Intrinsically Disordered SRC-3/AIB1 Protein Undergoes Homeostatic Nuclear Extrusion by Nuclear Budding While Ectopic Expression Induces Nucleophagy

SRC-3/AIB1 (Amplified in Breast Cancer-1) is a nuclear receptor coactivator for the estrogen receptor in breast cancer cells. It is also an intrinsically disordered protein when not engaged with transcriptional binding partners and degraded upon transcriptional coactivation. Given the amplified expression of SRC-3 in breast cancers, the objective of this study was to determine how increasing SRC-3 protein levels are regulated in MCF-7 breast cancer cells. We found that endogenous SRC-3 was expelled from the nucleus in vesicle-like spheres under normal growth conditions suggesting that this form of nuclear exclusion of SRC-3 is a homeostatic mechanism for regulating nuclear SRC-3 protein. Only SRC-3 not associated with CREB-binding protein (CBP) was extruded from the nucleus. We found that overexpression in MCF-7 cells results in aneuploid senescence and cell death with frequent formation of nuclear aggregates which were consistently juxtaposed to perinuclear microtubules. Transfected SRC-3 was SUMOylated and caused redistribution of nuclear promyelocytic leukemia (PML) bodies and perturbation of the nuclear membrane lamin B1, hallmarks of nucleophagy. Increased SRC-3 protein-induced autophagy and resulted in SUMO-1 localization to the nuclear membrane and formation of protrusions variously containing SRC-3 and chromatin. Aspects of SRC-3 overexpression and toxicity were recapitulated following treatment with clinically relevant agents that stabilize SRC-3 in breast cancer cells. We conclude that amplified SRC-3 levels have major impacts on nuclear protein quality control pathways and may mark cancer cells for sensitivity to protein stabilizing therapeutics.


Introduction
Nuclear steroid receptor coactivator-3 (SRC-3/AIB1/NCOA3) was originally identified as a gene amplified in nearly 10% of breast cancers [1] but has been shown to be overexpressed in 50-60% of cases with weak to absent expression in normal mammary tissue [2]. SRC-3 bound to liganded nuclear receptors on chromatin recruits histone acetyltransferases (HATs) including p300/CBP and p/CAF to the complex [3]. SRC-3 coactivates nuclear receptors including the estrogen receptor (ER) [4] as well as other transcription factors [5] and is necessary for the normal development of the mammary gland [6]. While SRC-3 correlates with expression of the ER and PR in breast tumors [7] it can also be overexpressed in ER-alpha-positive (ER + ) tumors [8].
Both the function and protein levels of SRC-3 are rigorously regulated. Multiple phosphorylations mediated by several different kinases regulate both SRC-3 function and degradation at several

MCF-7 Cells Extrude Endogenous SRC-3 through Nuclear "Budding" and Altered SRC-3: CBP Stoichiometry Regulates Nuclear Localization and Expulsion: Many Breast Tumors Express Amplified Levels of SRC-3
Among available human breast cancer cell lines, only MCF-7 and BT474 [25] demonstrate SRC-3 gene amplification and express high levels of the protein while the balance of breast cancer cell lines tested (10 ER + and one ER-negative (ER − )) have no gene amplification and moderate to low SRC-3 protein expression [26,27]. For this reason, we chose to use MCF-7 cells for these studies. Mechanisms controlling the nuclear content of this IDP outside of transcription-coupled degradation are not known. Since large complexes may undergo nucleo-cytoplasmic transfer through nuclear pore-independent mechanisms we first examined SRC-3 localization by IF of MCF-7 cells. IF of MCF-7 cells for SRC-3 indicated variation in expression across the population and with some cells displaying higher levels than others. Remarkably, a small proportion of cells (approximately 5% of MCF-7 cells) demonstrated large SRC-3 containing vesicle-like budding nuclear protrusions ( Figure 1A). Thus SRC-3 appears to undergo nuclear protein control through nuclear budding. We also analyzed SRC-3 by IF in T47-D and ZR-75-1 breast cancer cell lines which express less SRC-3 relative to MCF-7 cells [26,27]. Even with lower SRC-3 protein content, both T47-D and ZR-75-1 lines exhibited nuclear protrusions/blebs containing SRC-3 (Supplementary Figure S1). Analysis of anti-SRC-3 IHC staining of human breast cancer sections from the Protein Atlas collection (www.proteinatlas.org/ENSG00000124151-NCOA3/pathology/breast+cancer) also showed evidence of SRC-3-positive extranuclear budding at a frequency ranging from 2-5% of cells (Supplementary Figure S2). Therefore nuclear clearance of SRC-3 appears to involve a nuclear budding mechanism.
As discussed above SRC-3 binds and synergistically folds with the histone acetyltransferase CBP in the process of initiation of transcription of target genes. If excess SRC-3 exceeds the capacity of CBP, then misfolded SRC-3 might form aggregates and/or require nuclear sequestration or nuclear exclusion. To assess whether overexpressed SRC-3 colocalizes with CBP, we first performed IF after transfection of SRC-3 (tSRC-3) alone. Transfected SRC-3 formed large aggregates, only small regions of which overlap with endogenous CBP (eCBP) (Figure 1, panel i). When both CBP and SRC-3 were overexpressed in the same cell more co-aggregates were observed however SRC-3 aggregates not associated with CBP were perinuclear and underwent nuclear extrusion (Figure 1, panel ii). Thus, excess SRC-3 not associated with CBP forms independent aggregates that are expelled from the nucleus while large nuclear bodies containing CBP and SRC-3 are not, consistent with nuclear retention of folded complexes and expulsion of unfolded SRC-3. This result also indicates a requirement for stoichiometric nuclear expression of SRC-3 folding partners such as CBP, necessary to prevent SRC-3 aggregation. In panel (i) Z-stack images show SRC-3 aggregates overlapping with diffuse CBP (red arrows). CBP not associated with SRC-3 is indicated by the green arrow. In panel (ii) some aggregates merge as CBP and SRC-3 foci (white arrows), while others near the nuclear periphery contain only SRC-3 (red arrows) (100 × magnification).

Overexpressed SRC-3 Induces Cell Cycle Arrest, Senescence, and Apoptosis
Since it appeared that SRC-3 nuclear levels are regulated in part through vesicle-like budding we tested the effect of increasing SRC-3 protein in MCF-7 cells. However, attempts at establishing MCF-7(SRC-3) clones largely failed as cells did not survive the expansion process. We only were able to isolate six clones that survived through limited proliferation expressing varying levels of SRC-3 protein and consisting of small to medium-sized cells (Figure 2A). Relative to MCF-7(SRC-3) clones, on the same blot endogenous 160kDa SRC-3 was not detected at this exposure. Low molecular weight degradation products were clearly visible and are associated with turnover of SRC-3 in conjunction with the estrogen receptor in rapidly dividing cells cultured in estrogenic media [9]. However after passage, small and medium-sized clones expressing SRC-3 grew very slowly (see below) with various levels of SRC-3 degradation. Clone M3 appeared to have ceased expression of SRC-3 possibly due to the formation of senescence-associated heterochromatin (see below). Relative expression of SRC-3 protein in control and transfected colonies isolated at clonal selection is shown in Figure 2B. Previous In panel (i) Z-stack images show SRC-3 aggregates overlapping with diffuse CBP (red arrows). CBP not associated with SRC-3 is indicated by the green arrow. In panel (ii) some aggregates merge as CBP and SRC-3 foci (white arrows), while others near the nuclear periphery contain only SRC-3 (red arrows) (100× magnification).

Overexpressed SRC-3 Induces Cell Cycle Arrest, Senescence, and Apoptosis
Since it appeared that SRC-3 nuclear levels are regulated in part through vesicle-like budding we tested the effect of increasing SRC-3 protein in MCF-7 cells. However, attempts at establishing MCF-7(SRC-3) clones largely failed as cells did not survive the expansion process. We only were able to isolate six clones that survived through limited proliferation expressing varying levels of SRC-3 protein and consisting of small to medium-sized cells (Figure 2A). Relative to MCF-7(SRC-3) clones, on the same blot endogenous 160kDa SRC-3 was not detected at this exposure. Low molecular weight degradation products were clearly visible and are associated with turnover of SRC-3 in conjunction with the estrogen receptor in rapidly dividing cells cultured in estrogenic media [9]. However after passage, small and medium-sized clones expressing SRC-3 grew very slowly (see below) with various levels of SRC-3 degradation. Clone M3 appeared to have ceased expression of SRC-3 possibly due to the formation of senescence-associated heterochromatin (see below). Relative expression of SRC-3 protein in control and transfected colonies isolated at clonal selection is shown in Figure 2B. Previous reports have demonstrated that stable clones of transfected SRC-3 have distinct morphological changes such as increased cell size [28]. Indeed, SRC-3 transfected colonies contained large, flattened, sometimes multi-nucleated cells ( Figure 2C). While pcDNA3 control cells (C3) had normal cell cycle phase distributions, most SRC-3 overexpressing clones proliferated very slowly and had aberrant (greater than 2N) DNA content that varied widely ( Figure 2D,E), consistent with an aneuploid S/G2/M phase delay or arrest. galactosidase expression 72hrs post-Ad-SRC-3 infection while none of Ad-LacZ cells expressed this senescence marker ( Figure 2G). Transfection of cells with wtSRC-3 or a stable mutant of SRC-3 (S102A) also resulted in substantial cell death as indicated by cleaved PARP-1 ( Figure 2H). Thus, increasing SRC-3 protein above endogenous levels is highly detrimental to cell viability. h post-transfection with empty vector (EV), wtSRC-3 or the stable mutant SRC-3(S102A). Note that although highly expressed relative to wtSRC-3 and to the gel loading control, S102A does not induce transcription of cyclin E. Actin was used as a protein loading control.

Ectopically Expressed SRC-3 Protein Forms Nuclear Aggregates
To understand the mechanism of SRC-3-induced cytotoxicity/senescence we performed IF. Strikingly, transiently transfected SRC-3 was either homogeneously distributed in the nucleus or formed solid or ring shaped-nuclear aggregates ( Figure 3A). Alanine substitution mutants of SRC-3 at previously identified phosphoserines were all able to form aggregates as was a mutant deleted of the polyQ region [9].  . Images were acquired using a 10 × objective (EC Plan-Neofluar) with a sidemounted AxiocamHRm camera (Carl Zeiss). YFP was excited using the Colibri LED illumination system (LEDmodule 505nm, Carl Zeiss) and detected using the 46HEYFP filter (Carl Zeiss). Exposure times were 1 ms (brightfield/phase contrast) and 100ms (YFP) at 10 min intervals for 24 h and compiled into video files using Axiovision 4.8 software (Carl Zeiss). 20 min intervals are shown. Cells circled in blue showed continuous accumulation of SRC-3 while cells circled in orange appeared to resolve aggregates.

SRC-3 Overexpression Does Not Affect the Proteasome but Induces Autophagy
Cytoplasmic aggresomes include ubiquitin and proteasomal subunits [29] that eventually overwhelm the proteasome [17]. To assess whether nuclear SRC-3 overexpression affected global proteasome function, we stably transfected MCF-7 Images were acquired using a 10× objective (EC Plan-Neofluar) with a side-mounted AxiocamHRm camera (Carl Zeiss). YFP was excited using the Colibri LED illumination system (LEDmodule 505nm, Carl Zeiss) and detected using the 46HEYFP filter (Carl Zeiss). Exposure times were 1 ms (brightfield/phase contrast) and 100ms (YFP) at 10 min intervals for 24 h and compiled into video files using Axiovision 4.8 software (Carl Zeiss). 20 min intervals are shown. Cells circled in blue showed continuous accumulation of SRC-3 while cells circled in orange appeared to resolve aggregates.
Previous studies of a GFP-tagged disordered nuclear protein called GFP170* showed that small aggregates of GFP170 * form at or adjacent to PML bodies and then move toward each other and fuse to form larger aggregates accompanied by spatial rearrangements of the PML bodies [29]. Live cell imaging of SRC-3-YFP-transfected cells shows that aggregates SRC-3 foci formed rapidly (within 3 h) from the first appearance of puncta. In some cells, they coalesced and resulted in cell death (circled in blue) while in other cells they reached a maximum size then began to dissipate (circled in orange) ( Figure 3B).

SRC-3 Overexpression Does Not Affect the Proteasome but Induces Autophagy
Cytoplasmic aggresomes include ubiquitin and proteasomal subunits [29] that eventually overwhelm the proteasome [17]. To assess whether nuclear SRC-3 overexpression affected global proteasome function, we stably transfected MCF-7 cells with an unstable variant of GFP tagged with a nuclear localization motif (NLS-GFPu) [19] then infected these cells with adenovirus expressing SRC-3(Ad-SRC-3). Immunoblot showed the presence of NLS-GFPu in cells treated with MG132, a proteasome inhibitor, compared with vehicle-treated cells while no protein was detected following either Ad-SRC-3 or Ad-LacZ infection ( Figure 4A). To assess proteasome inhibition by SRC-3 overexpression in individual cells we performed IF for SRC-3 and GFP in Ad-LacZ and Ad-SRC-3 infected cells transfected with NLS-GFPu. GFP-fluorescence in Figure 4B shows that cells co-expressing the GFPu and transfected SRC-3 express similar levels as GFPu in control/LacZ-transfected cells quantified in Figure 4C. Thus SRC-3 nuclear aggregates do not block the 26S proteasome. Moreover, overexpression of SRC-3 did not induce markers of the unfolded protein response (UPR) in either MCF-7 cells or LCC9 tamoxifen-resistant derivative cells [30] including the absence of a P-PERK shift associated with inhibition of translation, induction of the UPR genes CHOP (cell cycle arrest), PDI (protein folding) [31] (and references therein) or calnexin which controls protein transport for ER-associated degradation [32] ( Figure 4D).
Since nuclear SRC-3 expression had no effect on proteasome activity or induce the UPR, we next tested whether it induced the autophagic degradation pathway. Adenoviral transduction of SRC-3 in MCF-7 cells stably expressing eGFP-LC3 resulted in perinuclear formation of autophagic vesicles in many cells overexpressing SRC-3 ( Figure 4E). Furthermore, SRC-3 transduced MCF-7 cells displayed increased formation of cleaved/phosphatidyl-ethanolamine conjugated LC3 (LC3-II) indicating the induction of autophagy ( Figure 4F). Transduced SRC-3 was further stabilized by the autophagy inhibitor, hydroxychloroquine (HCQ). Together these data indicate that overexpressed SRC-3 protein is proteotoxic and can form nuclear aggregates that induce an autophagic response.

SRC-3 Nuclear Aggregates Are Proximal to Microtubules and Overexpression Disrupts Both Mitotic Microtubule Dynamics and the Nuclear Membrane
To further characterize the striking cytotoxicity of excess SRC-3 in MCF-7 cells we examined mechanisms associated with protein aggregate formation and resolution. Small aggregates of cytoplasmic misfolded proteins form and are rapidly transported toward the microtubule (MT)-organizing center, where they coalesce to form aggresomes [29]. In Figure 5A, deconvoluted 3-D images showed that SRC-3 nuclear aggregates were invariably closely associated with perinuclear microtubules. To investigate whether SRC-3 overexpression would interfere with mitotic microtubule dynamics we treated cells with 10 nM taxol for 18-24 h post-infection with Ad-LacZ or Ad-SRC-3 and evaluated mitotic figures. Figure 5B shows that overexpressed homogeneously distributed SRC-3 disrupted the microtubular network resulting in a perinuclear concentration of tubulin in cells without or possibly prior to aggregate formation. In some cells tubulin was colocalized with SRC-3-containing nuclear cap-like structures in cells transfected with SRC-3 in the absence of aggregates. MCF-7 clones expressing SRC-3 had aberrant DNA content profiles which could be a consequence of microtubular dysfunction. The results in Figure 5C demonstrate a significant reduction in metaphase cells as determined by metaphase microtubule structures in cells infected with Ad-SRC-3 but not Ad-LacZ  Figure 5D. Degradation of the nuclear lamina by autophagy was initially discovered as a response to oncogenic stress-a process termed "nucleophagy" [33]. Since SRC-3 overexpression resulted in nuclear membrane perturbation we acquired Z-stack images of YFP-tagged SRC-3, tubulin and lamin B1 in cells overexpressing SRC-3. Figure 5E shows the distortion and disruption of nuclei accompanied the release of lamin B1 into the cytoplasm. A variety of lamin B1 nuclear disruptions patternswere observed accompanying increased SRC-3 expression. In Figure 5F a misshapen nucleus can be seen as well as regions of discontinuous lamin B1 IF in all cells where transfected SRC-3 is expressed.

SRC-3 Overexpression Redistributes PML Bodies and Induces PML
The synthetic disordered protein GFP170* forms aggregates that coalesce at PML bodies resulting in the redistribution of PML [29]. To determine the effect of SRC-3 on PML bodies we performed IF of MCF-7 cells transiently expressing Thus, the senescence and cytotoxicity induced by increasing SRC-3 protein levels in MCF-7 cells most likely derive from the effect of nuclear aggregate formation on microtubule function and nucleophagy-mediated cell death.

SRC-3 Overexpression Redistributes PML Bodies and Induces PML
The synthetic disordered protein GFP170* forms aggregates that coalesce at PML bodies resulting in the redistribution of PML [29]. To determine the effect of SRC-3 on PML bodies we performed IF of MCF-7 cells transiently expressing varying levels of transfected SRC-3. Similar to GFP170*, Figure 6A shows ring-like SRC-3 aggregates that form adjacent to PML bodies. In cells expressing homogenously distributed SRC-3, PML bodies were either undetectable ( Figure 6B (i,i') or PML was focused on nuclear membrane caps ( Figure 6B (ii,ii') (arrow)). In other examples, small SRC-3 aggregates were released from the nucleus along with PML. In the absence of SRC-3 aggregate formation PML formed diffuse masses near the nuclear membrane ( Figure 6C (arrows)). Thus, aggregate formation and homogenous increases in SRC-3 have distinctive effects on PML. Oncogenic ras [34] and myc [35] are IDPs that induce PML upon overexpression resulting in senescence. Consistent with the induction of senescence by SRC-3 (see Figure 1F), a strong increase in PML protein was observed 96hrs following transfection of SRC-3 ( Figure 6D). (i,i') or PML was focused on nuclear membrane caps ( Figure 6B (ii,ii') (arrow)).
In other examples, small SRC-3 aggregates were released from the nucleus along with PML. In the absence of SRC-3 aggregate formation PML formed diffuse masses near the nuclear membrane ( Figure 6C (arrows)). Thus, aggregate formation and homogenous increases in SRC-3 have distinctive effects on PML. Oncogenic ras [34] and myc [35] are IDPs that induce PML upon overexpression resulting in senescence. Consistent with the induction of senescence by SRC-3 (see Figure 1F), a strong increase in PML protein was observed 96hrs following transfection of SRC-3 ( Figure 6D).

Increased SRC-3 Protein Level Impacts the SUMO Pathway
SUMOylation has a critical role in proteostasis and regulates nuclear protein aggregation in neurodegenerative diseases [36]. SUMOylation is also involved in the formation of PML nuclear bodies [37,38] which were strongly reduced in SRC-3

Increased SRC-3 Protein Level Impacts the SUMO Pathway
SUMOylation has a critical role in proteostasis and regulates nuclear protein aggregation in neurodegenerative diseases [36]. SUMOylation is also involved in the formation of PML nuclear bodies [37,38] which were strongly reduced in SRC-3 overexpressing cells. PML bodies are themselves the major sites of SUMOylation of PML client proteins [39]. We, therefore, tested the involvement of the SUMO pathway following overexpression of SRC-3. Immunoprecipitated SRC-3 from SRC-3 transfected cells appears as a very high molecular weight smear in addition to the 160kDa band consistent with poorly detergent soluble SUMO-SRC-3 aggregates (Figure 7A,B). UBC9 is the only known E2 SUMO-conjugating enzyme [40]. Therefore, we tested the effects of SRC-3-GFP overexpression on UBC9 protein levels by IF. Figure 7C shows a representative SRC-3-GFP-positive cell in which UBC9 IF staining intensity was markedly reduced. Average IF pixel values quantified for UBC9 (CY3) in SRC-3-GFP-positive cells and GFP-negative cells are graphed ( Figure 7D) indicating a significant decrease in detectable UBC9 in SRC-3 overexpressing cells. Thus, ectopic SRC-3 protein is SUMOylated and accompanied by a reduction in detectable UBC9. Since transfected SRC-3 was SUMOylated and redistributed PML, we next determined its effects on the cellular distribution of SUMO-1by IF. The panels in Figure 7E depict the various structural and localization changes that occur in SRC-3 overexpressing cells. Figure 7E, panel i shows discrete SUMO-1 foci in an overexpressing cell. One cell shows an SRC-3 containing protrusion into the cytoplasm in association with SUMO-1. In some expressing cells SUMO-1 was restricted to the nuclear periphery ( Figure 7E, panels ii,iii). In Figure 7E, panel iii the enlarged cell shows large foci of SUMO-1. Interestingly, nucleoli numbers were reduced to one or were undetectable as assessed using DAPI-a phenomenon previously reported in cells lacking UBC9 [41].
Importantly, SUMOylation of nuclear membrane laminA/C has been shown to be required for the interaction between LC3 and lamin A/C to facilitate the process of nucleophagy in response to DNA damage [42]. Similar to Figure 4B showing tubulin-associated extranuclear chromatin, cytoplasmic chromatin surrounded by SUMO-1 is also present in panel iii (arrowhead).
The GTPase activating protein, ranGAP1, is a key enzyme in the ran-controlled receptor-mediated nuclear pore transport system [43]. SUMOylation of ranGAP1 is essential for nuclear membrane localization and it can be detected using anti-SUMO-1 ( [44] and references therein). IF for ran-GAP1 in SRC-3 transfected cells showed disruption in the normal localization at the nuclear membrane ( Figure 7F).

Pharmacologic Stabilization of SRC-3 Recapitulates Effects of Overexpression
In the experiments above we noted that overexpression of levels of SRC-3 that do not result in aggregation was sufficient to result in nucleophagic-like expulsion of SRC-3 from the nucleus. We reasoned that clinically relevant treatment modalities that are associated with senescence induction and or proteasome/autophagy inhibition have the potential to increase SRC-3 levels [45] and may, therefore, result in similar responses similar to ectopic expression. Proteasome and autophagy inhibitors have been utilized often in combination with other therapeutics, in vitro and clinically to induce cytotoxicity in breast cancer cells and tumors [46][47][48] albeit sometimes with limited success [49]. We tested the effects of proteasome and autophagy inhibition on levels and localization of SRC-3. Immunoblot analysis in Figure 8A shows that the proteasome inhibitor, bortezomib (B), only slightly increased SRC-3 in MCF-7 cells, possibly due to increased autophagic clearance as described elsewhere [50]. Unexpectedly, we found that HCQtreatment alone strongly reduced SRC-3 protein levels, a finding which may reflect increased proteasomal activity when the autophagic removal of proteasome components is inhibited [51]. The combination of B with HCQ increased the levels of SRC-3 protein and treatment for, 72 h with this combination also resulted in a significantly higher level of cytotoxicity compared to either agent alone ( Figure 8B). To assess the effects of the B+HCQ treatment on SRC-3 at the cellular level we performed IF. Figure 8C, panel (i) depicts SRC-3 IF in cells treated with vehicle and shows again that a subset of cells exhibit nuclear protrusions containing SRC-3 although even in high contrast epifluorescence images (panel i') virtually no SRC-3 was detected in the cytoplasm. In contrast a 72-h exposure to B or B+HCQ strongly increased the level of SRC-3 in the cytoplasm ( Figure 8C, panels ii/ii' and iii/iii'). Moreover, extrusion of chromatin into the cytoplasm was also observed in combination-treated cells ( Figure 8C, panel iii"). Thus, inhibition of the proteasome results in aberrant transfer of endogenous SRC-3 to the cytoplasm in approximately 2% of the MCF-7 cells examined. While overall protein stability and downstream effects are also factors in pharmacologic-mediated cytotoxicity of B+HCQ treatment, the presence of SRC-3 within nuclear protrusions and increased cytoplasmic localization demonstrates that SRC-3 is an important substrate for this form of nucleophagy-like egress in response to protein stabilization.
Tamoxifen (Tam) is an important therapeutic in ER+ breast cancer and has also been shown to stabilize SRC-3 protein [52] while promoting senescence in ER+ breast cancer cells [53]. The immunoblot in Figure 8D shows that SRC-3 levels are substantially increased in MCF-7 cells treated with 5 µM Tam within 2 h. Consistent with increased SRC-3 protein levels, Figure 8E shows enhanced SRC-3 IF in Tam-treated MCF-7 cells vs. control following a 3hr exposure to the drug. A cell shown in vehicle-treated cultures appeared to have higher SRC-3 expression relative to surrounding cells and both this cell and cells within the Tam-treated cultures exhibited nuclear SRC-3 protrusions typical of SRC-3 transfected cells. Notably, high expression of SRC-3 in Tam-treated cells was also associated with reduced nucleoli (arrows) as we observed in SRC-3 transfected cells. Similar to cells transfected with SRC-3 (Figure 7), numbers of nucleoli were significantly reduced after the 3 h exposure to Tam ( Figure 8F). Since transfected SRC-3 resulted in dispersion of PML nuclear bodies we performed IF for PML in Tam-treated cells. In Figure 8G Tam reduced the detection of discrete PML bodies compared to control. Quantification of PML bodies showed that Tam treatment resulted in a significant decrease compared to vehicle ( Figure 8H). Thus, drug-induced stabilization of SRC-3 has similar effects as SRC-3 transfection. control. Quantification of PML bodies showed that Tam treatment resulted in a significant decrease compared to vehicle ( Figure 8H). Thus, drug-induced stabilization of SRC-3 has similar effects as SRC-3 transfection.

Discussion
The AIB1 gene is amplified in approximately 5-10 percent of breast cancers while high expression is present in more than 60 percent of primary tumors, typically associated with ER + tumors [54]. SRC-3 is known to be tightly regulated by phosphorylation, dephosphorylation and SUMOylation events that affect both transcriptional activity and post-transcriptional degradation [9]. Since SRC-3 is highly disordered in the absence of folding with CBP [13], rigorous control of unbound protein levels is necessary to avoid potential proteotoxicity. Misfolded nuclear proteins are sequestered at PML bodies and SUMOylated. Indeed, SUMOylation of SRC-3 is associated with loss of SRC-3 coactivator function [11]. Thus SRC-3 may be SUMOylated in the absence of transcriptional activation to facilitate sequestration of unfolded protein. Consistent with this prediction, transfected SRC-3 was SUMOylated and often formed aggregates in MCF-7 cells consistent with previous reports that transfected SRC-3 appears as discrete foci in both HeLa cells [55] and an ER-negative breast cancer cell line [56].
Although it is unclear whether nucleophagy is a normal homeostatic process in eukaryotes [57] there is evidence that it is employed for the clearance of nuclear waste [58]. Interestingly, our finding that SRC-3 nuclear protrusions can be seen even in untreated MCF-7 cells and other breast cancer cell lines and tumors which may be indicative a form of nucleophagy or nuclear "budding" which contributes to the regulation of SRC-3 protein levels.
This mechanism may be particularly important for cancer cells that contain high levels of unfolded transcription factors such as SRC-3 and utilized under conditions where folding is reduced either due to stoichiometric imbalances or inhibition of transcription-coupled degradation. Nuclear budding has been proposed as an alternative mechanism of nuclear protein quality control analogous to Herpesvirus egress. This process involves budding from the nuclear membrane of aggregates or complexes too large for the nuclear pore in order to access the cytoplasmic autophagy machinery [59]. A similar process is utilized in yeast called "piecemeal microautophagy" [60].
Overexpression of SRC-3 can result in SUMOylation of misfolded protein within PML bodies resulting in SUMO-targeted ubiquitination and degradation. However, similar to nucleophagy, nuclear budding is proposed to transport large aggregates destined for autophagic degradation out of the nucleus independent of nuclear pores [61]. We speculate that nuclear budding/piecemeal microautophagy may be utilized for homeostatic removal of SRC-3 from the nucleus while increasing expression induces nucleophagy based on the various cellular phenotypes we observed. For example, nuclear oncogene-induced senescence has been associated with nuclear membrane lamin B1 autophagy [33] while SUMOylation of nuclear lamins occurs during DNA damage-induced nucleophagy [42]. Thus, lamin B1/RanGAP disruption and redistribution of SUMO to the nuclear lamina in SRC-3 overexpressing cells together provide evidence for nucleophagy which may result in gross nuclear disruption and cell death or controlled budding of DNA/SRC-3/tubulin nuclear protrusions ( Figure 5B).
We found that overexpression of SRC-3 also altered PML/SUMO localization and disrupted the localization of SUMO-RanGAP1 consistent with perturbation of the SUMO pathway. Similar to other disordered proteins, SRC-3 aggregates accumulate adjacent to PML bodies. However, in the absence of aggregate formation, PML bodies dispersed in response to overexpression of SRC-3. These observations parallel that of cells lacking SUMO-1 which results in aberrant nuclear pore localization of RanGAP1 and dispersion of PML bodies [62]. We speculate that high expression of unfolded transcription factors could contribute to the reduced numbers of PML bodies typical of transformed cells relative to normal cells. PML protein sometimes coalesced and accumulated at the nuclear membrane in some cells in cap-like formations in response to increased levels of SRC-3. Interestingly PML-II can displace lamins from the nuclear membrane [63] which was clearly evident in SRC-3-transfected cells ( Figure 5E,F) accompanied by the release of small SRC-3 foci. It is possible that PML may participate in the process of nucleophagy induced by IDPs by contributing to the disassembly of lamins in the nuclear membrane to allow access to the autophagosome. SUMO E2 enzyme phosphorylation by CDK1/cyclin B during mitosis increases its overall activity [64,65] when SUMOylation events are necessary for proper chromosome segregation [41,66,67]. Thus, the perturbation of the SUMO pathway after SRC-3 overexpression may have contributed to the aneuploidy we observed in MCF-7(SRC-3) colonies as could direct effects on microtubule function. The formation of cytoplasmic "aggresomes" [68] requires microtubules and dynein-based transport to coalesce aggregates. Remarkably, nuclear SRC-3 aggregates in MCF-7 cells were invariably juxtaposed with microtubules. How microtubules might affect nuclear protein aggregation is not clear. Components of the linker of nucleoskeleton and cytoskeleton (LINC) complexes spanning the inner and outer nuclear membranes interact with perinuclear microtubules that exert force on the nucleus through microtubule motors [69,70]. LINC components include nuclear envelope nesprins, which connect the nuclear lamina to the cytoskeleton and molecular motors [71]. Based on the proximity of SRC-3 aggregates to microtubules it is possible that these structures could form organizing centers for aggregation of unfolded SUMOylated nuclear proteins.
Stress on protein quality control mechanisms may ultimately participate in senescence induction when SRC-3 and other nuclear IDPs are stabilized by proteasome/autophagy inhibitors or Tam. In this regard, Tam treatment strongly reduced PML bodies which are dependent on SUMOylation of PML [72] suggesting diversion of SUMO from PML to SRC-3 and possibly other proteins. In the future, it will be of interest to determine if endogenous amplified expression of SRC-3 and other oncogenic IDPs are potential liabilities of cancer cells that can serve as markers of sensitivity to agents that stabilize these proteins or interfere with the processing of IDPs by blocking SUMOylation.