Replicative senescence at the molecular level in vitro

Basic eukaryotic cell cycle

The cell cycle of most cells consists of four coordinated processes: cell growth, DNA replication, distribution of the duplicated chromosomes to daughter cells and cell division. The completion of all these processes takes approximately 24 hours. The cell cycle is divided into two basic parts: mitosis and interphase. Mitosis (or M phase) occurs at the end of the cell cycle and lasts only about an hour. During mitosis, the separation of daughter chromosomes occurs followed by cytokinesis (cell division). The majority of the cell cycle is spent in interphase-the period between mitosis. During interphase both cell growth and DNA replication occur. The cell grows at a steady rate throughout interphase with most cells doubling in size between one mitosis and the next. DNA synthesis however occurs only in a defined phase of interphase. Within interphase there are three distinct phases. Following M phase is the G1 phase, S phase and G2 phase (Figure 1). During G1, the cell is metabolically active and continuously grows, but does not replicate its DNA. During S phase the complete genetic material of the cell must be duplicated. In G2, the cell continues to grow and proteins are synthesised ready for mitosis. Cells in G1, which are not ready to progress through the cell cycle, enter a resting stage known as G0 or quiescence. Most eukaryotic cells spend the majority of their time in a quiescent state.

Throughout the cell cycle there are a number of checkpoints which regulate cell progression from one phase to another. There is a G1 checkpoint that ensures everything is ready for DNA synthesis, a G2 checkpoint to determine whether the cell can proceed to M phase and a checkpoint within M phase to ensure the cell is ready to complete cell division.

Entry into each of the phases of the cell cycle are controlled by two classes of molecules, cyclins and cyclin dependent kinases (CDKs). Cyclins form the regulatory subunits and CDKs the catalytic subunits of an activated heterodimer; cyclins have no catalytic activity and CDKs are inactive in the absence of a partner cyclin. When activated by a bound cyclin, CDKs perform a biochemical reaction called phosphorylation that activates or inactivates target proteins to orchestrate coordinated entry into the next phase of the cell cycle. Different cyclin-CDK combinations determine which downstream proteins are targeted.

For example, during the latter stages of the G1 phase cyclin D1 forms a complex with CDK4, which subsequently phosphorylates and inactivates retinoblastoma (Rb) growth repression (Connell-Crowley et al, 1997). Conversely, growth arrest caused by DNA damage for example is the result of an up-regulation of CDK inhibitors such as p21 and p16 which bind to and inhibit the activity of CDK thereby preventing the phosphorylation of Rb (Aprelikova et al, 1995).

Figure 1: Basic overview of the eukaryotic cell cycle, showing each of the different phases

Replicative capacity of cells from disease states

The gradual appearance of senescent cells may contribute to the development of age-related disease. However, the presence of disease by other mechanisms may result in accelerated senescence. Disease may cause tissue damage which leads to cellular turnover for the purpose of replacing lost cells. This exhausts the replicative capacity of the cells and accelerates the appearance of senescent cells. For example Goldstein and co-workers (1978) looked at the replicative lifespan of fibroblasts from normal, prediabetic, diabetic donors. Diabetes mellitus is a common genetically determined disorder associated with reduced life expectancy. This study confirmed earlier findings that there is an inverse correlation between donor age and replicative lifespan, but emphasised the importance of physiological state of the donors. Normal cell strains showed significantly better growth capacity than diabetic and prediabetic cells. The results indicated that with an increasing predisposition to diabetes, there is a progressive decrease in replicative capacity.

Another group investigating atherosclerosis took vascular smooth muscle cells (VSMC) from human atherosclerotic plaques and grew them in culture (Bennett et al, 1998). Results showed that VSMCs taken from plaques have lower rates of proliferation and underwent senescence earlier than cells derived from normal vessels.

A more recent study looked at the replicative capacity of osteoblasts in Rheumatoid arthritis (RA) compared with Osteoarthritis (OA) (Yudoh et al, 2000). The results indicated that the replicative capacity of osteoblasts decreased gradually with donor age and this decrease was higher in RA patients than with OA patients at any donor age. They also reported an increase in senescent osteoblastic cells with age in both groups in which the rate of expression of senescent cells was higher in RA patients than with age-matched OA patients.

Tesco et al (1993) looked at the replicative capacity of fibroblasts in patients with familial Alzheimer’s disease (FAD) to examine whether features compatible with a systemic premature aging were present. Data showed that there was no significant difference in replicative capacity of fibroblasts between FAD patients and controls. This is not a surprising result, since the fibroblasts studied are unrelated to the development of FAD and if features of premature ageing were present they would have most likely manifested themselves as other diseases other than just Alzheimer’s. For example, Werner’s syndrome is a premature ageing disorder which displays a multitude of age-related afflictions including diabetes and heart disease (Kipling and Faragher, 1997). When fibroblasts were taken from patients with Werner’s syndrome and grown in culture, the number of population doublings achieved was smaller compared with normal cells of a similar chronological age (Martin et al, 1970)

These studies suggest that disease is an important factor contributing to the exhaustion of the replicative capacity of cells. However, it is possible that some diseases arise as a result of the gradual increase in senescent cells with time. It is also possible that unknown factors result in accelerated senescence, which subsequently manifests itself as a biological impairment or disease.

.

Factors, other than disease, which may contribute to cellular injury and cell loss, may be environmental such as UV radiation, chemical damage from smoking and foods, and normal biological damage from general wear and tear.

Relationship between replicative capacity and organismal ageing

Leonard Hayflick was the first to propose that the senescence of normal cells may contribute to the organismal ageing. Investigations into this proposal started by comparing the replicative potential of cells, usually fibroblasts, extracted from individuals at various ages.

The first of these studies showed an inverse relationship between donor age and the number of population doublings achieved in vitro (Martin et al, 1970). This study looked at the replicative lifespan of fibroblasts taken from 100 subjects with an age range from foetal to 90 years. These cells were cultured and the number of population doublings before entering senescence was recorded. The results showed that the replicative potential decreased as donor age increased. A later study showed similar results (Schneider, 1979). This study looked at the ability of fibroblasts taken from young (20-35 years) and old (65+ years) to proliferate in culture. It was reported that cell cultures from old human donors have a reduction in their proliferative capacity. A more recent study looked at the replicative capacity of human adrenocortical cells to proliferate as a function of donor age (Yang et al, 2001). Again, it was found that younger cells have a higher proliferative capability than the old. In this instance, population doubling fell from 50 for foetal cells to almost a total lack of division in culture from older cells.

To investigate the possible link between replicative lifespan and organismal ageing, a few studies compared replicative capacity with longevity in animals. One such study investigated the relationship between longevity of eight mammalian species (mouse, rat, rat-kangaroo, mink, rabbit, bat, horse and human) and the lifespan of normal fibroblasts in vitro (Röhme, 1981). It was reported that there was a direct relationship found between the longevity of the eight mammalian species and the replicative capacity of their cultured fibroblasts. A much later, but similar study, compared animal life spans and in vitro replicative capacity of skin fibroblasts in groupings of small, middle, large, and very large breeds of dogs of specific ages (Li et al, 1996). It was found that the life spans were inversely correlated to the frame sizes of the breeds. It was shown that all the small breeds studied have a longer life span than that of the large breeds. The replicative capacity of fibroblasts from the large dogs (Great Dane and Irish Wolfhound) was significantly decreased compared with that of the small dogs. The reasoning behind these observations may again be due to varying degrees of cell turnover between the species. Large dogs consist of more cells than small dogs and as a result more cell turnover was initially required in their development compared to small dogs. This increase in cell turnover would subsequently lead to a decrease in replicative potential and an increase in the rate of senescent cell formation.

Interestingly, a recent study looked at the replicative capacity of 124 skin fibroblast cell lines from donors of different ages which were medically examined and declared “healthy” (Cristofalo et al, 1998). Healthy people were used specifically as previous studies, discussed later, have shown that disease states may accelerate the reduction in replicative capacity. Results indicated that there was no significant correlation between the replicative capacity of the cell lines and donor age. In the same study, a comparison of multiple cell lines established from the same donors of different ages also failed to show any significant differences. It was concluded that the replicative capacity of fibroblasts in vitro does not correlate with donor age. However, differences in replicative capacity with age may only be observed as a result of increased cell turnover in response to disease and cellular injury. Therefore, a healthy old person who has had little or no cellular injuries or disease would have had little cell turnover and therefore have cells which may have a replicative capacity similar to someone much younger. Thus, this study supports the notion that replicative capacity is an indicator of biological age.

Replicative capacity of tissues from normal human populations

The maximum replicative potential for mitotic cells varies between different cell types. Some cell types, such as endothelial cells, may have a maximum replicative capacity around 30 cPD (cumulative population doublings) while other cell types such as embryonic fibroblasts may have a maximum replicative capacity of 100 cPD. For example, one early study looked at the replicative capacities of several different tissue types (skeletal muscle, bone marrow spicules and mesial of the midupper arm) taken from donors of the same age (Martin et al, 1970). It was found that the replicative capacity of these tissues, despite being taken from the same individual, displayed variation in their replicative capacity. Cultures derived from skin fibroblasts achieved the greatest number of population doublings, bone marrow spicules the least and skeletal muscle giving intermediate results. There are a number of explanations for these observations. The first is that all cells do have the same replicative capacity, but the replicative history (rate of cell turnover) of each tissue at the time of extraction is so different that such variation is observed. Some tissues may have undergone a higher rate of cellular turnover than others, thereby exhausting its replicative capacity earlier. The second is that the replicative history of each tissue is similar, but it is the length of the telomeres between tissues that differs. Some tissues may senesce sooner than others because they started out with shorter telomeres. It is unlikely that these explanations alone are correct. A combination of the two is the most likely cause for such variation in replicative capacity. Tissues differ in both their replicative history and replicative capacities.

The results also show that the replicative capacity of the same tissues between individuals of the same age also differs. This difference may again be due to the same differences which effect proliferative variability between different tissues of the same individual. For example, one individual may have a shorter replicative capacity in a particular tissue than another of the same age due to increases in cell turnover, maybe in response to disease or injury, or maybe differences in initial telomere lengths. Cultured human embryonic fibroblasts were found to senesce at 50±10 cPD (Hayflick and Moorehead, 1961). This meant that some cultures were senescent only after 40 cPD while others at 60 cPD. These differences in replicative lifespan may be a consequence of the stochastic mechanism which triggers a cell to senesce. Therefore, the difference in replicative capacities of the same tissues between individuals of the same age may also be due to the stochastic events which govern a cell becoming senescent. Thus, the replicative capacity of a tissue measures biological age and not chronological age. Unfortunately there have been few studies looking at the replicative capacity of different tissues from the same individuals. This would have given a better insight into the relationship between chronological and biological age.

Replicative senescence at the cellular level in vitro

Historical overview

Until the middle of the 20th century, it was widely held that normal mitotic tissue could not age in a degenerative sense because it had an indefinite capacity to proliferate. This view had developed for two reasons. Firstly, the Nobel laureate Alexis Carrel appeared to have demonstrated the long-term cultivation of normal chick fibroblasts for periods considerably in excess of the lifetime of the animal (Parker, 1938; Witkowski, 1990). Secondly, the technical difficulties associated with tissue culture techniques until the 1950s rendered the duplication of Carrel’s studies very difficult for all but a few highly specialized laboratories (Parker, 1938). Ageing was not the primary research interest of these centres. Thus, it was not until a series of classic experiments by Hayflick & Moorhead in the early 1960s which demonstrated that cultures of normal human fibroblasts did not have an infinite capacity to expand, brought about the idea of intrinsically immortal mitotic tissue into question. Their work demonstrated that, after a finite period of growth, cultures of normal human fibroblasts became completely composed of viable but non-dividing cells (Hayflick & Moorhead 1961; Hayflick 1965). These initial observations have been reproduced in hundreds of studies, and since that time virtually all human mitotic cell types subjected to rigorous study have been shown to undergo this cellular senescence in culture.

Dynamics of normal cell populations.

The early observation by Hayflick and Moorehead (1961) that cultured cells have a maximum limit on the number of divisions before entering senescence lead to the assumption that all cells in a given culture divide roughly the same number of times before entering senescence. Hayflick considered the senescence of cultures was marked by three distinct phases. In Phase 1 the initial culture, was considered to terminate with the formation of the first confluent sheet of cells. Phase 2 was characterised by vigorous growth requiring repeated subculture. In phase 3, the senescence of the culture was characterised by the cessation of mitosis. In this model it was assumed that cultures were composed of a homogenous population of cells which were either all growing (Phase 1 or 2) or all non-growing (Phase 3) and that failure to grow was due to cell death (Kalashnik et al, 2000). The notion that senescence was cell death was soon disproved with the demonstration that RNA synthesis occurred in these cells (Macieira-Coelho et al, 1966). Evidence against the idea that cultures were composed of homogenous populations was provided by a number of different studies. Cristofalo and Scharf (1973) demonstrated the presence of senescent cells in early passage cultures using long pulse-labelling experiments on embryonic fibroblasts. 3H-thymidine labels those cells which have entered S-phase (dividing cells), and since senescence cells are halted in G1 they cannot enter S-phase, and so the percentage of unlabelled cells can be calculated. It was shown that senescent cells are present in early passage cultures and that the percentage of senescent cells gradually increases with each serial passage of the culture. This observation was explained by experiments demonstrating that cultured fibroblasts are composed of cells which display variation in proliferative potential (Smith and Whitney, 1980). Related experiments also showed that two cells arising from a single mitosis differed in their ability to proliferate by as many as eight doublings (Jones et al, 1985). Using the miniclone technique the replicative capacity of individual cells growing in bulk culture can be measured as well as the sizes of colonies generated by dividing cells (Ponton et al, 1983). Results showed that the percentage of glial cells capable of dividing gradually decreases with every new passage. This data is based on the broad distribution of colony sizes which showed a shift from many large colonies to more small colonies as population doublings increased.

Modern techniques for measuring the dynamics of normal cell populations involve measuring not only the senescent fraction of cells, but also the proliferating and apoptotic fraction. The most commonly used method to visualise senescent cells both in culture and in vivo is the senescence-associated beta-galactosidase assay (SA-β-Gal) (Dimri et al, 1995). Although this is a safer method than using 3H-thymidine labels, its robustness as a biomarker is questionable since the assay is dependent upon lysosomal mass (and cell size) rather than growth state (Lee et al, 2006). Cellular proliferation markers such as bromodeoxyuridine (BrdU) and Ki67 are commonly used to label and calculate the proliferating fraction. For measuring the apoptotic fraction, terminal transferase dUTP nick end labelling (TUNEL) is a commonly used method. This assay can detect DNA fragmentation that results from apoptotic signaling cascades.

An example of these methods being used for determining the growth dynamics of human umbilical vein endothelial cells (HUVEC) can be observed in a paper by Kalashnik et al (2000). Results show a gradual decline in the growth fraction as measured by Ki67, an increase in the senescent fraction and the apoptotic fraction remaining unchanged with each serial passage.

These findings thus suggest that the mechanisms resulting in cellular senescence is a stochastic process. As the proliferative capacity of cells declines with age or with increasing population doublings, the mechanism leading to cellular senescence is one in which these stochastic events gradually increases.

Overview of mechanisms of ageing

Ageing is a multi-causal process with some ageing mechanisms being more predominant in some tissues than in others. This is due to the difference in the biological make-up of tissues which makes them more or less predisposed to a particular ageing mechanism. Since tissues are normally composed of a mixture of mitotic cells, post-mitotic cells and long-lived proteins that are functionally inter-related, there is going to be considerable overlap in the ageing mechanism within tissues. An alteration in one tissue component will likely have a direct impact on another. The outcome of such changes is going to be different from tissue to tissue due to impairment of specific structure-function relationships.

Mitotic cells

Senescent phenotype and biological impact

Senescent cells tend to adopt an extracellular matrix (ECM) degrading, proinflammatory phenotype (West et al, 1989; Kletsas et al, 2004). Senescent cells usually up-regulate matrix metalloproteinases (MMPs), enzymes capable of degrading proteins such as collagen and elastin which make up the extracellular matrix. Since the extracellular matrix (ECM) is important for providing support and anchorage for cells, separating different tissues and regulating intercellular communication, its degradation by MMPs is likely to impact all areas of ECM function. MMP activity is normally inhibited by TIMPs (tissue inhibitor of metalloproteinases), but research suggests that that these inhibitors themselves are down-regulated at senescence, thereby further contributing to matrix degradation (Hornebeck, 2003).

Senescent cells also secrete many cytokines which due to their diverse function could have multiple consequences on the ageing of tissues. These secreted proteins may not just impact on local tissue but also tissues found throughout the organism. The presence of cytokines can alter cell functions by up-regulating or down-regulating several genes and their transcription factors, resulting in the production of other cytokines and an increase in the number of surface receptors for other molecules (Gallin and Snyderman, 1999). The ability of cytokines to reach many tissues and have such diverse consequences on cell function suggests that only a small fraction of senescent cells may need to be present for there to be any significant impact on tissue impairment or disease development.

As discussed in post-mitotic ageing, the accumulation of senescent cells in some tissues is likely to reduce the number of cells which can provide support and protection to post-mitotic cells. Therefore, the appearance of senescence cells may have a direct impact on the impairment of post-mitotic tissues.

Some of the changes observed during cellular senescence are also likely to be cell type specific. Different cell types are going to have different transcriptional profiles since their functions are different and these differences may result in tissue specific impairment. For example, in senescent vascular endothelial cells, nitric oxide synthase (eNOS) activity has been found to be decreased (Matsushita et al, 2001; Minamino et al, 2002). Since nitric oxide (NO) is important in regulating vascular function, a decline in its production may have detrimental consequences. A reduction in NO production by eNOS for example has been suggested to be a significant risk factor for cardiovascular disease (Cannon, 1998). This decline in eNOS activity at senescence appears to be specific to vascular endothelial cells. Even if eNOS is produced by other cells types and a similar decline with age is observed, the consequence of such changes is going to be different, if any at all. This is due to alterations in specific structure-function relationships.

Overall, senescent cells within tissues are thought to contribute to the ageing process by:

1) Altering the behaviour of neighbouring growth-competent mitotic cells.
2) Degradation of structural components such as the extracellular matrix.
3) Reducing the pool of growth-competent mitotic cells.
4) Cellular dysfunction: inability to function properly.