Although senescence has long been implicated in aging-associated pathologies, it is not clearly understood how senescent cells are linked to these diseases. To address this knowledge gap, we profiled cellular senescence phenotypes and mRNA expression patterns during replicative senescence in human diploid fibroblasts. We identified a sequential order of gain-of-senescence phenotypes: low levels of reactive oxygen species, cell mass/size increases with delayed cell growth, high levels of reactive oxygen species with increases in senescence-associated β-galactosidase activity (SA-β-gal), and high levels of SA-β-gal activity. Gene expression profiling revealed four distinct modules in which genes were prominently expressed at certain stages of senescence, allowing us to divide the process into four stages: early, middle, advanced, and very advanced. Interestingly, the gene expression modules governing each stage supported the development of the associated senescence phenotypes. Senescence-associated secretory phenotype–related genes also displayed a stage-specific expression pattern with three unique features during senescence: differential expression of interleukin isoforms, differential expression of interleukins and their receptors, and differential expression of matrix metalloproteinases and their inhibitory proteins. We validated these phenomena at the protein level using human diploid fibroblasts and aging Sprague-Dawley rat skin tissues. Finally, disease-association analysis of the modular genes also revealed stage-specific patterns. Taken together, our results reflect a detailed process of cellular senescence and provide diverse genome-wide information of cellular backgrounds for senescence.
Human somatic cells have a limited capacity to divide in culture and eventually enter replicative senescence, a state of irreversible proliferative arrest (Hayflick, 1965; Cristofalo et al., 2004). In addition to losing the ability to divide, senescent cells exhibit diverse alterations in their cellular and biochemical features, the most well-established of which are an enlarged and flattened cellular morphology, senescence-associated β-galactosidase (SA-β-gal) activity, and an increase in the generation of reactive oxygen species (ROS) (Hwang et al., 2009). Enlarged cellular morphology is linked to increases in cellular mass, while a flattened morphology is associated with activity changes in cytoskeletal proteins such as intermediate filaments, actins, and microtubules, as well as with enhanced cell adhesion activity. SA-β-gal activity is associated with increased, but altered, lysosomal activity; ROS generation is mainly due to damaged mitochondrial activities (Hwang et al., 2009). In addition, we recently reported several mitochondrial and metabolic phenotypes in senescent cells: mitochondrial elongation, enhanced glycogenesis that may contribute to cell granularity, and increased lipogenesis accompanied by increases in organellar mass (Yoon et al., 2006; Seo et al., 2008; Kim et al., 2010). However, it still remains unclear, of all the senescence phenotypes, which one is the primary event and what are their molecular backgrounds.
At present, telomere shortening is the most plausible mechanism underlying the limited division capacity of senescent cells (Shay et al., 1991; Levy et al., 1992; Harley & Villeponteau, 1995). Although telomere shortening is believed to contribute to the determination of cellular lifespan (Morin, 1997), it is uncertain whether and how it contributes to organismal aging and aging-associated pathogenic conditions. Several recent reviews have indicated that inflammatory cytokines secreted from senescent cells contribute to age-related diseases such as cancer and degenerative diseases; the time-dependent release of cytokines may be linked to various age-related diseases (Freund et al., 2010; Campisi et al., 2011; Rodier & Campisi, 2011). However, systematic evaluation of time-dependent gene expression, including cytokine expression, has not been performed to date. We therefore carried out genome-wide gene expression profiling at various time points during the process of replicative senescence and evaluated the relevance of gene expression to senescence phenotypes, identifying the phase-specific genes that are differentially expressed during the replicative senescence process. We further analyzed our genome-wide data to determine potential links among gene expression, senescence, and age-related diseases.
Time sequence profiling of senescence phenotype acquisition during replicative senescence
To understand the order of general cellular senescence phenotypes, we first monitored SA-β-gal activity, the most typical senescence marker, during replicative senescence. Primary human diploid fibroblasts (HDFs) were continuously subcultured until they lost their division capacity (Kim et al., 2003). During this process, doubling times (DT, time duration for doubling of cell population on a culture plate) and population doubling numbers (PD, the passaged numbers of the cells obtained by sequential population doubling) were recorded as described in Materials and Methods, and cells were subjected to the SA-β-gal assay at various stages of the senescence process. Cellular DT remained at 2 days until their PD becomes 40 (PD40), and thereafter, it increased progressively (Fig. 1A), implying that a subset of cell population begins to lose their capacity for division after PD40. It is noteworthy to remind that the cell population is heterogeneous – gradually losing nonsenescent cells and gradually gaining senescent cells – as primary cells are passaged to replicative senescence.
Cells with a DT of 7 (DT7 cells) developed a low level of SA-β-gal activity; the population of SA-β-gal-positive cells progressively increased thereafter (Fig. 1B and Fig. S1A, Supporting information). In addition, high SA-β-gal activity was observed in DT10 cells, but in a small subset of the population. Interestingly, another significant and drastic increase occurred in high levels of SA-β-gal activities in the population of DT20 cells, resulting in significant increase in total SA-β-gal activities. However, low levels of SA-β-gal activities slightly decreased. These results suggest that low SA-β-gal-positive cells may convert to high SA-β-gal cells. However, possibility of direct emergence of high SA-β-gal cells cannot be ignored. Our analysis implies that the early delay of cell growth from DT2 to DT3 and to DT5 is not directly linked to SA-β-gal activity, and that independent molecular changes may be required for persistent delay of cell growth from DT2 to DT30, for gain of SA-β-gal activity, and for transition of low to high SA-β-gal activities, respectively.
Next, we evaluated other cellular senescence phenotypes, such as increases in cell granularity, cell size, and cell mass. When we analyzed the side scattering (an indication of cell size) and forward scattering (an indication of cellular granularity) of the HDFs, we found that both phenotypes increased progressively from DT3 and reached their maximum levels at DT12 (cell granularity) and DT10 (cell size) (Fig. 1C). Unexpectedly, cell area increased from DT5 (Fig. 2A), and total intracellular mRNA and protein levels increased from DT3, observations that are consistent with an increase in cell size (2B and 2C). These results imply that cellular flattening, which may be linked to cytoskeletal and cell adhesion activity (Hwang et al., 2009), develops later than the increase in cell size.
Reactive oxygen species are often considered to be the key modulators of senescence phenotypes, playing multifaceted roles as cellular signaling messengers as well as cell-damaging agents. We evaluated the time course of ROS production with DCF-DA fluorescence dye, which primarily measures hydrogen peroxide and the temporal link to the acquisition of other senescence phenotypes. Unexpectedly, intracellular ROS levels significantly increased in cells over DT2, much earlier than the increase in cell mass, although the ROS levels remained relatively low (Fig. 2D and Fig. S1B). The initial minor increase in intracellular ROS levels may therefore be related to both the increase in cell size and the slight delay in cell cycle progression from DT2 to DT3. This hypothesis is supported by our previous observation that ROS generation delays overall cell cycle progression (Byun et al., 2008). Another drastic increase in intracellular ROS levels was observed in cells beyond DT7 (Fig. 2D), implying a link between high levels of intracellular ROS and a gain of SA-β-gal activity. We summarize the order of the acquisition of senescence phenotypes in Fig. 2E.
Gene expression changes during replicative senescence
To explore the molecular background of replicative senescence, we performed time-series gene expression profiling of the cells (Fig. 3A). We first carried out unsupervised clustering of the most variable genes (standard deviation > 0.2, n = 6,845). The gene expression profile revealed four distinct modules (G1, G2, G3, and G4) that were prominently expressed at certain stages of the senescence process (Fig. 2A). Module G1 included genes highly expressed in DT2 cells (n = 862), while module G2 included genes expressed in DT2~DT7 cells (n = 2,012). Module G3 exhibited high expression during DT3~DT20 (n = 510). Finally, module G4 was expressed from DT10 to more than DT30 (n = 2,440; Table S1, Supporting information).
The functional characteristics of each module were evaluated by an analysis of Gene Ontology (Table 1). Module G1 was prominently enriched for cell cycle-related genes, indicating active cell proliferation during this phase. This activity was suppressed at DT3, suggesting that senescent growth arrest may occur at this time point. Module G2 included metabolic process–related genes and tRNA/RNA processing, while module G3 was associated with expression of inflammatory and immune-related functions such as defense response and antigen processing. Module G4 was enriched for genes related to cell death and cell growth regulation.
|G1||Cell cycle (ES = 53.47)|
|GO:0007049~cell cycle||1.59 × 10−63|
|GO:0022403~cell cycle phase||2.49 × 10−61|
|GO:0000279~M phase||5.93 × 10−61|
|DNA repair (ES = 27.53)|
|GO:0006259~DNA metabolic process||4.08 × 10−47|
|GO:0006281~DNA repair||1.63 × 10−25|
|GO:0006974~response to DNA damage stimulus||2.75 × 10−24|
|Microtubule-based process (ES = 13.43)|
|GO:0007017~microtubule-based process||1.93 × 10−18|
|GO:0007051~spindle organization||3.74 × 10−15|
|GO:0000226~microtubule cytoskeleton organization||4.50 × 10−14|
|G2||Metabolic process (ES = 8.55)|
|GO:0034660~ncRNA metabolic process||4.34 × 10−12|
|GO:0042254~ribosome biogenesis||3.41 × 10−10|
|GO:0022613~ribonucleoprotein complex biogenesis||3.74 × 10−10|
|tRNA process (ES = 6.56)|
|GO:0043039~tRNA aminoacylation||9.98 × 10−08|
|GO:0043038~amino acid activation||9.98 × 10−08|
|GO:0006418~tRNA aminoacylation for protein translation||9.98 × 10−08|
|RNA processing (ES = 5.37)|
|GO:0006396~RNA processing||7.76 × 10−10|
|GO:0000377~RNA splicing, via transesterification reactions with bulged adenosine as nucleophile||1.62 × 10−06|
|GO:0000375~RNA splicing, via transesterification reactions||1.62 × 10−06|
|G3||Inflammatory response (ES = 4.54)|
|GO:0006952~defense response||3.70 × 10−09|
|GO:0006954~inflammatory response||2.63 × 10−05|
|GO:0009611~response to wounding||1.97 × 10−03|
|Antigen processing and presentation (ES = 3.49)|
|GO:0002474~antigen processing and presentation of peptide antigen via MHC class I||4.76 × 10−05|
|GO:0048002~antigen processing and presentation of peptide antigen||6.02 × 10−04|
|GO:0019882~antigen processing and presentation||1.14 × 10−03|
|Apoptosis (ES = 3.05)|
|GO:0042981~regulation of apoptosis||1.30 × 10−05|
|GO:0043067~regulation of programmed cell death||1.63 × 10−05|
|GO:0010941~regulation of cell death||1.77 × 10−05|
|G4||Cell death (ES = 4.51)|
|GO:0008219~cell death||8.35 × 10−06|
|GO:0016265~death||1.20 × 10−05|
|GO:0006915~apoptosis||8.72 × 10−05|
|Cell growth (ES = 4.34)|
|GO:0001558~regulation of cell growth||4.62 × 10−07|
|GO:0030308~negative regulation of cell growth||2.85 × 10−05|
|GO:0008361~regulation of cell size||7.21 × 10−05|
|NF-kappaB cascade (ES = 4.27)|
|GO:0010647~positive regulation of cell communication||8.09 × 10−06|
|GO:0043123~positive regulation of I-kappaB kinase/NF-kappaB cascade||2.58 × 10−05|
|GO:0010740~positive regulation of protein kinase cascade||2.63 × 10−05|
We classified the process of replicative senescence into four phases (bottom of Fig. 3A). The early phase is governed by combined action of modules G1 and G2, resulting in active proliferation and active metabolism. The middle phase, controlled by modules G2 and G3, leads to the suppression of cell proliferation activity, but active metabolism is sustained. This finding corresponds to the senescence phenotypes of increases in cell mass and size as well as delayed cell growth (Fig. 2E). Modules G3 and G4 contribute to inflammation-related functions during the advanced phase. Last, the very advanced phase is only governed by module G4.
Expression of the genes encoding senescence-associated secretory phenotype (SASP) proteins
Analysis of the gene expression profile revealed that inflammatory functions were expressed by module G3. Inflammation-related factors released from cells are known to play key roles in senescence; together these factors constitute the SASP, which may be phase specific (Freund et al., 2010; Rodier & Campisi, 2011). We examined the gene expression profiles of the known SASP-related proteins and their isoforms (n = 162) at each phase of senescence process and found 55 phase-specific genes (Fig. 3B). IL10 and IL18 were included in module G1 expression, while IL15 was in module G2. In module G3, many interleukins (IL1B, IL6, IL8, and IL32) and chemokine ligands (CLXCLs 1, 2, 5, 6, 10, and 16) were included. Module G4 included the isoforms of insulin-like growth factor-binding proteins (IGFBPs 2, 3, 5, and 7), interleukins (IL1A, IL12A, and IL17D), and matrix metalloproteinase (MMP) genes (MMP1, MMP3, and MMP12). Our data demonstrate the overall phase-dependent expression pattern of SASP genes during senescence.
Overall activities of SASP-related genes are modulated by combined actions with phase-specific expressions of their receptors or their inhibitory proteins
Our analysis of SASP-related gene expression also revealed phase-specific expression patterns of the receptors for SASP. Therefore, we hypothesized that the combined actions of cytokines and their receptors, and MMPs and the tissue inhibitors of metalloproteinases (TIMPs) were important at each phase. Potential combined actions of interleukins and their receptors were estimated, based on their expression profiles (Fig. 4A,B). Interleukins can display four effects in senescent cells: an increased paracrine effect, an increase in paracrine/autocrine effects, a sensitivity increase, or a decrease in overall action. For example, IL1A expression increases in senescent cells, but the expression of its receptor decreases; this scenario may imply that the senescent cells can release IL1A but they cannot respond to it due to a lack of receptors. A recent study demonstrated that IL1A was not usually secreted, but rather remains cell surface-bound, making our speculation complicated (Orjalo et al., 2009). However, it is still acceptable that the surrounding nonsenescent cells can be the primary target to the released cytokines. Our scenarios can be well applicable to the action modes of all the other secretory interleukins. We further validated the expression of interleukins and their receptors at the protein level. As expected, IL1R1 and IL7R showed shifted expression during senescence (Fig 4C). The expression of IL6 and IL8 proteins peaked during advanced phases of senescence (DT10 and DT14; Fig. 4D). Due to low expression at the protein level, we were unable to use the same method to detect other interleukins, including IL1α, IL1β, and IL17.
Shifts were also observed in the relative expression levels of MMPs and TIMPs. Some MMPs/TIMPs (MMP11, MMP2, MMP20, MMP27, and TIMP1) were expressed during the early and middle phases, while others (MMP1, MMP12, MMP3, TIMP1, TIMP2, and TIMP3) were expressed in the advanced and very advanced phases of senescence (Fig. S2). The protein expression levels of the MMPs and TIMPs were validated by Western blot (Fig. 4C), and zymography also revealed a distinct isoform shift: bands 1 and 2 increased, and band 3 decreased (Fig. 4E).
We further validated the expression shift of SASP-related genes in vivo using skin tissues from young to old Sprague-Dawley rats. The expression levels of the SASP-related genes during the aging process revealed similar patterns of shift (Fig. 5A,B). Taken together, our data constitute detailed, genome-scale profiles of the senescence-related expression of inflammatory signals, observations that are in agreement with a previous report (Rodier & Campisi, 2011).
Gene expression profiles show phase-specific linkage to age-associated diseases
To pinpoint the regulatory signaling pathways involved in each phase of senescence, we analyzed the expression of canonical signaling pathways in the KEGG database (Table 2). Interestingly, inflammatory signaling pathways such as RIG-I (P = 7.7 × 10−4) and Toll-like receptor signaling (P = 2.0 × 10−3) were enriched in module G3 (Fig. S3A,B). In support of our finding, a recent report demonstrated that RIG-I is a key mediator of senescence-associated inflammation, particularly for IL6 and IL8 (Liu et al., 2011). The expression of lysosomal protein-encoding genes was enriched in module G4 (P = 1.0 × 10−9, Fig. S3C), which may correspond to the high activity of SA-β-gal at this time point. In addition, the p53 pathway was identified in module G4 (P = 3.5 × 10−4), suggesting a regulatory role in cell death signaling and aggressive senescence processes (Fig. S3D).
|Base excision repair||15||3.10.E-11|
|Disease||Insulin; blood pressure, arterial||4||2.95.E-03|
|Neural tube defects||10||9.30.E-03|
|Colon cancer rectal cancer||4||1.28.E-02|
|Pathways in cancer||48||2.80.E-03|
|KEGG_Pathway||Pathways in cancer||21||7.60.E-04|
|RIG-I-like receptor signaling pathway||9||7.70.E-04|
|Toll-like receptor signaling pathway||10||2.00.E-03|
|Disease||Cleft lip with cleft palate||11||2.96.E-03|
|p53 signaling pathway||21||3.50.E-04|
|Epithelial cell signaling in Helicobacter pylori infection||20||9.80.E-04|
|Phosphatidylinositol signaling system||21||1.10.E-03|
|Small cell lung cancer||22||2.50.E-03|
Finally, we evaluated the possible association of modules G1–G4 with age-related diseases (Table 2) to examine potential contribution of senescent cells to the diseases. Module G1 was associated with cancer, while module G2 was linked to insulin-related disease and neural tube defects. Module G3 exhibited an association with inflammatory diseases, and module G4 was associated with cardiovascular diseases and neurodegenerative diseases. Potential association of modular genes with canonical pathways and age-related diseases are summarized in Fig. 5C. Taken together, these observations suggest that stage-specific expression of gene modules of the senescent cells acquired during the organismal aging process might contribute to the development of age-related diseases.
Although various cellular senescence phenotypes are known, their temporal sequence and molecular underpinnings remain unclear (Hwang et al., 2009). In the present study, we evaluated the relevance of several cellular senescence phenotypes to the changes in gene expression during replicative senescence of HDF cells. Interestingly, the gene expression profile revealed four distinct expression modules, and Gene Ontology analysis was consistent with the changes in cellular senescence phenotypes. Combined action of modules G1 (DNA replication, cell cycle, and DNA repair) and G2 (metabolic process and ribosomal activity) is consistent with fast cell growth (short DT). This stage can be classified as early. However, it is difficult to understand how cells experience an increase in ROS during the early stage without further changes in phenotype; accumulation of ROS may trigger the expression of module G3. ROS can have also both beneficial and detrimental effects on cellular growth and fate, depending on the intracellular levels and location. The low increase in the early stage may play an essential positive role in maintaining cell growth activity.
The middle stage of senescence begins with the shutdown of module G1 expression and induction of the expression of module G3 (inflammation, antigen presentation, apoptosis), mainly via the combined action of modules G2 and G3. Active metabolism without expression of module G1 (delayed cell cycle progression) at this stage is enough to trigger increases in cellular mass and size. Involvement of module G3 genes at this stage is quite interesting, as cancer-related pathways and inflammatory signaling (for example, RIG-1 and Toll-like receptor signaling) may allow these senescence stages to serve as cellular backgrounds for tumor-promoting environments.
The advanced stage is controlled by modules G3 and G4 (cell death, cell growth, and NF-kB cascade), implying that cells in this stage exhibit cellular properties of both cell death and cell survival, in combination with inflammatory activities. Although the genes in these modules have well-known links to inflammatory diseases and cancer, it is not clear whether these genes contribute to cellular senescence phenotypes (high ROS generation and SA-β-gal activity) or vice versa. The fourth, very advanced stage of senescence is controlled by module G4, suggesting that the cells in this stage may be prone to cell death. Our measurements of the expression of genes in module G4 are consistent with an association with degenerative diseases; these cell death genes must be also linked with high SA-β-gal activity. Overall, our gene expression profiles are well matched to the senescence phenotypes.
Cellular senescence is the phenomenon by which normal diploid cells lose the ability to divide and proposed to be one of the etiologies of aging-associated diseases (Fossel, 2002; Carnes et al., 2008; Burton, 2009; Campisi et al., 2011), including cancer, cardiovascular disease, type II diabetes, Alzheimer's disease, and Parkinson's disease (Youdim & Riederer, 1993; Erusalimsky & Kurz, 2005; Sone & Kagawa, 2005; Campisi et al., 2011; Neill, 2012). Although the incidences of these diseases are greatly influenced by differences in the genetic backgrounds of individuals (Jasperson et al., 2010; Munter et al., 2010) and in the extrinsic factors that they experience, such as dietary habits (VanSaun et al., 2009; Tong et al., 2010) and environmental toxins (Guengerich et al., 1996; Fornai et al., 2005; Ling & Groop, 2009), it is widely accepted that the intrinsic aging process of the cell (or tissue) increases the organism's vulnerability to these diseases. These age-associated diseases are associated with different incidence profiles based on age. For example, the rate of incidence of most human cancers increases exponentially with age to 60–75 years, but thereafter, the rates off or may even decline (Benz, 2008). However, the incidences of Alzheimer's and Parkinson's diseases further increase after 80 years of age (Ahmed et al., 2010; Rocca et al., 2011). These observations imply that the biochemical or molecular backgrounds of aged tissues are quite different in middle-aged and much older people; cells (or tissues) may have disease-specific backgrounds that differentially trigger age-specific diseases.
Nonetheless, we cannot directly apply our stage-specific gene expression profiles to incidence and/or development of aging-associated diseases because all the tissues in human aging process consist of diverse status of or several kinds of cells. Then, how is cellular senescence linked to aging-associated diseases? There are no definitive answers to this question as yet. However, our present results suggest a few possible mechanisms. First, populations of cells in certain stages of senescence may drive local tissue environments that trigger and/or promote a disease, depending on the tissue. Therefore, mixed populations of senescent cells of different stages may complement each other to suppress or enhance senescence phenotypes. Previous reports have described the potential contributions of accumulated senescent cells to organismal aging and several pathogenic conditions (Youdim & Riederer, 1993; Kitada et al., 1995; Vasile et al., 2001; Castro et al., 2003; Erusalimsky & Kurz, 2005; Neill, 2012). Second, the senescent cell may be the point of origin of a disease or provide a disease-friendly environment. For example, the four potential effects of the interleukin-family genes may also be applicable to the actions of diverse growth factors or chemokines. The four different potential effects for all released factors may make important contributions to stage-specific disease association. Third, the contribution of the senescent cell to disease is not restricted to the expression of a single gene. Differential expression of MMP and TIMP isoforms during senescence highlights the importance of the combined action of these genes, which may be derived from different cells in different stages of senescence. Fourth, sequential induction of various gene profiles by certain gene modules may contribute to the development of disease. These progressive and consecutive features may be linked with the progressivity of most aging-associated diseases. We have diagrammed these three possible action modes depending on the types of cell population, which are mainly based on differential expressions of interleukins and their receptors (Fig. S4).
The last critical question is whether the gene expression modules and differential expression patterns observed here can be reproduced in organismal tissues. Somel et al. reported eight age-dependent expression patterns in human and macaque brains (Somel et al., 2010). Although that investigation did not focus on differential expression of gene isoforms, these previous observations partly support our hypothesis. We also confirmed differential expression profiles of several gene groups using skin tissues harvested from aging Sprague-Dawley rats. However, it is not clear whether our theory is applicable to other tissues; for example, we previously reported that the maximum level of enhanced glycogenesis, with increased expression of glycogen synthase and glycogen, occurred in the middle-aged rat liver (Seo et al., 2008). The second unsolved question is related to tissues that contain postmitotic cells, such as muscles and the brain. Several senescence traits have been reported in postmitotic cells (Porta et al., 1982; Schmucker & Sachs, 2002; Terman et al., 2003), but most of these traits are still not clearly characterized. Therefore, it is difficult to determine the link between cell senescence and age-related pathology in these tissues, although senescence of mitotic cells in these tissues may contribute to the overall pathogenic environment.
Our analysis was mainly focused on SASPs, remaining many questions of how SASPs are linked with the other key controllers of senescence, such as p16. It is well reported that p16 increase and cause senescence independent of telomere shortening (Itahana et al., 2003; Brookes et al., 2004; Herbig et al., 2004). When we looked through mRNA expression profile of p16 in our senescence system, we could not find any change in p16 mRNA expression. Nevertheless, its protein expression levels increased in middle, advanced, and very advanced stages of HDF senescence (Fig. S5). To explain this unexpected result, further detailed analyses on its molecular mechanism underlain should be addressed. In addition, p21 protein, another negative regulator of the cell cycle, showed a similar pattern, whereas PCNA protein (a proliferation marker) decreased reciprocally (Fig. S5), implying that expression of SASPs may be linked to the expression shift of these cell cycle regulators. Then, the link between SASPs' expressions and the senescence-associated factors controlling cell cycle progression, such as p16, p53, p21, should be elucidated. Despite the experimental limitations of this study due to lack of analysis for direct linkages to human aged tissues, our overall results support the hypothesis of a link between cell senescence and aging-associated pathology.
Cell culture and induction of replicative senescence
For replicative senescence, primary HDFs isolated from the foreskin of a 5-year-old boy according to the method described previously (Kim et al., 2003) were cultured in DMEM medium supplemented with 10% fetal bovine serum (FBS, GIBCO, Grand Island, NY, USA) and antibiotics at 37 °C in a humidified incubator with 5% CO2. To develop replicative senescence, confluent HDFs were evenly transferred into two new dishes, and the cells were cultured until getting confluent again to generate one population doubling. Cultures were maintained in a 100-mm dish, and in this study, confluent status of cultures means almost confluent culture condition without contact inhibition. The number (n) of PD was calculated using the equation, n = log2 Ne/Ns, where Ne and Ns are the numbers of cells at the end of cell culture and those seeded at the start of one passage, respectively. The numbers of population doublings (PD, times), the doubling time (DT, days), and total culturing period were continuously monitored. DT determined for this study includes the cells' adaptation time after re-seeding because confluently grown cells were divided into two new plates at every passage. Cells at several PD intervals were frozen and stored for further experiments. When necessary, the stored cells were re-cultured, passaged at least twice to stabilize the cells and to confirm their DT, and then almost confluently grown cells were subjected to further analysis. But, cells after PD80 kept cultured by splitting in 1:4 ratio for all the analysis without storage because these cells were slightly labile and susceptible to death after freezing and thawing, and reseeding.
Senescence-associated β-galactosidase (SA-β-gal) staining
SA-β-gal activity was assayed at the same time for all the stage of senescent cells at pH 6.0 as described previously (Seo et al., 2008). The stains were visualized 12 h after incubation at 37 °C. High stain level was determined by marking a strongly stained cell, which was clearly distinguished from the other low level stains (as shown in the inset of Fig. 1B) on the DT10 culture plate as high stain standard using Image J software (NIH, Bethesda, MA, USA). The numbers of the two different levels (low and high) of blue-colored and total cells were determined using Image J software; percentage of the cells stained blue was estimated to compare the degree of senescence-associated cells.
Estimation of intracellular ROS, cell area, cell size and cellular granularity
To estimate intracellular ROS level, cells were stained with 10 μm 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA, Molecular probes, Eugene, OR, USA) for 15 min at 37 °C before performing flow cytometric analysis with FACSCanto™ II (Becton Dickinson Corp., San Jose, CA, USA). Analysis of intracellular ROS levels for all the stage of senescent cells was performed at the same time. Mean values of arbitrary fluorescence unit of 10,000 cells were analyzed. Cell size and cell granularity of 10,000 cells were evaluated by analyzing forward scattering (FSC) and side scattering (SSC) of the stained cells, respectively, as previously described (Hwang et al., 2009). Cell area was measured using axiovision 4.8 software (Carl Zeiss AG, Gottingen, Germany) after drawing cell margin from images of cultured cells on plates.
Measurement of total cellular RNA and protein levels
After total cell number was counted, total RNA of the cells was extracted using Trizol (Invitrogen Life Technologies, Carlsbad, CA, USA), purified using RNeasy columns (Qiagen, Valencia, CA, USA) according to the manufacturers' protocol. After processing with DNase digestion, clean-up procedures, RNA samples were quantified, aliquoted, and stored at −80 °C until use. For quality control, RNA purity and integrity were evaluated by denaturing gel electrophoresis, OD 260/280 ratio, and analyzed on Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA).
After total cell number was counted, the cells were lysed on ice with 100 μL of lysis buffer (20 mm Tris–HCl, pH 8.0, 137 mm NaCl, 1% Triton X-100, 10% glycerol, 1 mm Na3VO4, 5 mm NaF, and 1 mm phenylmethylsulfonyl fluoride). After lysis, the material was sonicated for 2 min at 30-s intervals and centrifuged at 12,000 g for 10 min at 4 °C. Protein concentrations were determined using BCA protein assay reagent (Pierce Biotechnology Inc., Rockford, IL, USA) using bovine serum albumin as a standard. Total protein levels per cell (ng cell−1) were calculated.
Gene expression profiling and data analysis
Total RNAs isolated from three different HDF culture plates at each stage were mixed and amplified and purified using the Ambion Illumina RNA amplification kit (Ambion, Austin, TX, USA) to yield biotinylated cRNA according to the manufacturer's instructions. Briefly, 550 ng of total RNA was reverse-transcribed to cDNA using a T7 oligo (dT) primer. Second-strand cDNA was synthesized, in vitro transcribed, and labeled with biotin-NTP. After purification, the cDNA was quantified using the ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).
Seven-hundred and fifty nanogram of labeled cDNA samples was hybridized to each human HT-12 expression v.4 bead array for 16–18 h at 58 °C, according to the manufacturer's instructions (Illumina, Inc., San Diego, CA, USA). Detection of array signal was carried out using Amersham fluorolink streptavidin-Cy3 (GE Healthcare Bio-Sciences, Little Chalfont, UK) following the bead array manual. Arrays were scanned with an Illumina bead array Reader confocal scanner according to the manufacturer's instructions.
The quality of hybridization and overall chip performance were monitored by visual inspection of both internal quality control checks and the raw scanned data. Raw data were extracted using the software provided by the manufacturer [Illumina Genome Studio v2009.2 (Gene Expression Module v1.5.4)]. Array data were filtered by detection P-value < 0.05 (similar to signal to noise) in all samples (we applied a filtering criterion for data analysis; higher signal value was required to obtain a detection P-value < 0.05). Selected gene signal value was transformed by logarithm and normalized by quantile method. Functional characterization of the gene sets identified in aging processes was performed by using david software (http://david.abcc.ncifcrf.gov).
Detection of cytokines in media and serum
Human diploid fibroblasts at diverse phases were subcultured in 60-mm plates until they achieved confluence and then replaced with new medium without serum for 1 day. The medium was collected and centrifuged at 10800 g for 5 min, and the resulting supernatant was frozen at −80 °C for subsequent assay of the cytokines by Mosaic™ ELISA kit (cat MEA001, R&D systems Inc., Minneapolis, MN, USA). Images of the plates were taken by VersaDoc™ Imager (Bio-Rad, Hercules, CA, USA) and quantified using the q-view™ imager Software (Quansys Biosciences, Logan, UT, USA).
Rat serum was obtained from Sprague-Dawley rats (6 months old and 24 months old) by allowing the blood to clot for 30 min and subsequently centrifuging for 10 min at 1000 g and then applied to Milliplex MAP Rat Cytokine/Chemokine Magnetic Bead Panel kit (Cat. RECYTMAG-65K, EMD Millipore, Darmstadt, Germany).
Monitoring MMPs' activities using Zymography
Media MMPs' activities secreted from HDFs were assayed by zymography, as described by Herron et al. (1986) with slight modification. Briefly, the same conditioned media obtained for the cytokine assay were used. Aliquots (conditioned media) calculated to contain MMP released from equal cell lysate were applied to nonreducing sample buffer (without β-mercaptoethanol). Nonreducing SDS-PAGE gel containing 0.1% gelatin (12% polyacrylamide) was prepared. After electrophoresis, proteins on the gel were re-natured by washing in 2.5% Triton X-100 with gentle shaking for 30 min and activated by incubating for 48 h at 37 °C in substrate buffer (50 mm Tris–HCl, pH 7.5, 5 mm CaCl2, 0.02% NaN3). After washing the gels in distilled water, the gel was stained with 0.5% Coomassie Blue R-250 in 40% methanol and 10% acetic acid for 2 h. Gels were washed with destaining solution (40% methanol, 10% acetic acid). Proteins with MMP activity appeared as clear bands against dark background due to their degrading activity of gelatin.
Western blot analysis and antibodies
Cells were washed twice with PBS and lysed with lysis buffer [50 mm Tris–Cl, pH 7.5, 0.1 m NaCl, 1 mm EDTA, 1% (v/v) Triton X-100, 10 μg mL−1 aprotinin and 10 μg mL−1 leupeptin, and 1 mm PMSF]. Rat skin tissues (0.5 g) were crashed after freezing with liquid nitrogen, suspended in 1 mL of lysis buffer, and blended by tissue homogenizer (Polytron PT3000, Brinkman, Instruments, Westbury, NY, USA). A portion (30 μg) of the lysate was applied to Western blot analysis as previously described (Yoon et al., 2006). Antibodies against IL1R (cat.04-465) and MMP12 (cat. 6010) were purchased from Millipore (Billerica, MA), and antibodies for IL7R (sc-662) and MMP11 (sc-8836) were from Santa Cruz (Santa Cruz, CA). Antibodies for TIMP2 (ab109708) and TIMP4 (ab58425) were purchased from Abcam (Cambridge, UK).
This work was supported by the National Research Foundation of Korea (2012R1A5A2051425). We thank the Aging Tissue Bank of Pusan, Korea, for the supply of aged tissues.