Senescent cells develop a senescence-associated secretory phenotype (SASP). The factors secreted by cells with a SASP have multiple biological functions that are mediated in an autocrine or paracrine manner. However, the status of the protein kinase D1 (PKD1; also known as PRKD1)-mediated classical protein secretory pathway, from the trans-Golgi network (TGN) to the cell surface, during cellular senescence and its role in the cellular senescence response remain unknown. Here, we show that the activities or quantities of critical components of this pathway, including PKD1, ADP-ribosylation factor 1 (ARF1) and phosphatidylinositol 4-kinase IIIβ (PI4KIIIβ), at the TGN are increased in senescent cells. Blocking of this pathway decreases IL-6 and IL-8 (hereafter IL-6/IL-8) secretion and results in IL-6/IL-8 accumulation in SASP-competent senescent cells. Inhibition of this pathway reduces IL-6/IL-8 secretion during Ras oncogene-induced senescence (OIS), retards Ras OIS and alleviates its associated ER stress and autophagy. Finally, targeting of this pathway triggers cell death in SASP factor-producing senescent cells due to the intracellular accumulation of massive amounts of IL-6/IL-8. Taken together, our results unveil the hyperactive state of the protein secretory pathway in SASP-competent senescent cells and its critical functions in mediating SASP factor secretion and the Ras OIS process, as well as in determining the fate of senescent cells.
Cellular senescence is a stable proliferation arrest state triggered by a variety of stresses, such as replicative exhaustion, DNA damage, oxidative stress and oncogene activation (Kuilman et al., 2010). Senescent cells display many distinct features that are different to those in young cells in their morphology, gene transcription and expression, chromatin structure and metabolism. They also secret numerous cytokines, chemokines and other proteins, which is termed the senescence-associated secretory phenotype (SASP) (Salama et al., 2014). Acquisition of the SASP is mainly regulated at transcriptional level by nuclear factor (NF)-κB and C/EBPβ (Acosta et al., 2008; Kuilman et al., 2008). A persistent DNA-damage response (DDR) (Rodier et al., 2009), the p38α MAPK (also known as MAPK14) (Freund et al., 2011), IL-1α (Orjalo et al., 2009), protein kinase D1 (PKD1; also known as PRKD1) (Wang et al., 2014) and GATA4 (Kang et al., 2015) have been implicated in SASP induction by modulating NF-κB activity. SASP gene expression is also subjected to translational regulation by mTOR (Herranz et al., 2015; Laberge et al., 2015) and epigenetic regulation by macroH2A1 (Chen et al., 2015).
Cells with a SASP execute a tumor suppression function either by reinforcing senescence in an autocrine manner (Acosta et al., 2008; Kuilman et al., 2008) or by recruiting immune cells to clear premalignant senescent cells in a paracrine fashion (Kang et al., 2011). SASP factors also trigger tissue remodeling, and hence play a critical role in tissue repair (Demaria et al., 2014; Krizhanovsky et al., 2008) and embryonic development (Muñoz-Espín et al., 2013; Storer et al., 2013). In contrast, owing to the accumulation of senescent cells in later life, SASP factors can promote tumor formation and metastasis, and may contribute to the ‘inflammaging’ and many age-related pathologies (Campisi, 2013; van Deursen, 2014).
SASP factors exert their effects in both an autocrine and paracrine fashion. In general, most secretory proteins, including the core SASP components, such as IL-6 and IL-8 (hereafter IL-6/IL-8), are exported to plasma membrane via a conventional endoplasmic reticulum (ER)-to-Golgi secretory pathway (Brandizzi and Barlowe, 2013; Lee et al., 2004). It has been reported that the ER, Golgi and TGN are positioned close to the TOR-autophagy spatial coupling compartment (TASCC), and that IL-6 and IL-8 are enriched in the TGN and/or TASCC in Ras-induced senescent cells (Narita et al., 2011). The Golgi complex structure is also altered and dispersed in senescent cells (Cho et al., 2011). However, whether the activities or the quantities of the classical protein secretory pathway components are increased during cellular senescence and what the impact of this pathway may have on the cellular senescence and SASP-related functions remain largely unknown.
The serine/threonine kinase PKD1 plays pivotal role in mediating secretory protein trafficking from the TGN to the cell surface (Liljedahl et al., 2001; Malhotra and Campelo, 2011). First, PKD1 is recruited to the TGN by binding to diacylglycerol (DAG) (Baron and Malhotra, 2002) and ARF1 (Pusapati et al., 2010), and is then activated at TGN. The active TGN-associated PKD1 activates its downstream target PI4KIIIβ to trigger protein secretion to cell surface (Hausser et al., 2005; Valente et al., 2012). However, whether the PKD1-mediated protein secretory pathway from the TGN to the plasma membrane is involved in regulating SASP secretion and its related functions, such as reinforcing cellular senescence, has not been defined.
We report here that the classical protein secretory pathway components, including PKD1, DAG, PLD1, ARF1 and PI4KIIIβ, play a vital role in modulating secretion of SASP factors and their function. Perturbation of this pathway affects the establishment of Ras oncogene-induced senescence (OIS), and can induce death in senescent fibroblasts and cancer cells.
Golgi structure is altered and the protein secretory pathway at the TGN is activated in senescent cells
To explore whether the PKD1-mediated classical protein secretory pathway is involved in regulating SASP factor secretion and function, we used IMR90 cells stably expressing a 4-hydroxytamoxifen (4-OHT)-inducible ER-targeted Ras (ER:Ras) fusion protein (denoted ER:Ras IMR90 cells) as a model of OIS (Narita et al., 2011), and the human lung fibroblast 2BS cell line as a replicative senescence model. We first compared the Golgi structure of young and senescent cells by staining the Golgi for the specific TGN marker TGN46 (also known as TGOLN2). TGN46 is a type I integral membrane protein that is localized to the TGN and cycles between early endosomes and the TGN, and plays a critical role in exocytic vesicle formation (Prescott et al., 1997). In agreement with previous observations (Cho et al., 2011), young IMR90 and 2BS cells showed a compact TGN in the perinuclear region, while both Ras-induced senescent IMR90 cells and replicative senescent 2BS cells predominantly displayed a dispersed and protruding TGN morphology (Fig. 1A; Fig. S1A, second panel), implying that senescent cells are in an active secretory state. The senescent state of ER:Ras-IMR90 cells was characterized by p16 upregulation (also known as CDKN2A) and IL-6/IL-8 expression upon Ras induction (Fig. 1B), and the senescent state of 2BS cells was confirmed by senescence-associated (SA)-β-gal staining (Fig. S1B).
Since PKD1 is recruited to TGN to mediate protein secretion from TGN to cell surface (Liljedahl et al., 2001; Malhotra and Campelo, 2011), we next determined the TGN localization of PKD1 in young and senescent cells. The immunofluorescence signals of TGN46 and PKD1 overlapped both in young and Ras-induced senescent IMR90 cells (Fig. 1A), as well as in young and replicative senescent 2BS cells (Fig. S1A), which suggests that PKD1 is located at the TGN both in young and senescent cells. This result is consistent with previous reports showing that PKD1 is a TGN-localized kinase that mediates protein secretion (Liljedahl et al., 2001; Malhotra and Campelo, 2011).
Next, we determined the PKD1 activity state at the TGN in young and senescent cells. First, we utilized a high-content analysis (HCA) method to measure the mean intensity of the signal of PKD1 phosphorylated at Ser738 and Ser742 [p-PKD1 (Ser738/742)] within the TGN46-positive regions. The HCA quantification data showed that the PKD1 activities at TGN in both Ras-induced senescent IMR90 cells and replicative senescent 2BS cells were much higher than in their corresponding young cells (Fig. 1C,D; Fig. S1C,D). To further verify the increase of PKD1 activity at TGN in senescent cells, we isolated the Golgi from young and senescent IMR90 and 2BS cells, and analyzed PKD1 activity by immunoblotting with antibody specific for p-PKD1 (Ser738/742) or p-PKD1 (Ser916), respectively. PKD1 activation is dependent on the phosphorylation of two activation loop sites, Ser738 and Ser742 (Iglesias et al., 1998). In addition, the C-terminal Ser916 is an autophosphorylation site for PKD1, and its phosphorylation correlates with PKD1 catalytic activity (Matthews et al., 1999). The phosphorylation level of PKD1 on Ser916 or Ser738/742 was drastically increased in both senescent 2BS and IMR90 cells, whereas active PKD1 in corresponding young cells was barely detected (Fig. 1E,F). In addition, the level of PKD1 at Golgi in senescent cells was also significantly higher than in young cells (Fig. 1E,F), which suggests that more PKD1 is recruited to the TGN in senescent cells compared to in the young cells. The Golgi marker TGN46 and ER marker calnexin served as positive and negative control, respectively, to prove the purity of the Golgi isolation.
In the PKD1-mediated protein transport pathway from the TGN to plasma membrane, ARF1 is required for PKD1 location to the TGN (Pusapati et al., 2010) and PI4KIIIβ acts as a substrate of PKD1 (Hausser et al., 2005; Valente et al., 2012). To further compare the TGN status of young and senescent cells, we measured the ARF1 and PI4KIIIβ levels at TGN by using the HCA method. The HCA quantification data showed that both ARF1 and PI4KIIIβ levels significantly increased at the TGN in both Ras-induced senescent IMR90 cells and replicative senescent 2BS cells, when compared with their corresponding young cells (Fig. 1G–J; Fig. S1E–H).
Taken together, these results show that senescent cells display a dispersed and protruded TGN structure, and the activities or quantities of TGN-localized components of classical protein secretory pathway, such as PKD1, ARF1 and PI4KIIIβ are increased in senescent cells compared to in young cells, all of which indicate that senescent cells are in the active secretory state.
Pharmacological inhibition of the protein secretory pathway in senescent cells suppresses IL-6/IL-8 secretion and causes cytokine accumulation
To investigate whether the PKD1-mediated protein secretory pathway is involved in secretion of the core SASP factors IL-6/IL-8, we first individually inhibited the critical components of this pathway by using various specific inhibitors. In the PKD1-mediated protein transport pathway, DAG is required for PKD1 recruitment to the TGN by binding to proline 157 in the PKD1 C1a domain (Baron and Malhotra, 2002). To examine whether DAG plays a role in IL-6/IL-8 secretion, Ras-induced senescent IMR90 cells were treated with the pan-DAG scavenger L-cycloserine (L-CS) (Rémillard-Labrosse, 2009). DAG elimination by L-CS treatment substantially diminished IL-6/IL-8 secretory levels as detected by ELISA (Fig. 2A,B); however, the intracellular IL-6/IL-8 protein levels were elevated (Fig. 2C), when compared with those in untreated senescent IMR90 cells. These results demonstrate that removal of DAG reduces IL-6/IL-8 secretion and results in the intracellular accumulation of IL-6/IL-8 in senescent cells.
Next, we explored molecules involved in DAG formation at TGN in Ras-induced senescent IMR90 cells. Phospholipase C β3 (PLCβ3), phospholipase D1 (PLD1), sphingomyelin synthase 1 and 2 (SMS1/2), and intracellular Ca2+ are reported to engage in the DAG production at the TGN for the classical protein secretory pathway (Díaz Añel, 2007; Kunkel and Newton, 2010; Sonoda et al., 2007; Villani et al., 2008). In order to identify which of these molecules are responsible for Golgi DAG generation in senescent cells, we used the specific PLCβ3 inhibitor U73122 and its inactive analog U73343, as a negative control (Díaz Añel, 2007), PLD1 pathway inhibitors FIPI (Chen et al., 2012), 1-butanol and propranolol and their negative control 2-butanol (Sonoda et al., 2007), SMS1/2 inhibitor D609 and its negative control C-6NBD (Villani et al., 2008), as well as the intracellular Ca2+ chelator BAPTA and its agonist thapsigargin (THAP) (Kunkel and Newton, 2010), to treat Ras-induced senescent IMR90 cells. Among these, only the PLD1 inhibitors FIPI, 1-butanol and propranolol could significantly decrease IL-6/IL-8 secretion and cause intracellular accumulation of IL-6/IL-8 (Fig. 2D–F; Fig. S2A,B), and PLCβ3, SMS1/2, and Ca2+ inhibitors did not impede IL-6 secretion or cause elevated intracellular levels of IL-6/IL-8 (Fig. S2C–H). We noticed that the PLCβ3 activator m-3M3FBS treatment caused Ras-induced senescent IMR90 cell death in the dosage we used in this study, and hence resulted in the inhibition of both intracellular IL-6/IL-8 protein levels and the secretory level of IL-6 (Fig. S2C,D) (Díaz Añel, 2007). We also observed that the Ca2+ agonist thapsigargin decreased IL-6 secretion, which was in line with a previous report showing that thapsigargin could reduce SASP gene transcription, including that of IL-6/IL-8, via a thapsigargin-induced ER stress-triggered negative-feedback loop to mitigate SASP gene transcription (Chen et al., 2015).
We further utilized the specific ARF1 inhibitor brefeldin A (BFA) to treat Ras-induced senescent IMR90 cells and examine IL-6/IL-8 secretion (Robineau et al., 2000). BFA treatment remarkably blocked IL-6/IL-8 secretion and led to the intracellular accumulation of massive amounts of IL-6/IL-8 (Fig. 2G–I). BFA supplementation also disrupted the colocalization of PKD1 with TGN46 and caused PKD1 redistribution from the TGN to the cytosol in replicative senescent 2BS cells (Fig. S2I).
Taken together, these results demonstrate that pharmacological inhibition of the PKD1-mediated classical protein secretory pathway decreases IL-6/IL-8 secretion and causes the intracellular accumulation of IL-6/IL-8 in senescent cells.
Genetic impairment of classical protein secretory pathway suppresses IL-6/IL-8 secretion during Ras OIS
To further explore the role of PKD1-mediated protein secretory pathway in regulating SASP factor secretion more specifically, we genetically mutated or silenced the critical components of this pathway including PKD1, PLD1, ARF1 and PI4KIIIβ in ER:Ras IMR90 cells. A PKD1 C1a domain Pro157 to glycine mutation is known to result in the loss of DAG binding to PKD1 (Maeda et al., 2001). To probe whether the DAG-binding capacity of PKD1 is essential for IL-6/IL-8 secretion, this DAG-binding deficient mutant of PKD1 was cloned into the pHBLV lentiviral vector (named pHBLV-PKD1-P157G) and introduced into ER:Ras IMR90 cells, and then cells were given 4-OHT to induce Ras expression. Upon Ras induction for 5–6 days, the secretory levels of IL-6 and IL-8 in PKD1-P157G-expressing cells was notably reduced compared to that seen in the vector control cells (Fig. 3A; Fig. S3A). In contrast, PKD1 overexpression promoted IL-6/IL-8 secretory levels relative to the vector control cells (Fig. 3A; Fig. S3A), which is consistent with a previous report showing that PKD1 enhanced IL-6/IL-8 expression and secretion via elevation of NF-κB activity (Wang et al., 2014). These results suggest that PKD1-P157G suppressed IL-6/IL-8 secretion during the establishment of Ras OIS. Therefore, this data indicate that the DAG-binding capability of PKD1 is necessary for the IL-6/IL-8 secretory process during Ras OIS.
To validate the role of PLD1 in IL-6/IL-8 secretion during Ras OIS, we silenced it with two independent shRNAs, by cloning them in the pLL3.7 lentiviral vector (named pLL3.7-PLD1) and transfecting this vector into ER:Ras IMR90 cells. pLL3.7-PLD1 notably abated IL-6/IL-8 secretory levels compared to the mock cells (Fig. 3B,C, and Fig. S3B), which indicated that loss of PLD1 reduced IL-6/IL-8 secretion during Ras OIS.
Next, we deleted ARF1 in ER:Ras IMR90 cells with two independent shRNAs. ARF1 loss significantly reduced IL-6/IL-8 secretory levels compared to the vector control cells during Ras OIS (Fig. 3D,E; Fig. S3C). A PKD1 C1b domain proline 281 to glycine mutation causes the loss of ARF1 binding to PKD1 (Pusapati et al., 2010). To further decipher whether the ARF1-binding capacity of PKD1 is important for IL-6/IL-8 secretion in Ras OIS, we overexpressed the ARF1-binding deficient mutant of PKD1 (vector pHBLV-PKD1-P281G) in ER:Ras IMR90 cells. PKD1-P281G significantly reduced IL-6/IL-8 secretory levels during Ras OIS, when compared with mock cells and wild-type PKD1-expressing cells (Fig. 3A; Fig. S3A). These findings show that ARF1 is required for PKD1 recruitment to the TGN and plays a crucial role in mediating IL-6/IL-8 secretion during Ras OIS.
Finally, we depleted PI4KIIIβ in ER:Ras IMR90 cells with two independent shRNAs. PI4KIIIβ deletion significantly attenuated IL-6/IL-8 secretory levels compared to the vector control cells during Ras OIS (Fig. 3F,G; Fig. S3D). In summary, these data imply that the PKD1-mediated classical protein secretory pathway from the TGN to the cell surface participates in the regulation of SASP factor secretion when cells are undergoing Ras OIS, and genetic inhibition of this pathway suppresses IL-6/IL-8 secretory levels during Ras OIS.
Blockade of the PKD1-mediated protein secretory pathway retards Ras OIS
The finding that perturbing the PKD1-mediated classical protein secretory pathway significantly repressed IL-6/IL-8 secretion led us to aim to further delineate the role of this pathway in the establishment of Ras OIS, since it is known that secretory IL-6/IL-8 can reinforce cell senescence in a cell-autonomous fashion and disruption of IL-6/IL-8 autocrine network can attenuate OIS entry or maintenance (Acosta et al., 2008; Kuilman et al., 2008). To examine this, pLL3.7-PLD1, pLL3.7-ARF1, pLL3.7- PI4KIIIβ, pHBLV-PKD1-P157G and pHBLV-PKD1-P281G were introduced into ER:Ras IMR90 cells, and then the infected cells were induced to express Ras. After Ras induction for the indicated number of days, the cells were subjected to SA-β-gal activity, cell proliferation and BrdU incorporation assays to evaluate the cell senescent state. Relative to vector control cells, PKD1 overexpression induced a significantly higher level of SA-β-gal activity, a biomarker for senescent cells. Conversely, overexpression of PKD1-P157G and PKD1-P281G largely reduced SA-β-gal-positive staining cells (Fig. 4A; Fig. S3E). Moreover, PKD1-P157G and PKD1-P281G-expressing cells exhibited continuous cell proliferation under Ras induction within 6 days (Fig. 4B), and a significantly higher BrdU incorporation ratio after 6 days of Ras overexpression (Fig. 4C,D). In contrast, cell proliferation was halted after 3 days of Ras induction and DNA synthesis was dramatically reduced after 6 days of Ras expression in vector control cells, which indicates that cells entered the senescent state. PKD1-expressing cells entered the proliferation arrest state earlier than mock cells (Fig. 4B), which agrees with our previous report showing that PKD1 overexpression promotes Ras OIS (Wang et al., 2014). Likewise, PLD1, ARF1 and PI4KIIIβ deletion significantly decreased SA-β-gal activity levels, when compared with mock cells, under Ras induction (Fig. 4E; Fig. S3F). Furthermore, PLD1, ARF1 and PI4KIIIβ depletion conferred continuous proliferation on IMR90 cells and a much higher level of BrdU incorporation was seen when compared with that in mock cells under Ras induction, which indicates that Ras OIS was notably alleviated (Fig. 4F–H). These results demonstrate that the PKD1-mediated classical protein secretory pathway is involved in mediating Ras OIS entry, and when the protein secretory pathway is blocked, Ras OIS entry is delayed.
PKD1-mediated protein secretory pathway deficiency represses ER stress and autophagic activity during Ras OIS
It has been proposed that SASP-competent cells undergoing OIS induce ER stress and autophagic activity (Dörr et al., 2013). In light of this finding, we further explored whether inhibition of PKD1-mediated protein secretory pathway might affect these Ras OIS effects. Consistent with previous results (Dörr et al., 2013), ER stress was induced, as demonstrated by increased phosphorylation of JNK1 (also known as MAPK8) (denoted p-JNK) and upregulation of the cyclic AMP-dependent transcription factor ATF4, and autophagy was activated, as indicated by the elevation of autophagy marker LC3-II (the phosphatidylethanolamine conjugate of LC3; LC3 proteins are also known as MAP1LC3 proteins), in both pHBLV and pLL3.7 empty vector-expressing IMR90 cells after Ras induction for 5–6 days, when compared to IMR90 cells without Ras induction (Fig. 5A–E; Fig. S4, the left-hand three lanes). Conversely, PKD1 mutants PKD1-P157G and PKD1-P281G, as well as PLD1, ARF1 and PI4KIIIβ depletions, markedly abolished p-JNK, ATF4 and LC3-II expression (Fig. 5A–E; Fig. S4, the two right-hand lanes). In addition, the basal levels of JNK were barely changed in these cells. These results suggested that ER stress and autophagic activity triggered by the Ras induction were attenuated by deletion of PKD1-mediated protein secretory pathway. The overexpression of PKD1-P157G and PKD1-P281G, and the phosphorylation levels of PKD1, as well as the knockdown of PLD1, ARF1 and PI4KIIIβ, were confirmed by western blotting (Fig. 5A–E; Fig. S4). We also observed that inhibition of this pathway reduced the level of the senescent marker p16 induction compared to that seen in mock cells, which implies that Ras OIS entry is postponed, which is consistent with the SA-β-gal activity, cell proliferation and BrdU incorporation results (Fig. 5A–E; Fig. S4).
Targeting protein secretory pathway induces senescent cell death
The finding that perturbation of PKD1-mediated protein secretory pathway causes the massive intracellular accumulation of cytokines in senescent cells led us to further explore what consequence this may exert on the established senescent cells. To determine this, we exposed normal ER:Ras IMR90 cells without Ras induction and Ras-induced senescent ER:Ras IMR90 cells to BFA or L-CS, respectively. To avoid the potential toxicity to cells, high doses of BFA or L-CS were removed after 2 h treatment. A massive IL-6/IL-8 accumulation occurred in the two inhibitor-treated Ras OIS senescent cells compared to the untreated senescent cells (Fig. 6A,B, six right-hand lanes). In contrast, no IL-6/IL-8 were detected in normal ER:Ras IMR90 cells in the absence or presence of inhibitor treatment alone. We noticed that exposure to a high concentration of BFA induced IL-6 expression in ER:Ras IMR90 cells, but this did not occur after a low dose of BFA treatment (Fig. 6A, left-most six lanes). Therefore, the induction of IL-6 by a high level of BFA exposure may contribute to its toxic side effect. We next examined whether the inhibitor treatment-induced cytokine accumulation provoked death of the senescent cells. Indeed, we observed that inhibition of ARF1 or DAG triggered apoptosis of senescent cells, as indicated by the induction of cleaved caspase 3 (Dörr et al., 2013; Green and Llambi, 2015) (Fig. 6A,B). Conversely, hardly any apoptotic signal was detected in control cells treated with these inhibitors, even in cells treated with a high dose of BFA. To further confirm that BFA or L-CS treatment induces apoptosis in senescent cells but not in normal cells, we used Annexin V conjugated to FITC and propidium iodide (PI) to double stain the cells, and then performed flow cytometry detection. The FACS results demonstrated that the proportion of apoptotic cells in BFA or L-CS-treated Ras-induced senescent IMR90 cells was much higher than that in BFA- or L-CS-treated normal young IMR90 cells (Fig. S5A), which is consistent with the cleaved caspase 3 results. These results indicate that classical protein secretory pathway is not only responsible for SASP factor secretion, but also can serve as the target to selectively eliminate SASP-producing senescent cells.
To further extend our study, we treated the human MCF-7 breast cancer cell line with a low dose of doxorubicin (Dox) for a short period to induce MCF-7 cell senescence (Elmore et al., 2002), then followed with L-CS, FIPI or BFA addition to determine whether inhibition of the PKD1-mediated classical protein secretory pathway also abolishes IL-6/IL-8 secretion, and causes the intracellular accumulation of IL-6/IL-8, and eventually provokes apoptosis in therapy-induced senescent (TIS) cancer cells. The senescent state of Dox-treated MCF-7 cells was characterized by increased SA-β-gal activity (Fig. S5B). Similar to the Ras OIS results, L-CS, FIPI or BFA treatment of senescent MCF-7 cells significantly reduced IL-6/IL-8 secretory levels, and elevated the intracellular IL-6/IL-8 protein levels, when compared with the untreated senescent MCF-7 cells (Fig. S6). These results demonstrate that the PKD1-mediated classical protein secretory pathway also regulates IL-6/IL-8 secretion in TIS cells, and inhibition of this pathway causes the intracellular accumulation of IL-6/IL-8 in TIS cells.
Like with the Ras OIS results, BFA or L-CS treatment also induced senescent apoptotic cell death in MCF-7 cells, as characterized by the induction of cleaved PARP-1, due to the intracellular accumulation of IL-6/IL-8. By contrast, only full-length PARP-1 was detected and no cleaved PARP-1 was present in BFA or L-CS-treated normal MCF-7 cells (Fig. 6C,D). Since a relatively high basal level of IL-6 is expressed in MCF-7 cells (Simone et al., 2015), we also observed that BFA and L-CS exposure led to the IL-6 elevation in normal MCF-7 cells. However, these treatments could barely trigger cell death in normal MCF-7 (Fig. 6C,D). The apoptosis of BFA- or L-CS-treated senescent MCF-7 cells was further shown by Annexin V–FITC and PI double-staining results (Fig. S5C). To confirm the specific effects of BFA or L-CS treatment on inducing death in senescent MCF-7 cells, the IC50 values for BFA- and L-CS treatment of normal MCF-7 cells were analyzed by means of an MTT assay. As shown in Fig. S7A,B, after 24 h treatment, the IC50 values for BFA and L-CS treatment of MCF-7 cells were 4.2 μM and 78.8 mM, respectively, which was much higher than the corresponding concentrations of BFA (4 nM) or L-CS (2.5 mM) we used in the 24 h treatment of MCF-7 cells. As shown in Fig. S7A,B, there was negligible cell death observed when 4 nM BFA or 2.5 mM L-CS were used to treat normal MCF-7 cells for 24 h. Therefore, this result ruled out the possibility that BFA or L-CS treatment induced non-specific toxic cell death in MCF-7 cells. Taken together, our observations imply that protein secretory pathway in SASP factor-producing senescent cells may represent a senescence-specific target that selectively eliminates senescent cells.
In view of senescent cells developing the SASP, and SASP factor-associated autocrine and paracrine effects, it is plausible to speculate that cellular senescence will induce the activity of protein secretory pathway and that blockage of SASP factor secretion will affect functions of SASP factor. However, much effort had been made to elucidate the SASP regulation and function, with very little known about how SASP factor secretion is regulated during senescence, and what the role of the SASP secretory pathway might have on the cellular senescence. In this study, we show that the PKD1-mediated conventional protein secretory pathway from the TGN to the cell surface is activated during cellular senescence and is involved in regulating SASP factor secretion. Inhibition of this pathway affects the Ras OIS response, or induces various types of cell death in senescent cells.
The core SASP secretory cytokines IL-6/IL-8 establish a self-amplification network to reinforce cell senescence in an autocrine manner, and disruption of this network attenuates OIS entry and maintenance (Acosta et al., 2008; Kuilman et al., 2008). However, whether blockage of the IL-6/IL-8 secretory pathway will alleviate Ras OIS response had not previously been elucidated. We observed that impairment or deprivation of the essential components of classical protein secretory pathway, such as PKD1, DAG, PLD1, ARF1 and PI4KIIIβ, decreased IL-6/IL-8 secretory levels during Ras OIS and eventually retarded Ras OIS (Figs 3–5). These results imply that the PKD1-mediated classical protein secretory pathway is not merely involved in regulating protein secretion, but also involved in Ras OIS responses through mediating IL-6/IL-8 secretion and the establishment of the IL-6/IL-8 autocrine network. Previously, we identified PKD1 acted as a downstream effector of the reactive oxygen species (ROS)-protein kinase Cδ (PKCδ) cascade to mediate Ras OIS by regulating IL-6/IL-8 expression via modulation of NF-κB activity (Wang et al., 2014). Therefore, we propose that, during Ras OIS, the ROS–PKCδ–PKD1 axis modulates SASP factor expression, and the PKD1-mediated classical protein secretory pathway from the TGN to the cell surface regulates SASP factor secretion. Moreover, both processes are not exclusive and they all have significant impact on the Ras OIS response.
High production of SASP factors in SASP-capable senescent cells triggers hypermetabolic activity to meet the energy demand for massive secretory protein biosynthesis, folding, post-translational processing and secretion. However, this may exceed the cellular capacity to accurately cope with such high levels of protein processing, thereby causing ER stress and the unfolded protein response (UPR), and ultimately inducing autophagic activity and ubiquitylation to degrade misfolded proteins for cell survival. In contrast, senescent cells with a low level of SASP factor secretion are virtually absent of ER stress and autophagy (Dörr et al., 2013). Therefore, SASP factor production has been linked to hypermetabolism, ER stress, UPR and autophagy in senescent cells, and can be viewed as a manifestation of various senescence-inducing stresses, such as oncogenic stress, DNA damage stress and replicative exhaustion stress. Our results demonstrate that interruption of PKD1-mediated protein secretory pathway mitigates ER stress and autophagy during the establishment of Ras OIS through reducing senescence-associated IL-6/IL-8 levels (Figs 3 and 5). Thus, we speculate that protein secretory pathway may also serve as a stress transducer to mediate the cellular senescence response. Inactivation of this pathway may largely relieve senescence-inducing stresses, thereby delaying cellular senescence processes. In contrast, inhibition of autophagy in SASP factor-producing senescent cells led to further accumulation of misfolded proteins, which triggered senescent cell death (Dörr et al., 2013). Similarly, we also found that targeting of PKD1-mediated protein secretory pathway in Ras-induced senescent cells or therapy-induced senescent cancer cells not only reduced IL-6/IL-8 secretion, but also caused massive intracellular accumulation of IL-6/IL-8 (Figs 2 and 6; Fig. S6), which selectively induced senescent cell death (Fig. 6). Hence, inhibition of this pathway in SASP factor-producing senescent cells further amplified the stresses, thereby triggering senescent cell death. Based on these results, we propose a model as shown in Fig. 7 to summarize our findings.
Senescent cells accumulate in various tissues and organs with age (Collado et al., 2007). Selective elimination of senescent cells prevent or delay age-related disorders and extend healthspan in mouse model (Baker et al., 2016, 2011). In addition, cancer therapy-induced senescence (TIS) can induce senescence and SASP in tumor cells. Although TIS facilitates cancer treatment outcome because of its associated immune surveillance, the resident and viable senescent cells in vivo may cause cancer relapse later (Gonzalez et al., 2016). Therefore, selective removal of normal senescent cells may improve healthy aging, and selective eradication of TIS cells may promote the long-term efficacy of TIS. However, selective clearance of senescent cells is a challenging task. SASP-competent TIS cells exhibit hypermetabolic activity and autophagic activity, which present as a senescent cell-specific vulnerability that can be selectively targeted to induce senescent cell death (Dögrr et al., 2013). Recently, some senolytic drugs such as the combination of Dasatinib and Quercetin (Zhu et al., 2015), and ABT263 (Chang et al., 2016), were identified to be able to inhibit anti-apoptotic proteins BCL-2 and BCL-xL in senescent cells, and thereby selectively kill senescent cells. In our work, we found that SASP factor-producing senescent cells showed an increased activity of classical protein secretory pathway, which was not seen in normal cells, and thus may represent another senescent cell-specific weakness. Intervention of this pathway leads to intracellular accumulation of massive cytokines and ultimately induces OIS and TIS apoptotic cell death (Fig. 6). Therefore, our discovery may provide potential targets for selectively erasing senescent cells to mitigate deleterious effects of the SASP.
In summary, we show that the PKD1-mediated classical protein secretory pathway is activated in senescent cells and is essential for SASP factor secretion. Ablation of this pathway either delays the onset of Ras OIS, or triggers senescent cell death.
MATERIALS AND METHODS
IMR90 human diploid fibroblasts (HDFs) transduced with an ER:RAS (kindly provided by Masashi Narita, Cancer Research U.K., Cambridge Research Institute, Cambridge, UK), 293T cells, MCF-7 cells and human diploid 2BS fibroblast cells (National Institute of Biological Products, Beijing, China) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). To induce senescence, IMR90 cells expressing ER:H-RasV12 (ER:Ras-IMR90 cells) were given 100 nM 4-hydroxytamoxifen (4-OHT).
Plasmids, antibodies and reagents
Full-length protein kinase D1 (PKD1) was cloned by PCR from cDNA of 2BS cells, and then inserted into the EcoRI and BamHI restriction enzyme sites of the pHBLV-puro vector, using primers: forward, 5′-GGAATTCCATGAGCGCCCCTCCGGTCCTGCG-3′ and reverse, 5′-CGGATCCGTCAGAGGATGCTGACACGCTCACCG-3′.
pHBLV-PKD1-P157G and pHBLV-PKD1-P281G were obtained by PCR using pHBLV-PKD1 as template, and the primers of site-mutants were as follows: pHBLV-PKD1-P157G, forward: 5′-TCATTCATACAGAGCTGGAGCTTTCTGTG-3′ and reverse, 5′-CCAGCTCTGTATGAATGAACAAAGAGAGC-3′; pHBLV-PKD1-P281G, forward, 5′-TGTCATCCACTCCTACACCCGGGGCACAGTGTGC-3′ and reverse, 5′-CCCCGGGTGTAGGAGTGGATGACAAATGTGTGCG-3′.
The shRNA was designed according to the pLentiLox 3.7-puro vector instruction manual. The double-stranded RNAs were synthesized by GENEray Biotechnology of Beijing and then inserted into the HpaI and XhoI sites of the pLentiLox 3.7-puro vector. The human PLD1-specific shRNA target sequences used were: pLL3.7-PLD1, 5′-GUUAAGAGGAAAUUCAAGCTT-3′ and pLL3.7-PLD1#2, 5′-GGTGGGACGACAATGAGCATT-3′. The human ARF1-specific shRNA target sequences used were: pLL3.7-ARF1, 5′-GUGUACUCGACAGAUUAUCTT-3′ and pLL3.7-ARF1#2, 5′-CACCAUAGGCUUCAACGUGTT-3′. The human PI4KIIIβ-specific shRNA target sequences used were: pLL3.7- PI4KIIIβ, 5′-GGGAUGACCUUCGGCAAGATT-3′ and pLL3.7- PI4KIIIβ#2, 5′-GAGAUCCGUUGCCUAGAUGTT-3′.
The primary antibodies used for western blotting analysis were as follows: anti-PKD1 (Cat#2052), anti-p-PKD1 (Ser744/Ser748) (Cat#2054), anti-p-PKD1 (Ser916) (Cat#2051), anti-LC3-II (Cat#4108), anti-p-JNK (Thr183/Tyr185) (Cat#4671), anti-JNK (Cat#9252), anti-PLD1 (Cat#3832) anti-PARP1 (Cat#9532) and anti-cleaved caspase 3 (Cat#9664) from Cell Signaling Technology; anti-human IL-6 (Cat#MAB2061) and human IL-8 (Cat#MAB208) from R&D Systems; anti-ARF1 (Cat#20226-1-AP) and anti-p16 (Cat#10883-1-AP) from Proteintech; anti-ATF4 (Cat#ab184909), anti-Calnexin (Cat#ab22595) and anti-pro-caspase 3 (Cat#ab32150) from Abcam; anti-PI4KIIIβ (Cat#NBP2-12814) from Novus Biologicals; anti-TGN46 (Cat#AHP500) from AbD Serotec; and anti-GAPDH (Cat#AP0063) and anti-β-tubulin (Cat#BS1482M) from BioWorld. For western blotting, anti-GAPDH and anti-β-tubulin antibodies were used at 1:5000. All other primary antibodies were used at 1:1000. For immunofluorescence, all primary antibodies were used at 1:200.
BAPTA (used at 15 µM, Cayman), U73122 (used at 5 µM, Cayman), U73343 (used at 5 µM, Santa Cruz Biotechnology), D609 (used at 100 µM, Santa Cruz Biotechnology), C-6NBD (used at 5 µM, Cayman), m-3M3FBS (used at 100 µM, Santa Cruz Biotechnology), FIPI (used at 250 nM, MedChem Express) and thapsigargin (used at 5 µM, ENZO) were dissolved in DMSO (Sigma); Tamoxifen (used at 100 nM, Sigma) and BFA (used at 3.6 µM, Alomone) were dissolved in methanol; Propranolol (used at 50 µM, ENZO) and L-Cycloserine (used at 25 mM, Santa Cruz Biotechnology) were dissolved in sterile water. 1-butanol and 2-butanol (both used at 0.5%) were from Adamas.
Cells were washed twice with ice-cold 1× PBS, harvested, and lysed in radioimmune precipitation assay buffer (RIPA buffer; Applygen Technologies) with a phosphatase inhibitor tablet (Roche Diagnostics) and protease inhibitor mixture (Fermentas). Cell lysates were then centrifuged for 15 min at 15,000 g and 4°C, and the insoluble debris was discarded. Protein concentration was determined by using BCA protein assay reagent (Pierce). Cell lysate (20–40 μg) was subjected to 8–15% SDS-PAGE and proteins were transferred onto nitrocellulose membranes (Millipore). The primary antibodies used for western blot analysis were as described above.
The supernatants of IMR90 cells were collected after treatment as indicated. Human IL-6 and IL-8 levels were measured with an ELISA kit from Boster.
IMR90 cells or 2BS cells were seeded on glass coverslips or in 96-well plate (View Plate-96 F), washed with PBS, and then ﬁxed in 4% paraformaldehyde at room temperature (RT) for 10 min and washed with PBS. Next cells were permeabilized with 0.5% Triton X-100 in PBS (5 min, RT) and blocked with blocking buffer (5% BSA and 0.05% Tween 20 in TBS) for 1 h at RT. After that cells were incubated with the primary antibodies diluted in TBST for 4°C overnight. After washing with PBS and incubated with secondary antibodies diluted in 5% milk for 1 h in dark and RT, cells were analyzed by a Confocal Laser Scanning Microscope (Olympus FV1000). Images were processed with Adobe Photoshop.
Imaging was carried out using an automated high-throughput microscope (High Content Screening System Operetta, Perkin Elmer). HCA was carried out using the Harmony software (version 3.5; Perkin Elmer). Briefly, DAPI staining (blue) of the nuclei was used to identify cells, TGN46 staining (green) to define TGN, and p-PKD1 (Ser738/742) staining (red) to detect PKD1 activity. The nuclei were segmented using the building block ‘Find Nuclei’. The cell area was defined using the building block ‘Find Cytoplasm’. The normalized total TGN46 staining intensity value for each young or senescent cell was obtained by subtracting the cytoplasmic intensity (background intensity) from raw intensity values. The minimum and maximum threshold filters were set to determine positive p-PKD1 (Ser738/742) signals on the TGN46-staining regions. The total p-PKD1 (Ser738/742) staining intensity value for each young or senescent cell was obtained by measuring the intensity values of nine positive p-PKD1 (Ser738/742) signal spots on the TGN46 staining fields and summing them. The intensity ratio of p-PKD1 (Ser738/742) to TGN46 was calculated to represent the relative PKD1 activity on the TGN for each young or senescent cell. A minimum of 50 cells were measured in each duplicate experiment to calculate a mean ratio. The mean intensity ratio values per cell were used to calculate fold changes among young and senescent cells. Error bars and P-values were derived from independent biological replicates. The antibodies used for the analysis were validated with robust controls (shRNAs or overexpression) to assess their specificity. Statistical analysis of immunofluorescence data was conducted in Microsoft Excel 2013.
Isolation of Golgi
The Golgi isolation was performed by using the Golgi apparatus isolation kit (GENMED SCIENTIFICS Inc.) according to the manufacturer's instruction. Briefly, cells were washed twice with cold 1× PBS and resuspended in 10 ml of Reagent A, and then centrifuged at 300 g for 10 min at 4°C. Next, 5 ml of reagent lysis buffer was added to cells with and mixed thoroughly by vortexing. Lysates were centrifuged twice at 1000 g for 10 min at 4°C to remove cell nucleus and undissolved cells, and the supernatant was centrifuged at 10,000 g for 20 min at 4°C, and then the supernatant was centrifuged at 100,000 g for 60 min at 4°C to obtain microsomes. The precipitate was resuspended in 1 ml reagent F and 5 ml reagent G, and centrifuged for 2.5 h at 365,000 g at 4°C to pellet Golgi. Golgi pellets were resuspended in reagent F and centrifuged at 100,000 g for 10 min at 4°C to obtain a pure component. The supernatant was discarded and 0.5 ml reagent F was added into the Golgi preparation for long-term stable conservation.
SA-β-gal activity assay
For SA-β-gal staining, cells were washed twice with ice-cold 1× PBS, fixed for 10 min at room temperature in 3% formaldehyde, and washed twice with 1× PBS. The cells were then incubated overnight at 37°C without CO2 in a freshly prepared SA-β-gal staining solution.
Cell growth curves
Cell proliferation was assayed using the 3-(4,5-dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide (MTT) method. Cells were seeded into a 96-well plate at a density 2×103 cells per well and cultured for periods ranging from 1 to 6 days. The medium was changed every 24 h. At the indicated times, an aliquot of cells were stained with 25 μl of MTT solution (5 mg/ml in 1× PBS; Sigma) for 4 h and then disrupted with DMSO. The optical density at 570 nm was determined.
BrdU incorporation assay
IMR90 cells or 2BS cells were seeded on glass coverslips with different treatments. Next, cells were incubated with bromodeoxyuridine (BrdU) for 16 h and then stained with anti-BrdU antibody (1:1000; Cat# 6813, Cell Signaling Technology) for immunofluorescence.
Annexin V–FITC and PI apoptosis detection assay
IMR90 cells or MCF-7 cells were gently trypsinized and washed once with serum-containing medium, then collected by centrifugation (300 g for 5 min). Cells were re-suspended in 1× binding buffer, and Annexin V–FITC and PI were added. Cells were incubated at room temperature for 5 min in the dark, then Annexin V–FITC and PI staining of the cells was analyzed by flow cytometry.
An unpaired two-tailed Student’s t-test was used to determine the significance of differences between samples indicated in figures. Results are depicted as mean±s.d. P<0.05 (*) or P<0.01 (**) were considered significant.
We thank Dr Masashi Narita for plasmids and cell lines.
Conceptualization: J.C.; Methodology: Y.S., P.W., H.S., Z.S., C.X.; Software: G.L.; Validation: J.C.; Formal analysis: Y.S., P.W., H.S., Z.S.; Investigation: Y.S., P.W., H.S.; Resources: G.L.; Data curation: Y.S., P.W., H.S.; Writing - original draft: P.W., J.C.; Writing - review & editing: J.C.; Visualization: T.T.; Supervision: T.T.; Project administration: G.L.; Funding acquisition: T.T., J.C.
This work was supported by the Ministry of Science and Technology of the People's Republic of China (MOST) (grants 2013CB530801 and 2014CB910503), and the National Natural Science Foundation of China (NSFC) (grant 81370455).
The authors declare no competing or financial interests.