Depletion of Deoxyribonucleotide Pools Is an Endogenous Source of DNA Damage in Cells Undergoing Oncogene-Induced Senescence

Abstract

In normal human cells, oncogene-induced senescence (OIS) depends on induction of DNA damage response. Oxidative stress and hyperreplication of genomic DNA have been proposed as major causes of DNA damage in OIS cells. Here, we report that down-regulation of deoxyribonucleoside pools is another endogenous source of DNA damage in normal human fibroblasts (NHFs) undergoing HRAS^G12V-induced senescence. NHF-HRAS^G12V cells underexpressed thymidylate synthase (TS) and ribonucleotide reductase (RR), two enzymes required for the entire de novo deoxyribonucleotide biosynthesis, and possessed low dNTP levels. Chromatin at the promoters of the genes encoding TS and RR was enriched with retinoblastoma tumor suppressor protein and histone H3 tri-methylated at lysine 9. Importantly, ectopic coexpression of TS and RR or addition of deoxyribonucleosides substantially suppressed DNA damage, senescence-associated phenotypes, and proliferation arrest in two types of NHF-expressing HRAS^G12V. Reciprocally, short hairpin RNA-mediated suppression of TS and RR caused DNA damage and senescence in NHFs, although less efficiently than HRAS^G12V. However, overexpression of TS and RR in quiescent NHFs did not overcome proliferation arrest, suggesting that unlike quiescence, OIS requires depletion of dNTP pools and activated DNA replication. Our data identify a previously unknown role of deoxyribonucleotides in regulation of OIS.

Oncogene-induced senescence (OIS) represents an important fail-safe mechanism that suppresses proliferation of premalignant cells.^1–3 Compelling evidence suggests that the response to DNA damage is one of the intrinsic processes required for the induction of OIS.^4–7 It was shown that aberrant activation of HRAS in human fibroblasts induces hyperreplication of genomic DNA, which leads to alterations in progression of DNA replication fork, generation of single- and double-strand DNA breaks (SSBs and DSBs, respectively), and activation of DNA damage response (DDR).⁶ SSBs induce DDR by engaging serine/threonine-protein kinase ATR (ataxia telangiectasia and Rad3-related protein) that transmits signaling to checkpoint kinase 1 (CHK1).⁸ CHK1 phosphorylates CDC25 protein, one of the key regulators of cell cycle progression, and targets it for degradation.⁸ DSBs initiate DDR that depends on another serine/threonine protein kinase, ataxia telangiectasia mutated (ATM).⁹ Activation of ATM results in phosphorylation of several targets, including the histone H2A variant H2AX,¹⁰ p53 tumor suppressor,^11,12 and CHK2 kinase.¹³ Both ATR and ATM signaling pathways are activated in normal human fibroblasts (NHFs) undergoing HRAS^G12V-induced senescence,^4–7 whereas ATM and CHK2 are required for this senescence because their individual short hairpin RNA (shRNA)-mediated inhibition enabled NHFs to overcome proliferation arrest and other senescence-associated phenotypes.^6,7

At the same time, studies conducted in yeasts and mammalian cells report that stalling of DNA replication fork and activation of ATR/CHK1 and ATM/CHK2 pathways can be induced by pharmacologic depletion of all or selected nucleotide pools.^14,15 In the present study, we investigated endogenous processes that caused DNA damage in human fibroblasts undergoing OIS and demonstrated that DNA damage at least partially originates from underexpression of key enzymes involved in de novo deoxyribonucleoside biosynthesis and subsequent depletion of endogenous dNTP pools. We propose that nucleotide deficiency caused by aberrant expression of activated HRAS contributes to OIS.

Materials and Methods

Cell Lines and Populations

HNFs WI-38 were purchased from ATCC (Manassas, VA). BJ-ET-RAS^G12V-ER^TAM fibroblasts were a gift from Dr. Andrei Gudkov (Roswell Park Cancer Institute, Buffalo, NY). Cells were cultured in Dulbecco’s modified Eagle’s essential minimal medium supplemented with 10% fetal calf serum, 2 mmol/L glutamine, and 100 U/mL penicillin G plus 100 μg/mL streptomycin.

Lentiviral Constructs and Infection

Lentiviral infection protocols and vectors containing cDNAs of HRAS^G12V were described previously.¹⁶ pLKO1 vector containing shRNA for ribonucleotide reductase (RR) 2 was purchased from Sigma-Aldrich (St. Louis, MO). cDNA for thymidylate synthase (TS), RR1, and RR2 were amplified by reverse transcription polymerase chain reaction from total RNA isolated from human melanoma cells and cloned in pLV-SV-puro expression vector (a gift from Dr. Peter Chumakov, Cleveland Clinic, Cleveland, OH).

Assays for Cell Proliferation and Senescence

For the proliferation assay, cells were plated in 96-well plates at ∼50% confluence 2 days before the assay. Cells were incubated with a nucleoside analog of thymidine, 5-ethynyl-2′-deoxyuridine (EdU), for 60 minutes, followed by fixation and staining for EdU-incorporated cells with the use of the ClickiT EdU Assay kit (Invitrogen, Carlsbad, CA).

For the senescence assay, cells were plated in 12-well plates, fixed, and incubated at 37°C with staining solution containing the X-Gal substrate (BioVision, Mountain View, CA). The development of blue color was detected visually with a microscope.

Immunoblot Analysis, Immunofluorescence, and Comet Assay

For Western blot analysis, antibodies were used for the following human proteins: ERK1 and phospho-ERK1/2, RR1, and RR2 from Santa Cruz Biotechnology (Santa Cruz, CA; sc-42, sc-93, sc-7383, sc-11733, and sc-10844, respectively), TS (ab7398-1) from Abcam (Cambridge, MA), thymidine kinase 1 (34003) from EMD Chemical Group (San Diego, CA), and ATM (2873) and phospho-ATM-Ser1981 (4526) from Cell Signaling Technology (Danvers, MA). Membranes were developed with alkaline phosphatase-conjugated secondary antibodies and the α-Innotech FluorChem HD2 imaging system (R&D Systems, Inc., Minneapolis, MN). For immunofluorescence-based analysis, formaldehyde-fixed cells were stained with H2AX-γ–specific antibodies (05-636) from Millipore (Billerica, MA) according to the manufacturer’s recommendations, followed by staining with DAPI. Alkaline comet assay was performed with Trevigen (Gaithersburg, MD) CometAssay HT kit (4252-040-K) according to the manufacturer’s recommendations.

Nucleotide Triphosphate Quantification

Nucleotide triphosphates were extracted and assayed as described previously.^17,18 All measurements were performed in quadruplicate, and data were recorded as SD of the mean. For convenience, data are presented in the respective figures in terms of arbitrary units. However, the actual pool sizes determined, in pmol/cell, were comparable with values previously reported for cultured mammalian cells.¹⁸

Chromatin Immunoprecipitation

Interactions between retinoblastoma (RB) tumor suppressor protein, tri-methylated lysine 9 of histone H3 (H3K9^3Me), and the promoters of TS, RR1, and RR2 were assessed with the EZ-Chip kit from Millipore according to the manufacturer’s recommendations with the following antibodies: IgG (sc-2027; Santa Cruz Biotechnology), RB (sc-73598; Santa Cruz Biotechnology), or H3K9^3Me (07-442; Millipore). The following primers were used for the PCR: TS, 5′-TTCCCGGGTTTCCTAAGACT-3′ and 5′-TGGATCTGCCCCAGGTACT-3′; RR1, 5′-GCTGACAGGGCGGAAG-3′ and 5′-GGAAGGGGATTTGGATTGTT-3′; RR2, 5′-CCTCACTCCAGCAGCCTTTA-3′ and 5′-CACCAACCTCGTTGGCTAAG-3′.

Fluorescence-Activated Cell Sorting Analysis

Cells were pelleted, resuspended in phosphate-buffered saline containing 0.5% fetal calf serum (PBS-S), and fixed in 70% ethanol. The cells were washed in PBS-S, resuspended in PBS-S containing propidium iodide (Sigma-Aldrich), incubated for 30 minutes at 37°C, and analyzed with a FACScan fluorescence-activated cell sorter (BD Biosciences, San Jose, CA).

Results

HRAS^G12V-Overexpressing Senescent WI-38 Cells Possess Low dNTP Pools and Underexpress Key Enzymes Required for Deoxyribonucleoside Biosynthesis

To determine whether NHFs undergoing OIS possessed low levels of endogenous nucleotide pools, we assessed amounts of intracellular ribonucleoside and deoxyribonucleoside triphosphates in WI-38 cells transduced with control or HRAS^G12V-expressing lentiviral vector 6 days after infection, that is, soon after the emergence of senescent phenotypes (Figure 1A). We found that, although overexpression of HRAS^G12V depleted both pools of nucleotide triphosphates, levels of rNTPs were affected less severely than dNTP levels (Figure 1B). Unlike de novo ribonucleotide biosynthesis pathways that include multiple enzymatic activities, the entire de novo biosynthesis of deoxyribonucleosides was largely controlled by only two enzymes. RR converted ADP, CDP, GDP, and UDP to dADP, dCDP, dGDP, and dUDP, respectively.¹⁹ TS catalyzed the de novo synthesis of dTMP from dUMP.²⁰ Thus, we measured expression levels of TS and both subunits of RR (RR1 and RR2) in control and HRAS^G12V-expressing WI-38 cells by Western blot analysis 6 days after infection. We found that like dNTP pools, the levels of these enzymes were significantly depleted in senescent fibroblasts (Figure 1C). In addition, the expression of a gene encoding thymidine kinase 1, a key enzyme required for the salvage deoxyribonucleoside synthesis,²¹ was also down-regulated in senescent cells.

Figure 1.

Senescent HRAS^G12V-WI38 fibroblasts possess low NTP pools and underexpress key enzymes for deoxyribonucleoside metabolism. A: Cells were infected with empty lentiviral vector (V) or lentiviral vector encoding HRAS^G12V (R). Six days after infection, cells were fixed and processed to visualize SA-β-Gal activity. Numbers indicate a percentage of SA-β-Gal–positive cells. The numbers were derived from counting cells in multiple view fields. B: Cells infected as in A were harvested 6 days after infection, and intracellular NTPs were extracted and quantified. Amounts of each NTP were normalized by the amounts of a corresponding NTP detected in vector cells. C: Cells infected as in A were harvested 6 days after infection and lysed, and total protein extracts were probed in Western blot analysis with antibodies shown on the right. D: Cells infected as in A were collected, fixed, stained with propidium iodide, and analyzed by flow cytometry.

Coexpression of TS, RR1, and RR2 or Addition of Deoxyribonucleosides Reduces DNA Damage and OIS in NHFs

The highest expression of TS and RR has been reported in the S phase of the cell cycle.^22,23 Therefore, depletion of TS, RR, and dNTP pools in HRAS^G12V-WI-38 cells may have been a consequence of G₀/G₁ or G₁/S cell cycle arrest induced by HRAS^G12V (Figure 1D). Alternatively, this depletion may have been a cause of proliferation arrest and senescence in cells overexpressing HRAS^G12V. To discern between these possibilities, we attempted to restore TS and RR levels in HRAS^G12V-WI-38 fibroblasts after assessment of senescence-associated phenotypes. To this end, we concurrently transduced WI-38 cells with cDNAs for TS, RR1, and RR2 (hereafter TRR) or in parallel with three empty vectors (hereafter 3V). Two days after infection, 3V and TRR cells were superinfected with viruses that contained empty vector or HRAS^G12V-expressing vector. Activity of HRAS^G12V in 3V and TRR cells was monitored by measuring the levels of phosphorylated ERK1/2 that appeared to be the same in both types of cells (Figure 2A). Expression of TS, RR1, and RR2 proteins was followed by Western blot analysis. As shown in Figure 2A, superinfection of HRAS^G12V in TRR cells did not significantly change total (exogenous and endogenous) amounts of these proteins. Consequently, dNTP levels in HRAS^G12V-TRR cells were 2- to 2.5-fold higher than dNTP levels in HRAS^G12V-3V cells (Figure 2B). Moreover, the number of cells exhibiting DNA damage detected by the comet assay or by staining with H2AX-γ–specific antibodies or the number of SA-β-Gal-positive cells was also two- to threefold lower in HRAS^G12V-TRR populations than in HRAS^G12V-3V populations (Figure 2C). Most importantly, inhibition of proliferation by HRAS^G12V in TRR cells was only partial compared with the complete and stable proliferation arrest of HRAS^G12V-3V cells (Figure 2, D and E, and Supplemental Figure S1A). Thus, restoration of TS and RS levels in cells overexpressing HRAS^G12V resulted in the substantial suppression of senescence phenotypes.

Figure 2.

Coexpression of TS, RR1, and RR2 reduces OIS phenotypes in HRAS^G12V-WI38 fibroblasts. A: Cells were infected with control vectors three times (3V) or with vectors expressing cDNA for TS, RR1, and RR2 cDNAs (TRR). Forty-eight hours later, cells were superinfected with the empty vector (V) or HRAS^G12V (RAS). Six days after second infection, cells were harvested and lysed, and total protein extracts were probed in Western blot analysis with antibodies indicated on the left. B: Cells were infected as in A. Six days after the second infection, intracellular dNTPs were extracted and quantified. Amounts of each dNTP from HRAS^G12V-infected cells were normalized by the amounts of corresponding dNTP detected in 3V-Vector cells or TRR-Vector cells. C: Cells were infected as in A. Six days after infection, cells were fixed and processed to visualize SA-β-Gal activity, fixed and stained with H2AX-γ–specific antibodies and DAPI, or subjected to the comet assay (the type of assay is shown on the left). Numbers indicate percentage of positive cells in each assay. The numbers were derived from counting cells in multiple view fields. D and E: Cells were infected as in A. Two days after second infection, cells were plated in 12-well plates in triplicate and counted every other day for 6 days. Numbers below the graph correspond to the days after infection. In parallel, at day 7 after infection, cells were assayed for proliferation rates (EdU incorporation). In each population, the number of positive cells was divided by the number of total cells.

To independently verify the importance of maintaining high dNTP pools for suppression of DNA damage and OIS, we attempted to restore intracellular dNTPs in WI-38-HRAS^G12V cells by supplementing their culture media with deoxyribonucleosides, starting the next day after infection with HRAS^G12V-encoding virus, that is, before the emergence of senescence-associated phenotypes. Culture media with or without deoxyribonucleosides was replaced with the fresh ones, respectively, with or without deoxyribonucleosides, every 2 days for the duration of the experiment. Incubation with deoxyribonucleosides did not affect the activity of HRAS^G12V, as was determined by the absence of changes in the amounts of phosphorylated ERK1/2 between treated and untreated populations of cells (Figure 3A). However, this treatment did increase intracellular amounts of dNTP pools 1.5- to 2.5-fold (Figure 3B) and decreased the number of H2AX-γ- and comet-positive cells and cells that exhibited SA-β-Gal activity in the treated populations (Figure 3C). Accordingly, HRAS^G12V was less effective in suppressing proliferation of WI-38 cells growing in the medium supplemented with deoxyribonucleosides compared with the untreated cells that underwent proliferation arrest (Figure 3, D and E, and Supplemental Figure S1B). Therefore, even partial restoration of suppressed dNTP pools via either overexpression of TS and RR or incubation with exogenous deoxyribonucleosides reduced DNA damage and the full-scale manifestation of OIS in WI-38 cells.

Figure 3.

Exogenous deoxyribonucleosides reduce OIS phenotypes in HRAS^G12V-WI38 fibroblasts. A: Cells were infected with empty lentiviral vector (V) or lentiviral vector encoding HRAS^G12V (R). Next day after infection, 100 μmol/L of each deoxyribonucleoside was either added (+dN) or not to the culture media, and media were replaced with the fresh one with or without deoxyribonucleosides, respectively, every 2 days. Six days after infection, cells were harvested and lysed, and total protein extracts were probed in Western blot analysis with antibodies indicated on the left. B: Cells were infected and treated as in A, followed by dNTP extraction and quantification 6 days after infection. All dNTP amounts were normalized by the amounts detected in “vector” cells. C: Cells were infected and treated as in A. Six days after infection, cells were fixed and processed to visualize SA-β-Gal activity or fixed and stained with H2AX-γ–specific antibodies and DAPI or subjected to the comet assay (the type of assay is shown on the left). Numbers indicate percentage of positive cells. The numbers were derived from counting cells in multiple view fields. D and E: Cells were infected and treated as in A. Two days after infection, cells were plated in 12-well plates in triplicate and counted every other day for 6 days. Numbers below the graph correspond to the days after infection. In parallel, at day 6 after infection, cells were assayed for proliferation rates (EdU incorporation). In each population, the number of positive cells was divided by the number of total cells.

To generalize our findings, we used a different type of NHF, primary BJ fibroblasts ectopically expressing the human telomerase reverse transcriptase gene and tamoxifen-inducible RAS^G12V-ER^TAM chimeric gene (BJ-ET-RAS^G12V-ER^TAM cells). On activation of HRAS^G12V via addition of 4-hydroxytestosterone (Figure 4A), proliferation of these cells was reduced gradually until reaching an arrest at approximately day 8 (Figure 4B). The arrest was accompanied by changes in cell structure, increased SA-β-Gal activity, and increased DNA damage that was visualized by antibodies to H2AX-γ (Figure 4C). At the same time, continuous presence of deoxyribonucleosides in growth media, although not affecting HRAS^G12V activation (Figure 4A), reduced inhibition of cell proliferation and other senescence-associated phenotypes (Figure 4, B and C) similarly to WI-38 cells.

Figure 4.

Exogenous deoxyribonucleosides reduce OIS phenotypes in BJ-ET-RAS^G12V-ER^TAM fibroblasts. A: Untreated cells (UN), cells treated with 0.1 μmol/L 4-hydroxytestosterone (4-OHT), or with 4-hydroxytestosterone and 100 μmol/L of each deoxyribonucleoside (4-OHT+dN). Seven days after adding the chemicals, cells were harvested and lysed, and total protein extracts were probed in Western blot analysis with antibodies indicated on the left. B: Cells were plated in 12-well plates in triplicate and treated as in A and counted every 2 days for 8 days. Numbers below the graph correspond to the days after infection. C: Cells were treated as in A for 7 days. Cells were fixed and processed to visualize SA-β-Gal activity or fixed and stained with H2AX-γ–specific antibodies and DAPI (the type of staining is shown on the left). Numbers indicate percentage of positive cells. The numbers were derived from counting cells in multiple view fields.

Depletion of TS and RR2 Causes Senescence in NHFs

To understand whether down-regulation of TS and RR was sufficient to cause DNA damage and emergence of senescent phenotypes in NHFs with normal, not hyperstimulated, DNA replication, we depleted amounts of TS and RR in WI-38 fibroblasts via coinfection with vectors that carried shRNAs for TS and RR subunit R2 and compared these cells with cells infected with control shRNA vector. Depletion of TS and RR2 (Figure 5A) led to stable proliferation arrest (Figure 5B and Supplemental Figure S1C), albeit with a longer latent period compared with HRAS^G12V-WI-38 cells (9 to 11 days versus 4 to 6 days). Arrested fibroblasts exhibited DNA damage as was visualized by staining with antibodies to H2AX-γ and elevated activity of SA-β-Gal (Figure 5C); however, the proportion of SA-β-Gal- and H2AX-γ-positive cells was also lower compared with WI-38 cells arrested by overexpression of HRAS^G12V (compare Figures 2, 3, and 5).

Figure 5.

Deficiency of TS and RR2 induces senescence in WI-38 fibroblasts. A: Cells were infected with control shRNA lentiviral vector (Cl) or with combination of vectors carrying shRNAs for TS or RR2 shRNAs (T+R2). Two days after infection, cells were harvested and lysed, and total protein extracts were probed in Western blot analysis with antibodies indicated on the left. B: Cells were infected as in A. Two days after infection, cells were plated in 12-well plates in duplicate and counted every other day for 9 days. Numbers below the graph correspond to the days after infection. C: Cells were infected as in A. Ten days after infection, cells were fixed and processed to visualize SA-β-Gal activity or fixed and stained with H2AX-γ–specific antibodies and DAPI (type of assay is shown on the left). Numbers indicate percentage of positive cells. The numbers were derived from counting cells in multiple view fields.

Overexpression or TS and RR or Addition of Exogenous Deoxyribonucleosides Does Not Suppress Quiescence in NHFs

We were interested in addressing the role of TS and RR overexpression in suppressing quiescence of NHFs, a form of temporary arrest of DNA replication that was not associated with DNA damage and could be achieved by serum deprivation.^4,24 Incubation of WI-38 cells in medium devoid of fetal bovine serum for 48 hours resulted in substantial down-regulation of TS, RR1, and RR2 (Figure 6A); however, no increase was detected in the number of H2AX-γ-positive cells (Figure 6B), suggesting that no DNA damage was present. Importantly, continuous supplementation with deoxyribonucleosides of fetal bovine serum-free culture media did not even partially affect proliferation arrests as was determined by EdU incorporation at day 5 after replacing the culture medium (Figure 6C). Accordingly, the same assay showed that proliferation of TRR-expressing WI-38 cells (described above) was suppressed by serum withdrawal as efficiently as the proliferation of control 3V cells (Figure 6C). Thus, we concluded that, unlike OIS, overexpression of TS and RR or supplementation with deoxyribonucleosides did not suppress quiescence in NHFs.

Figure 6.

Coexpression of TS, RR1, and RR2 or addition of exogenous deoxyribonucleosides does not affect quiescence of WI-38 fibroblasts. A: Cells were incubated in culture medium containing 10% or 0% fetal bovine serum (FBS) for 2 days. Cells were collected and lysed, and total protein extracts were probed in Western blot analysis with antibodies indicated on the left. B: Cells treated as in A were fixed and stained with antibodies for H2AX-γ. The numbers were derived from counting cells in multiple view fields. C: Uninfected WI-38 cells, 3V WI-38 cells, and TRR WI-38 cells were grown in the media containing 10% (black bars) or 0% (white bars) FBS supplemented or not with 100 μmol/L deoxyribonucleosides (+dN) for a period of 5 days. Cell proliferation was assessed by EdU incorporation assay. The numbers were derived from counting cells in multiple view fields.

Promoters of TS, RR1, and RR2 Interact with RB and Undergo Heterochromatization in Senescent HRAS^G12V-WI38 Cells

Transcription levels of RR1, RR2, and TS can be up-regulated by ectopic expression of E2F1.²⁵ Promoters of some E2F-responsive genes have been shown to interact with RB tumor suppressor proteins in NHFs undergoing HRAS^G12V-induced senescence.²⁶ Thus, to identify whether silencing of TRR genes in HRAS^G12V-WI38 senescent fibroblasts occurs via interactions with RB, we performed chromatin immunoprecipitation in these and vector cells with control antibodies, antibodies to RB, or H3K9^3Me, a hallmark of heterochromatin²⁷ and senescence-associated heterochromatin foci.²⁶ After precipitation, the chromatin was de-crosslinked, and the purified DNA was probed in semiquantitative PCR with primers that spanned E2F1-binding sites in the promoters of RR1, RR2, and TS genes. By this method, we demonstrated substantial enrichment of the promoter regions of all three genes in chromatin precipitated with RB- or H3K9^3Me-specific antibodies from HRAS^G12V-senescent but not vector cells (Figure 7), suggesting that RB binding and formation of heterochromatin contributed to silencing of TRR genes in senescent HRAS^G12V-WI38 cells.

Figure 7.

Promoter regions of TS, RR1, and RR2 are enriched with RB and H3K9^3Me. A: Cells were infected with empty lentiviral vector or lentiviral vector encoding HRAS^G12V. Six days after infection, cells were harvested and lysed, and total protein extracts were probed in Western blot analysis with antibodies shown on the left. B: Cells were infected as in A. Six days after infection, cells were cross-linked and lysed, and chromatin was immunoprecipitated with the antibodies shown on the top, followed by the reversal of the cross-linking and DNA isolation. DNA obtained from chromatin immunoprecipitated with the antibodies designated on the top, or 0.1% of the input DNA (Inp.), was used in PCR with the primers specific to the promoter regions of the genes shown on the left.

Discussion

DNA damage induced by ectopic expression of activated RAS or RAF oncoproteins in normal mammalian cells has been considered as a main cause of senescence, a fail-safe mechanism for suppressing tumor development at a premalignant stage.^1,2 In the present study, we tested the hypothesis that at least one of the main mediators of HRAS^G12V DNA damage- and senescence-inducing activity is down-regulation of dNTP pools. Decrease in the amounts of intracellular deoxyribonucleotides in OIS cells may cause DNA damage and affect DNA replication fork progression (which was described in fibroblasts undergoing HRAS^G12V-induced senescence⁶) via several interrelated mechanisms. The most obvious one is that dNTP deficiency slows processivity or even completely arrests DNA polymerase. However, this is an unlikely scenario because inhibition of DDR proteins ATM or CHK2 abolishes HRAS^G12V-induced senescence in NHFs, presumably via a dNTP-independent mechanism.^4–7 Most likely, stalling or collapse of the replication fork in a senescent cell occurs because of the DNA damage sites preexistent to the current DNA replication, and these preexistent lesions may originate from dNTP depletion. For example, it has been shown that inhibition of TS activity results in depletion of dTTP pools and in symmetrical increase in the amounts of deoxyuridine triphosphates, which are used instead of dTTP for DNA polymerization.²⁸ The incorporated uracil is removed via the uracil base excision repair system; however, if the number of incorporated uracil bases is high, the repair machinery is overwhelmed and generates abasic sites, as well as DSBs and/or SSBs,¹⁵ which would ultimately cause replication stress and would affect the replication fork progression during the following round of DNA replication. Similarly, inhibition of RR and subsequent depletion of dATP, dGTP, and dCTP pools leads to nucleotide misincorporation and activation of the mismatch repair system²⁹ that, if overwhelmed, would also generate DSBs and/or SSBs.³⁰ In addition, both base excision repair and mismatch repair systems use de novo DNA synthesis to fill the gaps generated in the course of DNA repair.³¹ Low levels of dNTPs would affect fidelity of this synthesis and subsequently DNA repair in general.

The proposed mechanisms of generation of DNA damage in OIS cells are in good agreement with previously published observations. For instance, as was mentioned above, a large number of HRAS^G12V-expressing cells were arrested with partly replicated DNA, suggesting stalling or a collapse of replication fork.⁶ DNA hyperreplication (multiple firing of the same replication origin) was proposed to be a primary cause for this phenotype.⁶ Stalling or collapse of the replication fork and subsequent induction of the S-phase checkpoint can be induced by pharmacologic inhibition of TS or RR via mechanisms described above in cells with normal, not hyperstimulated, DNA replication. Therefore, it is conceivable that, under the conditions of low dNTP pools, multiple origin firings and subsequently multiple replication forks would generate even higher DNA damage. Accordingly, fragile sites in OIS cells were more prone to undergoing mutations than any other regions of the genome.⁶ Similarly, pharmacologic inhibition of RR has been shown to disproportionate increase mutagenesis of genomic regions that contain fragile sites.^32,33

We demonstrated that even partial restoration of depleted intracellular dNTP pools is sufficient for substantial suppression of DNA damage and senescence. Despite significant ectopic overexpression of TS and RR or incubation with excessive amounts of deoxyribonucleosides, we were unable to completely restore dNTP pools in cells undergoing senescence. Complete reconstitution may require involvement of salvage pathway enzymes (especially for phosphorylation of exogenously added nucleosides) that, like thymidine kinase 1²¹ (Figure 1C), may be suppressed in senescent cells or enzymes involved in biosynthesis of ribonucleotides. Interestingly, the genomes of herpes viruses that replicate their DNA in quiescent cells with depleted dNTP pools often encode enzymes for both de novo and salvage pathways, including homologues to cellular RR1 and RR2, TS, thymidine kinase, dihydrofolate reductase, and dUTP pyrophosphatase.³⁴ However, we could not recapitulate expression of all of these genes in our experimental settings because overexpression of additional cDNAs in WI-38 cells affected their normal proliferation rates.

It is known that transformed cells use aerobic glycolysis as a major pathway for biosynthesis of adenosine triphosphate, a process called the Warburg effect.³⁵ Several groups reported that the enhanced rates of glycolysis occur largely because of the increased demand of a transformed cell for macromolecule components, including nucleotides that can be synthesized from glycolytic metabolites.^36,37 Bypassing senescence is considered an initial step in oncogenic transformation, and our data suggest that elevation of nucleotide levels is an important prerequisite for this step. Thus, switching to aerobic glycolysis to increase nucleotide pools may be required already at early stages of tumorigenesis.

Our results suggest that both nucleotide depletion and active DNA replication are required for efficient induction of DNA damage and OIS. Concurrent depletion of TS and RR2 in WI-38 fibroblasts with normal DNA replication induced senescence with longer latent period than overexpression of HRAS^G12V, and senescent populations contained lower fractions of H2AX-γ-positive and SA-β-Gal-positive cells than WI-38 cells overexpressing HRAS^G12V (with presumably hyperreplicated DNA) (Figures 2 and 5). Furthermore, serum withdrawal-arrested WI-38 cells, with no DNA replication, do not undergo DNA damage and senescence despite down-regulation of TS, RR1, and RR2 (Figure 5). Accordingly, ectopic overexpression of these enzymes in quiescent WI-38 cells or exogenous addition of deoxyribonucleosides into their media did not overcome this form of proliferation arrest.

We demonstrated that promoters of TS, RR1, and RR2 in HRAS^G12V-WI38 senescent cells interacted with RB tumor suppressor proteins and were enriched with heterochromatin marker H3K9^3Me (Figure 7). RB has been shown to play an essential role in HRAS^G12V-induced senescence by binding to the promoters of several E2F-responsive genes and by establishing repressive chromatin at the target loci.²⁶ Thus, our findings may offer a novel connection between RB activation and repression of nucleotide metabolism, ultimately resulting in the increased DNA damage and senescence. Taken together, our data provide a previously unidentified mechanism for the induction of DNA damage and OIS.

Supplemental Data

Supplemental Figure S1

Cells transduced with the indicated constructs were collected and fixed at day 7 after infection (A), at day 6 after infection (B), or at day 10 after infection (C). Fixed cells were stained with propidium iodide and analyzed by flow cytometry. “dN” indicates addition of deoxyribonucleoside mixture the next day after infection.

Supplemental Data

Supplemental material for this article can be found at http://dx.doi.org/10.1016/j.ajpath.2012.09.011.