1.4. GENETIC BASIS FOR REPLICATIVE CELLULAR SENESCENCE OF HUMAN CELLS
Normal cells in culture can be grown for only a limited number of cell divisions after which they exhibit morphological changes and cease proliferation, a process termed replicative cellular senescence or cellular aging (113). Hayflick and Moorhead (114) reported this finding with human fibroblasts over 30 years ago, and it has been subsequently confirmed by many investigators using cells from different tissues and species. The failure of cells to grow beyond this limit is an inherent property of normal cells that cannot be explained simply by inadequate media components or growth conditions (113, 114). The key determinant in the lifespan of cells in culture is the number of cell doublings, not the length of time in culture (113). Normal cells transplanted serially in vivo also exhibit a finite lifespan, indicating that cellular senescence is not a cell culture artifact (115).
Several lines of evidence suggest that the aging of cells in culture may be related to the aging of the organism from which they were derived (113, 116). These lines of evidence, although not conclusive, provide provocative support for the hypothesis that aging of cells is related to the aging process of the organism.
Escape from cellular senescence is an important step in neoplastic progression of human and rodent cancers (117). Many, but not all, tumor cells can be grown indefinitely in culture. Cells that have escaped senescence are termed immortal. It is not clear whether the failure of some tumor cells to grow in culture is a technical artifact or an indication that escapes from senescence is not required for these cancers. A model to explain these observations is discussed later in this review.
Normal cells escape senescence following treatment with diverse carcinogenic agents, including chemical carcinogens, radiation, viruses, and oncogenes (117). This observation suggests that immortalization is important in carcinogenesis. While immortality is not sufficient for neoplastic transformation, most immortal cells have an increased propensity for spontaneous, carcinogen-induced or oncogene-induced neoplastic progression (117). Therefore, escape from senescence is a preneoplastic change that predisposes a cell to neoplastic conversion. It is clear that immortal cells are further along the multistep pathway to neoplasia than normal cells. Because cellular senescence limits the growth of cells, it is reasonable that senescence might be one mechanism by which tumor suppressor genes operate (117, 118).
1.4.1. GENETIC BASIS FOR CELLULAR SENESCENCE
Two major theories of cellular senescence have been debated for many years (119). One is the error catastrophe or damage model, which proposes that the random accumulation of damage or mutations in DNA, RNA, or proteins lead to the loss of proliferative capacity. The experimental evidence supporting the error accumulation hypothesis has been criticised (119). A second hypothesis is that senescence is a genetically programmed process. Strong experimental support for a genetic basis for senescence is provided by studies of Pereira-Smith and Smith (120) and by Sugawara et al. (121), which are discussed below.
It is possible to fuse cells of different origins and to select the hybrid cells using biochemical markers for drug sensitivity or resistance that differ in the parental cells. When cells with a finite lifespan are fused to immortal cells with an indefinite lifespan, the majority of these hybrids senesce (120, 122). Even hybridization of two different immortal human cell lines with each other can result in senescence, indicating that different complementation groups exist for the senescence function lost in these cells (120). By fusing different immortal human cell lines with each other, Pereira-Smith and Smith (120) established four complementation groups, suggesting that loss or inactivation of one of multiple genes allow cells to escape from senescence. If this hypothesis is correct, it should be possible to map the genes involved in cellular senescence.
Studies with hamster and human interspecies hybrids and microcell-mediated chromosome transfer experiments have mapped putative senescence genes to specific human chromosomes (121, 123-126).
The initial mapping by Sugawara et al. (121) of a senescence gene to chromosome 1 was demonstrated by three, independent experimental approaches: I) interspecies hybrids of normal human cells with immortal hamster cells that escaped senescence showed non-random losses of human chromosome I, 2) inter species cell hybrids with human cells carrying a t(I;X) chromosome that allow selective pressure for retention of the long arm of chromosome 1 had a high frequency of senescent hybrids, and 3) microcell-mediated chromosome transfer experiments demonstrated that introduction of a single copy of human chromosome I, but not other chromosomes, restored the program of senescence in certain immortal cell lines.
Using the technique ofmicrocell-mediated chromosome transfer, we and others have mapped senescence genes to over ten human chromosomes (123, 126) (Table 1.II). These results have led us to propose the following hypothesis: cellular senescence is controlled by genes that are activated or whose functions become manifest at the end of the lifespan of the cell. Defects in the function of these gene products allow cells to escape the program of senescence and become immortal. Immortalization relieves one constraint on tumor cell growth, allowing malignant progression.
According to this hypothesis, a family of senescence genes exists, and immortalization occurs due to defects in these genes. This theory is supported by the complementation studies of Pereira-Smith and Smith (120), which show that different immortal cell lines, when fused together, can complement each other and senesce. This theory also explains our results that introduction of a specific human chromosome causes senescence in some cell lines but not others.
Mapping Locations of Putative Human Senescence Genes by Microcell-Mediated Chromosome Transfer
|1 (lq25/1q42)||Unknown (telomerase regulators)||Syrian hamster fibrosarcoma (lOW)|
endometrial carcinoma (HHUA)
endometrial carcinoma (Ishikawa)
|2||Unknown||cervical carcinoma (SiHa)|
|3||Unknown (telomerase regulator)||renal cell carcinoma (RCC23, KCI2)|
|4||Unknown||cervical carcinoma (HeLa)|
bladder carcinoma (182)
|6||Unknown||SV 40 immortalized cells|
|7||Unknown||chemically induced immortalized human cells (KMST-6, SUSM-I)|
|11||Unknown||bladder carcinoma (HIS)|
|13||Rb||many cell lines|
|17||p53||many cell lines|
|18||Unknown||endometrial carcinoma (HHUA)|
|X||Unknown||Chinese hamster (Ni2)|
ovarian carcinoma (Hoc8)
breast carcinoma (ELCO)
One of the chromosomal regions to which a senescence gene has been mapped by chromosome transfer is chromosome 9p21 (127). The p16 tumor suppressor gene maps to this region, making it a candidate for the senescence activity mapped to this region. The p16 protein is an inhibitor of the cyclin dependent kinases CDK4 and CDK6, key regulators of the G 1 phase of the cell cycle. Furthermore, p16 is mutated in a number of immortal cancer cells. We therefore examined whether p16 is modulated during cellular aging and found that it is upregulated in senescent human cells and functions as a major inhibitor of the cyclin dependent kinases in senescent cells (128). Because p16 is mutated in many cell lines and reintroduction of pl6 into immortal cell lines induces cellular senescence (129-131), we conclude that pl6 is the senescence gene mapped to 9p21. These observations validate the approach of mapping senescence genes by microcell-mediated chromosome transfer.
Two other cell cycle control genes, p53 and Rb, also have properties of senescence genes. Downregulation of their expression by antisense methods results in extension of the lifespan of normal human cells (132), and reintroduction of the genes into certain immortal cells causes cessation of growth and morphological changes similar to senescent cells (133, 134). According to our hypothesis, specific genes are involved in one or more pathways leading to a program of senescence. In normal cells, Rb and p53 proteins are negative regulators of the cell cycle, which is controlled by other proteins to allow cell cycle progression following mitogenic signals. In senescent cells, a program is activated that blocks entry into the DNA synthesis phase of the cell cycle, and the cells become irreversibly growth arrested in the G1 phase.
Rb and p53 may participate in one or more pathways that activate or affect the senescence program, and deletions or mutations of p53 or Rb genes could result in the inability to activate or mediate the senescence program. In addition, p53 and Rb can be normal in some immortal cells, and genes that control their phosphorylation or other posttranslational modifications may be defective.
In addition to regulation of cell cycle control, we have observed that another function for a senescence gene mapped by microcell-mediated chromosome transfer is repression of the enzyme telomerase, which is responsible for the maintenance of telomeres in most immortal cell lines. Telomeres are specialised structures at chromosome ends that consist of tandemly repeated DNA sequences - (TTAGGG)n in mammals -and telomere-associated proteins, such as TRFI and TRF2 (135). Telomeres have been suggested to function as a mitotic clock that determines a cell's lifespan (136, 137). In normal human somatic cells, which have no detectable telomerase activity, the telomeric DNA ends progressively shorten with each cell division. This is due to the inability of conventional DNA polymerases to replicate the ends of linear DNA molecules completely (136, 137) or to some other DNA end degradation mechanism (138).
The telomere hypothesis for cellular senescence suggests that cells undergo senescence when their telomeres reach a critically short length (136, 137). Immortal cancer cells have mechanisms that compensate for telomere shortening, most commonly through the activation of telomerase (139), allowing them to stably maintain their telomeres and grow indefinitely. Three components of the human telomerase enzyme have beenidentified: the human telomerase RNA component hTERC, also known as hTR, which acts as an intrinsic template for telomeric repeat synthesis (140); TEPl (148), a telomerase-associated protein l/telomerase protein component 1 (also known as TP1/TLPl), which is similar to the Tetrahymena telomerase protein p80 (141, 142); and a telomerase catalytic subunit containing reverse transcriptase motifs, human telomerase reverse transcriptase (hTERT), also known as hEST2/hTRT,
which was isolated based on its similarity to its counterparts in yeast and Euplotes aediculatus (143-148). Of great interest are how these components regulate the activity of human telomerase and how they are regulated by factors that could playa role in cellular senescence, immortalization, and carcinogenesis. hTERC and hTEPl are expressed in both normal (mortal) and immortal cells. Expression of hTERT, however, is detected in telomerase positive cells but not in telomerase negative cells. Furthermore, introduction of the hTERT gene under the control of a constitutive promoter results in telomerase activity in cells that lack telomerase activity. Thus, hTERT is the catalytic subunit of telomerase, and repression of hTERT expression must be key in negatively regulating telomerase and thereby lifespan in normal cells.
We observed that the introduction of a normal human chromosome 3 into the renal cell carcinoma cell line RCC23 by microcell-mediated chromosome transfer results in a cellular senescence-like cell growth arrest. Introduction of chromosome 3 induces repression of telomerase activity and shortening of telomeres (149). We further established a causative role for telomerase repression in normal human chromosome 3-induced cellular senescence (150) and, most importantly, observed a link between a cellular senescence-inducing gene and transcriptional repression of the catalytic subunit of telomerase (hTERT) (Fig. 1.6). Cloning of this repressor protein may provide understanding of a key event in cancer.
1.4.2. PATHWAYS TO CELLULAR SENESCENCE
We observed that introduction of different chromosomes can independently indace senescence in the same immortal cell line (151). Using a human endometrial carcinoma cell line, senescence genes for these cells were mapped to chromosomes land 18 (151). This finding implies that multiple pathways of senescence exist and that immortal cells arise due to defects in each of these pathways. The senescence program could be activated by a single pathway and immortal cells would arise due to mutations in any of the genes that encode proteins involved in this pathway.
Fig. 1.6. Transcriptional repression of hTERT gene by a telomerase repressor gene on chromosome 3.
Fig. 1.7. Model for cellular senescence.
An alternative hypothesis is that the senescence program is activated by multiple, independent pathways (Fig. 1.7). Immortal cells would require at least one mutation in each pathway. Mutations that affect only a single pathway would not result in cells that were immortal, but the cells might have an extended life - span. For example, SV40 virus infection (which inactivates Rb and p53 proteins) results in extended lifespan but not immortalization of infected cells (152). Additional genetic changes, possibly loss of chromosome 6, are required for immortalization of SV40-infected human cells (153). Reintroduction of a normal chromosome 6 results in senescence of SV40 immortal cells (154). Antisense downregulation of Rb and p53 mRNAs also results in extension of the lifespan of human cells without immortalization (132).
The multiple pathways to senescence hypothesis is also consistent with the multistep nature of immortalization observed in chemically induced immortalization (5). In addition, the inability to assign certain immortal cells to a single complementation group further supports this hypothesis (155,156).
An important aspect of the hypothesis of multiple pathways to cellular senescence is that it explains why many tumor-derived cells are not immortal. Hayflick has shown that cells from adults can be grown in culture for 14 to 29 population doublings (113). If all the changes necessary for tumorigenic conversion were to accumulate in an adult cell without loss or gain of lifespan potential (which may be unlikely), then this cell could grow to form a tumor of only 16,354 cells (14 doublings or 214 cells) to 5.4x108 cells (29 doublings or p9 cells). It is estimated that a tumor formed after 30 cell doublings would be approximately 1 cm3 size (Fig. 1.8).
Fig. 1.8. Cellular population doubling potential versus tumor size. Adapted from DeVita et al. (174)
Interestingly, Paraskeva and coworkers have shown that colon adenomas of < 1 cm3 in size are rarely capable of indefinite growth in vitro .adenomas of >1 cm3 are often immortal (157-159). This observation supports the hypothesis that escape from senescence is a requirement for growth beyond a certain size or cell number. A tumor of < 1 cm3 may be but not generally. For the tumor to expand, an extension of the lifespan would be necessary. Extension of the lifespan to 40 population doublings (pd) would yield a tumor of 1 kg, whereas an extension to 50 pd would yield of 1000 kg, which would very certainly be sufficient to kill the host. Thus, immortalization is not as important as extension of lifespan for neoplastic progression. Mutation of a senescence gene in one pathway may result in an extended lifespan without immortalization according to the multiple pathways model. This hypothesis may explain why tumor cells are not always immortal.
1.4.3. CELLULAR SENESCENCE IN TELOMERASE POSITIVE HAMSTER CELLS
Unlike normal human cells, we have observed that normal, diploid Syrian hamster embryo cells (SHE) express the enzyme telomerase but still undergo senescence at the end of their replicative lifespan. After 20-30 pd these cells cease proliferating, enlarge in size, exhibit a pH 6.0 senescence-associated b-galactosidase activity (160), and fail to phosphorylate the RB protein (161) or enter into S - phase after serum stimulation. We have observed that SHE cells express telomerase throughout their replicative lifespan and the average telomere length does not appear to decrease, remaining at about 23 kb in senescent cells. In addition, individual clones of SHE cells also have telomerase activity and telomeres that do not decrease in length, ruling out the possibility that there is a rare, immortal sub-population of telomerase-expressing cells that is lost during passaging.
There are four possible ways to interpret these findings. First, SHE cell telomerase may not completely maintain telomeres, allowing a slight reduction in telomere length that is below the resolution of the TRFs but sufficient to cause loss of one or a few telomeres that are very short (probably >5 kb) in young cells, and leading to senescence. Zijlmans (162) found that telomere lengths ranged from 5 kb to 80 kb in three strains of inbred mouse. The similarity in appearance in TRFs of hamster and mouse telomeres suggests that hamster telomeres may also have a wide variability in length. The second possibility is that not all telomeres are maintained equally, allowing progressive reduction in one or a few telomeres in each cell while the majority remain long. Since Zijlmans (162) reported similar rate of loss in all telomeres, this seems unlikely. Third, sudden DNA damage could cause loss of a telomere.
There is evidence that oxidative damage may cause DNA breaks that are more severe in the telomeres than the rest of the chromosomes (163). Even in cells that express telomerase, complete and sudden loss of a telomere may be capable of causing senescence through a telomere loss signaling pathway. The final possibility is that SHE cells may senesce in response to some signal that is not related to telomere length. Although there is substantial evidence supporting the hypothesis that telomere loss is a senescence signal in human cells, we do not yet know if there are other signals that can independently initiate senescence pathways. Oxidative damage, which alters cellular poroteins. lipids and nucleic acids, and DNA damage have been shown to cause cells to enter a state that is indistinguishable from cellular senescence (164-166).
These agents, however, may also cause breakage of DNA in the telomeres thereby causing senescence through a telomere loss pathway. Loss of CpG methylation (167) has also been consistently found in senescent cells and has been hypothesised to initiate senescence by activation of senescence genes. If SHE cells do not senesce as a result of telomere shortening, they may provide a way to study other senescence pathways that occur in the absence of telomere loss.
A major theory for the cause of cellular senescence is the telomere shortening hypothesis (136, 137). This theory provides an explanation for the mitotic clock in replicative cellular senescence and is supported by several lines of evidence including:
1) telomere length correlates with lifespan of human cells,
2) extension of telomere length increases the lifespan of cancer cells in hybrids with normal cells,
3) inhibition of telomerase activity and reduction of telomere length by introduction of chromosome 3 into a renal cancer cell induces cellular senescence, and
4) introduction of catalytic subunit of telomerase (hTERT) into normal cells extends lifespan and possibly induces immortality (168).
However, there is also evidence that suggests that signals other than telomere shortening may be the cause of cellular senescence. This evidence includes the observations that:
1) many microcell hybrids and cell-cell hybrids between normal cells and immortal cells senesce without repression of telomerase,
2) multiple pathways to senescence exist, suggesting multiple inducers,
3) normal, diploid Syrian hamster embryo cells senesce while expressing telomerase activity and without detectable telomere shortening, and
4) other possible inducers of cellular senescence have been proposed, including oxidative stress, terminal differentiation and alterations in DNA methylation.
If there is a cause of cellular senescence other than telomere shortening, what possibilities exist? As listed in Table 1.III, a number of other types of damage occur in senescent cells. Among these oxidative stress is an interesting candidate. Several lines of evidence support the role of oxidative damage as a possible cause of replicative cellular senescence. These include:
1) oxidative stress is a major source of spontaneous DNA and protein change in cells and organisms,
2) mutations in genes involved in response to oxidative stress cause increased longevity in vivo in C. Elegans,
3) growth in high oxygen atmosphere causes a decrease in lifespan of cells in culture, while growth in low oxygen causes an increase in lifespan, and
4) Chen and Ames (165) demonstrated that hydrogen peroxide treatment induces a senescence-like state in human fibroblasts.
Possible Mechanisms for Induction of Cell Senescence
Induction of Cellular Senescence by Activation of Signal Transduction Pathways
|Hydrogen Peroxide||Chen and Ames (165)|
|Gamma Irradiation||Di Leonardi et al. (171)|
|Hyperoxia||Von Zglinicki et al. (163)|
|Histone Deacetylase Inhibitors||Ogryzko et al. (172)|
|Oncogenic Ras||Serrano et al. (169)|
|Bleomycin/Actinomycin DNR||Robles and Adami (173)|
|Phosphatidylinositol 3-kinase Inhibitors||Tresini et al.(170)|
Recently, several investigators reported that short-term treatments of young cells with different agents, including several oxidative stress-inducing and DNA damaging treatments, can induce senescent-like phenotypes (Table 1.IV). This is based primarily on the observations of induction of irreversible, growth arrest states with senescent cell-like morphologies and expression of low pH β-galactosidase staining, which is a putative marker of senescent cells (160). One criticism of these types of studies is that the induction of a senescent-like state may not truly recapitulate the senescent cell phenotype or be related to the normal causes of cellular senescence. However, further studies of this type will be important to further define senescent-specific markers and to understand the mechanisms of induction of senescence-associated markers.
Are oxidative stress and telomere shortening independent inducers of cellular senescence? It is possible that oxidative stress induces cellular senescence by affecting the rate of telomere shortening. von Zglinicki et al. (163) found that cells grown in a high oxygen atmosphere experience increased telomere loss, as well as increased single-stranded telomere breaks, and senesce prematurely with the same average telomere length as normal senescent controls.
However, the intriguing possibility also exists that oxidative damage pathways may have targets other than DNA. Support for this hypothesis is provided in some studies including:
1) Serrano et al. (168) found that retroviral transfection of oncogenic ras into normal human, rat, and mouse cells causes a senescence-Iike state within 3 population doublings accompanied by the expression of senescence - associated genes including senescence-associated p-galactosidase,
2) Chen and Ames (165) found that hydrogen peroxide is able to induce a senescence-Iike state in human fibroblasts, and hydrogen peroxide is known to activate the ras pathway in the cell membrane, suggesting that an important possible target of oxidative damage may be lipids in the cell membrane, and
3) Tresini et al. (170) found that reversible suppression of PI3 kinase, also involved in cell-membrane signal transduction, causes a senescence-Iike state that is dependent on the continued presence of the PI3 kinase inhibitor.
Further studies will be required to determine if telomere shortening is the central signalling mechanism for induction of cellular senescence or whether other telomere-independent inducers are also involved.
Cellular replicative senescence is controlled by multiple genes. Important regulatory genes in this process are mutated or inactivated in immortal cancer cells that have escaped the senescence program. This program can be restored in these cells by introduction of normal chromosomes, allowing mapping and cloning of these genes. Two functions of these genes are elucidated from these studies -regulation of cell cycle control and regulation of telomerase. Multiple pathways are also suggested by the ability of independent chromosomes to induce senescence. Telomere shortening is an important inducer of cellular senescence. However, some hybrids senesce without telomere shortening or repression of telomerase, and normal, diploid hamster cells senesce without apparent telomere shortening. These findings suggest the action of other inducers, such as oxidative stress, are important activators of cellular senescence pathways.