a corresponding increase in syncytia formed (Figs 2C and 1I). In total, 138±5 cells (mean±s.e.m) made up a 22,500 µm² epithelial region at 5 days, which was reduced to 50±5 cells in 50-day-old flies. With this decrease in cell number, we expected a corresponding decrease in total number of epithelial nuclei, but strikingly found that number of epithelial nuclei remained constant at ∼188 nuclei (Fig. 2D). The nuclear morphology in aging cells was indistinguishable from those in young flies, suggesting there was no apoptosis (Fig. 2A,B, see DAPI inset). In addition, the epithelial nuclei remained diploid in 40-day-old flies indicating the nuclear DNA content did not change with age (Fig. 2E).

To further assess the role of cell death, we overexpressed cell survival genes to determine whether inhibition of apoptosis could reduce generation of polyploid cells with age. Two caspase pathway inhibitors, the baculoviral anti-apoptotic gene p35 (p35^OE), and microRNA bantam (ban^OE) were overexpressed with the epi-Gal4/UAS system in Drosophila (Kester and Nambu, 2011; Thompson and Cohen, 2006). Overexpression of these caspase inhibitors did not inhibit age-induced polyploidy (Fig. 2F). We still observed a significant increase in the formation of multinucleated epithelial cells by 40 days with no significant difference when compared to the control (epi-Gal4/w¹¹¹⁸) flies (Fig. 2G). In conclusion, cell death via apoptosis does not appear to be necessary for age-induced polyploidy in Drosophila.

Multinucleated cells can arise by endomitosis or cell fusion (Bailey et al., 2021; Peterson and Fox, 2021). Endomitosis occurs when cells enter the cell cycle, but fail to complete M phase. In particular, failed cytokinesis would generate a binucleated cell or multiple endomitotic cell cycles could generate a multinucleated cell. Alternatively, two or more cells can fuse together to generate bi- and multi-nucleated cells, similar to what occurs in the development of myotubes that make up the Drosophila musculature (Lee and Chen, 2019). To elucidate the mechanism of polyploid generation with age, we knocked down the mitotic regulators cdc2 (Cdk1) and stg (cdc25) with the epi-Gal4/UAS-RNAi system to continuously reduce the mitotic regulators over the aging time course. Thus, if there were a burst of mitotic activity over the 40 days it could be reduced. However, we found that binucleated and multinucleated cells were still prominent in older flies despite knockdown of either cdc2 or stg (Fig. S2A). Therefore, the inhibition of endomitosis was not sufficient to block the generation of multinucleated cells with age, suggesting that syncytia do not arise by endomitosis (Fig. S2B,C). Next, we tested whether polyploid epithelial cells might arise by cell fusion.

The dBrainbow lineage system was recently used to detect multinucleated rectal papillae cells in Drosophila (Peterson et al., 2020). The Cre recombinase-based fluorescent labeling technique mosaically labels the Drosophila tissue of interest through Gal4/UAS-dBrainbow expression. The constitutive Cre-based recombination of the dBrainbow cassette then results in expression of EGFP–HSV, EBFP2–HA or mKO2–Myc (Hampel et al., 2011). In Drosophila, multicolor labeling has been shown to only occur in the rectal papillae cells that exhibited cytoplasmic sharing during development (Peterson et al., 2020). No multicolor cells were observed in fly brain, intestine nor ovary follicle cells. Taking a similar approach, we expressed UAS-dBrainbow cassette with the epi-Gal4 strain, but omitted staining for mKO2–Myc fluorescence, so we could co-stain for FasIII and validate the syncytia boundaries. Therefore, a bicolored cell co-expressing both EGFP–HSV and EBFP2–HA would be indicative of a cell fusion event (Fig. 3A). In young 5-day-old fly tissue expressing both markers, we observed distinct patches of EGFP–HSV and EBFP2–HA with only an occasional bicolored region (Fig. 3B,D). By contrast, in 40-day-old flies, there was extensive overlap of EGFP–HSV and EBFP2–HA fluoros indicative of cell fusion. Young flies had on average 1.5±1.0 bicolored regions, whereas old flies had 16.9±3.0 bicolored regions per 10,000 µm² epithelial area (mean±s.e.m.; Fig. 3C,D).

We then examined whether bicolored regions matched the FasIII boundaries containing more than one nucleus. At 5 days, only ten bicolored cells were identified in three out of the ten flies assayed, and all but one cell had two or more nuclei as expected (Fig. 3B,D–F). At 40 days, 60% of cells (50 out of 84) had more than two nuclei, and cell size scaled proportionally with the number of nuclei per cell (Fig. 3E, R²=0.85). Surprisingly, we found that 40% of bicolored epithelial cells appeared to be mononucleated (Fig. 3F). We examined the location of the apparent mononucleated bicolored cells and found that they were always in clusters or adjacent to a bicolored region containing a syncytium (Fig. 3C,G). This suggests that cytoplasmic sharing proceeds cell junction breakdown, which has been observed in development of the recta pupal cells as well as during wound healing in the Drosophila epithelium (Peterson et al., 2020; White et al., 2023 preprint). Therefore, in the fly ventral epithelium, syncytia arise with age through cell fusion, which might proceed cell border breakdown.

Cytoskeletal remodeling is known to play a key role in cell fusion, which can be inhibited by expression of a dominant negative Rac GTPase, Rac^DN (Fernandes et al., 2005; Losick et al., 2013). Indeed, the continuous overexpression of Rac^DN with epi-Gal4/UAS system was sufficient to strongly reduce generation of syncytia cells, with significant reduction of binucleated and multinucleated cells in 40-day-old flies (Fig. 3H,I). Strikingly, the 40-day-old epithelium expressing Rac^DN resembled the young 7-day-old control epithelium with preservation of mononucleated epithelial cells. Only 5% of the epithelium was composed of multinucleated cells in the 40-day-old Rac^DN flies compared to the 50% observed in control flies (Fig. 3H,J). We also measured epithelial nuclear ploidy to see whether there was a compensatory increase in nuclear DNA ploidy, but found that mononucleated cells remained diploid (2C) in Rac^DN epithelial cells (Fig. 3K). Finally, we evaluated whether inhibition of epithelial cell fusion could extend the lifespan of the flies, but the continuous expression of Rac^DN instead reduced mean lifespan by 7 days (Fig. S3A; Table S9).

Loss of mobility is a conserved characteristic of aging in both humans and model organisms including C. elegans, where the velocity of organismal movement has been used as an indicator of healthspan (Hahm et al., 2015). The abdomen of the fruit fly also needs to be able to stretch and bend in response to changing physiological and environmental conditions. We recently discovered that epithelial tension contributes to the ability of the male fly to efficiently bend its abdomen (Losick and Duhaime, 2021). Therefore, we assayed whether syncytia also arose with age in the ventral epithelium of the male fly. Indeed, we found that syncytia arose with age, similar to what occurred in females, and were dependent on Rac GTPase in male flies as well (Fig. S3B–D; Table S9). Thus, we next addressed whether the biomechanical fitness of the fly abdomen in the male was dependent on age-induced polyploidy.

For this assay, abdominal movements of immobilized male flies were live imaged, and the number of bends per minute quantified. A bend was defined as a male curling its abdomen from fully extended to a 45° angle or greater, as measured from the midline axis of its body (Fig. 4A). As expected, aging caused a significant decline is abdominal bending efficiency (Fig. 4B). Young 7-day-old flies had a mean bending efficiency of 16.7 bends per minute, whereas this declined in 40-day-old flies to 8.95 bends per minute. The mean bending angle remained constant at 81° in both young and old flies (Fig. 4C). Next, we asked whether inhibition of cell fusion via expression of Rac^DN would help to maintain abdomen movement with age. Contrary to what we found in old control animals, no significant difference was observed between the bending efficiency in young and old flies when epithelial cell fusion was inhibited through constitutive expression of Rac^DN (Fig. 4B). The mean bending angle was also significantly enhanced by Rac^DN expression, providing animals with more torque (Fig. 4C). Overall, the change in bending efficiency suggests that epithelial multinucleation might cause a reduction in epithelial tension with age.

Epithelial tension can be inferred by measuring the relative recoil velocity following a laser ablation cut of the epithelial non-muscle myosin II protein network via expression of the Drosophila myosin regulatory light chain (called Sqh) fused to GFP (Sqh–GFP) (Shivakumar and Lenne, 2016). As we reported previously, the relative recoil velocity was determined by ablating a 1 µm line through the Sqh–GFP-enriched region in the abdominal epithelium (Losick and Duhaime, 2021). This led to an initial mean recoil velocity of 14.2 µm/s in epithelium from young 7-day-old animals, similar to our previous results demonstrating that adult epithelium is under tension (Fig. 4D,E). Remarkably, epithelial tension was significantly reduced by ∼8-fold to 1.8 µm/s in 40-day-old flies. To determine whether age-associated syncytia formation was the cause of the reduced epithelial mechanics, we ablated both epithelium of young and old flies expressing Sqh–GFP and Rac^DN to block cell fusion. Ectopic expression of Rac^DN did not alter epithelial tension in young 7-day-old flies, as the mean initial recoil velocity was not significantly different at 11.9 µm/s. However, ablation of Sqh–GFP regions in 40-day-old Rac^DN flies that retained a mononucleated epithelium maintained the young mean initial recoil velocity of 14.0 µm/s. Taken together, this analysis suggests that inhibition of cell fusion is sufficient to restore epithelial mechanics and maintain Drosophila abdominal movement over the lifespan of the fly.

Mechanotransduction is critical for tissue homeostasis and function, as abnormal tissue mechanical properties can lead to defects in tissue integrity and contractility, which has been recently shown to be an indicator of tissue healthspan (Essmann et al., 2020; Heer and Martin, 2017). Given that we found that age-induced polyploidy disrupts epithelial mechanics, we first defined how the expression and activity of non-muscle myosin II changed with age using Sqh–GFP. Myosins have been found to form supracellular networks that anchor through adherens junctions to facilitate many tissue morphogenetic events, including ventral furrow formation in Drosophila (Yevick et al., 2019). Therefore, we assessed whether generation of multinucleated cells with age alters the myosin supracellular network in coordination with the loss of cellular junctions.

As we observed previously, Sqh–GFP is polarized towards the nuclear side of epithelial cells and forms a concentrated band in both parallel and perpendicular cell rows within the ventral abdomen of 7-day-old fruit flies (Fig. S4A; Fig. 5A) (Losick and Duhaime, 2021). This myosin organization is disrupted in 40-day-old animals, where Sqh–GFP becomes more diffuse and dispersed across the epithelium (Fig. S4A). Despite the disrupted organization, the total myosin protein expression did not change with age and surprisingly did not correlate with loss of epithelial junctional integrity (Fig. S4B). Some of the syncytia cells had dispersed Sqh–GFP, whereas in others, bands were still intact. Therefore, we focused on quantifying the dispersion based on the localization in parallel and perpendicular cellular bands versus interior cellular regions (Fig. S4A). Using this analysis, we found a significant reduction in the ratio of Sqh–GFP with age for both the parallel and perpendicular bands versus the interior region, with no significant difference between them (Fig. S4C). We focused our analysis on the alteration in parallel Sqh–GFP bands to the interior regions, which were more prominent, for further analysis.

Inhibiting cell fusion associated with age, via expression of Rac^DN, did not significantly alter Sqh–GFP total expression nor localization (Fig. 5A–C). Sqh–GFP localization still became dispersed with age even in animals where epithelial junctions were maintained. Given that myosin localization could not explain the change in epithelial tension, we next addressed whether myosin activity is altered with age and is dependent on cell fusion. Phosphorylation of the regulatory light chain of myosin plays an important role in epithelial contractions and, as we reported previously, is required for tension in adult fly epithelium (Losick and Duhaime, 2021). Myosin II phosphorylation can be detected by immunostaining for phosphorylated (P-)Myo II, which we have previously shown is dependent on Rho kinase (Rok) (Losick and Duhaime, 2021). Interestingly, epithelial specific Rok^RNAi also results in a reduction in the Sqh–GFP banding ratio as well as enhancement of epithelial multinucleation (Fig. S4D–I), suggesting that myosin organization is dependent on myosin activity. Indeed, we found that P-Myo intensity significantly declines with age in the adult epithelium (Fig. 5D,E). P-Myo staining intensity was reduced 2-fold in 40-day-old compared to 7-day-old animals, and inhibition of cell fusion through the constitutive expression of Rac^DN was sufficient to maintain P-Myo staining in the epithelium of 40-day-old flies. Therefore, age-induced polyploidy leads to a decline in epithelial mechanics through loss of myosin activity, which can be restored via the Rac-dependent inhibition of cell fusion.

This discovered link between mechanotranduction and epithelial multinucleation led us to investigate age-associated diseases known to be linked with mechanosensory proteins and correlate with a corresponding increase in generation of syncytia. The butterfly-shaped pigmentary pattern dystrophy, a rare human disease, is characterized by deposits at the level of the RPE in the macular region that resemble the wings of a butterfly. Heterozygous missense mutations in α-catenin 1 (α-Cat1, encoded by CTNNA1), a mechanosensory protein, are associated with this disease (Saksens et al., 2016; Tanner et al., 2021). Many features of the disease are recapitulated in the Ctnna1^Tvrm5 mouse model, including multinucleation of RPE cells. In 4-month-old C57BL/6J mice, the RPE is made up of centrally located, predominantly binucleated cells with only 4% of multinucleated RPE cells (Fig. 6A,C). This observation is consistent with a previous study, which documented predominantly binucleated RPE cells in young mice (Chen et al., 2016). Whereas in 4-month-old Ctnna1^Tvrm5 mutant mice, we determined that the RPE is made of 21% multinucleated cells, which covers 51% of the central epithelium (Fig. 6B–D). To determine whether this is analogous to our Drosophila model, we knocked down αCat (αcat^RNAi) in fruit fly epithelium to determine whether multinucleation could be induced. αCat was expressed at a low, but detectable level in adult fly epithelium and could be efficiently knocked down within 7 days of RNAi expression (Fig. S5A–D). Indeed, we found a ∼4-fold increase in multinucleated epithelial cells with genetic knockdown of αcat in fly epithelium suggesting that αcat is required to prevent epithelial multinucleation in both flies and mice (Fig. 6E,F,I).

The mechanism that leads to generation of multinucleated cells in the RPE remains unknown; however, using our fly model, we can easily assay whether αcat^RNAi syncytia arises by cell fusion. Indeed, we found that Rac^DN expression rescued the multinucleation caused by knockdown of αcat as well as reducing the tissue area composed of multinucleated cells at 7 days in fly epithelium (Fig. 6E–J). αCat is part of the adherens junction complex, which includes E-cadherin, β-catenin and p120-catenin (Leckband and de Rooij, 2014). To verify that junctional remodeling with age was not unique to septate junctions, we compared co-stained Drosophila expressing the E-cadherin marker Shg–GFP with anti-FasIII immunostaining (Fig. S5E,F). Shg–GFP and FasIII colocalized nearly perfectly on the epithelial junctions in both young and old flies. Therefore, age-induced polyploidy leads to the remodeling of both epithelial septate and adherens junctions.

Next, we assayed whether epithelial-specific knockdown of shg would also result in generation of epithelial syncytia, similar to that observed in αcat^RNAi knockdown flies. Indeed, we observed an ∼10-fold increase in multinucleated epithelial cells at 7 days when shg was significantly knocked down (Fig. S5G–K). Taken together, these findings suggest that reduction in the adherens junction protein expression might serve as a signal to stimulate cell fusion with age. Interestingly, αCat, which primarily localizes to abdominal epithelial nuclei, shows a 2-fold age-related reduction in expression in 40-day-old flies compared to in 7-day-old animals (Fig. 6K,L). In conclusion, we have characterized a new Drosophila model to study the mechanisms of age-induced polyploidy revealing that the biomechanics of the young epithelium can be maintained through inhibition of cell fusion.

Polyploidy has been found to increase with normal aging in a variety of animal tissues, including the brain, eye, heart and liver (Chen et al., 2016; Ikebe et al., 1986; Matsumoto et al., 2021; Nandakumar et al., 2020; Senyo et al., 2013). Here, we find that polyploid cells also arise in an age-dependent manner in fruit fly epithelium (Fig. 6M). An advantage of using the fruit fly as a model is that age-induced polyploidy arises by 20 days, instead of the months to years in mice and humans. Importantly, the fly epithelial cells become multinucleated with age and resemble the cells formed by the wound-induced polyploidization response we previously described (Losick et al., 2013). However, the age-induced polyploid cells arise solely by cell fusion and not via the endocycle as well.

In normal aging and age-associated diseases, multinucleated cells are a common feature of human and murine ocular tissues made up of post-mitotic cells. The cornea endothelium is terminally differentiated and becomes multinucleated with age and during the age-associated disease Fuchs endothelial corneal dystrophy (Ikebe et al., 1986; Losick et al., 2016). Likewise, the RPE is made up of terminally differentiated cells that become multinucleated with normal aging and in macular diseases (Chen et al., 2016; Saksens et al., 2016; Zhang et al., 2019). In C57BL/6J mice, aging causes a decrease in RPE cell number and increase in cell size, characterized by multinucleated cells as early as 6 months of age (Chen et al., 2016). However, because the total number of RPE nuclei did not change with age, this led the authors to the conclude that the multinucleated RPE cells arose by failed cytokinesis. We also observed that total epithelial nuclear number did not change with age in the fly epithelium; however, our studies suggest that failed cytokinesis is not the mechanism underlying cellular multinucleation. To date no study in the eye has directly assayed for cell fusion or endomitosis (failed cytokinesis) in vivo, so the mechanism of epithelial multinucleation remains unknown.

In addition, we show that multinucleation in Drosophila epithelium is dependent on cell fusion using cell–cell junctional protein markers as well as mosaic dBrainbow labeling. The loss of protein junctions with age is consistent, as both septate junctions (anti-FasIII) and E-cadherin (Shg–GFP) markers, which colocalize on epithelial junctions, are lost with age. However, the dBrainbow mosaic labeling revealed that cell fusion events might be more widespread within this tissue, as bicolored mononucleated cells that retain 50% or more FasIII junctional staining were observed. This suggests that either the septate junctions, although detectable, might be nonfunctional or that other mechanisms promote syncytia formation with age. For example, the fly rectal papillae cells were discovered to become multinucleated via remodeling of the gap junctions of the cell, whereas their septate junctions remained intact (Peterson et al., 2020). This suggests that alternative pathways might be playing a role in the observed effects in abdominal epithelium, and further investigation might be warranted.

Remodeling of the actin cytoskeleton is critical for many cell–cell fusion events from fertilization to muscle development (Brukman et al., 2019). In Drosophila muscle, actin polymerization in the attacking myoblast cell propels membrane protrusions into the founder muscle cell facilitating the close contact required to initiate a fusogenic synapse (Kim and Chen, 2019). Cell fusion also requires mechanical forces, including pushing and resisting cellular forces. In particular, the mechanotransduction proteins, Myosin II and Spectrins, accumulate at fusogenic synapse to mediate myoblast fusion (Duan et al., 2018; Kim et al., 2015). Here, we find that conserved regulators of mechanotransdution are also required to regulate cell fusion with age, albeit through a reduction, rather than enhancement of tension. In epithelium, αCat is required to link cadherins to the actin cytoskeleton. In mice, the Ctnna1^Tvrm5 mutant is associated with a missense Leu436Pro substitution that maps to the M-region, a force-sensing module in αCat, which is sensitive to tension in both mice and flies (Leckband and de Rooij, 2014; Saksens et al., 2016; Sheppard et al., 2023). Interestingly, we found that αCat declines with age and its knockdown is sufficient to generate multinucleated cells in young flies. Likewise, inhibition of myosin activity through knockdown of Rok, as well as a reduction in P-Myo, leads to the enhancement of epithelial multinucleation. This suggests a link between a reduction in epithelial tension and cell fusion (Fig. 6M). Overexpression of Rac^DN is sufficient to inhibit fusion with age, revealing that P-Myo and epithelial tension can be restored when cell fusion is inhibited. Still, it remains to be determined how Rac^DN interferes with epithelial cell fusion and whether it is direct via action on an actin-based fusion pore or via an indirect mechanism. In Drosophila, recent studies have indicated a role for autophagy in epithelial cell fusion during homeostasis and aging, so it conceivable that autophagosome trafficking might be dependent on actin cytoskeleton to remodel cellular junctions (Scherfer et al., 2013; Kakanj et al., 2022).

Taken together, we have identified a detrimental consequence of age-induced polyploidy as epithelial multinucleation leads to decline in epithelial tension, contributing to reduction in organismal movement with age. This is likened to recent reports in C. elegans that have shown how biomechanics are altered with age due to changes in cuticle stiffness (Rahimi et al., 2022). In Drosophila, we find that the change in epithelial mechanics with age is a result of change in epithelial ploidy given that inhibiting cell fusion serves as a means maintain the epithelial tension and animal movement with age. Hence, the biomechanical healthspan of the fruit fly can be preserved through inhibition of age-induced polyploidy.

Drosophila melanogaster strains used in this study were reared on standard corn syrup-soy food (Archon Scientific) at 25°C, 60% humidity and in a 12-h-light–12-h-dark cycle. The following Drosophila strains were obtained from Bloomington (b) or created in our lab using the strains indicated: GMR51F10-Gal4, referred to as epi-Gal4 (b38793), w¹¹¹⁸ (b3605), CantonS (b64349), UAS-cdc2^RNAi (b28368), UAS-stg^RNAi (b29556), UAS-Rac^DN (b6292), UAS-αcat^RNAi (b38987), UAS-rok^RNAi (b28797), UAS-dBrainbow (Chr. 3) (b34513), UAS-dBrainbow (Chr. 2) (b34514), Crey/FM7;; 51F10-Gal4/TM3 (derived from epi-Gal4 (b38793), Crey (b851), and Sn28/FM7;; TM2/TM6b balancer (obtained from Dr Tina Tootle, University of Iowa, Iowa City, USA), sqh-GFP (b57144), shg-GFP (b60584), UAS-αcat^OE (b58787), UAS-p35^OE (b5073), and UAS-ban^OE (b60672). All flies used in this study were female and the control, unless otherwise noted, was the heterologous epi-Gal4/w¹¹¹⁸ strain. Epi-Gal4/UAS system was used to drive gene expression or RNAi knockdown unless otherwise noted.

Mouse strains used in this study were C57BL/6J-Ctnna1^Tvrm5/PjnMmjax and C57BL/6J (The Jackson Laboratory, stock #43572 and #000664, respectively) were bred and maintained under standard conditions of 12:12 light-dark cycle in the Research Animal Facility at The Jackson Laboratory. Mice were provided with NIH31 (6% fat chow) diet and HCl acidified (pH 2.8–3.2) water ad libitum and maintained in pressurized individual ventilation cages, which were regularly monitored to maintain a pathogen-free environment. The C57BL/6J-Ctnna1^Tvrm5/PjnMmjax mice were maintained on C57BL/6J background and confirmed for the absence of Crb1rd⁸ mutation. All animal experiments were performed according to approved guidelines.

Newly eclosed flies were collected and aged (males and females together), for the time as denoted in figure legend. Female flies (unless specified as male) were dissected as previously described (Bailey et al., 2020). Briefly, abdomens were fixed in 4% paraformaldehyde, permeabilized in 1× PBS with 0.3% Triton X-100 and 0.3% BSA, then stained overnight at 4°C using primary antibodies. In this study, mouse anti-FasIII (DSHB, 7G10, AB_528238, 1:50), rabbit anti-Grh (AB_2568305, 1:300; Losick et al., 2016), rabbit anti-GFP (Thermo Fisher Scientific, A-11122, AB_221569, 1:2000), rabbit anti-phospho-Myosin (Cell Signaling, AB_330248, 1:50), rat anti-HA (Roche, 11867423001, AB_390918, 1:100), and mouse anti-α-catenin (DSHB, D-Cat, AB_532377) primary antibodies were used. Secondary antibodies from Thermo Fisher Scientific included donkey anti-rabbit-IgG conjugated to Alexa Fluor 488 (A21206, AB_2535792), goat anti-mouse-IgG conjugated to Alexa Fluor 568 (A11031, AB_144696), goat anti-mouse-IgG conjugated to Alexa Fluor 633 (A-21146), goat anti-rat-IgG conjugated to Alexa Fluor 568 (A-11077) and goat anti-rat-IgG conjugated to Alexa Fluor 633 (A21094, AB_2535749) used at 1:1000 dilution. All tissues were stained with DAPI at 1 µg/ml and mounted in Vectashield (Vector Laboratories, H1000-10) on a glass coverslip and slide, with the inner tissue facing out.

Samples from the experiment involving Brainbow were imaged using the 40× oil (Zeiss Immersol 514F) objective on a Zeiss LSM 880 Airyscan. A Z-stack was set using 1μm per slice and Zen Black was used as the acquisition software. All other abdomens were imaged using a Zeiss AxioImager M2 with ApoTome and a 40x dry objective, with a Hamamatsu Orca-Flash 4.0 camera. A Z-stack was set using 0.50–0.55 μm per slice and Zen Blue was used as the acquisition software. For all samples, using the NIH FIJI software, images were flattened using SUM slices tool for all DAPI channels and intensity measurements and MAX intensity for all other visualizations.

The cell area was measured using FasIII (septate junctions) to determine cell boundaries. An area of 10,000 µm² or 22,500 µm² at least 25 µm away from the dissected edge was chosen. The image was converted into a binary black–white image using the threshold tool, and any missing segments of cell boundary as determined by the analyzer while comparing the thresholded image and the original FasIII image were drawn in using the pencil tool, set at 3 µm. Any portion of the image that did not have a FasIII signal at least twice the brightness of the background was not considered a border. Areas of staining more than 3 µm apart were not considered a continuous border. A region of interest (ROI) map was generated using the analyze particles tool, which was used to calculate the cell area. Any cells for which edges extended beyond the area being analyzed were excluded. For each cell, the number of nuclei per cell was found by merging the FasIII and Grh (an epithelial nuclear marker) channels and counting the number of nuclei present in each cell.

The area covered by multinucleated cells was calculated by summing the areas of each cell with three or more nuclei and dividing by the total area of all cells measured in a given image, then multiplying by 100.

dBrainbow mosaic labeling was used to quantify the number of fused cells per 100 µm² epithelial area. Bicolored regions were identified in epithelial regions co-expressing EGFP–HSV (Alexa Fluor 488) and EBFP2–HA (Alexa Fluor 568) fluorescence. Bicolor cell areas and nuclear number were further quantified by measuring the FasIII borders and DAPI-stained nuclei. In determining the percentage of samples that contained Brainbow-labeled cells in close proximity, any 332.8 μm² area that contained both EGFP–HSV (Alexa Fluor 488) and EBFP2–HA-(Alexa Fluor 633 or 568) fluorescence were considered in close proximity, and up to two regions per fly (one on either side of the midline) could be chosen. Any regions that were scratched or damaged during dissection were excluded.

Fluorescence intensity was measured for αCat protein expression. To do so, 50 ROIs were drawn around the Grh⁺ epithelial area and transferred to the αCat channel. The integrated density of αCat was calculated to determine the average fluorescence intensity for each Drosophila abdominal epithelium. For Sqh–GFP expression, quadrilateral ROIs were drawn around non-intersecting regions of Sqh–GFP signal in banding regions. ROIs fell into one of two groups based on dorsal-ventral orientation (‘Parallel’ or ‘Perpendicular’ with respect to the dorsal-ventral axis). An adjacent area of low/no Sqh–GFP expression was then selected to serve as a background signal measurement (‘Interior’). The total expression in each banding region ROI was divided by the area of said ROI and compared to the same measurements in the corresponding interior ROI, yielding a ratio of expression within the banding region to expression in the Interior (n=20 regions per fly were quantified). Total expression for both Sqh–GFP and P-Myo was quantified from 10,000 µm² regions per fly.

Epithelial ploidy was calculated according to Bailey et al. (2020). Samples from young and old flies were dissected and stained as described above, then imaged with the same exposure time. In FIJI, ROIs were drawn around each nucleus in the Grh (epithelial nuclear marker) channel. This ROI map was superimposed on the sum of slices of the Z-stack for the DAPI channel. Nuclei that overlapped with fat body or other tissue were excluded, then the area and integrated density was measured for each nucleus. The intensity was calculated by subtracting the background, then comparing the intensity of nuclei from old flies to those of the young flies whose nuclei were previously shown to be diploid (Losick et al., 2013).

Lifespan was determined by placing 10 newly eclosed females and 5 males each into a total of 8–10 food vials. The numbers of live female flies were counted each day. Flies were transferred to new food every 2–3 days, and maintained with male flies for duration of the lifespan assay. A log rank (Mantel–Cox) test was used for lifespan analysis.

The abdominal bending assay was performed as previously described (Losick and Duhaime, 2021). Briefly, analysis was performed on male flies, aged as described on date of scoring. Males were immobilized by securing them to a Sylgard dish by gently adhering their thorax to the dish with an insect pin. Males were video recorded for 1 min using a SZX7 stereomicroscope with DP22 camera and Cellsense software (Olympus), and the angle of each bend was measured in Fiji software. A bend was defined as a male curling its abdomen from a fully extended position to a curled position at 45° or greater angle as measured from the midline axis of its body, and then returning to its resting elongated position. Flies that did not move at all during the imaging period were excluded from the analysis.

Adult flies were aged for either 7–9 days or were 40 days old and dissected as described previously (Bailey et al., 2020). Freshly dissected tissue was mounted in halocarbon oil 27 (Sigma). Laser ablation was performed using an Intelligent Imaging Innovations (3I) integrated Ablate module, Yokogawa spinning disk confocal on a Zeiss Axio Observer with a 40× water objective. The 3I ablate module (Slidebook software) was set to a power of 75 out of 200 with 50 repetitions at a laser power of 20%. Time-lapse images were recorded within a single plane to obtain the fastest frame rate. After selecting the focal plane, an ablation ROI was drawn perpendicular (width-wise) to the brightest visible Sqh–GFP band and subsequently cut. Laser ablation results in the recoil of the surrounding tissue whose initial velocity (V₀), calculated by displacement distance/time, is proportional to the magnitude of the resting tension. Therefore, relative initial retraction velocities were compared for each sample and condition and correlated with relative tissue tension.

For RPE flatmounts, 4-month-old C57BL/6J-Ctnna1^Tvrm5/PjnMmjax and age-matched C57BL/6J mice were asphyxiated by CO₂ inhalation. The enucleated eyes were marked with an orange dot dorsally to orient the eyes and placed in ice-cold 4% paraformaldehyde (PFA) solution. The extraocular tissue and the anterior segment of the enucleated eyes were removed. The eye cups, which were fixed overnight in 4% PFA at 4°C, were washed in 1× TBS and the neural retina was separated from the RPE-choroid-sclera. Six radial cuts were made toward the optic nerve to flatten the posterior eyecup. RPE cell borders were stained with phalloidin (F-actin) and nuclei were stained by incubating with DAPI for 2 days at 4°C with agitation. Both phalloidin and DAPI were prepared in 0.3% Triton X-100 in 1× TBS. The RPE-choroid-sclera tissue was then moved through a glycerol gradient of 10%, followed by 20% and 50%, and allowed to equilibrate in each gradient for ∼8 h at 4°C with constant agitation. The whole RPE-choroid-sclera tissue was then flattened, with the RPE side up, onto slides, mounted in Vectashield (Vector Laboratories) overlaid with a coverslip, and examined under a fluorescent Zeiss Axio Observer.Z1 Microscope (Carl Zeiss AG). Identical imaging parameters were applied to both Ctnna1^Tvrm5 and C57BL/6J control RPE flat mounts. For cell and nuclear count, identical regions were selected in oriented eyes from both strains, and the cells and nuclei within those cells were counted by selecting a uniform area of 350 µm² for each image, using the same protocol as described above for the Drosophila cell and nuclear count, in FIJI software.

All Drosophila experiments analyzed contained a minimum of 10–15 biological replicates (fruit flies) obtained from two or more experiments, unless otherwise noted. Mouse RPE analysis used three biological replicates (two female and one male mouse). Data represents the mean±s.em. and statistical analysis was performed with GraphPad Prism 9. A unpaired two-tailed t-test with Welch's correction was used for pairwise analysis and two-way ANOVA with Tukey's multiple comparisons test unless otherwise noted in the figure legend. Lifespan analysis was undertaken using a log-rank (Mantel–Cox) test. P-values are denoted as: ns, not significant (P>0.05), *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001.

We would like to thank the members of the Losick lab for critical review of this manuscript and the fly community, particularly the Bloomington Drosophila Stock Center (NIH P40OD018537), the Developmental Studies Hybridoma Bank (created by the NICHD and maintained at The University of Iowa, Department of Biology, Iowa City, IA 52242, USA) and the TRiP Center at Harvard Medical School (NIH/NIGMS R01-GM084947) for providing transgenic stocks or additional reagents used in this study. Images were acquired using equipment of the Light Microscopy Facility at the MDI Biological Laboratory, which is supported by the Maine INBRE grant (GM103423) from the National Institute of General Medical Sciences at the National Institutes of Health as well as Bret Judson and the Boston College Imaging Core for infrastructure and support.