IκBζ is a regulator of the senescence-associated secretory phenotype in DNA damage- and oncogene-induced senescence

In addition to apoptosis, cellular senescence has been recognized as a potent tumor-suppressive mechanism that is induced by various stress stimuli including the exposure to chemotherapeutic drugs or irradiation, telomer erosion or abnormal oncogene activation (Adams, 2009; Kuilman et al., 2010; Campisi, 2013; Acosta and Gil, 2012). A hallmark of senescent cells is a sustained cell cycle arrest that is established by the p53 and retinoblastoma tumor-suppressive pathways and maintained by an upregulation of the p16^Ink4a and p21^WAF/CIP1 cyclin-dependent kinase (CDK) inhibitors. Cells undergoing senescence display a number of characteristic features, such as a flattened and enlarged morphology and increased senescence-associated β-galactosidase (SA-β-gal) activity, which is often used as a marker to detect senescent cells (Dimri et al., 1995). Senescent cells also show widespread chromatin modifications, known as senescence-associated heterochromatin foci (SAHF), which contain chromatin repression marks, such as phosphorylated heterochromatin protein 1 γ (pHP1γ), HMGA1, histone H3 trimethylation at lysine-9, or the histone variant macroH2A (Narita et al., 2003; Adams, 2007; Sulli et al., 2012).

In addition, it has been identified that senescent cells secrete a number of growth and immune factors, a feature termed senescence-associated secretory phenotype (SASP) or senescence-messaging secretome (Coppé et al., 2008; Coppé et al., 2010; Kuilman and Peeper, 2009). Secreted factors include several interleukins, chemokines, growth factors and matrix metalloproteinases. It was proposed that the SASP may act as a doubled-edged sword and, depending on the physiological context, can exert tumor-promoting or suppressive activities. Several SASP factors have the ability to promote tumor cell proliferation, invasion or angiogenesis (Krtolica et al., 2001). On the other hand, more recent evidence shows that interleukin-6 (IL6) or IL8, which are conserved key factors of the SASP, could reinforce the senescent growth arrest by promoting reactive oxygen production or by enhancing the DNA damage response in a positive feedback loop (Kuilman et al., 2008; Acosta et al., 2008). Finally, these cytokines or other secreted factors have been shown to attract immune cells in an inflammatory response, leading to the elimination of senescent tumor cells (Kang et al., 2011; Xue et al., 2007; Hoenicke and Zender, 2012).

Despite the potential importance of the SASP, so far little is known about its mechanism of regulation, its relation to DNA damage or its impact on different cancer types. It was observed that inhibition of the DNA damage-responsive kinases ATM and Chk2 prevents the release of some SASP components, whereas p53 obviously restrains SASP formation (Rodier et al., 2009). Important transcriptional activators of the SASP are nuclear factor kappaB (NFκB) and CCAAT/enhancer-binding protein (C/EBPβ) as well as p38 kinase, which appears to be required for sustained NFκB activation (Kuilman et al., 2008; Rovillain et al., 2011; Chien et al., 2011; Jing et al., 2011; Freund et al., 2011). Furthermore, membrane-bound IL1α has been proposed as a general autocrine regulator of the SASP, because blockade of IL1 signaling reduced IL6 and IL8 secretion in senescent human fibroblasts (Orjalo et al., 2009). It was reported that transcription of these cytokines was dependent on IL1-mediated NFκB activation. Yet, it is largely unknown how NFκB, which is certainly a master regulator of the SASP, is activated during senescence.

NFκB is classically activated by the IκB kinase (IKK)-induced phosphorylation and subsequent proteasomal degradation of its cytoplasmic inhibitor IκBα, which enables nuclear translocation of NFκB and transcriptional activation of a diverse array of target genes involved in various biological processes (Oeckinghaus et al., 2011). Recent evidence, however, suggests that the activation of NFκB target genes is more complex and dependent on the particular target gene context or stimulus, which is thought to facilitate a selective gene regulation in distinct physiological settings (Smale, 2011). Whereas rapid activation of primary response genes is directly induced by the classical NFκB pathway, expression of so-called secondary response genes is delayed and requires prior protein synthesis of additional co-regulators. In this context, we and others have recently identified the inhibitor of NFκB ζ (IκBζ, encoded by the gene Nfkbiz), an atypical nuclear IκB protein, which is not regulated by phosphorylation-induced degradation and can act as a repressor but, more importantly, also as an activator of a selective subset of NFκB target genes (Totzke et al., 2006; Yamazaki et al., 2001; Kitamura et al., 2000). IκBζ itself is a primary response target gene and, by association with the NFκB subunit p50, is thought to exert its transcription-enhancing activity on secondary response genes mainly at the level of chromatin remodeling (Kayama et al., 2008; Yamazaki et al., 2008).

In the present study, we found that IκBζ is upregulated in several models of senescence. Moreover, we demonstrate that expression of IL6 and IL8, two highly conserved SASP cytokines, requires IκBζ in both DNA damage- and oncogene-induced senescence, thus establishing IκBζ as an essential novel regulator of SASP formation.

Previously, we reported that ionizing radiation induces senescence rather than apoptosis in the p53 wild-type-expressing human breast cancer cell line MCF7 (Essmann et al., 2005). In order to establish MCF7 cells as a model for irradiation-induced senescence, we investigated further senescence markers, in addition to the enlarged morphology and SA-β-galactosidase activity (Fig. 1A). Immunofluorescence microscopy of γ-irradiated MCF7 cells revealed typical punctuate structures of phosphorylated heterochromatin protein 1γ (pHP1γ) in the nuclei (Fig. 1A), and real-time PCR (RT-PCR) demonstrated an ∼10-fold higher expression of the CDK inhibitor p21 (Fig. 1B) 5 days post irradiation. Furthermore, mRNA expression of the central SASP cytokines IL6 and IL8 (Fig. 1B) as well as of other SASP components (data not shown) was significantly enhanced in γ-irradiated cells. We confirmed cytokine secretion by analyzing cell culture supernatants in flow cytometric bead assays that also showed enhanced expression of IL6 and IL8 (Fig. 1C). Interestingly, in addition to the established SASP factors, also expression of IκBζ, which was almost undetectable in control cells, was increased about 5-fold in senescent MCF7 cells (Fig. 1B).

In order to verify the results from irradiation-induced senescence, we additionally investigated therapy-induced senescence and therefore incubated MCF7 cells with bleomycin, a DNA-damaging chemotherapeutic drug. Similar to irradiation, bleomycin treatment induced the typical senescence alterations, including an enlarged morphology, increased SA-β-gal activity as well as a strong nuclear recruitment of pHP1γ (Fig. 1D). Moreover, compared to control cells, RT-PCR analyses revealed a more than 15-fold higher expression of p21 as well as a strong induction of IL6 and IL8 mRNA and protein expression (Fig. 1E,F). Importantly, the expression of IκBζ was increased 3.6-fold, indicating that upregulation of IκBζ is independent of the senescence-inducing stimulus.

To confirm that IκBζ is generally induced during senescence, we further investigated oncogene-induced senescence by Ras^G12V expression (Serrano et al., 1997). To this end, we took advantage of the pInducer lentiviral vector system (Meerbrey et al., 2011) and generated MCF7/Ras^G12V cells that expressed oncogenic Ras^G12V protein under the control of the tetracycline-responsive element (TetOn). Doxycycline treatment resulted not only in a strong induction of Ras^G12V, but concomitantly induced senescence alterations including increased p21 expression, SA-β-gal activity and the characteristic pHP1γ staining (Fig. 2A–C). Furthermore, in concert with p21, IL6 and IL8 mRNA expression were significantly induced (Fig. 2D). The expression of the senescence markers was again associated with a strongly increased IκBζ expression, indicating that, regardless of the stimulus, IκBζ is induced in both DNA-damage and oncogene-induced senescence.

IκBζ has previously been identified as an important regulator of IL6 expression in the Toll-like receptor/IL1 receptor pathway (Yamamoto et al., 2004; Seshadri et al., 2009). To investigate a functional role of IκBζ for SASP formation, we transfected MCF7/Ras^G12V cells with an siRNA targeting IκBζ. Owing to the long time course of the experiments, siRNA transfection only resulted in an ∼50% reduction of IκBζ expression (Fig. 2E). Nevertheless, despite the incomplete knockdown, downregulation of IκBζ upon Ras-induced senescence resulted in a significantly reduced IL6 and IL8 expression compared to a non-targeting control siRNA (Fig. 2E). In contrast, mRNA expression of Ras and p21 remained unaffected by the IκBζ-specific siRNA during doxycycline treatment (Fig. 2E).

In order to corroborate IκBζ-dependent expression of SASP components, we next generated a HeLa cell line for the doxycycline-inducible expression of IκBζ. Doxycycline treatment of the resulting cell line HeLa/TetOn-IκBζ triggered a robust protein expression of IκBζ (Fig. 3A). At the mRNA level, IκBζ expression was increased more than 20-fold, which was further induced during irradiation-induced senescence (Fig. 3B,C). In line with the previous experiments, transgenic expression of IκBζ resulted in strongly enhanced IL6 and IL8 expression (Fig. 3D,E). Unlike the SASP components, p21 was only slightly but not significantly affected by doxycycline-induced IκBζ expression (Fig. 3F). Thus, in different senescence models modulation of IκBζ levels affects SASP cytokine expression.

To further substantiate the role of IκBζ for SASP formation, we employed mouse embryonic fibroblasts (MEFs) from IκBζ-deficient (Nfkbiz^−/−) mice. MEFs from wild-type (WT) and Nfkbiz^−/− mice were induced to undergo senescence by irradiation or incubation with bleomycin. Both treatments resulted in senescence as verified by a β-galactosidase assay of cytosolic extracts (Fig. 4A,D). In line with the results in MCF7 cells, neither irradiation nor bleomycin treatment resulted in a significant difference in p21 induction in IκBζ-deficient and WT MEFs (Fig. 4B,E). However, IκBζ-deficient MEFs were almost completely unable to upregulate IL6 and the murine IL8 orthologue CXCL1, whereas both cytokines were strongly induced in WT MEFs upon irradiation. In addition, IκBζ deficiency completely abolished the induction of IL6 and CXCL1 upon bleomycin treatment. These results were additionally confirmed by the measurement of IL6 and CXCL1 cytokines in the culture supernatant of control and senescent WT and Nfkbiz^−/− MEFs. The amount of IL6 and CXCL1 protein was exclusively increased in supernatants from senescent WT MEFs but not from Nfkbiz^−/− MEFs (Fig. 4C,F). Thus, these data further underscore our results of the knockdown experiments in MCF7 cells and unambiguously establish IκBζ as an indispensable regulator for the central SASP components IL6 and IL8/CXCL1.

In human fibroblasts expression of SASP cytokines was previously proposed to depend on autocrine IL1α signaling (Orjalo et al., 2009). To delineate the position of IκBζ in IL1α signaling and SASP formation, we first incubated MCF7 cells with IL1α, which clearly induced IκBζ as well as IL6 and IL8 (Fig. 5A). IL1α-mediated induction of IκBζ as well as of the cytokines was strongly impaired by the clinical IL1 receptor antagonist Anakinra (Dinarello et al., 2012). In contrast, blockade of the IL1 receptor pathway by Anakinra did neither affect induction of IκBζ nor IL6 and IL8 expression in irradiated MCF7 cells (Fig. 5B). In addition, upon incubation of MCF7 cells with bleomycin, the induction of IκBζ or the SASP components remained largely unaffected by Anakinra (Fig. 5C), indicating that IL1α is not required for senescence-induced IκBζ or cytokine expression.

As IL1α has been proposed to mediate SASP induction (Orjalo et al., 2009), we wished to further corroborate our data in another experimental system and therefore used MEFs deficient in MyD88, an essential component of the IL1 signaling pathway (Gay et al., 2011). As expected, IL1α induced upregulation of IκBζ, IL6, and CXCL1 in WT MEFs, but not in MyD88^−/− cells (Fig. 5C). In contrast, we detected no significant differences in the upregulation of IκBζ and the cytokines between WT and MyD88^−/− MEFs upon senescence induction by either irradiation or bleomycin treatment (Fig. 5D,E). Hence, our data suggest that not only IκBζ can be regulated independently of IL1 signaling, but that also for expression of the SASP components IL1 signaling is dispensable.

In addition to NFκB, other transcription factors might be involved in the regulation of SASP formation. To identify which SASP components are modulated by IκBζ, we performed RT-PCR analyses for additional SASP genes in WT and Nfkbiz^−/− MEFs. Both bleomycin treatment as well as irradiation led to strong induction of the mRNA levels of monocyte chemoattractant protein-1 (MCP1, CCL2) in WT but not IκBζ-deficient MEFs (Fig. 6A), indicating that MCP1 is regulated by IκBζ. In contrast, there were no significant differences in the senescence-associated induction of IL15 mRNA levels (Fig. 6B). Moreover, we found that, in addition to IL6, CXCL1/IL8 and MCP1, bleomycin-induced transcription of MCP2 and MIF was suppressed in IκBζ-deficient as compared to WT cells (Fig. 6C). In contrast, the mRNA levels of other SASP products, such as TIMP1, TIMP2 and fibronectin, were even more strongly induced in IκBζ-deficient cells, which might be due to the fact that IκBζ can additionally function as an inhibitor of particular target genes (Totzke et al., 2006; Yamazaki et al., 2001). In conclusion, IκBζ drives the expression of a subset of SASP factors, but these include in particular potent chemokines.

The present study demonstrates that the atypical nuclear IκB protein, IκBζ, is an essential mediator required for the induction of conserved SASP cytokines. Although a few recent studies have already suggested an involvement of NFκB, in particular of its subunit p65, in SASP formation (Rovillain et al., 211; Chien et al., 2011; Jing et al., 2011; Freund et al., 2011), it has become clear that NFκB target genes can be classified in two groups, namely primary and secondary response genes, of which the latter require additional co-factors for transcriptional activation (Smale, 2011). Our results suggest that several SASP components, including IL6, IL8, MCP1 and others belong to those secondary response genes that require prior induction of IκBζ. IκBζ itself is a primary NFκB target, whose transcriptional induction depends on the proteasomal degradation of cytoplasmic IκBα and subsequent NFκB p65 activation, suggesting that expression of the SASP components requires a two-step mechanism (Fig. 7). It is currently thought that such gene-specific regulation and mechanistic diversity between a primary and secondary NFκB response ensure a selectivity and precise control of NFκB target gene activation in distinct physiological settings, including apoptosis, inflammation, senescence and other NFκB-controlled processes.

We studied different models of senescence in human and murine cells and show that IκBζ is induced in both oncogene- and DNA damage-induced senescence. Although we have tried several commercial antibodies, immunoblot analyses did only yield reliable data on IκBζ protein expression in the HeLa overexpression system. This might be caused by the fact that, in contrast to acute proinflammatory stimulation with e.g. IL1 or LPS, SASP formation requires a more sustained but weaker inflammatory and DNA damage response (Adams, 2009; Kuilman et al., 2010; Campisi, 2013; Acosta and Gil, 2012). In fact, the induction of secondary response genes is generally delayed compared to primary NFκB response genes that are immediately induced. Nevertheless, we show that even a partial RNAi-mediated knockdown of IκBζ is sufficient to reduce mRNA and protein levels of IL6 and IL8 during SASP formation. Vice versa, the inducible expression of IκBζ led to a strong potentiation of IL6 and IL8 expression during senescence. Even more convincingly, we show that during senescence of IκBζ-deficient MEFs induction of IL6 and IL8 expression was completely abolished.

Although the secretory features of the SASP are probably cell type- and stimulus-specific, IL6 and IL8 have been identified as the most conserved and robustly expressed SASP cytokines (Campisi, 2013; Kuilman and Peeper, 2009). In tumors these factors can attract immune cells leading to immune surveillance and subsequent elimination of senescent cells. In addition to the proinflammatory effects, IL6 and IL8 have been shown to reinforce the growth arrest and thereby contribute to the senescence process. For instance, it was demonstrated that depletion of IL6 abolished oncogene-induced senescence and suppressed SAHF formation and p15^Ink4b expression (Kuilman et al., 2008). In addition, IL8 can increase reactive oxygen production and exacerbate DNA damage (Acosta et al., 2008). Although we have not studied such amplification loops in detail, we found no significant difference in the expression of β-galactosidase or p21 in the presence or absence of IκBζ. Thus, autocrine or paracrine effects might also depend on the senescent tumor cell type.

In addition to IL6 and IL8/CXCL1, MCP1 was identified as an IκBζ-dependent SASP cytokine, a finding that is in line with our observation that MCP1 expression in macrophages strictly requires IκBζ (Hildebrand et al., 2013). MCP1 is one of the most potent chemokines and involved in the recruitment of macrophages, favoring tumor cell clearance and cancer regression (Deshmane et al., 2009). On the other hand, MCP1 expression has been identified as a prometastatic factor during senescence of melanoma cells (Ohanna et al., 2011).

The SASP comprises more than 40 different factors that are certainly not all controlled by IκBζ. In fact, we found that senescence-associated expression of TIMPs and fibronectin was apparently even suppressed by IκBζ. In addition to NFκB, C/EBPβ has been implicated in SASP expression (Kuilman et al., 2008). Interestingly, NFκB and C/EBPβ often show synergistic effects in diverse settings of gene regulation (Oeckinghaus et al., 2011). Moreover, there is a close interdependence of C/EBPβ and IκBζ in the regulation of secondary response genes. Recent analyses surprisingly revealed that IκBζ is required for the recruitment of C/EBPβ to the lipocalin-2 promoter, which is an established IκBζ target gene (Yamazaki et al., 2008). In addition, Kuilman et al. showed by chromatin immunoprecipitation that C/EBPβ was present at both the IL6 and IL8 promoters, suggesting a cooperative role of IκBζ and C/EBPβ for SASP induction (Kuilman et al., 2008).

The mechanism of how DNA damage triggers NFκB activation and subsequent induction of primary target genes such as IκBζ is unclear. Although p53 is not required for SASP formation, available evidence suggests that NFκB activation is mediated by the DNA damage-responsible ATM kinase, resulting in IKK activation and IκBα degradation (Miyamoto, 2011). In this context, however, different mechanisms of ATM activation, involving its interaction with IKK subunit NEMO as well as with SUMO ligase PIASγ, RIG1, PARP1 and presumably additional molecules have been proposed (Miyamoto, 2011; Liu et al., 2011; Hinz et al., 2010; Ohanna et al., 2011). Thus, although the detailed mechanisms of NFκB activation during senescence are unclear, genotoxic stress seems to be linked to NFκB activation via ATM. Interestingly, it was reported that ATM signaling does not regulate the entire SASP, although ATM was required for the secretion of IL6 and IL8 (Rodier et al., 2009).

Senescence is accompanied by massive alterations in chromatin structure and the epigenetic silencing of E2F-regulated genes driving cell proliferation. It is worth mentioning that IκBζ-mediated gene expression largely depends on histone modification and nucleosome remodeling of secondary target genes (Kayama et al., 2008; Yamazaki et al., 2008). It will therefore be interesting to investigate whether IκBζ also influences repressive chromatin marks, such as H3K9 trimethylation in genes of SASP factors.

So far, NFκB has been implicated in tumor biology mainly due to its anti-apoptotic effect involving the transcriptional activation of several apoptosis inhibitors, such as FLIP, XIAP and several anti-apoptotic Bcl2 proteins (Perkins, 2012). In view of this anti-apoptotic but potentially pro-senescent role, activation of NFκB might be considered as a double-edged sword. Using gene expression arrays of WT and IκBζ-deficient cells we interestingly found that, in contrast to chemo- and cytokines, IκBζ is not involved in the transcriptional regulation of anti-apoptotic genes (unpublished data). Thus, it is tempting to speculate that also in this respect IκBζ might confer specificity to the NFκB response.

MCF7 cells were maintained in RPMI-1640 medium (PAA Laboratories, Linz, Austria), supplemented with 10% fetal calf serum (FCS; PAA Laboratories) and antibiotics (MycoZapPlus-CL; Lonza, Cologne Germany). WT, Nfkbiz^−/− (Shiina et al., 2004) and MyD88^−/− (Adachi et al., 1998) MEFs were cultured in DMEM (PAA Laboratories) containing 10% FCS and antibiotics. HeLa/TetOn-IκBζ and MCF7/TetOn-Ras^G12V cells were maintained in RPMI-1640 medium supplemented with 10% tetracycline-free FCS (PAA Laboratories), 400 µg/ml neomycin (PAA Laboratories) and antibiotics. For IL1α stimulation, MEFs were cultured in the presence of murine IL1α (100 ng/ml; ImmunoTools, Friesoythe, Germany) for 1 hour and collected by scraping, before mRNA expression was analyzed by quantitative (q) RT-PCR. All chemicals were purchased from Sigma (Munich, Germany). Monoclonal mouse anti-β-actin (AC-74, Sigma), monoclonal mouse anti-p21 (Becton Dickinson, Heidelberg, Germany) and polyclonal rabbit antisera against Ras (C-20, Santa Cruz Biotechnology, Heidelberg, Germany) and IκBζ (Hildebrand et al., 2013) were used for immunoblot analysis. Immunofluorescence staining was performed using polyclonal rabbit anti-phospho-serine-83 HP1γ (Abcam, Cambridge, UK) and monoclonal mouse anti-α-tubulin (Sigma). Horseradish-coupled secondary antibodies to mouse, rabbit and goat IgG were purchased from Promega (Mannheim, Germany). Secondary 488-/594-Alexa-Fluor-coupled antibodies to mouse and rabbit IgG were purchased from Molecular Probes (Life Technologies, Darmstadt, Germany).

IκBζ and Ras^G12V cDNAs were PCR-amplified using the primers IκBζ (5′-CACCATGATTGTGGACAAGCTGCTGGAC-3′; 5′-CTAATACGGTGGAGCTCTCTGCTGAATGG-3′) and Ras^G12C (5′-CACCATGACGGAATATAAGCTGGT-3′; 5′-TCAGGAGAGCACACACTTGCAGCTC-3′). PCR products were ligated into pENTR-D-TOPO (Life Technologies) according to the manufacturer's protocol yielding pENTR/IκBζ and pENTR/Ras^G12V. The pInducer20/TetOn-IκBζ and pInducer20/TetOn-Ras^G12V lentiviral vectors were generated using gateway cloning technology (Life Technologies) by recombining pInducer20 plasmid (Meerbrey et al., 2011) with pENTR/IκBζ and pENTR/Ras^G12V, respectively, using LR-Clonase II (Life Science Technologies) according to the manufacturer's protocol. Vector sequences were confirmed by sequencing.

HEK293FT cells were cultured in DMEM supplemented with 10% FCS, 1% L-glutamine, 1% sodium pyruvate, 1% non-essential amino acids, neomycin (400 µg/ml) and antibiotics. Cells were seeded 24 hours before transfection in culture dishes at a density of 6×10⁵ cells/cm². The pInducer20/TetOn-IκBζ and pInducer20/TetOn-Ras^G12V plasmids were co-transfected with lentiviral assembly ΔR8.9 and envelope (VSV)g plasmids cells using jetPEI reagent (PEQLAB, Erlangen, Germany). 48 hours post transfection virus-containing culture supernatant was collected and concentrated to 250 µl with filter vials (Sartorius AG, Göttingen, Germany). MCF7 or HeLa cells were transduced with virus-containing supernatant in the presence of polybrene (4 µg/ml; Sigma, Munich, Germany) for 48 hours. Transduced cells were selected in the presence of neomycin (600 µg/ml) for 7 days and analyzed for target gene overexpression upon addition of doxycycline (2 µg/ml; Fagron, Barsbüttel, Germany) by immunoblotting.

For siRNA transfection of MCF7/Ras^G12V cells 1×10⁵ cells/cm² were seeded in culture dishes in the presence of doxycycline (200 ng/ml) to induce Ras^G12V expression and either IκBζ ON-TARGET Plus Smartpool siRNA or non-targeted (NT) Smartpool ON-TARGET plus control siRNA (Thermo Fisher Scientific, Bonn, Germany) was delivered using Dharmafect I reagent (Thermo Fisher Scientific) after 24 and 72 hours. Cells were grown for additional 48 h, collected by scraping and analyzed for mRNA expression by qRT-PCR. For IL1α stimulation 1×10⁵ MCF7/WT cells per cm² were seeded and transfected with siRNA after 24 hours. 48 hours post siRNA transfection recombinant IL1α (100 ng/ml; ImmunoTools) was added to the culture medium for 1 hour. Then, cells were harvested by scraping and mRNA expression was analyzed by qRT-PCR. Anakinra (Swedish Orphan Biovitrum; Stockholm, Sweden) was added at a concentration of 1 µg/ml 24 hours prior to irradiation or treatment with bleomycin or IL1α and replenished daily.

Cells were γ-irradiated with 20 Gy (MCF7), 15 Gy (MEF), and 10 Gy (HeLa). Therapy-induced senescence was achieved by treatment with bleomycin (50 µg/ml). Cells were collected 5 days post senescence induction by scraping. For Ras-induced senescence MCF7/TetOn-Ras^G12V cells were incubated for 3 days in the presence of doxycycline (200 ng/ml) and for further 4 days in the absence of doxycycline. Samples from cells cultured tetracycline-free medium for identical periods of time served as control.

Cells were washed in ice-cold PBS and resuspended in lysis buffer [1% Nonidet P-40, 20 mM HEPES (pH 7.9), 2 mM PMSF, 350 mM NaCl, 1 mM MgCl₂, 0.5 mM EDTA, 0.1 mM EGTA and 0.5 mM DTT] supplemented with complete protease inhibitor cocktail (Roche, Mannheim, Germany). Protein concentrations were determined using the BCA assay (Thermo Fisher Scientific), and 15 µg of protein per lane were loaded onto standard SDS-PAGE gels. After electrophoresis proteins were transferred onto polyvinylidenedifluoride membranes (Amersham Biosciences, Freiburg, Germany) by tank blotting. Membranes were blocked in PBS containing 4% BSA and 0.05% Tween 20) for 1 hour, followed by an overnight incubation with the primary antibody in blocking buffer at 4°C. After washing the membrane thrice in blocking buffer, the secondary antibody (1∶5000) was applied for 1 hour. Proteins were visualized using ECL reagents (Amersham Biosciences).

Total RNA was isolated from cells using the RNeasy kit (Qiagen, Hilden, Germany) according to manufacturer's protocol. Complementary DNA synthesis and qRT-PCR were performed as described (Graupner et al., 2011). Primers for SASP analyses were purchased from Qiagen (QuantiTect Primer Assays) or Sigma. For IκBζ the following primers were used: human IκBζ (5′-CCTTTCAAGGTGTTCGGGTA-3′, 5′-CAAGCAGGTCCATCAGACAA-3′), mouse IκBζ (5′-TATCGGGTGACACAGTTGGA-3′, 5′-TGAATGGACTTCCCCTTCAG-3′). Results were normalized to GAPDH and analyzed by the ΔΔC_t method to give fold mRNA expression compared with untreated control samples.

Cells were seeded 24 hours before senescence induction by γ-irradiation, bleomycin stimulation or Ras^G12V overexpression in six-well plates at a density of 5×10⁴ cells/cm². Five days later SA-β-gal activity was assessed using the Senescence Cells Histochemical Staining Kit (Sigma) according to the manufacturer's protocol. Alternatively, SA-β-gal activity was assayed in cell lysates. To this end, cells were harvested by scraping, washed in ice-cold PBS and lysed for 30 minutes on ice in lysis buffer (20 mM Tris-HCl pH 7.4, 0.2% Triton X-100) supplemented with complete protease inhibitor cocktail. 5 µl of protein lysate were incubated with 45 µl reaction buffer (66 mM Na₂HPO₄, 66 mM NaCl, 33 mM citric acid, freshly prepared 2 mM 4-methylumbelliferone) at 37°C for 1 hour. Reaction was stopped by adding 200 µl 0.2 M Na₂CO₃ followed by centrifugation at 14,000 rpm for 5 minutes. Fluorescence of samples was detected at 355 nm excitation and 460 nm emission in an Infinite M200 plate reader (Tecan, Männedorf, Switzerland). Based on a methylumbelliferone standard curve, enzyme activity was calculated and normalized to protein concentrations.

Cells were seeded 24 hours before senescence induction onto coverslips (5 mm²) in 12-well plates. Five days after senescence induction by γ-irradiation or bleomycin treatment and 7 days after oncogene-induced senescence induction the cells were washed with PBS and fixed with ice-cold fixation solution (50% acetone, 50% methanol) for 5 minutes. Cells were washed twice with PBS, followed by incubation for 1 hour in blocking buffer (4% BSA and 0.05% saponin in PBS) at room temperature. The primary antibody diluted (1∶500) in blocking buffer was incubated at 4°C overnight. After washing the cells thrice in blocking buffer, the appropriate Alexa-Fluor-coupled secondary antibody (1∶500 in PBS) was applied for 1 hour. The cells were washed three times in PBS and incubated afterwards in PBS containing 100 ng/ml 4′,6-diamidino-2-phenylindol (DAPI; Life Technologies) for 5 minutes. Coverslips were mounted in fluorescence mounting medium (DAKO, Hamburg, Germany) and analyzed using a DMI6000 fluorescence microscope (Leica, Wetzlar, Germany).

Data are presented as the means ± s.d. or s.e.m. from at least three independent experiments. Statistical significance was calculated using Student's t-test. Values of P<0.05 were considered significant: *P<0.05; **P<0.005, ***P<0.0005.

We thank Stephen J. Elledge, Roland Piekorz, Masami Morimatsu and Gudrun Totzke for providing various materials.