Bypass of Senescence After Disruption of p21CIP1/WAF1 Gene in Normal Diploid Human Fibroblasts

See allHide authors and affiliations

Science  08 Aug 1997:
Vol. 277, Issue 5327, pp. 831-834
DOI: 10.1126/science.277.5327.831


Most somatic cells die after a finite number of cell divisions, a phenomenon described as senescence. The p21CIP1/WAF1 gene encodes an inhibitor of cyclin-dependent kinases. Inactivation of p21 by two sequential rounds of targeted homologous recombination was sufficient to bypass senescence in normal diploid human fibroblasts. At the checkpoint between the prereplicative phase of growth and the phase of chromosome replication, cells lacking p21 failed to arrest the cell cycle in response to DNA damage, but their apoptotic response and genomic stability were unaltered. These results establish the feasibility of using gene targeting for genetic studies of normal human cells.

The replicative life-span of somatic cells reflects the number of cell divisions, not chronological time, and may contribute to organismic aging (1). Shortening of telomeres may be the molecular mechanism that triggers an irreversible arrest, referred to as senescence, of the prereplicative phase of growth in the cell cycle (G1) (2). Genes that have been implicated in regulating senescence include tumor suppressors p53 (3) and RB1 (4), cyclin-dependent kinase (Cdk) inhibitors p21CIP1/WAF1 (5) and p16INK4a (6), and several currently unidentified genes (7). Viral oncoproteins that interfere with p53 and RB1 cause bypass of senescence and extended life-span, followed by a decline designated as crisis (8). Two limitations have hampered studies of human senescence. First, viral oncoproteins may not completely inactivate their targets. Second, studies in rodents cannot be extrapolated to humans because of interspecies differences in the mechanisms of senescence and immortalization (9).

Introduction of null mutations into a cellular gene is a direct and unambiguous way to test the function of that gene. We were concerned that normal human diploid fibroblasts (HDFs) would senesce before two sequential rounds of gene targeting could be completed. Therefore, we developed strategies for efficient gene targeting in somatic cells (10), established a culture system that allows high single-cell cloning efficiency (11), and generated a new cell strain (LF1) (Fig. 1A) (12).

Figure 1

Extension of in vitro life-span after loss of p21 expression. (A) Experimental outline. (B) Passage history of LF1: 20% O2(▵); 5% O2 (○). Senescence was reached at passage 48 (16). (C) Passage history of HO7.2-1 (○) and its nontargeted siblings (•). All cell strains are clonal and originated in the same experiment (21). (D) Data for +/− cell strains 5.4-1 (▵), 8.2-2 (▿), and 9.3-1 (○) replotted to illustrate nonuniform kinetics of senescence. (E) Stained dishes of senescent +/+ and +/− cell strains. (i) LF1 (+/+; passage 48) 23 days after plating (day 201; B); (ii) same culture 102 days after plating (day 280); (iii) 5.4-1 (+/−; passage 3) 42 days after plating (day 75; D); (iv) 9.3-1 (+/−; passage 5) 50 days after plating (day 93). Plates i and ii were ∼80% confluent; with the exception of the colonies (dark-staining areas) plates iii and iv were ∼20% confluent. (F) Growth curves of LF1 (+/+; passage 13) (□), HE1.3-2 (+/−; passage 5) (▵), HO7.2-1 (−/−; passage 5) (○), and 9.4 (+/−; passage 3) (▿). Doubling times were 18, 25, 39, and 61 hours, respectively (allR 2 values were ≥0.98). Growth of 9.4 was measured six passages before senescence.

LF1 cells (5 × 107 cells) at passage 7 were electroporated with a targeting vector containing a neomycin (neo)-resistance gene (Fig.2A) (13). Twenty colonies were obtained and expanded into clonal cell strains. Southern (DNA) blotting analysis showed that three clones (HE1.2-1, HE1.3-2, HE3.2-1) (14) contained one targeted p21 gene copy (Fig. 2B). HE1.3-2 cells (5 × 107 cells) were electroporated with a vector containing a hygromycin (hyg)-resistance gene (Fig.2A), yielding 24 clones, one of which (HO7.2-1) had targeted the second p21 gene copy (Fig. 2B), and two of which had retargeted theneo-targeted gene copy (15, 16). Protein immunoblotting analysis confirmed that the HO7.2-1 clone did not express p21 protein (Fig. 2C).

Figure 2

Gene targeting of p21 locus. (A) Schematic representation of wild-type (top),neo-targeted (middle), and hyg-targeted (bottom) gene copies. Exons, filled boxes; Southern (DNA) hybridization probes, solid bars. (B) Southern (DNA) hybridization analysis. Cell strains are listed on top. Size markers (kbp) are on the right. Lanes 1 to 5 (Bam HI), first round of targeting; 1.1-2 is a nontargeted sibling. Lanes 6 to 9 (Bgl II), second round of targeting; 6.2-1 is a nontargeted sibling. Lanes 10 to 15 (Bam HI), LOH in +/− cell strains; 10.2, 2.4, 6.2-3, and 9.3-2 are nontargeted siblings of HO7.2-1. (C) Protein immunoblotting analysis. Cell strains are listed on top. Size markers (kD) are on the left. Equal amounts of protein were loaded in all lanes. 6.4, 9.3-1, 9.4, and 5.4-1 are nontargeted siblings of HO7.2-1. Samples were harvested at the following passages: HE1.3-2, passage 9; HO7.2-1, passage 13; 6.4, passage 8; 9.3-1, passage 7; 9.4, passage 7. Lanes 6 and 7, 5.4-1 harvested at passage 4 and passage 7, respectively.

The hygromycin-resistant colonies were expanded into cell strains and passaged until senescence (Fig. 1C) (17). The 21 nontargeted (p21 +/−) strains senesced between passages 2 and 10 (mean passage 6.76 ± 2.55 SD) (18), whereas the HO7.2-1 strain did not cease proliferation until passage 19, when it displayed signs of crisis (19). During the period of extended life-span, no cell death was evident in HO7.2-1 cultures. Because in our experimental regimen one passage is equivalent to a minimum of two population doublings (PD) (11), loss of p21 resulted in quantitatively the same life-span extension (20 to 30 PD) as the introduction of SV40 large tumor antigen (T-Ag) (20).

Most p21 +/− strains senesced with unusual kinetics (Fig. 1D). Rather than the rapid and irreversible decline in growth characteristic of normal HDFs (Fig. 1B) (21) many +/− strains resumed growth after an initial decline in proliferation. Cultures with a high proportion of senescent cells spontaneously gave rise to colonies of healthy cells (Fig. 1E); it was possible to clone some of those colonies and propagate them for several passages. The same phenomenon occurred with an independently neo-targeted p21 +/− strain (HE3.2-1) that had been transfected and selected for hygromycin resistance (15). Spontaneous life-span extension was not observed with the parental LF1 strain, nor has it been reported to occur spontaneously in cultures of normal HDFs.

Southern (DNA) hybridization revealed that the apparent extension of life-span was accompanied by loss of heterozygosity (LOH) at the p21 locus (Fig. 2B). The intensity of the wild-type band diminished gradually with passage. Protein immunoblotting analysis confirmed loss of p21 expression (Fig. 2C). In contrast, p21 expression persisted at late passage in p21 +/− strains displaying kinetics of senescence indistinguishable from normal HDFs. p21 was lost in 12 of 12 cultures displaying extended life-span but persisted in 3 of 3 cultures with normal kinetics of senescence.

Cells expressing SV40 T-Ag continue to shorten their telomeres during the extended life-span phase (22). Early-passage HO7.2-1 cells had shorter telomeres than senescent LF1 cells, and telomeres continued to shorten as HO7.2-1 cells approached crisis (Fig.3A). Both the reduction of telomere length and the weak hybridization signal are indicative of cells in crisis. Neither presenescent LF1 cells nor HO7.2-1 cells during their extended life-span phase expressed telomerase activity (23,24).

Figure 3

Telomere length (A) and p16 expression (B) during aging of p21 +/+ and −/− cells. Cell strains are listed on top. Passage number (p) is indicated below each lane. Telomere length was measured by the terminal restriction fragment assay (22). One microgram of DNA was loaded in each lane. Size markers (kbp) are on the left. Expression of p16 protein was measured by immunoblotting. Equal amounts of protein were loaded in all lanes.

The Cdk inhibitor p16 is up-regulated in senescent cells (6) and thus has been linked with entry into senescence. Expression of p16 increased with age in LF1 cells and continued to increase in HO7.2-1 cells during their extended life-span (Fig. 3B). Thus, loss of p21 allows cells to bypass senescence in spite of expression of p16.

There was no correlation between expression of p21 and growth rate: the +/− strain 9.4 (which expressed p21) (Fig. 2C) grew more slowly than the −/− strain HO7.2-1, which in turn grew more slowly than its +/− parent HE1.3-2 (Fig. 1F). Furthermore, there was little difference in growth between strain 9.4 and several of its siblings that lacked p21. Thus p21 is not a growth-rate determinant in normal HDFs.

Cell cycle profiles were identical in exponentially growing LF1 and HO7.2-1 cultures (Fig. 4A), but LF1 cells responded more completely to serum withdrawal. Therefore, although p21 evidently contributes to the ability of normal human cells to become quiescent, additional factors also must participate. Consistent with the phenotype of murine p21 null cells (25), HO7.2-1 cells did not arrest at the DNA damage-induced G1checkpoint (Fig. 4B). Loss of p21 in a human adenocarcinoma cell line enhances apoptosis (26). HO7.2-1 cells grew more slowly but became apoptotic at the same rate as LF1 cells (Fig. 4, C and D) (27).

Figure 4

Phenotypes of p21 +/+ and −/− cells. (A) Cell cycle distribution. Exponential cells: cultures were passaged 1:4 twice at <50% confluence. Quiescent cells: cultures at 80% confluence were incubated in 0.25% serum for 60 hours. Cells were stained with propidium iodide for flow cytometry. S, phase of chromosome replication; M, mitosis; G2, period between S phase and onset of M. (B) S-phase entry. Quiescent cells were stimulated (15% FBS) and after 2 hours were irradiated with 12 gray of gamma rays. LF1 and HO7.2-1 samples were collected for flow cytometry 16.5 and 21.5 hours after stimulation, respectively. Analogous results were obtained when cisplatin was used to induce DNA damage (15). (C and D) Apoptosis. Cisplatin (cis; 10 μg/ml) was added at 50% confluence (C) or 2 hours after serum stimulation (D). Total cells (adherent plus floating cells) were counted with a Coulter counter. Analogous results were obtained with etoposide (15).

To investigate the role of p21 in genomic instability, amplification of the CAD (trifunctional enzyme carbamoyl-phosphate synthetase, aspartate transcarbamylase, dihydroorotase) gene was selected by using three times and nine times the median lethal dose of the drug N -(phosphonacetyl)-l-aspartate (27). Treatment of HO7.2-1 cells (1 × 106 cells at each concentration of drug) over a 3-month period produced no colonies (15).

Our results clearly establish the importance of p21 as a key regulator of senescence. We find that of the three major functions of p53—checkpoint control, apoptosis, and genome stability—in normal human cells, p21 has a role in only the first one. It is highly probable that the ability of p21 null cells to become apoptotic may prevent them from developing into cancers. The methods we have developed allow direct manipulation of genes in normal human cells.

  • * Present address: Department of Pathology, New York University Medical Center, New York, NY 10016, USA.

  • To whom correspondence should be addressed. E-mail: John_Sedivy{at}


View Abstract

Navigate This Article