We observed that 60% of the population of replicatively senescent normal human foreskin fibroblasts HCA2 consisted of cells with 4N DNA content (Figure ), while the remaining cells were in G1 stage. Intrigued by the result we repeated propidium idodide (PI) staining on replicatively senescent WI-38 and IMR-90 cells, which are the two commonly used normal human fibroblast strains. The senescent WI-38 and IMR-90 cells contained 37% and 39% of cells with 4N DNA content, respectively (Figure ).
Figure 1 Senescent human fibroblast cultures contain a large fraction of putative G2-arrested cells with 4N DNA content. (A) Propidium iodide (PI) staining and flow cyctometric analysis of HCA2 normal human foreskin fibroblasts. Cells entered senescence at PD73. (more ...)
We hypothesized that these cells are transiently arrested in G2 stage and have not yet progressed to mitosis. We then incubated HCA2 senescent cells for additional 10 weeks counting from the time the cultures entered senescence and proliferation ceased and repeated cell cycle analysis weekly. No changes in cell cycle distribution were observed (Figure ). This result suggests that the observed cell cycle distribution with 2N and 4N populations is a stable terminally growth arrested state for the human fibroblasts.
Although the cell cycle distributions we observed (Figure ) closely resembled G1/G2 population, we considered two alternative explanations for the origin of the 4N cells. First, these cells may be polynucleated, containing two nuclei with 2N DNA content. Microscopy analysis of senescent cells stained with DAPI showed that senescent cell population contained 5% of cells with more than one nucleus (Figure ), which cannot account for the observed 36-60% of cells with 4N DNA content. The second possibility is that the 4N cells are tertaploid cells in G1 stage. This implies a very high level of polyploidy in replicatively senescent fibroblasts. To test this possibility, we examined presenescent HCA2 cells at PD63, which are still actively dividing, for the presence of tertaploid G2 fraction with 8N DNA content. The fraction of such cells was 0.3%, indicating that the presenescent culture contains a very low number of tetraploid cells (Figure ). This result argues against the hypothesis that 4N population observed in senescent cultures consists of G1 arrested tetraploids. We then performed in situ staining of young and replicatively senescent HCA2 cells with a probe to the centromeric region of chromosome 8. Diploid G1 cells are expected to show two signals corresponding to the two homologues of chromosome 8, diploid G2 cells are also expected to show two signals corresponding to two pairs of closely positioned sister chromatids, while tetraploid G1 cells are expected to show four signals. Eighty five percent of young cells showed two signals, and 11% showed four signals. Sixty percent of senescent cells showed two signals, and 28% showed four signals (Figure ). As described above, HCA2 senescent culture contained 40% of cells with 2N DNA content and 60% of cells with 4N DNA content (Figure ), thus tetraploid G1 cells can potentially explain only less than half of the population of cells with 4N DNA content. Combined with the lack of tetraploid G2 population in pre-senescent cultures (Figure ), this result indicates that the 4N cell population is unlikely to be composed of G1-arrested tetraploid cells. To explain the origin of the cells with four dispersed centromeres, we hypothesize that when senescent cells undergo prolonged arrest in G2 stage, sister chromatids may separate resulting in appearance of cells with 4N DNA content and four separated chromatids. We further analyzed the protein level of cyclins in replicatively senescent cells. In agreement with previous findings [7
], the level of cyclin D1 which promotes progression through G1/S phases was elevated, while cyclin B1, promoting G2/M transition, was absent in senescent cells (Figure ). This cyclin profile is consistent with G2-arrest.
The senescent cell population with 4N DNA content is not due to polynucleated cells or tetraploid cells.
Collectively, our results indicate that replicatively senescent cell populations arrest in both G1 and G2 stages. One potential implication of G2 arrest in senescent cells is that many of these cells retain sister chromatids available as a backup or as a DNA repair template. Another implication is that the 4N DNA content may contribute to the complex changes in expression patterns observed in senescent cells.
It was proposed that replicative senescence represents a state where the growth-promoting TOR pathway is activated, while the cell cycle is blocked leading to cellular hypertrophy and the typical enlarged phenotype of senescent cells [10
]. Indeed, treatment with TOR inhibitor rapamycin suppressed senescence in mouse cells [11
], and inhibition of TORC1 attenuated replicative and RAS-induced senescence in human cells [12
]. Our finding that a large fraction of replicatively senescent cells arrests in G2 phase supports this view of replicative senescence. We propose that as cells enter replicative senescence they progress to G2 phase due to activated TOR pathway but cannot enter to mitosis. As a result, replicatively senescent cells accumulate in G2 phase, and a mixture of G1/G2 – arrested cells represents the terminal cell cycle arrest in replicatively senescent cultures.