Clearance of protein aggregates during cell division

Abstract


Introduction
Nascent polypeptides fold into three-dimensional structures to perform their biological functions (Hipp et al., 2019).Misfolded proteins tend to have their hydrophobic residues exposed and result in aggregation of proteins (Hartl et al., 2011).Intramolecular beta-sheets in some proteins are also structural elements prone for protein aggregation (Tyedmers et al., 2010).Protein misfolding and aggregation not only disrupt the native function of the protein but also could interfere other proteins' functions by co-aggregation (Olzscha et al., 2011;Woerner et al., 2016).This challenges proteome homeostasis (proteostasis) and causes proteostatic stress in cells, which is a main driver of cellular aging and neurodegenerative disorders (Balch et al., 2008;Labbadia and Morimoto, 2015).
Cells evolve several protein quality control (PQC) systems to maintain their proteostasis in which components of PQC systems facilitate refolding, degradation, and spatial deposition of misfolded protein aggregates (Jayaraj et al., 2020).Molecular chaperones are key players in the PQC system.
They are involved to fold nascent polypeptides, target misfolded proteins for degradation by ubiquitin-proteasome system or autophagy, and promote solubilisation of protein aggregates (Tyedmers et al., 2010).Eukaryotic cells confine misfolded proteins and smaller aggregates into distinct cellular deposition sites, which reduce the reactivity of these harmful species to the proteome (Kaganovich et al., 2008).These deposition sites include aggresomes or aggresome-like induced structures locating near centrosomes, INQ (intranuclear quality control compartment) residing in the nucleus, cytosolic CytoQ, and the perivacuolar IPOD (insoluble protein deposit) (Kaganovich et al., 2008).Confinement of protein aggregates in a deposition site facilitates retention of aggregates in short-lived cells (mother cells) during asymmetric cell division (Aguilaniu et al., 2003;Fuentealba et al., 2008;Rujano et al., 2006;Singhvi and Garriga, 2009;Zhou et al., 2014).
The endoplasmic reticulum (ER) processes one-third of cellular proteome and possesses stronger ability to maintain aggregation-prone proteins in a non-toxic state than the cytosol owing to its specific molecular chaperone environment (Vincenz-Donnelly et al., 2018).Chaperones in the ER such as HSPA5/BiP regulates the unfolded protein response (UPR), which blocks instant protein synthesis and activates transcription of genes involved in PQC (Preissler and Ron, 2019;Wiseman et al., 2022).In addition, the proteostatic stress in the ER triggers ER-associated degradation (ERAD) and ER-phagy to remove misfolded proteins (Mochida and Nakatogawa, 2022;Olzmann et al., 2013).
Despite the high capacity of proteostatic stress response in the ER (Rousseau et al., 2004;Vincenz-Donnelly et al., 2018), excessive misfolded proteins still lead to formation of protein aggregates in the ER (Melo et al., 2022;Miyata et al., 2020).In response to protein aggregation in the ER, BiP functions as a disaggregase to promote solubilization of these aggregates.In cells undergoing asymmetric cell division such as in the budding yeast, the ER diffusion barrier and the ER stress surveillance (ERSU) pathway ensure ER protein aggregates are retained preferentially in the shortlived mother cells during cell division (Clay et al., 2014;Pina and Niwa, 2015).How ER protein aggregates are regulated in dividing cells, especially those that do not undergo asymmetric cell division, is less well studied.We addressed the fate of ER protein aggregates in human cells undergoing cell division symmetrically and identified a clearance mechanism of protein aggregates when cells are exiting mitosis.

Results
Targeting a proteostasis reporter to the ER results in protein aggregate formation in the nucleus.
To investigate how cells maintain their proteostasis during cell division, we employed a proteostasis reporter consists of a firefly luciferase mutant prone for protein mis-folding and protein aggregation fused to a green fluorescent protein (FlucDM-eGFP) (Gupta et al., 2011).Next, we targeted the reporter to the ER by fusing a ER-targeting sequence at its N-terminus (ER-FlucDM-eGFP).
Interestingly, we found that ER-FlucDM-eGFP assembled into visible protein aggregates in the nucleus when stably expressed in mammary epithelial MCF10A cells in the absence of any perturbation of the proteome stability (Figure 1A).The protein aggregates were below detection limit of our microscopy in first 2 days after lentivirus transduction, but progressively accumulated in cells and stabilized in the population several days after lentivirus transduction.This contrasted with a previous study in which the ER-FlucDM-eGFP was transiently expressed in cells and hence did not allow long term tracking of the protein (Sharma et al., 2018).Luciferase activity measurement showed that ER-FlucDM-eGFP has significantly reduced enzymatic activity and was not functional possibly due to mis-folding and protein aggregation (Figure 1B).Moreover, we observed the colocalization of ER-FlucDM-eGFP aggregates and thioflavin T (ThT), a universal dye to detect amyloid fibril, further supporting these aggregates were misfolded protein aggregates (Figure S1B).Thus, expression of ER-FlucDM-eGFP results in proteostatic stress and allows us to study how protein aggregates are regulated in human cells.
Formation of protein aggregates in the nucleus by ER-FlucDM-eGFP was not limited to MCF10A cells.The protein aggregates were also detected in the nucleus when ER-FlucDM-eGFP was expressed in other cell lines such as A549, MDA-MB-231 and U2OS cells (Figure 1A and S1A).
Similar observations were found when the reporter was fused to the red fluorescent protein mCherry or when a different N-terminal signal peptide sequence was used.In addition, when ER-FlucDM was replaced by ER-HaloDM, which is a modified haloalkane dehalogenase prone to mis-folding and aggregation (Melo et al., 2022), we observed similar protein aggregates in the nucleus (Figures 1C    and S1C).However, only very low number of cells formed aggregates in the nucleus when intact GFP was targeted to the ER (Figure S1D).Thus, ER-FlucDM-eGFP formed protein aggregates in the nucleus independent of cell line, fluorescent protein types, specific ER targeting sequence and FlucDM.
Despite localizing in the nucleus, we found that the protein aggregates were surrounded by the ER membrane as stained by the ER-tracker (Figure 1D).To further confirm this, we sectioned the cells and used electron microscopy to observe the protein aggregates.We found that the aggregates were surrounded by a single layer membrane closed to the inner membrane of the nucleus (Figure 1E).Thus, the aggregates are intra-nuclear membranous structure surrounded by the ER membrane.

Differential and slow turnover of protein aggregates in cells at interphase and mitosis.
Pathological proteins aggregates such as Z-α1-antitrypsin accumulated in the ER exhibited minimal turnover (Dickens et al., 2016).To test whether ER-FlucDM-eGFP aggregates has a reduced dynamics as well, we performed fluorescence recovery after photobleaching (FRAP) in cells expressing both ER-FlucDM-eGFP and ER-FlucDM-mCherry.The ER-FlucDM-eGFP signal served as a tracker for the location of aggregates and the ER-FlucDM-mCherry was used to probe the aggregate dynamics.We photobleached ER-FlucDM-mCherry with 561 nm laser while ensuring ER-FlucDM-eGFP signal was not affected.In interphase cells, ER-FlucDM-mCherry aggregates intensity barely recovered after 4.5 min post-bleaching, indicating that ER-FlucDM-mCherry aggregates possessed minimal turnover similar to that of in other protein aggregates such as Z-α1antitrypsin (Figure 2A).Similarly, ER-HaloDM-eGFP aggregates in the nucleus also displayed low recovery after photobleaching (Figure S2A).When ER-FlucDM-mCherry aggregates were photobleached in mitotic cells, the recovery of intensity was also much lesser than the ER-FlucDM-mCherry signal present in the ER network (Figures 2B and 2C).However, ER-FlucDM-mCherry aggregates in mitotic cells showed a higher recovery than that of in interphase cells (Figure 2D; aggregates of interphase cells recovered from 0.131 ± 0.093 to 0.157 ± 0.095; aggregates of mitotic cells recovered from 0.196 ± 0.137 to 0.356 ± 0.151).
Consistently, the single-cell analysis revealed that in mitotic cells, there was an increase of the recovery intensity of ER-FlucDM-mCherry in late time-point as compared to that of in early timepoints after photobleaching.The increased recovery intensity was not observed in ER-FlucDM-mCherry aggregates of interphase cells (Figure S2B).Taken together, ER-FlucDM-mCherry aggregates in the nucleus have low turnover and the aggregate displays increased turnover in mitotic cells.

Protein aggregates are cleared during cell division.
Since aggregates displayed some levels of turnover during mitosis (Figures 2B, 2C and 2D), we next investigated the aggregate behaviour throughout cell division.To this end, we labelled cells expressing ER-FlucDM-eGFP with SiR-Tubulin to distinguish cells at various stages based on the microtubule organization.We categorized cells into interphase and prophase, prometaphase, metaphase, anaphase, telophase and early G1 cells based on their microtubule structures (Figure 3A).Next, we analyzed the number and area of aggregates in cells of each category.Interestingly, we observed that the number and area of aggregates in telophase and early G1 cells were significantly lower than in interphase cells, suggesting that aggregates are cleared gradually when cells progress through mitosis and cytokinesis (Figure 3B).
To test whether this was the case, we employed time-lapse imaging to profile aggregates in cells undergoing division.To increase the number of dividing cells during imaging, we synchronized cells at G2/M and released them into mitosis prior to imaging.We observed that upon entry into mitosis, the ER-FlucDM-eGFP aggregates were released from the nucleus during nuclear envelope breakdown and gradually decreased in numbers as cells progress through mitosis and cytokinesis (Figure 3C and Video 1).In majority of early divided G1 cells, there was essentially no detectable aggregates (Figure 3D).Interestingly, the aggregates released from the nucleus during mitosis were still surrounded by the ER membrane as indicated by the ER-tracker (Figures S3A and S3B).The decrease of aggregates appears to happen specifically in dividing cells as the number and area of aggregates remained largely unchanged in interphase cells (Figures 3C and 3D).Furthermore, aggregates formed in cells expressing ER-HaloDM-eGFP decreased in numbers when cells progress through mitosis and cytokinesis (Figure S3C and Video 2), suggesting that the clearance of protein aggregates during cell division was not specific to ER-FlucDM-eGFP.

Effects of ER stress inducers on aggregate clearance in mitotic cells.
Expression of ER-FlucDM-eGFP results in proteostatic stress in cells.Consistently, we observed that gene expression for chaperones and co-chaperones in the ER were upregulated in these cells (Figure 4A), indicating cells involved ER proteostasis control during adaptation.We speculated that acute perturbation to ER homeostasis could exacerbate the proteostatic stress in dividing cells expressing ER-FlucDM-eGFP and affect aggregate clearance during mitosis.To test if this was the case, synchronized cells expressing ER-FlucDM-eGFP were released into mitosis in the presence of either DMSO (control) or 1 M thapsigargin (Thaps), which blocks the ER calcium ion pump and causes ER stress.In cells at prometaphase and metaphase (35 min after release), DMSO and Thaps treatments showed comparable number and area of aggregates in cells (Figures 4B and 4C).
However, in cells at the late stage of cell division (65 min after release), there were significantly higher number of aggregates retained in the cytosol of Thaps-treated cells than DMSO (Figures 4B    and 4C).Consistently, time-lapse microscopy revealed that ER-FlucDM-eGFP aggregates retained in the cytosol after cytokinesis in Thaps-treated cells, whereas the aggregates were largely cleared in control cells completing cytokinesis (Figures 4D, 4E, S4A and Video 4).Thus, acute treatment of ER stress inducer Thaps in dividing cells prevents clearance of ER-FlucDM-eGFP aggregates.
A previous study showed that cells with protein aggregates treated for hours with Thaps or Tunicamycin (Tuni), which inhibits protein glycosylation in the ER, promotes aggregate clearance in cells (Melo et al., 2022).To test if prolonged treatment of cells expressing ER-FlucDM-eGFP with Tuni could promote aggregate clearance, we pretreated cells blocked at G2/M with varying concentrations of Tuni for 3 hours and released cells into mitosis.Interestingly, prolonged pretreatment of Tuni prior to mitosis promoted aggregate clearance in dividing cells (Figure S4B).Thus, acute treatment of cells expressing ER-FlucDM-eGFP with ER stress inducers prevents aggregate clearance in dividing cells while prolonged treatment with ER stress inducers promotes aggregate clearance.

Hsp70 family protein BiP is required for aggregate clearance during cell division.
Since we observed that the expression of a chaperone Hsp70 (HSPA5/BiP) was upregulated in cells expressing ER-FlucDM-eGFP (Figure 4A) and Hsp70 is involved in elimination of protein aggregates (Melo et al., 2022;Nillegoda et al., 2015), we next tested whether the clearance of ER-FlucDM-eGFP aggregates in dividing cells was mediated by Hsp70 family proteins.To this end, cells entering mitosis were treated with either DMSO (control) or VER-155008 (VER), an inhibitor of HSP70's ATPase activity (Samanta et al., 2021).We found that VER-treated dividing cells fixed after 65 minutes released from the G2/M boundary retained high number of ER-FlucDM-eGFP aggregates in the cytosol compared to control treatment (Figures 5A and 5B).Time-lapse microscopy showed that when cells were treated with VER, ER-FlucDM-eGFP aggregates released from the nucleus upon entry into mitosis were not cleared as to the same extent as control cells (Figure 5C and Video 5).HSPA5/BiP is a major Hsp70 family member in regulating ER proteostasis.We next tested whether inhibition of BiP function could affect clearance of aggregates in dividing cells.Similar to VER treatment, inhibition of BiP's ATPase activity by YUM-70 (Samanta et al., 2021) led to a dosedependent accumulation of ER-FlucDM-eGFP aggregates in cells (Figures 5D and 5E).Moreover, by using immunofluorescence staining, we found a strong colocalization of BiP and ER-FlucDM-eGFP, consistent with a role of BiP in regulating ER-FlucDM-eGFP aggregates (Figures S5A and    S5B).Collectively, our data demonstrated that the clearance of ER-FlucDM-eGFP aggregates was regulated by the HSP70 family proteins, particularly BiP.

Aggregate clearance is a mitotic exit event but is independent of the anaphase promoting complex (APC/C).
Protein aggregates are subjected to proteasomal degradation in cells (Tyedmers et al., 2010).To test if aggregate clearance during cell division involves proteasomes, we inhibited the proteasome activity with MG132 in dividing cells expressing ER-FlucDM-eGFP.After treating cells released from the G2/M boundary with low or high concentrations of MG132 for 55 minutes, majority of mitotic cells were arrested at metaphase and accumulated high number of aggregates as compared to control cells treated with DMSO (Figures S6A and S6B).Next, we inhibited the proteasome with MG132 in cells at late anaphase or telophase and examined the aggregates 35 minutes after MG132 treatment (Figures S6C and S6F).In cells completing cytokinesis, there were significantly high number of aggregates in the early divided cells treated with MG132 versus the control (Figures S6D, 6E, S6G and S6H) Since protein aggregates are cleared when cells progress through mitosis and cytokinesis, which are the cell cycle stage with low Cyclin B/CDK1 activity, we speculated that inactivation of CDK1 activity in MG132-treated and metaphase-arrested cells could lead to clearance of ER-FlucDM-eGFP aggregates in the cytosol.To test if this was the case, we released cells into mitosis in the presence of MG132 for 55 minutes and treated metaphase-arrested cells with RO-3306 to inhibit Cdk1 activity (Figure 6A).Interestingly, ER-FlucDM-eGFP aggregates were cleared in cells treated with RO-3306 while the aggregates were retained in cells treated with DMSO (Figures 6B and S6I).
Consistently, live cell imaging demonstrated similar accelerated clearance in Cdk1 inhibited cells (Figures 6C and 6D).Thus, clearance of the aggregates happens when cells exit from mitosis.
We established that clearance of aggregates during cell division requires proteasomes and CDK inactivation, it is possible that the anaphase promoting complex (APC/C), which is a E3 ubiquitin ligase regulating mitotic exit (Watson et al., 2019), is involved in clearing the aggregates.To test if this was the case, we inhibited APC/C activity using a cocktail of APC/C inhibitors (APC/Ci) consists of Apcin and ProTAME in dividing cells expressing ER-FlucDM-eGFP (Sackton et al., 2014).
Interestingly, inhibition of APC/C did not affect aggregate clearance as APC/Ci-treated cells that have arrested at metaphase have comparable number of aggregates as in control DMSO-treated cells (Figures 6E and 6F).When cells at late anaphase or telophase were treated with APC/Ci, the newly divided cells have similar number of aggregates as in control cells (Figures 6G,6H and 6I).Lastly, we verified the efficacy of APC/C inhibition using cells expressing the APC/C reporter derived from Geminin (Figure S6J).Consistently, APC/Ci treatment has prevented the removal of APC/C reporter in newly divided cells indicating that the APC/C was inhibited by the dosage of APC/Ci used in our experiments (Figure S6K).Taken together, our data showed that clearance of ER-FlucDM-eGFP aggregates happens when cells exit from mitosis, is proteasome dependent but does not involve APC/C.

Discussion
We showed that stably expression of ER-FlucDM-eGFP leads to proteostatic stress and formation of protein aggregates in human cells without the use of additional proteostatic stressors.This fluorescent based live-cell imaging reporter enables us to study how protein aggregates are regulated in dividing cells, which are sensitive to genetic or environmental perturbations.
Conventional methods that use proteasomal inhibitors to cause proteome imbalance are not ideal to investigate proteostasis in dividing cells given an essential role of protesomes in driving mitotic exit (Ghislain et al., 1993;Glotzer et al., 1991).Additionally, methods such as heat shock also perturb mitotic progression, especially in human cells (Kakihana et al., 2019).By using ER-FlucDM-eGFP that does not perturb cell cycle progression, we reveal an unexpected process of clearing protein aggregates in cells progressing through mitosis and cytokinesis.
Previous studies showed that the proteome of mitotic cells have higher structural stability and less aggregation prone compared to that of in interphase cells (Becher et al., 2018;Wirth et al., 2013).
Post-translational modifications such as phosphorylation of mitotic proteins are suggested to be responsible for the protein stability during cell division (Becher et al., 2018).The aggregate clearance we identified in this study could represent an active mechanism that increases proteome stability in dividing cells.We found that aggregate clearance requires Cdk1 inactivation, which is a hallmark of mitotic exit, however, does not involve APC/C that is important in driving mitotic exit through protein degradation.It is possible that APC/C targets mostly cytosolic proteins whereas ER-FlucDM-eGFP aggregates are confined in a membrane during mitosis.Furthermore, the clearance of protein aggregates coincides with ER reorganization that happens throughout mitosis and cytokinesis, which may contribute to the aggregate clearance (Bergman et al., 2015;Schlaitz et al., 2013).
Misfolded or damaged proteins form protein aggregates and are spatially organized into specific subcellular sites to sequester them from other cellular processes (Hill et al., 2017).These aggregate deposition sites are localized in the cytosol or in the nucleus to confine misfolded proteins and aggregates for elimination from cells (Kaganovich et al., 2008).Similarly, we identified confinement of ER-FlucDM-eGFP aggregates in the nucleus and are surrounded by a single layer membrane.
These intranuclear membranous structures contain ER membranes and BiP, which is a key Hsp70 family chaperone in the ER and translocates into the nucleus under stress (Liu et al., 2023).How protein aggregates are targeted and assembled into the intranuclear membranous structure awaits future investigation.Interestingly, upon entry into mitosis, concomitant with the nuclear membrane breakdown, the membranous structure is released from the nucleus and the aggregates are cleared when cells progress though mitosis till early G1 phase of the daughter cells.The confinement and clearance of protein aggregates during cell division presumably protect dividing cells from harmful effects of protein misfolding and aggregation and to ensure reliable inheritance of genetic materials.
It has been reported that aggregates are asymmetrically retained in one of the daughter cells that are usually short-lived (Rujano et al., 2006).The aggregate clearance mechanism we report here may provide an additional cellular defense strategy to overcome proteostatic stress and to maintain proteome integrity during cell division.Hsp70 chaperones regulate proteostasis by preventing protein aggregation, promoting disaggregation and subjecting solubilized aggregates to degradation or refolding (Rosenzweig et al., 2019).A recent study showed that BiP solubilizes aggregates and drives aggregate clearance in the ER (Melo et al., 2022).Our study of mitotic aggregate clearance in dividing cells further extend the role of BiP in disaggregation.Consistently, we found that Cdk1 inactivation in the presence of the proteasomal inhibitor could still lead to aggregate clearance, suggesting a partial role of the proteosome in clearing aggregates and the disaggregation breaks down the aggregates into smaller and possibly soluble species.Interestingly, Melo et al. showed that aggregate clearance is promoted by treating cells with ER stressors as the treatment triggers UPR, which in turn leads to higher activity of BiP to cope with the ER stress.We observed differential responses of aggregate clearance to ER stressors depending on the duration of the treatments (Figure 4).It is possible that acute treatment of the ER stressor on dividing cells does not provide sufficient time for cells to accumulate abundant amount of BiP given that mitotic cells are not very transcriptionally active (Palozola et al., 2017).Also, degradation of BiP could be accelerated upon the treatment of ER stressors, which in turn undermines the abundance of BiP in disaggregation (Shim et al., 2018).Thus, short term treatment of the ER stressor in cells exacerbates the ER stress and affects aggregate clearance during cell division, whereas longer treatment of the ER stressor promotes aggregate clearance.Mechanistic understanding of aggregate clearance during cell division could offer new insights to understand disaggregation in proteostasis control.

Lentivirus packaging
Lentiviral particles were generated using the 3rd generation lentiviral packaging system.Packaging plasmids pRSV-Rev, pMDLg/pRRE, pMD2.G (kindly provided by Didier Trono) and the transfer plasmid were chemically transfected using GeneTwin transfection reagent (Biomed, TG101) into the HEK293Ta packaging cell line (Genecopoeia, LT008).The cell culture medium was harvested after 2 days and filtered through a 0.45 μm filter.The filtrate was concentrated using a 5X precipitation solution (250 g/L PEG 8,000 and 43.83 g/L NaCl in ddH2O) for overnight, and precipitated by centrifuging at 4000 ×g for 25 min at 4°C.Lentiviral particles were resuspended in phosphate buffered saline (PBS) and added to the culture medium, which were supplemented with 1 μg/mL polybrene (HanBio).

Luciferase activity test
Each well of 96-well plates (Biosharp, BS-MP-96W for luminescence, BS-MP-96B for fluorescence) was seeded with 10,000 cells in 100 μL of medium.After incubation for 18 hours, 5 μM MG132 was added to the MG132+FlucDM-eGFP group, while other untreated wells remained unchanged.
Following an additional 6 hours, cells were washed once with PBS.
For luminescence detection, cell lysis was achieved by adding 50 μL of Steady-Glo Luciferase Assay System buffer (Promega, E2510) to the wells, followed by a 15-minute incubation at room temperature.Luminescence in each well was then recorded three times over 5 minutes using the Spark® Multimode Microplate Reader (Tecan, 1000 ms exposure).Luminescence intensity in each well was calculated by normalizing these three measurements and subtracting the background signal from empty wells.
To determine relative specific luciferase activities, the luminescence values were divided by the fluorescence intensity of eGFP.Cells were immediately immersed in PBS after the PBS wash.Within 3 minutes, eGFP fluorescence intensity was measured by selecting the GFP channel of the Tecan microplate reader mentioned above.Fluorescence intensity in each well was calculated by subtracting the background signal from empty wells.

Mitotic cell preparation
To increase the quantity of dividing cells, cells were first seeded on imaging chambers (ibidi, 80826) and were synchronized to the G2/M boundary by treating cells with 7.5 μm RO-3306 (Selleckchem, S7747) for 16 to 20 hours.Cells were released to enter into mitosis synchronously by washing once with pre-warmed fresh medium.
To stain DNA, DAPI Staining Solution (Sangon Biotech, E607303, 5 mg/L) was diluted in PBS to 50 ng/mL and applied to cells for 10 minutes.After the DAPI solution was removed, fixed cells were washed with PBS twice before imaging.

Transmission Electron Microscopy (TEM)
To prepare samples for TEM, MCF10A cells expressing ER-FlucDM-eGFP were synchronized to the G2/M boundary by treating cells with RO-3306 for 16 to 20 h and were released into mitosis by removing RO-3306.After 45 min released from the G2/M boundary, cells were trypsinized for 10 min before centrifugation.The cell pellet was fixed by 2.5% paraformaldehyde (in PBS) for 30 min at room temperature before overnight at 4°C.Fixed pellet was further fixed by 1% osmium tetroxide and then 2% uranium acetate.Ethanol and acetone were used to dehydrate samples before embedding them into epoxy resin.Sectioned samples were observed by Tecnai G2 spirit 120kV transmission electron microscopy (Thermo FEI).

Live-cell staining
The live-cell staining dyes used are as follow: 100 nM SiR-Tubulin (Cytoskeleton, CY-SC002), 1 μM ER-tracker Red (Beyotime, C1041S) or 10 μM Thioflavin T (ThT) in the medium to label cells for about 30 min in the incubator before imaging.For ThT staining, about ten percent aggregates could be stained by ThT in ER-FlucDM-mCherry cells.

Spinning-disk confocal microscopy imaging
The hardware configuration of the spinning-disk confocal microscopy system was as described previously (Wang et al., 2023).Imaging utilized a z-step size of 0.5 μm, with a x-y plane resolution of 183.3 nm/pixel for the 60× lens and 110 nm/pixel for the 100× lens.Fluorophores were excited using laser lines at wavelengths of 405, 488, 561, or 640 nm.
For FRAP analysis, images were acquired with 488 nm and 561 nm laser lines through the 100× lens.The procedure involved acquiring three pre-bleach images every 30 seconds, followed by bleaching the mCherry signal using the 561 nm laser.Subsequently, one image was acquired immediately post-bleaching, followed by ten images acquired every 30 seconds.In total, fourteen time-point images were acquired, and the total duration of each FRAP experiment was 6.5 minutes.
For FRAP analysis of ER-eGFP and ER-HaloDM-eGFP, region-of-interests (ROI) in projected images were manually selected and the average fluorescence intensity of ROIs was measured in

Fiji.
For FRAP analysis of aggregates in cells co-expressing ER-FlucDM-eGFP and ER-FlucDM-mCherry (both interphase and mitosis), the aggregates were first segmented by Weka Segmentation.The unsegmented images from the mCherry channel were processed by Analyze Particles (Fiji/Analyze/Analyze Particles) based on the corresponding segmented images from the GFP channel.For fluorescence intensity quantification, background in all FRAP experiments were subtracted by one hundreds.Quantification and statistical analysis Datasets were analysed by Student's t test or Mann-Whitney test.Prism 8 (GraphPad) was usedfor the statistical analysis and data plotting.Statistical details are indicated in the figure legends.

Figure legends Figure 1 .
Figure legends

Figure 2 .
Figure 2. Differential aggregate dynamics in cells at interphase and mitosis.