They described a similar ranking of the reactivity of the compounds. Absolute sensitivity depended on the ROG/NOX ratio. At ROG/NOX = 4 (similar to Los Angeles), the reactivity of HCHO to methanol was between 20 and 70 to 1, about the same as in our study. 18. This work was supported by the Califomia Air Resources Board and the South Coast Air Quality Management District.
or extreme example of a quiescent cell. Within this context, one or more growthrelated proto-oncogenes might be under dominant repression in senescent cells. However, quiescent and senescent fibroblasts showed similar basal and serum-inducible expression of the c-myc and c-H-ras proto-oncogenes (6), both of which stimu14 August 1989; accepted 3 November 1989 late proliferation and are required for the ability to leave Go (12). Because serum generally induces these genes in the early and midportions of the interval preceding DNA synthesis (the GO/GI interval), it has Repression of c-fos Transcription and an Altered been suggested that senescent fibroblasts arrest growth in late GI (6, 9). On the other Genetic Program in Senescent Human Fibroblasts hand, based on morphology and the pattem of proteins synthesized throughout the proTARA SESHADRI AND JUDITH CAMPISI* liferative life-span, it has also been suggested that senescent fibroblasts have undergone Normal cells in culture invariably undergo senescence, whereby they cease prolifera- terminal differentiation (11). The growth tion after a finite number of doublings. Irreversible changes in gene expression arrest therefore may be one manifestation of occurred in senescent human fetal lung fibroblasts: a non-cell cycle-regulated mRNA a more complex phenotypic change. Here, we describe several differences in was partially repressed; an unusual polyadenylated histone mRNA was expressed; although serum induced c-H-ras, c-myc, and ornithine decarboxylase mRNA normally, gene expression between quiescent and seornithine decarboxylase activity was deficient; and serum did not induce mRNA for a nescent fibroblasts. These include gene rereplication-dependent histone and for the c-fos proto-oncogene. The loss of c-fos pression and novel gene induction that ocinducibility was the result of a specific, transcriptional block. The results suggest that cur only in senescent cells. Most striking, csenescent fibroblasts were unable to proliferate because of, at least in part, selective fos, which in quiescent fibroblasts is induced repression of c-fos; moreover, the multiple changes in gene expression support the view early in the GO/GI interval (14) and is essential for proliferation (15), is repressed at the that cellular senescence is a process of terminal differentiation. level of transcription in senescent cells. Our FUNDAMENTAL FEATURE OF NORlow but readily measurable frequencies. Rel- results suggest that the growth arrest shown mal cells is their limited ability to ative to rodent fibroblasts, human fibro- by senescent fibroblasts is distinct from proliferate in culture. Cells capable blasts are exceedingly resistant to transfor- growth-arrest states described for early pasof proliferation in vivo often go through an mation by chemical carcinogens, radiation, sage cells, and c-fos repression may at least initial mitotic period in culture, but invari- and oncogenic viruses (3, 4). partially explain why senescent cells fail to ably there is a gradual decline in cell diviSenescence is not programmed cell death. proliferate. To gain insight into the nature of cellular sion; in cells from humans and certain other Senescent fibroblasts remain viable for long species, the decline is virtually irreversible periods (many months), during which they senescence, we examined the pattem of gene and complete (1). This progression has been synthesize RNA and protein (5, 7). Early expression in human fibroblasts at the extermed the finite life-span phenotype or passage fibroblasts enter a reversible, proli- tremes of their proliferative life-span. Hucellular senescence. feratively quiescent state (Go) when de- man fetal lung fibroblasts [strain WI-38 (1)] Cellular senescence has been studied most prived of growth factors; they resume prolif- were used at early passage, when the cells extensively in cultures of human fibroblasts. eration when appropriate growth factors are had undergone 70% (typiGenerally, human fibroblasts senesce after resupplied (8). However, senescent cells cally 75 to 85%) were capable of DNA 20 to 60 population doublings (PDs). Be- cannot be stimulated to enter the S phase of synthesis, and at late passage (senescence) the cell cycle by any combination of growth >48 PDs, when 70% at early passage and 30-fold 16 to mRNA rose 5- to 7-fold within 8 hours, in 24 hours after serum stimulation (Figs. 1 essential agreement with Rittling et al. (6). and 2A), coincident with S phase (22). Thus, senescent fibroblasts fully express at During S phase, all four replication-depenleast four mRNAs and retain the signaling dent histones (H3, H2A, H2B, and H4) pathways for their induction by serum. were synthesized at a high rate (Fig. 2B). In Nonetheless, the constitutively expressed senescent cells, histone 3 mRNA was nearly pHE-7 mRNA declined sharply when the undetectable, regardless of the culture conditions (Figs. 1 and 2A). Moreover, none of cells became senescent. Despite equal basal and induced levels of the replication-dependent histones were odc mRNA at early and late passage, odc synthesized at a high rate at any time tested, enzyme activity was at least twenty-fold although other nuclear proteins were synlower in senescent cells cultured in 10% thesized at comparable rates at early and late serum; after serum deprivation and stimula- passage (Fig. 2B). These results support the tion, the induced activity was two- to three- finding that DNA synthesis and histone fold lower (Table 1). Regardless of passage expression remain coupled as neonatal hunumber, odc activity could not be increased man fibroblasts senesce (23) and suggest by guanosine triphosphate (GTP) (Table 1), that, like pHE-7, the replication-dependent which in some cell lysates activates the en- histone genes are repressed in senescent zyme (20). Recently, a disparity between cells. In senescent cells only, the murine hisodc mRNA and enzyme activity was shown in senescent IMR-90 cells, another human tone-3.2 probe (which detects murine and fetal fibroblast strain (21). The results sug- human histone 3 variants) detected a slower gest that, despite normal mRNA abun- migrating RNA, faintly detectable in total dance, odc is down-regulated in senescent RNA. Its abundance did not change after cells by a translational or post-translational serum deprivation or stimulation (Fig. 2A), and it was 5- to 20-fold less prevalent than mechanism. In contrast to the near normal induction replication-dependent histone 3 mRNA in of c-H-ras, c-myc, and odc mRNA in senes- early passage cells. It appeared to be closely cent cells, there was no induction of the S related to the replication-dependent mRNA phase-specific histone 3 mRNA. At early because the hybridization signal persisted Senescent cells
c-ras-Ha
pHE-7
Actin
c-myc
Table 1. Activity of odc in early- and late-passage fibroblasts. Lysates were prepared from early- and late-passage cells and assayed for odc activity as described (21). When harvested, the cells were either subconfluent in 10% serum (Exponential), serum-deprived (Serum-deprived), or serumstimulated after serum deprivation (Stimulated), as described in the text. Where indicated, 0.1 mM GTP was added to the lysate before addition of the substrate. Odc activity is expressed as picomoles of ['4C]C02 released from ["4C]omithine per hour per milligram of protein. Three experiments with slightly different time points gave similar results. Shown are the data from one experiment; triplicate determinations varied by less than 10%. Growth condition
Odc activ-
+GTP
ity odc
Histone 3.2
Fig. 1. Basal and serum-stimulated mRNA expression in early- and late-passage fibroblasts. Exponentially growing (EXP), serum-deprived (Q) and serum-stimulated [30 min to 30 hours (h)] WI-38 human fetal lung fibroblasts at early and late (senescent) passage were prepared as described in the text.
Total cellular RNA was isolated from 5 x 106 to 1 x 107 cells on 150-mm dishes and analyzed by Northern blotting with DNA probes prepared by random priming or RNA probes prepared by in vitro transcription (17, 25, 35-37). Relative differences in mRNA abundance were quantitated by laser densitometry of autoradiograms of the Northern blots (17, 36). 206
Early passage 45 Exponential 1 Serum-deprived 2 Stimulated, 4 hours 141 Stimulated, 8 hours 131 Stimulated, 16 hours 7 Stimulated, 24 hours Late passage
Exponential Serum-drprived Stimulated, 4 hours
Stimulated, 8 hours Stimulated, 16 hours Stimulated, 24 hours
42
1 1 146 130 6
1
1
or extreme example of a quiescent cell. Within this context, one or more growthrelated proto-oncogenes might be under dominant repression in senescent cells. However, quiescent and senescent fibroblasts showed similar basal and serum-inducible expression of the c-myc and c-H-ras proto-oncogenes (6), both of which stimu14 August 1989; accepted 3 November 1989 late proliferation and are required for the ability to leave Go (12). Because serum generally induces these genes in the early and midportions of the interval preceding DNA synthesis (the GO/GI interval), it has Repression of c-fos Transcription and an Altered been suggested that senescent fibroblasts arrest growth in late GI (6, 9). On the other Genetic Program in Senescent Human Fibroblasts hand, based on morphology and the pattem of proteins synthesized throughout the proTARA SESHADRI AND JUDITH CAMPISI* liferative life-span, it has also been suggested that senescent fibroblasts have undergone Normal cells in culture invariably undergo senescence, whereby they cease prolifera- terminal differentiation (11). The growth tion after a finite number of doublings. Irreversible changes in gene expression arrest therefore may be one manifestation of occurred in senescent human fetal lung fibroblasts: a non-cell cycle-regulated mRNA a more complex phenotypic change. Here, we describe several differences in was partially repressed; an unusual polyadenylated histone mRNA was expressed; although serum induced c-H-ras, c-myc, and ornithine decarboxylase mRNA normally, gene expression between quiescent and seornithine decarboxylase activity was deficient; and serum did not induce mRNA for a nescent fibroblasts. These include gene rereplication-dependent histone and for the c-fos proto-oncogene. The loss of c-fos pression and novel gene induction that ocinducibility was the result of a specific, transcriptional block. The results suggest that cur only in senescent cells. Most striking, csenescent fibroblasts were unable to proliferate because of, at least in part, selective fos, which in quiescent fibroblasts is induced repression of c-fos; moreover, the multiple changes in gene expression support the view early in the GO/GI interval (14) and is essential for proliferation (15), is repressed at the that cellular senescence is a process of terminal differentiation. level of transcription in senescent cells. Our FUNDAMENTAL FEATURE OF NORlow but readily measurable frequencies. Rel- results suggest that the growth arrest shown mal cells is their limited ability to ative to rodent fibroblasts, human fibro- by senescent fibroblasts is distinct from proliferate in culture. Cells capable blasts are exceedingly resistant to transfor- growth-arrest states described for early pasof proliferation in vivo often go through an mation by chemical carcinogens, radiation, sage cells, and c-fos repression may at least initial mitotic period in culture, but invari- and oncogenic viruses (3, 4). partially explain why senescent cells fail to ably there is a gradual decline in cell diviSenescence is not programmed cell death. proliferate. To gain insight into the nature of cellular sion; in cells from humans and certain other Senescent fibroblasts remain viable for long species, the decline is virtually irreversible periods (many months), during which they senescence, we examined the pattem of gene and complete (1). This progression has been synthesize RNA and protein (5, 7). Early expression in human fibroblasts at the extermed the finite life-span phenotype or passage fibroblasts enter a reversible, proli- tremes of their proliferative life-span. Hucellular senescence. feratively quiescent state (Go) when de- man fetal lung fibroblasts [strain WI-38 (1)] Cellular senescence has been studied most prived of growth factors; they resume prolif- were used at early passage, when the cells extensively in cultures of human fibroblasts. eration when appropriate growth factors are had undergone 70% (typiGenerally, human fibroblasts senesce after resupplied (8). However, senescent cells cally 75 to 85%) were capable of DNA 20 to 60 population doublings (PDs). Be- cannot be stimulated to enter the S phase of synthesis, and at late passage (senescence) the cell cycle by any combination of growth >48 PDs, when 70% at early passage and 30-fold 16 to mRNA rose 5- to 7-fold within 8 hours, in 24 hours after serum stimulation (Figs. 1 essential agreement with Rittling et al. (6). and 2A), coincident with S phase (22). Thus, senescent fibroblasts fully express at During S phase, all four replication-depenleast four mRNAs and retain the signaling dent histones (H3, H2A, H2B, and H4) pathways for their induction by serum. were synthesized at a high rate (Fig. 2B). In Nonetheless, the constitutively expressed senescent cells, histone 3 mRNA was nearly pHE-7 mRNA declined sharply when the undetectable, regardless of the culture conditions (Figs. 1 and 2A). Moreover, none of cells became senescent. Despite equal basal and induced levels of the replication-dependent histones were odc mRNA at early and late passage, odc synthesized at a high rate at any time tested, enzyme activity was at least twenty-fold although other nuclear proteins were synlower in senescent cells cultured in 10% thesized at comparable rates at early and late serum; after serum deprivation and stimula- passage (Fig. 2B). These results support the tion, the induced activity was two- to three- finding that DNA synthesis and histone fold lower (Table 1). Regardless of passage expression remain coupled as neonatal hunumber, odc activity could not be increased man fibroblasts senesce (23) and suggest by guanosine triphosphate (GTP) (Table 1), that, like pHE-7, the replication-dependent which in some cell lysates activates the en- histone genes are repressed in senescent zyme (20). Recently, a disparity between cells. In senescent cells only, the murine hisodc mRNA and enzyme activity was shown in senescent IMR-90 cells, another human tone-3.2 probe (which detects murine and fetal fibroblast strain (21). The results sug- human histone 3 variants) detected a slower gest that, despite normal mRNA abun- migrating RNA, faintly detectable in total dance, odc is down-regulated in senescent RNA. Its abundance did not change after cells by a translational or post-translational serum deprivation or stimulation (Fig. 2A), and it was 5- to 20-fold less prevalent than mechanism. In contrast to the near normal induction replication-dependent histone 3 mRNA in of c-H-ras, c-myc, and odc mRNA in senes- early passage cells. It appeared to be closely cent cells, there was no induction of the S related to the replication-dependent mRNA phase-specific histone 3 mRNA. At early because the hybridization signal persisted Senescent cells
c-ras-Ha
pHE-7
Actin
c-myc
Table 1. Activity of odc in early- and late-passage fibroblasts. Lysates were prepared from early- and late-passage cells and assayed for odc activity as described (21). When harvested, the cells were either subconfluent in 10% serum (Exponential), serum-deprived (Serum-deprived), or serumstimulated after serum deprivation (Stimulated), as described in the text. Where indicated, 0.1 mM GTP was added to the lysate before addition of the substrate. Odc activity is expressed as picomoles of ['4C]C02 released from ["4C]omithine per hour per milligram of protein. Three experiments with slightly different time points gave similar results. Shown are the data from one experiment; triplicate determinations varied by less than 10%. Growth condition
Odc activ-
+GTP
ity odc
Histone 3.2
Fig. 1. Basal and serum-stimulated mRNA expression in early- and late-passage fibroblasts. Exponentially growing (EXP), serum-deprived (Q) and serum-stimulated [30 min to 30 hours (h)] WI-38 human fetal lung fibroblasts at early and late (senescent) passage were prepared as described in the text.
Total cellular RNA was isolated from 5 x 106 to 1 x 107 cells on 150-mm dishes and analyzed by Northern blotting with DNA probes prepared by random priming or RNA probes prepared by in vitro transcription (17, 25, 35-37). Relative differences in mRNA abundance were quantitated by laser densitometry of autoradiograms of the Northern blots (17, 36). 206
Early passage 45 Exponential 1 Serum-deprived 2 Stimulated, 4 hours 141 Stimulated, 8 hours 131 Stimulated, 16 hours 7 Stimulated, 24 hours Late passage
Exponential Serum-drprived Stimulated, 4 hours
Stimulated, 8 hours Stimulated, 16 hours Stimulated, 24 hours
42
1 1 146 130 6
1
1
