Specialization of chromatin-bound nuclear pore complexes promotes yeast aging

The nuclear pore complex (NPC) mediates nearly all exchanges between nucleus and cytoplasm, and changes composition in many species as the organism ages. However, how these changes arise and whether they contribute themselves to aging is poorly understood. We show that in replicatively aging yeast cells attachment of DNA circles to NPCs drives the displacement of the NPCs’ nuclear basket and cytoplasmic complexes. Remodeling of the NPC resulted from the regulation of basket components by SAGA, rather than from damages. These changes affected NPC interaction with mRNA export factors, without affecting the residence of import factors or engaging the NPC quality control machinery. Mutations preventing NPC remodeling extended the replicative lifespan of the cells. Thus, our data indicate that DNA circles accumulating in the mother cell drive aging at least in part by triggering NPC specialization. We suggest that antagonistic pleiotropic effects of NPC specialization are key drivers of aging.


Introduction 1
Nuclear pore complexes (NPCs), which mediate transport of cargos between the nucleus and the 2 cytoplasm, undergo substantial changes during aging from yeast to mammals (Rempel et al., 2020). 3 In post-mitotic cells such as neurons, core NPC components are very stable, tend to become oxidized 4 over time and progressively lose functionality (Savas et al., 2012;Toyama et al., 2013). Accordingly, 5 in the neurons of old rats NPCs become increasingly leaky with age, affecting the proper retention of 6 nucleoplasmic proteins in the nucleus (D' Angelo et al., 2009). On the opposite, yeast NPCs are 7 targeted by a quality control machinery that removes damaged or misassembled NPCs (Webster et 8 al., 2014). Accordingly, no oxidized NPCs are observed in replicative aging yeast cells (Rempel et al., 9 2019). Still, proteomics analysis of the old yeast cells indicate that they progressively change 10 composition, losing their nuclear basket and cytoplasmic complexes (Janssens et  variety of cellular processes beyond nucleocytoplasmic exchange, from the regulation of gene 17 expression to DNA repair, it is plausible that defects in NPC function could affect cellular physiology 18 and contribute to the ageing process. However, a clear view of how age affects NPCs and their 19 functionality has been lacking so far, precluding a robust understanding of whether and how NPCs 20 indeed contribute to the aging process. 21 The budding yeast Saccharomyces cerevisiae undergoes both post mitotic and replicative aging 22 (Longo et al., 2012). In the first case, also called chronological aging, starved yeast cells progressively 23 lose viability, with their mortality increasing exponentially with time (Fabrizio and Longo, 2003). In 24 the second case, cells that proliferate in rich nutrient conditions divide asymmetrically through the 25 budding of rejuvenated daughters off the surface of their aging mother cell, which retain and 26 accumulate diverse aging factors (Mortimer and Johnston, 1959; reviewed in Denoth Lippuner et al., 27 2014). After undergoing 20-30 divisions and generating as many daughter cells, these mother cells 28 stop dividing and ultimately die. Although a number of the involved aging factors and some of the 29 mechanisms of their retention in the mother cell have been characterized, little is known about how 30 they affect the viability of old mother cells. Guarente, 1997)), which are generated in virtually all mother cells at some point in their lifespan 33 through excision of one or more rDNA repeat(s) from the rDNA locus, are a particularly prominent 34 aging factor. These episomes lack a centromere and are very efficiently retained in the mother cell 35 at mitosis. As they replicate once per division cycle, they accumulate exponentially in the nucleus of 36 the old mother cell over time. At the end of their life, the mother cell may contain up to thousand 37 ERCs, which represent about as much DNA as the rest of the genome (Denoth-Lippuner et al., 2014; 38 Morlot et al., 2019). Several lines of evidence establish that ERCs accumulation promote aging of 39 the cell. Mutants that form ERCs at a reduced rate, such as the fob1∆ mutant cells, show an 40 extended replicative lifespan (RLS; (Defossez et al., 1999). In reverse, cells in which recombination in 41 the rDNA is derepressed, such as cells lacking the sirtuin Sir2, show a higher rate of ERC formation, 42 accumulate them faster and are short-lived (Kaeberlein et al., 1999). Finally, artificially introducing 43 an ERC (Sinclair and Guarente, 1997) or any other replicating DNA circle in a young cell (Denoth-44 Lippuner et al., 2014) causes premature aging. However, how ERCs promote aging is not known. 45 The tight retention of the ERCs in the mother cell depends on their attachment to NPCs  Thus, these accumulating NPCs may contribute to the mechanisms by which ERCs promote aging of 20 the cell. 21 The nuclear pore complex is a ~100 megadalton entity comprising repetitions of more than 30 22 different subunits, called nucleoporins (short Nups) ( S-phase and remains confined in the mother cell upon cell division, where it  1  accumulates as the cell ages through successive rounds of cell division. Within 8-10 division cycles,  2  the accumulating DNA circles cluster together to form a bright patch at the nuclear periphery. The  3 attached NPCs accumulate with them in the nuclear envelope adjacent to the cluster (Denoth-4 Lippuner et al., 2014). Upon fusion of GFP to core nucleoporins, these NPCs are visible as a cap of 5 enhanced fluorescence density (Fig. 1C, Figure 1-supplement 1A), different from young cells without 6 accumulation of DNA circles, where nucleoporins are rather homogenously distributed in the nuclear 7 envelope (Figure 1-supplement 1B). In the few cells that accumulate the circle in the population, this 8 systems conveniently allows imaging circle-bound NPCs at single cell level, to discriminate them 9 from unbound NPCs in the remainder of the same nuclear envelope. 10 Thus, we used this engineered DNA circle to quantify the local enrichment of single nucleoporins in 11 the NPC cap compared to the rest of the nuclear envelope, to study the composition of DNA circle 12 bound NPCs. We tagged 17 nucleoporins with GFP, representative of different NPC subcomplexes, 13 in cells where the DNA circle cluster was labelled in red, owing to the expression of TetR-mCherry 14 ( Fig. 1B, C). In these images, all the stable core Nups, i.e. the scaffold components (outer ring: 15 Nup84, Nup133; inter ring: Nup170, Nic96) and the components of the transport channel (FG-Nups: 16 Nup53, Nup59) accumulated in the cap to similar extents. Quantification of their indicated a median 17 of 2.4-fold enrichment, compared to their localization elsewhere in the nuclear envelope; Fig. 1D, E). 18 All tested core nucleoporins showed the same enrichment in the cap covering the DNA circles, 19 indicating that the circles bind intact NPC cores. The enrichment level of core Nups in the cap served 20 therefore as reference to determine whether peripheral Nups were stoichiometrically associated 21 with core NPCs bound to circles or depleted from those pores. 22 In striking contrast to core Nups, most peripheral subunits on both sides of the NPC were found to 23 not accumulate in the cap. Four out of the five components of the nuclear basket and four out of the 24 six components of the cytoplasmic complexes were excluded from the cap (Fig. 1C The basket and cytoplasmic complexes are displaced from NPCs in wild type cells aging under 3 physiological conditions. 4 To evaluate whether remodeling of the circle-bound NPCs was indeed reflecting what happens in 5 cells undergoing unperturbed aging, we next asked whether the basket and cytoplasmic complexes 6 dissociates from NPCs of old cells not carrying our reporter circles. For some reason, accumulating 7 ERCs are rather dispersed throughout the nuclear periphery and form clusters only episodically, such 8 that NPC caps are less prominently observed in these cells. Clustering of the reporter circle might be 9 stabilized by the fluorescent label (mCherry still has a low affinity for itself) or ERCs have means to 10 escape clustering. In order to study if NPC remodeling happens under these conditions, we co-11 labelled pairs of Nups with distinct fluorophores and characterized their co-incorporation into NPCs 12 by analyzing the spatial correlation of their fluorescence in the nuclear periphery as the cells aged 13 ( Fig. 2A). Indeed, the signals of the two labelled Nups should correlate well with each other as long 14 as both colocalize to NPCs, and poorly if any one of them is displaced from NPCs ( Fig. 2A) . 15 In order to acquire images of old mother cells, the labelled cells were loaded on a microfluidic chip 16 (Jo et al., 2015) and imaged for 26 hours (21 divisions in average). A continuous flow of fresh 17 medium provided nutrients to the trapped mother cells and flushed their daughter cells out, 18 allowing the continuous imaging of the isolated mother cells. Bright field images were taken every 19 15 minutes to monitor budding events and record the replicative age of each cell. The ERCs mediate the displacement of the basket from NPCs of old mother cells 38 To determine whether this remodeling of the NPC upon aging is driven by the accumulation and 39 attachment of endogenous DNA circles, we first asked whether delaying ERC accumulation also 40 delayed the removal of the basket from NPCs. Thus, we deleted the FOB1 gene, which promotes ERC 41 formation, and tested whether this restored the presence of the basket components Nup60 and 42 Mlp1 at NPCs of old mutant cells (Fig. 3A). Strikingly, no significant dissociation of the basket from 43 NPCs was observed in old fob1∆ mutant cells compared to wild type at the same age (same number 44 of divisions) or to young fob1∆ mutant cells (Fig. 3A). Moreover, deleting the SGF73 gene, which 1 docks SAGA to NPCs and mediates circle anchorage to nuclear pores, also restored the localization of 2 the basket to NPCs in aged cells (Fig. 3B). Thus, ERC presence and attachment is required for the 3 basket to be displaced from NPCs. 4 Finally, as ERC anchorage displaces the basket from a substantial fraction of the NPCs in old wild 5 type mother cells, their daughters, which do not inherit ERCs are expected to rapidly restore the 6 proper localization of the basket proteins. Indeed, when comparing signal correlation between the 7 basket proteins Nup60 or Mlp1, with the core Nup, Nup159, colocalization between basket and core 8 nucleoporins was extensive in the rejuvenated daughter cells of old mothers, similar to what is 9 observed in both the young mothers and their daughters (Fig. 3C, D). The correlation between 10 basket and core Nups was only reduced in aged mothers. Altogether, we conclude that the 11 formation of ERCs and their attachment to NPCs drives the displacement of the nuclear basket from 12 pores. ERC accumulation is a direct cause for the displacement and loss of the nuclear basket from 13 NPCs in physiologically aging yeast cell. This event probably mediates the subsequent displacement 14 and loss of the cytoplasmic complexes (see below). 15 we next asked whether the attachment of DNA circles and basket removal induces a defect that 21 would be recognized as damage by the cell. To address this question, we investigated whether the 22

The cells do not recognize remodeled NPCs as defective
circle-bound NPCs recruit Chm7 more frequently than bulk NPCs. We recorded the localization of 23 Chm7, tagged with GFP, in cells loaded with DNA circles and asked whether it accumulated in the 24 cap (Fig. 4A). The cells containing a cluster of DNA circles were categorized in the following classes: 25 1) the cells showing at least one Chm7 focus in the NPC-cap, 2) those showing a Chm7 focus 26 somewhere else in the nuclear envelop and 3) those showing no visible Chm7 foci in the nuclear 27 envelope (Fig. 4A). Interestingly, most cells formed either no Chm7 focus at the nuclear periphery 28 (category 3, 41% of cells, n = 190 cells with circle cluster) or the focus formed was not associated 29 with the cluster of DNA circles (category 2, 23% of the cells). Only about a third of the cells (category 30 1, 36%) formed a Chm7 focus adjacent to the circle-cluster, i.e. where the NPC-cap is located. Note 31 that these Chm7-labeled foci are much smaller than the NPC-cap (Fig. 1C, 4A), indicating that if any, 32 only few of the NPCs in the cap are targeted by this machinery. Importantly, based on fluorescence 33 intensity measurements using Nup84-GFP in DNA circle loaded cells (Fig. 1C), we estimate that about 34 45% (+/-12%, n = 10) of the NPCs is sequestered in the NPC-cap at that stage. Thus, this analysis did 35 not reveal any substantial enrichment of Chm7 overlapping with the clusters of DNA circle; the 36 occurrence of Chm7 in the NPC cap seemed to be rather coincidental. We concluded that the circle-37 bound NPCs are not detected as defective by the cell more or less than the other NPCs in the rest of 38 the nuclear envelope. 39 The remodeled NPCs are not particularly old 40 Since the anchorage of a DNA circle to an NPC causes its retention in the mother cell, we next 41 wondered whether this could cause the progressive accumulation of older NPCs in aged mother 42 cells. To test this possibility, we measured the relative age of NPCs in old mothers and their 43 daughter cells using the tandem fluorescent protein timer, consisting of mCherry (mCh) and 44 superfolder GFP (sfGFP) (Khmelinskii et al., 2012). Due to different maturation kinetics between the 45 two fluorophores, a newly synthesized protein appears first in the green channel before acquiring 1 the red fluorescence over time. As the turnover rate of Nup170 is very low in NPCs (D'Angelo et al.,  2 2009), older pores with tagged Nup170 are expected to emit more red fluorescence than green in 3 comparison to newly assembled pores. To see if old mothers are enriched in red-shifted old pores, 4 we loaded the cells expressing Nup170-mCh-sfGFP on the microfluidic chip and imaged them as 5 above (Fig. 2B) as the cells aged (Jo et al., 2015). The fluorescence channels were recorded after 2 6 and 26h. We did observe a tendency for young cells to put slightly more red shifted NPCs in the bud 7 than in the mother cell (Fig. 4B, not statistically significant, n=31 mother-bud pairs), as was shown 8 before (Khmelinskii et al., 2012). Over time we observed a highly significant increase of the red 9 fluorescence signal relative to the green signal in old compared to young cells (n=29 old cells, 10 p<0.001), indicating that old cells actually accumulate old pores over time. Although we observed a 11 trend for the NPCs of the daughters of old mother cells to be slightly younger, indicative for some 12 retention of old pores in the mother cell, this difference was not significant in our analysis (Fig. 4B). 13 Thus, together these data make three points. First, the data support earlier findings indicating that 14 pre-existing NPCs in young cells are not particularly retained in the mother cell and even that their 15 segregation might be biased towards the daughter cell (Khmelinskii et al., 2012). Second, the data 16 establish that over time the mother cell does accumulate older NPCs, possibly with the accumulation 17 of ERCs. Third, these older NPCs are however not tightly retained in the old mother cell. As circle-18 bound NPCs are retained in the mother cell ( nuclear envelope, it does happen. Thus, we concluded that although circle-bound NPCs 23 accumulating in mother cells are rather old in average, they are not significantly older than those of 24 the rejuvenated daughter cells. Thus, together our data do not support the notion that the age of 25 the old mother NPCs drives their remodeling. 26

Nucleoporin acetylation promotes NPC remodeling 27
Since DNA circles attach to NPCs via the SAGA complex, we next wondered whether basket removal 2014, 2009). Therefore, we asked as a proof of principle whether preventing this modification by 38 substituting lysine-467 by arginine was sufficient to abrogate the exclusion of Nup60 from the cap 39 (Fig. 1C, E). Strikingly, not only Nup60-K467R now accumulated in the cap nearly as much as core 40 nucleoporins, but the cells expressing this protein also restored the localization of the cytoplasmic 41 complexes (Nup116, Nup42, Gle1 and Gle2) to cap NPCs (Fig. 5B, C). In contrast, the basket 42 component Mlp1 remained displaced, indicating that the anchorage of DNA circles might also lead to 43 other modifications in different basket proteins, independently of that of Nup60 at lysine-467. Thus, 44 we concluded that the acetylation of Nup60 on K467 upon circle attachment drives Nup60's 45 displacement from NPCs and subsequently that of the cytoplasmic complexes as well. How Nup60's 46 presence at the nuclear pore regulates the recruitment or stabilization of the cytoplasmic complexes 1 on NPCs is not known at this stage. Possibly a signal is transduced from the nuclear to the 2 cytoplasmic side of the NPC, or the clustering of the pores could drive the displacement of the 3 cytoplasmic complexes. Furthermore, additional events displace the Mlp1/2 proteins as well, 4 possibly also through their own acetylation or that of Nup2. 5 Thus, altogether we found no indication for the anchorage of DNA circles causing NPC deterioration. 6 Rather, our data indicate that circle anchorage promotes the displacement of the nuclear basket 7 through post-translational modifications, and that this in turn causes the detachment of the 8 cytoplasmic complexes. Considering the fact that Nup60 acetylation contributes to the regulation of 9 gene expression (Kumar et al., 2018), and that the residence of the basket to NPCs is dynamic also in 10 young cells, the removal of the basket upon circle attachment might reflect a common physiological 11 process taking place each time chromatin interact in a SAGA-dependent manner with interphase 12 NPCs. The fact that DNA circles remain attached to NPCs during mitosis, whereas chromosomes 13 detach from the nuclear periphery at that stage (Kanoh, 2013), might be caused by their inability to 14 undergo condensation and recruit the deacetylase Hst2 (Kruitwagen et al., 2018). Further studies 15 will be needed to determine the role of acetylating other Nups beyond Nup60 in basket 16 displacement, circle anchorage and aging, and whether Hst2 reverts the acetylation of these 17 proteins at chromosome-attached NPCs during mitosis. 18

Basket displacement promotes aging 19
Thus, together our results indicate that DNA circles modulate the organization of NPCs as they 20 attach to them, leading to the accumulation of remodeled NPCs as circles accumulate and the cell 21 progresses in replicative aging. Therefore, we wondered whether the effect of DNA circles on NPC 22 organization is at least a part of the mechanisms by which ERCs promote cellular aging. Thus, we 23 next investigated whether interfering with NPC remodeling has an impact on the lifespan of the 24 cells. We took advantage of a new microfluidic chip design, that could retain more efficiently cells 25 during their entire lifespan (i.e. >95% of the cells were kept until cell death (Fig. 6A, B showed an extended longevity (20 divisions, Fig. 6C functional relevance of this observation is unknown but has suggested that these NPCs might be 40 functionally specialized. Thus, we suggest that the accumulation of basket-less NPCs with age does 41 not reflect an accumulation of defective NPCs per-se, but rather leads to an imbalance of specialized 42 over non-specialized NPCs in old cells (Fig. 6D). 43 mRNA export and mRNA surveillance factors are specifically displaced from circle-bound NPCs 44 Thus, we reasoned that NPC specialization might have some regulatory effect on their function. 1 Nuclear pore complexes mediate the transport of cargos between nucleoplasm and cytoplasm and 2 therefore transport factors transiently localize to NPCs in young and healthy cells (Derrer et al.,  3 2019; Kumar et al., 2002). Any effect of NPC remodeling on the localization of these factors to NPCs 4 may reflect changes in their dynamics within NPCs and hence, on NPC functionality. Therefore, we 5 characterized how circle anchorage affected the recruitment of transport factors and other 6 associated proteins to NPCs. We labeled a broad panel of transport and associated factors with GFP 7 and quantified their nuclear localization in respect to the DNA circle cluster, as in Fig. 1 characterized. Consistent with circle-bound NPCs lacking a basket, the two basket-associated 10 proteins Esc1 and Ulp1 (both involved in telomeric silencing and mRNA surveillance, (Bonnet et al.,11 2015)) were excluded from NPC caps (Fig. 7A, B). In striking contrast, none of the 10 importins tested 12 were displaced from the cap (Fig. 7B). Likewise, the exportins Xpo1 (Crm1), Msn5 and Los1, which 13 ensure the export of proteins and tRNAs, accumulated with the cap to similar extent as the core 14 Nups. Since nearly all these proteins accumulated to the same extent as core NPCs in the cap 15 compared to elsewhere in the nuclear envelope, we concluded that basket-less pores interact with 16 them with similar dynamics as bulk NPCs. Two importins were in average significantly further 17 enriched at the cap, namely Srp1/Kap60 and Kap123. This may indicate either that these two 18 nucleoporins shuttle more intensely through basket-less NPCs, or that they linger longer in them. 19 Strikingly, Kap123 mediates the nuclear import of ribosomal proteins, promoting their subsequent 20 assembly into ribosomes, and the import of histones H3 and H4. Both ribosome and nucleosome 21 assembly have been shown to become reduced in aged yeast cells (reviewed in (Matos-Perdomo  22 and Machín, 2019)). 23 More strikingly, 5 exportins were specifically depleted from the cap (Fig. 7B), indicating that their 24 interaction with core NPCs was substantially decreased. Interestingly, all of these exportins are 25 involved in mRNA export. Thus, all seven proteins that we find to be excluded from the cap are 26 involved in this process (Bonnet et al., 2015;Iglesias et al., 2010;Stewart, 2010). We postulate that 27 circle-bound NPCs are specifically inhibited for their function in mRNA export. 28

Conclusion 29
In summary, our data indicate that old yeast cells accumulate an increasing proportion of NPCs 30 depleted of the nuclear basket and cytoplasmic complexes and that may have a reduced capacity for 31 mRNA export. Once the proportion of basket-associated NPCs goes below some threshold, the 32 resulting imbalance become deleterious for the cell. Importantly, our study establishes that the 33 stoichiometry changes observed with age are not due to NPC deterioration but rather to their 34 specialization as they become increasingly decorated with DNA circles, mainly ERCs. The function of 35 this specialization is not known but observed also in young cells, for example near the nucleolus. 36 Whether nucleolus-associated NPCs are depleted of other factors, such as the cytoplasmic 37 complexes, has not been reported but our study suggests that it might be the case. Furthermore, 38 the dissociation of the basket from NPCs appears to have broad functions beyond nucleolar NPCs in 39 young cells since the acetylation of Nup60 is involved in the regulation of diverse loci on 40 chromosomes, such as the activation of cell cycle genes (Kumar et al., 2018), and the SAGA complex 41 is involved in the rapid activation of many genes in response to environmental changes such as heat 42 shock, inositol starvation or changes in the available carbon source (Huisinga and Pugh, 2004; 43 Kremer and Gross, 2009). In many of these instances, SAGA-dependent regulation involves the 44 recruitment of the target locus to the nuclear periphery and their anchorage to NPCs. Thus, the 45 modifications observed at circle-bound NPCs reflect physiological changes that are common in 46 young cells and amplified through ERC accumulation in old cells. Future studies will require to 1 determine how this imbalance affects cellular viability. 2 The notion that the age-dependent remodeling of NPC is not due to damages but to regulatory steps 3 has consequences for our understanding of the evolution of aging. Indeed, a leading theory for the 4 apparition of the aging process in evolution is that it is the result of traits and processes that are 5 selected for their selective advantage early in life, despite of deleterious effects later-on (Kirkwood 6 and Rose, 1991). Antagonistic pleiotropy has been difficult to document beyond the well accepted 7 idea that sparing on quality control and damage-repair liberates resources for the generation of 8 progeny early in life, at the expense of longevity (Ackermann et al., 2007;Austad and Hoffman, 9 2018; Williams, 1957). The case of NPCs and their role in SAGA-dependent gene regulation might 10 depart from this notion by suggesting that the trade-off here is not a matter of quality control but of 11 adaptability. Indeed our data suggest that the remodeling of NPCs with age is a secondary 12 consequence of the role of chromatin-NPC interaction in the rapid and strongly response of cells to 13 environmental challenges. There is currently no evidence for DNA circles contributing to aging in other organisms but there is 20 strong evidence that chromatin interaction with the nuclear periphery is affected by age in many cell 21 types. Most remarkably, progerin, a progeriatric isoform of Lamin A and causing the Hutchinson-22 Gilford progeria syndrome in humans, affects the recruitment of heterochromatin to the nuclear Thus, we suspect that the effects of aging on NPCs and the role of NPCs in aging might be strikingly 26 similar between yeast and mammals. 27

Materials and methods 1 -Strains and plasmid 2
All the yeast strains and plasmids used in this study are listed in table S1 and are isogenic to S288C. 3 GFP-tag and knock-out strains were generated using classical genetic approaches (Janke et al., 4 2004)). All cultures were grown using standard conditions, in synthetic drop-out medium (SD-5 medium; ForMedium, Norfolk, UK) or indicated otherwise, at 30°C. 6 The non-chromosomal DNA circle was obtained from the Megee lab (Megee and Koshland, 1999) 7 and contains an array of 256 TetO repeats, the centromere is flanked by target site for the R- For fluorescent microscopy, yeast cells were precultured for minimally 24h in synthetic drop-out 27 medium. 1 ml of cells from exponential growing cultures with OD<1 were concentrated by 28 centrifugation at 1.000xG, resuspended in ~5 ul of low fluorescent SD-medium, spotted on a round 29 coverslip and immobilized with a SD/agar patch. The cells were imaged in z-stacks of 6 slices with 0.5 30 μm spacing, with a 100×/1.4 NA objective on a DeltaVision microscope (Applied Precision) equipped 31 with a CCD HQ2 camera (Roper), 250W Xenon lamps, Softworx software (Applied Precision) and a 32 temperature chamber set to 30°C. 33 To accumulated DNA circles in the nuclei of aging mother cells, yeast cells were pre-cultured for 24h 34 in SD-URA at 30°C and then shifted to SD-LEU medium supplemented with 1 µM β-Estradiol the lifespan analyses, a chip with a new cell trapping design was used (Fig. 6A, B), to ensure excellent 2 retention of old cells (see below). 3 To start the experiment, yeast cells were pre-cultured for 24h in SD-full supplemented with 0.1% 4 Albumin Bovine Serum (protease free BSA; Acros Organics, Geel, Belgium). Young cells from a 5 exponentially growing culture were captured in the traps of the microfluidic chip; the chip was 6 continuously flushed with fresh medium at a constant flow of 10 μl/min, using a Harvard PHD Ultra 7 syringe pump (Harvard Apparatus, Holiston, MA, USA) with two or four 60mL BD syringes, with inner 8 diameter 26.7 mm (Becton Dickinson, Franklin Lakes, NJ, USA). Bright field images were recorded 9 every 15 min. throughout the duration of the entire experiment. To measure the nucleoporin 10 colocalization or pre-mRNA translation, fluorescent images only after 2h, 12h, 26h or/and 50h. For 11 imaging we used an epi-fluorescent microscope (TiE, Nikon Instruments, Tokyo, Japan) controlled by 12 Micro-Manager 1.4.23 software (μManager, PMID 25606571), with a Plan Apo 60x 1.4 NA objective. 13 For fluorescence illumination of the GFP and mCherry labeled proteins, a Lumencor Spectra-X LED 14 Light Engine was used. Stacks of 7 slices with 0.3 μm spacing were recorded during. The age of the 15 cell was defined by the number of daughter cells that emerged during the budding cycles. 16 For the nucleoporin correlation analysis, a cell of interest was manually selected if it stays in the 17 focal plane in the bright field channel. Its age was determined and a segmented line was drawn 18 through the nuclear envelope in an image in the focal plane, using Fiji/ImageJ 1.51n (Schindelin et 19 al., 2012), and the intensity profiles were recorded for both fluorescence channels. The Pearson 20 correlation between the intensity profiles was calculated and plotted in R (R Development Core 21 Team, 2011). 22 For the tandem fluorescent protein timer analysis, late mitotic cells were selected after 2h and 26h 23 incubation in the chip. Its age was determined and a segmented line was drawn through the nuclear 24 envelope in an image in the focal plane, as described above. The average background corrected 25 intensity for GFP and mCherry was calculated and plotted in R. 26 To obtain a reliable lifespan curves, the majority of the cells should be retained until cell death to 27 prevent biasing the data. Although different microfluidic dissection platforms have been developed, 28 it is still a challenge to reach high enough retention efficiency in the microfluidics chip for life span 29 analysis. Here we used an improved design of yeast cell traps, having small "claws" at both sides, 30 preventing the escape of bigger cells at higher age (Fig. 6A, B). This allowed us to retain >95% of the 31 cells during their full lifetime. Only bright field images were recorded every 15 min. throughout the 32 entire experiment of 70-80 hours. All cells in a field of view were analyzed, the replicative lifespan 33 was determined for each single cell by counting the budding cycles before cell death.    6 log2-scale. The median accumulation of the core nucleoporins is indicated (green line). p-value 7 stands for student's t-test between the specific nucleoporin and pooled data of all core nups 8 together, no p-value is indicated if the difference is not significant; the sample size per strain is 9 indicated (n). and a mCherry-tagged Nup as a reference (e.g. Nup159-mCherry, see material and methods). The 6 Pearson correlation between the two intensity profiles is calculated, and used as a measure for 1 nucleoporin colocalization. (B) Fluorescent images of young and old cells in the yeast aging chip, with 2 nucleoporins labeled with GFP (green) and the reference Nup159 with mCherry (red). The age of 3 each cell is indicated. Scale bars are 5 µm. (C) Quantification of the degree of colocalization between 4 target and reference nucleoporin is plotted for young and old wt. The p-value stands for the 5 student's t-test between young and old cells. The sample size (n) and the median age is indicated. 6 (D) Fluorescent images of young and aged cells in the yeast aging chip, as in A, but with Nup84-7 mCherry as a reference nucleoporin. Scale bars are 5 µm. (E) Quantification of the co-localization 8 between nucleoporins in young and old cells, as in D, but with Nup84-mCherry as a reference. 9 1 Figure 3. Endogenouse DNA circles drive nuclear basket displacement in wild type aged cells. (A)  2 Quantification of the degree of colocalization between target and reference nucleoporin is plotted 3 for young and old wt, as was done in Fig. 2C, but in fob1Δ cells. The sample size (n) and the median 4 age is indicated. The p-value stands for the student's t-test between young and old cells. (B) Same as 5 A, but in sgf73Δ cells. (C) Fluorescent images of young and old mitotic cells in the yeast aging chip. 6 Scale bars are 5 µm. (D) Same as in Fig. 2C, but comparing nucleoporin colocalization in young and 7 old mother (M) with their corresponding daughter (D) cell. The p-value stands for the student's t-test 8 between mother and daughter cell. The sample size (n) and the median age is indicated. 9 10 nuclei in yeast cell with accumulated DNA circles labeled with TetR-mCherry and Chm7-GFP. 12 Occurance of Chm7-GFP localization is categorized 1) at least one Chm7 focus in the NPC-cap, 2) a 13 Chm7 clusters are labeled with TetR-mCherry (red). Scale bar is 2 µm. Same as A, but with Nup60-GFP and 4 nup60-K467R-GFP. (C) Quantification of GFP-labeled nucleoporin accumulation in the cap, in nup60-5 K467R (brown lines), compared to wt (black lines) on a log2-scale, as in Fig. 2B. Wt data is a copy 6 from Fig. 2B. The p-value stands for the student's t-test between accumulation ratio of a specific 7 nucleoporin in wt and nup60-K467R, no p-value is indicated if the difference is not significant. The 8 sample size (n) is indicated. accumulation is plotted for associated factors (grey), import factors (red) and export factors 6 (orange), and is compared to accumulation of the pooled NPC core subunits NPC (green line, 7 duplicated from Fig. 2B), no p-value is indicated if the difference is not significant. The sample size 8 (n) is indicated. 9