VCP maintains nuclear size by regulating the DNA damage-associated MDC1–p53–autophagy axis in Drosophila

The maintenance of constant karyoplasmic ratios suggests that nuclear size has physiological significance. Nuclear size anomalies have been linked to malignant transformation, although the mechanism remains unclear. By expressing dominant-negative TER94 mutants in Drosophila photoreceptors, here we show disruption of VCP (valosin-containing protein, human TER94 ortholog), a ubiquitin-dependent segregase, causes progressive nuclear size increase. Loss of VCP function leads to accumulations of MDC1 (mediator of DNA damage checkpoint protein 1), connecting DNA damage or associated responses to enlarged nuclei. TER94 can interact with MDC1 and decreases MDC1 levels, suggesting that MDC1 is a VCP substrate. Our evidence indicates that MDC1 accumulation stabilizes p53A, leading to TER94K2A-associated nuclear size increase. Together with a previous report that p53A disrupts autophagic flux, we propose that the stabilization of p53A in TER94K2A-expressing cells likely hinders the removal of nuclear content, resulting in aberrant nuclear size increase. Cells maintain a constant cytoplasm to nucleus volume ratio, although the role of DNA damage is not well explored. Here, the authors use Drosophila to connect TER94, the fly homolog of VCP, to disruption of DNA damage repair, leading to ubiquitinated Mu2 protein accumulation and enlarged nuclei.

I t has long been appreciated that cells keep a relatively constant karyoplasmic ratio, i.e., the proportion of nuclear volume and cytoplasmic volume 1 , implying maintaining such a physiological consistency has important roles. Indeed, aberrant nuclear size has served as a pathological hallmark in several types of malignant tumors 2,3 . However, it is unclear whether the change in nuclear size has a direct role in causing diseases or it represents a byproduct of another cellular defect that causes diseases. To address this, it is critical to understand the regulatory processes, as well as the cellular stresses, that impact nuclear size.
Studies from multiple systems have revealed that nuclear size is proportionally scaled to the cytoplasmic volume 1,[4][5][6][7][8][9] . This reliance on cytoplasmic volume suggests that nucleus can sense determinants for nuclear size control residing outside the nucleus. The import of nuclear lamina proteins through the nuclear pore complexes (NPCs) has been implicated in the nucleus growth during interphase 10,11 . The availability of ER membrane, the source for reassembled nuclear envelope at the end of cell division 12 , also constrains the nucleus size and shape after mitosis 13 . Recent genetic screens have further implicated nucleocytoplasmic transport, RNA processing, and nucleocytoskeletal interactions as key mechanisms for controlling nucleocytoplasmic ratio 14,15 . Cellular stresses like DNA damages can influence transcription and cause accumulation of certain nuclear proteins, although the impact of DNA lesions on nuclear size has not been thoroughly examined. In this study, we investigate the molecular mechanisms linking nuclear size control and DNA damage using Drosophila TER94, the fly homolog of VCP, which is known to function in both DNA damage response (DDR) and nuclear envelope reassembly after mitosis (see below).
We have previously used Drosophila to model IBMPFD (Inclusion body myopathy with early-onset Paget disease and frontotemporal dementia), a multi-system degenerative disorder associated with specific mutations in VCP 16 . VCP ATPases form hexamers and extract ubiquitinated proteins from organelles and protein complexes for proteasome-mediated degradation, refolding, or liberation 17 . VCP has been implicated in numerous processes. In the cytoplasm, VCP responds to stresses from the endoplasmic reticulum (ER) by participating in the ER-associated degradation (ERAD) of misfolded ER proteins to maintain the homeostasis 18 . A similar requirement for VCP function has also been reported for removing damaged proteins during mitochondrial quality control 19 . VCP is essential for the maturation of autophagosomes and their subsequent fusion with lysosomes 20,21 . During mitotic division, the nuclear envelope and Golgi reassembly also require VCP 22,23 . In the nucleus, this ubiquitindependent protein segregase/unfoldase has been reported to remove or degrade several chromatin-associated proteins involved in DDR 24 . With our Drosophila model, we show that IBMPFD alleles affect energy consumption 16 . Whether Drosophila TER94 has a nuclear function is not clear.
Machinery repairing DNA lesions operates with a precise spatiotemporal sequence, completing the tasks of lesion sensing, signaling, and repairing. To achieve this, DDR proteins recruited to the DNA damage sites for one task need to be displaced to allow the assembly of complexes for a different task. This dynamic regulation involves several post-translational modifications, including phosphorylation, ubiquitination, and SUMO (small ubiquitin-like modifier)-conjugation 25 . For instance, in response to double-strand breaks, the MRN (MRE11-RAD50-NBS1) sensor complex is loaded onto the damage sites, facilitating the recruitment of phosphatidylinositol-3 kinase family protein ATM (known as Tefu in fly) and ATM-dependent phosphorylation of the histone variant H2AX (known as H2AV in fly) 26 . The phosphorylated H2AX (γH2AX, known as γH2AV in Drosophila) then recruits MDC1 (known as Mu2 in fly), which serves as an adaptor for additional factors to remodel chromatins near the DNA damage sites 27 . A SUMO-targeted E3 ligase RNF4 (known as Dgrn in fly) mediates the ubiquitination of MDC1, which promotes its subsequent dissociation from chromatin 28 . The removal of MDC1 is required for the recruitment of downstream signaling and repair proteins, such as 53BP1 and BRCA1 28,29 . However, how ubiquitin-conjugated MDC1 are removed from chromatin is not understood.
In this study, we present evidence that loss of TER94 function leads to nuclear size increase. Consistent with the many known VCP nuclear substrates 30 , ubiquitinated proteins accumulate inside the nucleus of TER94 dysfunction cells, and the extent of accumulation correlates positively with the degree of nuclear enlargement. This TER94-associated nuclear size increase is modified by Mu2 RNAi and overexpression, linking DDR to this process. Genetic evidence further suggests the p53A isoform acts downstream of TER94 and Mu2 in this aberrant nuclear size increase. TER94 interacts physically with Mu2 and negatively regulates Mu2 level, suggesting Mu2 is a substrate of TER94. Mu2 positively regulates p53 protein level and is capable of directly interacting with p53. In addition, TER94 dysfunction, in a Mu2and p53-dependent manner, causes accumulation of p62, a wellknown indicator of disrupted autophagic flux. Together, our findings suggest that TER94 dysfunction disrupts DDR, causing an accumulation of Mu2 proteins, which stabilize p53 and perturb autophagy to enlarge the nuclei.

Results
Disruption of TER94 function causes nuclear enlargement in postmitotic cells. VCP is an essential gene, and whole-animal knockouts of VCP function in both fly and mouse cause embryonic lethality 31,32 . Thus, to analyze the cellular roles of VCP in adults, we chose an alternative approach of expressing UAS-controlled dominant-negative transgenes with tissuespecific GAL4 drivers. We have previously generated UAS-TER94 K2A (K248A & K521A) and UAS-TER94 E2Q (E302Q & E575Q), which overexpress Drosophila VCP mutants affecting ATP-binding and ATP-hydrolysis, respectively 33,34 . These variants with disabled ATPase activity are thought to act as dominant-negatives by forming non-functional hexamers with endogenous wild-type TER94, thereby depleting the functional TER94 pool 18,35 . When expressing TER94 K2A with Rh1-GAL4 (hereafter referred as Rh1 > TER94 K2A ), a driver active in postmitotic outer photoreceptor cells R1-R6 (R cells, Fig. 1a) from the late pupal stage on, we observed an age-dependent increase in the nuclear size ( Supplementary Fig. 1). We measured the R cell nuclear sizes by quantifying antibody-labeled nuclear areas in confocal sections and normalizing them to the cross-section areas of rhabdomeres, cylindrical photo-sensory organelles present in all R cells, from the same cells. As revealed by anti-Lamin antibody staining, which labels the nuclear lamina, the R cell nucleus transforms from a fringe-like shape ( Supplementary Fig. 1a, 70% pupal) to a round contour ( Supplementary Fig. 1a, 0-day adult) at the pupae-to-adult transition. This nuclear morphology transformation was similar in Rh1 > LacZ control and Rh1 > TER94 K2A , indicating that expression of dominant-negative TER94 does not hinder nuclear maturation during development. Instead, Rh1 > TER94 K2A nuclei in 2-day-old adults, compared to those of Rh1 > LacZ, exhibited a 35% increase in size ( Supplementary Fig. 1). From this point on, while Rh1 > LacZ nuclear size remained constant, the Rh1 > TER94 K2A nuclei continued to enlarge, as they were 117% larger than the control in 5day-old adults (Fig. 1b, c and Supplementary Fig. 1), and 129% larger than the control in 8-day-old adults ( Supplementary  Fig. 1).
To ensure these measurements from single confocal images accurately reflect the nuclear sizes, we compared the volumes of 3D representations of photoreceptor nuclei, reconstructed from Z-stacks of anti-Lamin sections (Supplementary Movie 1). The analysis revealed that in 5-day-old adults, TER94 K2A expression, while having no effect on the nucleus shape (Fig. 1d), approximately doubled the nuclear volume (Fig. 1d). Rh1 > TER94 K2A stained with anti-NPC (nuclear pore complex) antibody yielded nuclear enlargement as well, indicating the abovementioned phenotype was not caused by Lamin mis-localization ( Fig. 1e). This expanded nuclear size phenotype is specific, as the cell size revealed with a plasma membrane-associated mCherry and rhabdomere size remained unaltered by TER94 K2A expression ( Supplementary Fig. 2). Together, these results showed that TER94 K2A expression causes a post-development and progressive nuclear size enlargement.
This expansion of nuclear size was not seen in cell overexpressing TER94 A229E (Fig. 1b, c), an IBMPFD-associated variant, which we have previously shown to act as a dominantactive mutant 16 , suggesting that aberrant nuclear enlargement is not a cause to this dementia. Overexpression of TER94 E2Q , a different dominant-negative allele, also caused nuclear enlargement, indicating that this phenotype was not specific to the TER94 K2A variant (Fig. 1e). To further confirm that a loss of TER94 function causes this nuclear size aberrance, we generated mosaic eyes containing mutant clones of TER94 K15502 (a strong hypomorphic allele), and showed that homozygous TER94 K15502 cells, marked by the absence of GFP, exhibited nuclei larger than those of neighboring normal cells (Fig. 1f).
To ensure the nuclear size increase is not restricted to the outer R cells, we used Rh4-GAL4 driver, which is active in~70% of R7 cells (Fig. 1a), to express UAS-TER94 K2A and UAS-GFP-NLS (UAS-GFP-NLS was included to identify the TER94 K2Aexpressing R7 population). Consistent with the Rh1 > TER94 K2A results, R7 nuclei expressing TER94 K2A (Fig. 1g, arrows) were larger than those without TER94 K2A expression (Fig. 1g, arrowheads). Expressing TER94 K2A in other non-neuronal postmitotic cells, such as cone cells and indirect flight muscle cells, recapitulated this aberrant increase of nuclear size (Supplementary Fig. 3). Altogether, these data suggest that TER94 AAA ATPase has a fundamental role in regulating nuclear size in postmitotic cells.
Loss of TER94 function causes accumulations of autophagic markers and ubiquitinated nuclear proteins. VCP hexamer is known to process ubiquitinated substrates for degradation in the ubiquitin-proteasome system (UPS), and has been linked to autophagy, DDR, ERAD, and apoptosis 17 . To determine whether disruption of any of the aforementioned processes contributes to the nucleus expansion, we monitored the reporter activities for these pathways in Rh1 > TER94 K2A retina. Given that TER94 targets ubiquitinated substrates and ubiquitin-conjugation is required by both UPS and autophagy, we stained Rh1 > TER94 K2A adult retina with FK2, an antibody that recognizes mono-and polyubiquitinated conjugates 36 . At the late-pupal stage, at which the nuclear size of Rh1 > TER94 K2A outer R cells appeared normal, FK2 signal was seen in the nuclei of only a few cells (Fig. 2a). As the flies aged, the FK2 signal intensity and number of cells with nuclear FK2 staining increased, suggesting that TER94 K2A expression causes an accumulation of ubiquitinated substrates. In comparison, no nuclear FK2 signal and only occasional cytoplasmic FK2-positive puncta were detected in Rh1 > LacZ (Fig. 2a). This accumulation of ubiquitinated nuclear proteins was caused by loss of TER94 function, confirmed by the FK2 immunostaining of mosaic tissues bearing homozygous TER94 K15502 clones (Fig. 2c). Importantly, the strength of FK2 signal correlated strongly with the nuclear size increase (Fig. 2a, b). This correlation, along with the fact that the appearance of FK2 staining precedes nuclear enlargement, suggests the failure in removing ubiquitinated nuclear proteins in Rh1 > TER94 K2A cells is a cause for the aberrant nuclear expansion.
In 2-day-old Rh1 > TER94 K2A adults, while a significant portion of nuclei were already enlarged, the activity of CD3δ-YFP, an ERAD reporter requires UPS degradation, remained low, implying that ERAD/UPS impairment is unlikely to be a cause ( Supplementary Fig. 4a). The signal of CD8-PARP-Venus, an apoptotic reporter, was absent in 2-day-old Rh1 > TER94 K2A adults and could only be detected in a few Rh1 > TER94 K2A cells 1 week after the onset of nuclear expansion ( Supplementary  Fig. 4b). Similarly, another independent apoptotic reporter, cleaved Caspase-3, was appeared only in a subset of 5-day-old R7 cells expressing TER94 K2A by Rh4-GAL4 ( Supplementary  Fig. 4c). These observations preclude the notion of apoptosis being a cause for enlarged nuclei.
Disruption of nuclear TER94 function in DNA damage repair affects nuclear size. VCP has known functions in both the cytoplasm and nucleoplasm 17 . Given the correlation between ubiquitinated protein accumulation in the nucleus and the nuclear expansion, we speculated that these were due to a deficit of TER94 function in the nucleus. To test this hypothesis, we took advantage of a fortuitous finding of ours that overexpression of Derlin-1, an ER membrane protein that recruits TER94 to the ER vicinity upon ER stress, depletes the nuclear pool of endogenous TER94 40 (Fig. 3a, b). Indeed, overexpression of Derlin-1 caused nuclear expansion (Fig. 3a, b), supporting the notion that this nucleus enlargement was caused by a depletion of nuclear TER94. As a control, expression of Derlin-1 ΔSHP , a Derlin-1 construct incapable of binding to TER94 40 , had no effect on the nuclear TER94 level and the nuclear size (Fig. 3a, b). Furthermore, coexpression of wild-type TER94 in Rh1 > Derlin-1 suppressed the nuclear expansion phenotype ( Supplementary Fig. 6), presumably by replenishing the level of TER94 in the nucleus.
Within the nucleus, VCP function has been implicated in cell cycle and DDR regulations 41,42 . Using EdU incorporation assay, Drosophila-specific FUCCI (Fluorescent Ubiquitination-based Cell Cycle Indicator) system (Fly-FUCCI) 43 , and another cell cycle reporter S/G2/M-Green, we showed that Rh1 > TER94 K2A cells remained mitotically inactive ( Supplementary Fig. 7), indicating that the nuclear morphological change is not caused by cell cycle re-entry. To test the impact of TER94 inactivation on DDR, we stained Rh1 > TER94 K2A for γH2AV, a phosphorylated histone H2A variant formed at DNA damage sites to facilitate DDR processing 44,45 . Like in mammals, robust γH2AV signals react to genotoxic insults in flies, as tested by H 2 O 2 treatment (Fig. 3c). In the absence of external mutagen treatment, this DDR marker signal remained unchanged in Rh1 > LacZ cells, whereas γH2AV staining increased in the nuclei of TER94 K2A -expressing cells (Fig. 3d, e). More importantly, the elevation of γH2AV staining in Rh1 > TER94 K2A correlated with the onset of nuclear size expansion. Like the nuclear enlargement phenotype, increased γH2AV staining was detected in homozygous TER94 K15502 clones in mosaic eyes (Fig. 3f).
the cells' ability to repair DNA damages. In addition, we tested another DDR marker, PCNA (Proliferating Cell Nuclear Antigen), a DNA polymerase δ processivity factor essential for both DNA replication and repair 46 , and observed an elevation of PCNA in Rh1 > TER94 K2A (Fig. 3g). As we have excluded the possibility of Rh1 > TER94 K2A initiating DNA replication, this PCNA observation, along with the abovementioned γH2AV results, indicates that disruption of TER94 function perturbs the DDR process, possibly causing an accumulation of unrepaired DNA lesions.
To pinpoint the source of these unrepaired DNA lesions, we monitored the level of γH2AV in Rh1 > TER94 K2A flies raised in the dark, which minimizes their exposure to radiation and UV light. Compared to Rh1 > TER94 K2A grown in the regular 12 h light/dark cycle, those raised in a constant dark condition showed reduced γH2AV accumulation and nuclear expansion (Fig. 3h). In addition, expression of human superoxide dismutase (hSOD1), a scavenger of reactive oxygen species, in Rh1 > TER94 K2A background reduced γH2AV levels (Fig. 3i). The results from these two independent regimens of mutagen exposure minimization suggest that the increased γH2AV in Rh1 > TER94 K2A is a consequence of accumulating unrepaired DNA damages from environmental genotoxic agents.
Accumulation of Drosophila MDC1 phenocopies TER94 K2Aassociated nuclear enlargement. The increase of DDR markers, as well as FK2 accumulation (Fig. 2), in Rh1 > TER94 K2A suggests that, in the absence of TER94 function, one or a few DDR proteins accumulated, leading to enlarged nuclei. To test this hypothesis, we asked whether reduction of certain DDR proteins, which might be TER94 substrates, could suppress the nucleus expansion in Rh1 > TER94 K2A . Conversely, if accumulation of certain DDR proteins is responsible for the nuclear size increase in Rh1 > TER94 K2A , direct overexpression of these DDR proteins should recapitulate the enlarged nuclei phenotype. Using these two criteria, we screened DDR genes by RNA interference (RNAi) and Rh1-GAL4-driven overexpression. RNAi-mediated knockdowns of several candidates, including tefu, l(3)mbt, mu2, bon, dgrn, and p53 ( Fig. 4a and Supplementary Fig. 9a) showed suppression of the nuclear enlargement, although only Rh1 > eGFP-Mu2, among the tested constructs, presented a nuclear expansion phenotype similar to those seen in TER94 K2A (Fig. 4b, c and Supplementary Fig. 9b). We made UAS-Mu2 transgenic flies and showed Rh1 > Mu2 photoreceptors contained enlarged nuclei (Fig. 5a), verifying that this Mu2-overexpression phenotype was unaffected by the GFP moiety. Rh1 > eGFP-Mu2, similar to Rh1 > TER94 K2A , showed increased γH2AV staining (Fig. 4f, g), consistent with our notion that accumulation of DDR proteins hinders DNA damage repair. Moreover, the eGFP-Mu2 signal intensity in Rh1 > eGFP-Mu2 correlated positively with the nuclear size increase (Fig. 4h), strengthening our hypothesis that the inability to remove Mu2 from the chromatin in TER94 K2A results in nuclear expansion. Co-expression of Mu2 in cells deficient in TER94 function resulted in strong nuclear FK2 signals ( Supplementary Fig. 10), suggesting Mu2 is one of the ubiquitinated proteins accumulated in Rh1 > TER94 K2A . However, Rh1 > Mu2 by itself did not increase FK2 staining in the nucleus (Supplementary Fig. 10), suggesting that accumulation of ubiquitinated proteins per se is not a cause of nuclear enlargement.
The similarities between TER94 K2A and Mu2-overexpression phenotypes suggest that these two genes act on the same pathway.
Indeed, several lines of evidence suggest that Mu2 is a substrate of TER94. Co-overexpression of wild-type TER94 reduced eGFP-Mu2 signals in the nuclei, and conversely, reduction of TER94 function by RNAi increased nuclear eGFP-Mu2 signals (Fig. 4b, d, i). This influence of TER94 on Mu2 level was functional, as TER94 WT overexpression suppressed and TER94 knockdown enhanced Mu2induced nuclear enlargement, respectively (Fig. 4b, c, e). It is worth noting that, as both TER94 and Mu2 were expressed from a heterologous promoter (from UAS directed by Rh1-GAL4), this effect of TER94 on apparent Mu2 expression is unlikely to be transcriptional. Indeed, RT-PCR showed the level of Mu2 transcripts was unaffected in TER94 K2A -expressing cells (Fig. 4j). Nevertheless, Western blot of TER94 K2A lysates showed an increase in the endogenous Mu2 (Fig. 4k), indicating that loss of TER94 function increased Mu2 protein level via a post-transcriptional mechanism. The in situ proximity ligation assay (PLA) showed that TER94 interact with Mu2 in vivo (Fig. 4l). The fact that TER94 regulates Mu2 abundance, along with this observation that these two proteins are in close proximity, suggests that Mu2 is a potential TER94 substrate.
p53A acts downstream of MDC1 in TER94 dysfunctioninduced nuclear expansion. To identify downstream effectors that contribute to this nuclear size increase, we looked for proteins that are known to interact with Mu2 in literature and have significantly modified the TER94 K2A -caused nuclear expansion in the abovementioned RNAi screen. One such candidate is p53, which is known to interact with MDC1 in mammals 47,48 . In addition, RNAi-mediated downregulation of p53 suppressed both TER94 K2A and Mu2-induced nuclear expansion (Figs. 4a and 5a). To independently confirm this role of p53 in nuclear enlargement, we overexpressed TER94 K2A or Mu2 in homozygous p53 null (p53 5A-1-4 ; labeled as p53 −/− ) 49 background. Similar to the knockdown, a complete removal of p53 function suppressed TER94 K2A -and Mu2-associated nuclear expansion (Fig. 5b, c). These genetic interactions place p53 function downstream of TER94 and Mu2 in this context.
The Drosophila p53 gene encodes four isoforms, A/C, B, D, and E (the isoform C the same as isoform A), structurally differing by the length of the N-terminal transactivation domain (TAD) 50 . While the abovementioned genetic interactions link p53 to the TER94 K2A -associated nuclear size increase, it is unclear whether this particular p53 role is isoform-specific. Quantitative Western analysis of Rh1 > TER94 K2A adult head extract detected a significant increase of p53A (Fig. 5d). To verify which p53 isoform(s) facilitates this TER94 K2A -dependent nuclear expansion, we took advantage of the observation that p53 −/− Fig. 3 Blocking TER94 function hampers DNA damage repair. a Confocal images of 5-day-old Rh1 > LacZ, Rh1 > Derlin-1, and Rh1 > Derlin-1 ΔSHP retinas stained with phalloidin (red), anti-Lamin (green), and anti-TER94 (magenta). b Left: Analysis of TER94 subcellular distribution. The analyzed range is indicated by the white lines shown in panel a. Image J measures the intensities of anti-Lamin and anti-TER94 signals along the lines. The peaks of the anti-Lamin signal define the borders of the nuclei. Right: Quantification of the cross-section area of R1-R6 nuclei (top) and the nuclear TER94 intensity (cumulated pixel of TER94 along the line divided by the line distance defined by anti-Lamin-marked two peaks, bottom). c Confocal images of wild-type larval eye discs with or without 50 µM H 2 O 2 treatment stained with anti-Lamin (red) and anti-γH2AV (green). d Time-course analysis of the change in nuclear size and the level of γH2AV from Rh1 > LacZ and Rh1 > TER94 K2A adult eyes stained with phalloidin (red), anti-Lamin (magenta), and anti-γH2AV (green). e Quantification of the anti-γH2AV signal intensity over time from Rh1 > LacZ and Rh1 > TER94 K2A . f An adult TER94 K15502 clone, marked by the absence of myr.GFP (arrowheads), stained with anti-Lamin (magenta) and anti-γH2AV (red). R cell boundaries are outlined in dash lines. g Confocal images of Rh1 > LacZ and Rh1 > TER94 K2A adult retinas stained with phalloidin (red), anti-Lamin (magenta), and anti-PCNA (green). h Left: Confocal images of 5-day-old Rh1 > TER94 K2A retinas from flies raised in normal 12 h light/12 h dark (L/D) or complete dark (D/D) conditions after birth stained with phalloidin (red), anti-Lamin (magenta), and anti-γH2AV (green). Right: Quantification of the γH2AV intensity (top) and the cross-section area of R1-R6 nuclei (bottom). i Left: Confocal images of 5-day-old retinas from Rh1 > TER94 K2A co-expressing either LacZ or hSOD1 stained with phalloidin (red), anti-Lamin (magenta), and anti-γH2AV (green). Right: Quantification of the γH2AV intensity (top) and the cross-section area of R1-R6 nuclei (bottom). The number of independent experiments performed: 2 (a), 4 (c), 6 (f), 4 (g). Quantification details of b, e, h, and i are listed in Statistics and reproducibility. Scale bars: 5 µm (a, d, f), 20 µm (c), 10 µm (g, h, i).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24556-0 ARTICLE suppressed the Mu2-associated nuclear enlargement (see above). In Rh1 > eGFP-Mu2; p53 −/− retina, expression of p53B with Rh1-GAL4 driver had no effect on the nuclear size (Fig. 5c), indicating that this isoform has little, if any, role in this context. In contrast, expression of p53A in the same background restored the nuclear size increase (Fig. 5c), strongly suggesting that this isoform contributes to the TER94 K2A -linked nuclear expansion phenotype.
To ask whether stabilized p53A can directly influence nuclear size, we overexpressed Myc-tagged p53A with Rh1-GAL4. Although Rh1-GAL4 is active in all outer R cells, only a small subset of them exhibited detectable levels of Myc-p53A (Fig. 5g), consistent with the notion that p53 proteins are labile. Still, those with sufficient Myc-p53A accumulation exhibited enlarged nuclei, demonstrating that p53A expression alone can increase nuclear size. Importantly, Rh1 > Myc-p53A showed no increase in γH2AV ( Supplementary Fig. 11), suggesting that accumulation of DNA lesions and nuclear enlargement can be uncoupled. TER94 K2A nuclear expansion is associated with p53-mediated autophagy flux impairment. Expression of Drosophila p53A in photoreceptors has been reported to impair autophagic flux 54 . As p53A contributes specifically to the nuclear enlargement, we hypothesized that the Atg8a and p62 accumulation observed in TER94 K2A (Fig. 2d, e) might represent arrested autophagosomes. In this scenario, disrupted autophagic flux could cause the nuclear size increase in Rh1 > TER94 K2A by failing to remove damaged nuclear components. To ask whether this autophagic disruption in Rh1 > TER94 K2A depends on Mu2 or p53, we monitored the signal intensities of GFP-Atg8a or GFP-p62 in Rh1 > TER94 K2A eyes co-expressing RNAi constructs against p53 or Mu2. Compared to Rh1 > TER94 K2A alone, knockdowns of p53 and Mu2 significantly reduced the GFP-Atg8a and GFP-p62 signals (Fig. 6a-c). Antibodies staining for endogenous Atg8a and p62 levels confirmed this dependence of Rh1 > TER94 K2A on p53 or Mu2 ( Supplementary Fig. 12), demonstrating that TER94 K2A perturbs autophagy through a Mu2-and p53-dependent manner.
To further understand the nature of this autophagic disruption in Rh1 > TER94 K2A , we used a tandem GFP-mCherry-Atg8a fluorescent reporter, which allows us to determine, by a ratiometric analysis of GFP and mCherry fluorescence, the state of autophagy progression. As the GFP moiety in the chimera is quenched by the acidic environment during autophagosome-lysosome fusion, this reporter in Rh1 > LacZ control exhibited a signature puncta proportion in various stages: 58% in stage I (autophagosomal formation; yellow),~6% in stage II (fusion with lysosomes; red), and~30% in stage III (lysosomal degradation; small red). Treating the flies with rapamycin, which induces autophagy, decreased the proportion of puncta in stage I to~13%, with a concomitant increase of stage II puncta to~62%. In contrast, treatment with chloroquine, known to inhibit lysosomal fusion 55 , increased the proportion of stage I puncta to~83%. In Rh1 > TER94 K2A ,~93% of tandem reporter puncta were in stage I. This phenotypic similarity to chloroquine-treated retina ( Fig. 6d-g) suggests that TER94 dysfunction causes an arrest of autophagy progression. Notably, we observed diffused signals of tandem reporter (Fig. 6d) in the nuclei, a phenotype also seen with the GFP-Atg8a reporter (Fig. 2d). Because LC3/Atg8a can reside in the nucleoplasm and cytoplasm, we surmised that TER94 K2A hinders its exportation from the nucleus, a process promoted by autophagy activation 56,57 . To ensure that this disruption of autophagic flux is specific to this particular TER94 function, we showed that compared to Rh1 > LacZ control (~59%, Fig. 6h-j), the proportion of stage I puncta was also higher (~88%) in Rh1 > Mu2.
To confirm the autophagy flux impairment and nuclear size enlargement are facilitated by p53A, we compared the endogenous p62 levels and the nuclear sizes in R cells expressing Mu2 alone with those co-expressing p53A. Compared to LacZ control, Mu2 overexpression alone showed a slight but reproducible increase of p62. The p62 signal in Rh1 > Mu2 was strongly enhanced when p53A were co-expressed. Co-expression of p53B showed little effect on p62 signal, demonstrating the specificity of the enhancement by p53A (Fig. 6k). Likewise, the moderately enlarged nuclear size in Rh1 > Mu2 expanded further with p53A co-expression (Fig. 6k). Together, these results show that TER94 K2A and Mu2 overexpression cause autophagic flux defect and nuclear enlargement through p53A.

Discussion
We have used the Drosophila eye to investigate the physiological function of VCP and the mechanism by which specific VCP alleles contribute to neurodegeneration 16 . Here, using both cellspecific expression of dominant-negative TER94 mutants and FLP-induced TER94 k15502 mutant clones, we show that removal of TER94 function in postmitotic cells leads to an aberrant nuclear expansion. This nuclear size increase is not restricted to photoreceptors, as the expression of TER94 K2A in other cell types generates a similar phenotype, suggesting that the role of TER94 in nuclear size regulation is fundamental to most cells. Altered nuclear size has been associated with diseases like cancer, although the mechanistic link between aberrant nuclear size and cell pathophysiology is not well understood. While multiple explanations likely exist for aberrant nuclear enlargement, the known roles of VCP in degenerative disorders and the specificity Fig. 4 TER94 influences Mu2 level in regulating nuclear size. a Quantification of the R1-R6 cross-section nuclear area from Rh1 > TER94 K2A flies coexpressing UAS-RNAi constructs targeting indicated genes. The purple lines indicate the difference of the mean equals 1 (mean diff. = 1) as compared to the control (Rh1 > TER94 K2A , LacZ). The DDR genes with mean diff. ≥1 are marked in yellow. The number of independent nuclei measured are listed in Statistics and reproducibility. b Confocal and 3D rendering images (individual nuclei are color-coded) of 7-day-old (confocal) or 5-day-old (3D rendering) retinas from flies expressing indicated transgenes stained with phalloidin (red) and anti-Lamin (magenta). c-e Quantification of the cross-section area of R1-R6 nuclei (c), eGFP intensity (d), and nuclear volume (e). f Confocal images of 7-day-old retinas expressing indicated transgenes stained with anti-Lamin (red) and anti-γH2AV (magenta). g Quantification of the anti-γH2AV intensity. h Pearson's correlation of nuclear size and eGFP-Mu2 intensity (r: coefficient). i Western analysis of eGFP-Mu2 level (anti-GFP) in indicated TER94 backgrounds. Anti-Histone H3 and anti-β-Actin serve as loading controls. j RT-PCR measurement of mu2 and TER94 expression from indicated genotypes. The control (Ctrl) groups omitted templates. rp49 serves as control. k Western analysis of Mu2 level in tissues expressing LacZ or TER94 K2A . Anti-Histone H3 and anti-β-Actin serve as loading controls. l Larval eye discs from flies expressing indicated transgenes under the control of GMR-GAL4 are stained with anti-GFP (magenta) and anti-V5 (red) (left panel). Larvae bearing the same genotypes are subjected to in situ PLA using anti-GFP and anti-V5 antibodies (right panel). The duolink signal reveals proteins in the same complex. The lines in dot plots from a represent mean ± SE. The number of independent experiments performed: 3 (b, i), 2 (j, k, l). Quantification details of c, d, e, and g are listed in Statistics and reproducibility. Scale bars: 10 µm (b, f), 20 µm (l).
of TER94 K2A nuclear enlargement suggest this is an ideal model system to explore the underlying mechanism.
As the molecular function of VCP has been extensively studied, we suspect that the nuclear enlargement is linked to one of its characterized roles. We have previously shown that overexpression of Derlin-1, a direct interactor of TER94, retains TER94 in the cytoplasm 40 . Derlin-1-dependent depletion of nuclear TER94 causes enlarged nuclei, suggesting that this nuclear enlargement is caused by a disruption of TER94 function in the nucleus. In support of this, TER94 K2A -expressing R cells show elevated levels of ubiquitinated proteins inside the nucleus, and the level of ubiquitination correlates well with the severity of the nuclear size. Our results suggest that the clearance of some ubiquitinated nuclear proteins is blocked by TER94 K2A , thereby causing nuclear enlargement. While accumulation of nuclear proteins, caused by a disruption of bulk nucleocytoplasmic transport, has been shown to affect N/C ratio in yeast 15 and mammalian cells 58 , our analysis suggests that these accumulated ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24556-0 nuclear proteins in Rh1 > TER94 K2A are limited to DDR factors.
Within the nucleus, VCP has been shown to rapidly accumulate at DNA damage site, facilitating the extraction of chromatintethering proteins such as L3MBTL1 and KAP1 to assist the recruitment of 53BP1 or BRCA1, respectively 59,60 , and the removal of Ku80 from DNA at the end of non-homologous end joining repair 61 . Knockdowns of fly L3MBT and Ku80 suppress the Rh1 > TER94 K2A -associated nuclear enlargement, linking DNA repair to this aberrant nuclear size. This hypothesis is further supported by our analysis of MDC1, a scaffold protein that bridges the injured DNA and DDR factors 52,62,63 . Mu2, the fly homolog of MDC1, modifies the phenotype of Rh1 > TER94 K2A , and Mu2 overexpression recapitulates the phenotype of Rh1 > TER94 K2A , placing Mu2 downstream of TER94 in this context. As TER94 is pleiotropic, this evidence of Mu2 overexpression phenocopies TER94 K2A supports that DDR participates specifically in nuclear expansion. MDC1 is indispensable for DDR; however, prolonged retention of MDC1 on DNA damage site impedes the access of downstream signaling or repairing proteins 28,29,64 . In support of the notion that a coordinated removal of MDC1-γH2AX association is critical for maintaining genomic stability, we observed persistent γH2AV signals in Mu2overexpressing cells. Interestingly, Mu2 overexpression enlarges the nuclei without increasing FK2 staining, suggesting it is the elevated Mu2 level inside the nucleus, not the accumulation of ubiquitinated nuclear proteins per se, that causes TER94 K2Adependent nuclear expansion. In addition, as the SUMO-targeted ubiquitin ligase RNF4 and the deubiquitinase Ataxin-3 have been shown to regulate the removal of MDC1 from chromatin 65 , it is likely that the ubiquitinated nuclear proteins observed in Rh1 > TER94 K2A include Mu2. Consistent with this, we demonstrate that TER94 negatively influences Mu2 level via a posttranscriptional manner, and co-expression of Mu2 significantly increases FK2 signals in Rh1 > TER94 K2A . These observations, combined with the result of in situ PLA demonstrating that TER94 interacts with Mu2, imply that Mu2 is a direct substrate of TER94. Thus, we propose that TER94-mediated degradation of Mu2 is a crucial step permitting the DNA repair machinery to engage the lesion site.
During DDR, MDC1 serves as a binding platform for recruiting downstream DDR factors 52,62 including p53. Our observation that p53-RNAi suppresses both TER94 K2A and Mu2induced nucleus expansion places p53 downstream of these two proteins in promoting nuclear size increase. The PLA assay confirms an interaction between Mu2 and p53. Furthermore, the Western blotting results show Mu2 overexpression boosts p53 levels, suggesting that this Mu2-p53 interaction stabilizes labile p53 proteins. In mammalian systems, p53 protein level under normal condition is kept low, as most of p53 proteins are destined for degradation via a continuous ubiquitination by E3 ubiquitin ligase MDM2. In the presence of DNA damage, phosphorylation of p53 by various kinases could disturb this MDM2-p53 interaction and stabilize p53 66 . It has been reported that MDC1 is indispensable for CHK2-mediated p53 phosphorylation and stabilization in response to ionizing radiation in mammalian cells 52 . The observation that Mu2 overexpression positively and Mu2-RNAi negatively regulates p53 protein level, respectively, suggests a similar mechanism may exist in Drosophila. In support of this, MDC1 is known to interact with p53 via the BRCT domain 47,48 , which is conserved in Drosophila Mu2 53 . It is thought that the accumulated Mu2 represents a signal from stalled DDR, and this Mu2-p53 interaction allows p53 to elicit responses to protect the genome. Our demonstration that manipulating p53 level modifies nuclear size under TER94 K2A or Mu2 overexpression suggests that alteration in nucleus structural maintenance could be a response to stalled DDR.
LINC complexes, implicated in N/C control 14 , comprise inner and outer nuclear membrane proteins and function to bridge the nuclear lamina and the cytoskeleton 67 . In yeast and mammalian cells, chromatin associated with DSBs exhibited increased mobility and this phenomenon requires LINC complexes 68 , raising a potential link between chromatin repair to nuclear architecture. Our analysis suggests p53 may participate in this or a similar process, possibly functioning as a sensor of unrepaired DNA damages for machinery that regulates the nuclear size. Indeed, 53BP1, a p53 binding protein, decorates the damaged foci 68 , although whether p53 has a direct role in this process remains to be shown.
The p53 locus is capable of generating multiple isoforms through the usage of alternative promoters, splicing sites, and translational initiation sites 69 . Our study suggests the nuclear expansion, caused by TER94 dysfunction or Mu2-overexpression, selectively requires p53A. Importantly, expression of p53A alone can increase nuclear size in the absence of excessive DNA damages, placing its role downstream of unrepaired DNA damages in TER94 K2A -dependent nuclear expansion. A recent report suggests that Drosophila p53A and p53B have opposing effect on autophagy and apoptosis, and it is the balance between these p53 isoforms that mediates this cellular life-and-death decision under stress 54 . Our observation that p53A is selectively required for this context suggests that deregulated DDR, caused by either TER94 dysfunction or Mu2 accumulation, tips the balance to favor p53A-dependent response. In any case, our work provides a new direction in resolving p53-regulated cellular events and emphasizes the importance of unveiling factors controlling p53 abundance in an isoform-specific fashion.
Drosophila p53A have recently been reported to disrupt autophagy flux 54 . As our analysis implicates p53A, it seems plausible that a disruption of autophagy has a role in this TER94 K2A -associated nuclear enlargement. Indeed, TER94 K2A cells exhibit age-dependent p62 accumulation, a well-established indicator of autophagy deficiency 39 . Furthermore, downregulation of Mu2 or p53 ameliorates the p62 accumulation in Rh1 > TER94 K2A , indicating that this autophagy deficiency is mediated by Mu2 and p53. In addition to regulating chromatin modification 70 , autophagy has been shown to remove nuclear components upon genotoxic insults 71 . Thus, p53A expression may disrupt autophagy-mediated removal of nuclear components, thereby increasing the nuclear size. Taken together, we propose a model (Fig. 6l) that VCP dysfunction causes aberrant accumulation of MDC1 on the DNA damage foci. The MDC1 accumulation then stabilizes p53A, which hinders the autophagydependent removal of nuclear materials and alters the nuclear size.
Nearly 150 years ago, Sir Lionel Beale, the father of cytology, described different nuclear morphology in various tumor types and recognized its prognostic value 72 . Nuclear morphometry, including changes in nuclear size or shape, and the presence of nuclear lobulation, have since been extensively used to grade different tumors and optimize clinical treatment. As altered VCP expression is also associated with poor prognosis and sometimes linked to the metastatic potential in several cancer types [73][74][75][76][77][78] , our study provides a plausible explanation of how VCP dysfunction, the nuclear size change, and defective DDR could be linked. Furthermore, the identification of MDC1 is a key player in stabilizing p53 during DDR, and by which to modulate autophagy and nuclear size opens a new direction for the field studying p53mediated cellular responses. Our findings might help in developing new targets for clinical applications.
To generate somatic eye clones of TER94 K15502 , hs-FLP; FRT42D, GMR-myr. GFP virgins were crossed with FRT42D, TER94 K15502 / CyO males. The F 1 progeny were heat-shocked in a 37°C water bath for 1 h at second or third instar larval stage to induce mitotic recombination. An additional 1-hour heat-shock a 37°C on the following day was administered to increase the frequency of recombination. Eye clones were examined among F 1 adults without CyO.
To generate pUAST-Mu2 construct, Drosophila Mu2 was excised from cDNA clone LD44171 (Drosophila Genomics Resource Center, Indiana, USA) and subcloned into pUAST as a KpnI-XbaI fragment. The pUAST-Mu2 construct was verified by sequencing prior to transgenic fly production by P element-mediated transformation 80 (WellGenetics, Taiwan).
Antibody production and immunohistochemistry. To generate polyclonal anti-TER94, anti-p53, and anti-Mu2 antibodies, polypeptides corresponding to the amino acids 701-801 of TER94, amino acids 373-495 of p53, and the amino acids 627-947 of Mu2, respectively, were bacterially expressed, purified, and used for immunization (GeneTex). Whole-mount preparation of fly eyes and eye discs were Quantification of the nuclear size. For all the quantifications of cross-section areas, multiple R cell nuclei, each represented by a dot in figure panels, from at least 2 independent eyes of each group were analyzed with Fiji, an image processing package based on Image J 1.53e. The nuclear and the rhabdomeral areas, represented by Lamin and phalloidin staining, respectively, are manually outlined, and the area of each nucleus is normalized to the area of its corresponding rhabdomere as a relative nuclear cross-section size. For the measurement of nuclear volume, Zstacks of continuous cross-section images of 1 μm thickness that cover the anti-Lamin signals were collected. Raw data were further processed with Avizo 9.4 (Thermo Fisher Scientific) for nucleus volumetric segmentation and volume/ morphology quantification. We first use magic wand tool to segment nucleus in 3D, and further use Material Statistic function for each segmented nucleus volume quantification. To quantify the morphology changes, we use Label Analysis function and define a new parameter "sphericity" Ψ ¼ π

Ap
, where V p is the volume of each segmented nucleus and A p is the surface area of the same nucleus. Sphericity can represent if the 3D shape of the nucleus close to a perfect sphere (sphericity of a perfect sphere is 1). The formula we used for Pearson's correlation analysis of nuclear cross-section size and FK2 intensity/ eGFP-Mu2 intensity is r ¼ n ∑xy ð ÞÀð∑xÞð∑yÞ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , in which r is correlation coefficient, x is the nuclear cross-section size, y is the FK2 intensity or eGFP-Mu2 intensity, and n is the number of paired measurement.
Subcellular fractionation. Subcellular fractionations to isolate nuclear and cytosolic proteins were performed as described by Nabbi and Riabowol 83 with modifications. Briefly, fly heads of appropriate genotypes were homogenized in ice-cold PBS containing 0.1% NP-40 supplemented with protease inhibitor cocktail (Roche). While 25% of the homogenates (the whole cell lysates) were saved on ice, the remaining homogenates were centrifuged for 30 s at 9300 x g. The resulting supernatant (the cytosolic fraction) was kept on ice, whereas the pellet was resuspended in ice-cold PBS containing 0.1% NP-40 with protease inhibitor cocktail. The resuspended pellet was re-centrifuged for 30 s at 9300 x g and the resulting pellet was resuspended in ice-cold PBS containing 0.1% NP-40 with protease inhibitor cocktail (the nuclear fraction). The whole cell lysates and the nuclear fraction were sonicated on ice with two 8-s pulses at 20 kHz, followed by a centrifugation for 1 min at 16,100 x g to remove the debris. The final whole cell lysates and the nuclear fractions were obtained by collecting the supernatants after centrifugation.
EdU incorporation assay. To determine whether cells re-entry cell cycle, Click-iT ® EdU (Invitrogen) was used following the manufacturer's instructions. Larval or adult retinas were dissected in PBS and then incubated in S2 medium only (control) or S2 medium with 20 µM EdU for 1.5 h. After incubation, the tissues were fixed with 4% paraformaldehyde in PBS and permeabilized with 0.5% Triton ® X-100. Mouse anti-Lamin (1:20, ADL67.10, DSHB) antibody was used to label the nucleus. Zeiss LSM-510 confocal microscope was used for collecting the images.
Gamma irradiation. Flies were irradiated with an open beam Cobalt-60 gamma source for the total dose of 1 Gy at the dose rate of 5.412 Gy/h. Control adult flies were mock irradiated.
Quantification of a tandem GFP-mCherry-Atg8a reporter. GFP-mCherry-Atg8a structures were analyzed using ImageJ as previously described 84 . Briefly, after feature extraction of puncta on the raw images, the GFP and mCherry intensity in each pixel across the center of the puncta was measured in arbitrary units (a.u.). The mean fluorophore intensity of the control group in an experimental set was used to define the gating thresholds of GFP and mCherry. Plot profile was used to visualize mCherry and GFP intensity from several Atg8a-positive structures to reveal dynamics of the reporter in the tested conditions; high intensity of both fluorophores indicates autophagosomes, high mCherry and low GFP indicates fusion with the lysosome as GFP is quenched in this organelle, and low mCherry reflects degradation within the lysosome. Quantification of percentage of puncta in each quadrant and the ratiometric analysis were derived from three independent experiments.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All data generated or analyzed during this study are included in this published article (and its supplementary information files). The raw data underlying Figs. 1c, d, f, 2b, 3b, e, h-i, 4a, c-e, g-k, 5a-e, g, 6c, e-k and Supplementary Figs. 1b, 2b, d, 6b, c, 8b, d, 9b, 11 are available in the Source Data file provided with this paper. All unique materials generated are readily available from the authors. Source data are provided with this paper.