Gene Expression Responses to DNA Damage Are Altered in Human Aging and in Werner Syndrome

Gene expression responses to DNA damage are altered in human aging and in Werner Syndrome

Kasper J Kyng1,2, Alfred May1, Tinna Stevnsner2, Kevin G Becker3, Steen Klvra˚ 4 and Vilhelm A Bohr*,1

1Laboratory of Molecular Gerontology, National Institute on Aging, National Institutes of Health, 5600 Nathan Shock Drive, Baltimore, MD 21224, USA; 2Danish Center for Molecular Gerontology, Department of Molecular Biology, University of Aarhus, DK-8000 Aarhus C, Denmark; 3Gene Expression and Genomics Unit, National Institute on Aging, National Institutes of Health, Baltimore, MD 21224, USA; 4Institute for Human Genetics, University of Aarhus, Denmark

The accumulation of DNA damage and mutations is syndromes, caused by heritable mutations inactivating considered a major cause of cancer and aging. While it is proteins that sense or repair DNA damage, which known that DNA damage can affect changes in gene accelerate some but not all signs of normal aging (Hasty expression, transcriptional regulation after DNA damage et al., 2003). Age is associated withan increase in is poorly understood. We characterized the expression of susceptibility to various forms of stress, and sporadic 6912 genes in human primary fibroblasts after exposure to reports suggest that an age-related decrease in DNA three different kinds of cellular stress that introduces repair may increase the susceptibility of cells to agents DNA damage: 4-nitroquinoline-1-oxide (4NQO), c-irra- causing DNA damage. Reduced base excision repair has diation, or UV-irradiation. Each type of stress elicited been demonstrated in nuclear extracts from aged human damage specific gene expression changes of up to 10-fold. skin (Xu et al., 2000), and inefficient repair withage A total of 85 genes had similar changes in expression of after nitroquinoline-N-oxide was found in mice (Zahn 3–40-fold after all three kinds of stress. We examined et al., 2000) and UV repair has been shown to decrease transcription in cells from young and old individuals and at high age in peripheral lymphocytes (Lambert et al., from patients with Werner syndrome (WS), a segmental 1979). DNA damage accumulation withage does not in progeroid condition with a high incidence of cancer, and itself suggest decreased repair capacity, and other found various age-associated transcriptional changes experiments showed no age-associated change in DNA depending upon the type of cellular stress. Compared to repair (Collier et al., 1982). Since repair is not perfect, young individuals, both WS and old individuals had mutations in DNA are expected to accumulate over time similarly aberrant transcriptional responses to c- and even withnormal repair capacity. One study demon- UV-irradiation, suggesting a role for Werner protein in strated an age-associated shift in the pattern of DNA stress-induced gene expression. Our results suggest that repair: in old age, strand break repair mechanisms aberrant DNA damage-induced gene regulation may became more important, while replication repair was contribute to the aging process and the premature aging reduced (Niedermuller et al., 1985). This was interpreted in WS. as a consequence of altered genetic expression during the Oncogene (2005) 24, 5026–5042. doi:10.1038/sj.onc.1208692; aging process. Recently, Seluanov et al. (2004) studied published online 16 May 2005 the ability of young, presenescent, and senescent normal human fibroblasts to repair double-strand breaks in Keywords: microarray; Werner syndrome; DNA transfected DNA by using a fluorescent reporter damage; aging substrate. They found that DNA end joining becomes less efficient and more error-prone during cellular senescence and proposed this as a likely mechanism underlying the age-related genomic instability and Introduction higher incidence of cancer in the elderly. It has been suggested that aging is characterized by a DNA damage induced by cellular metabolism and distinct reprogramming of gene expression (Murano environmental stress is thought to contribute to aging et al., 1991; Lee et al., 1999, 2000; Ly et al., 2000; Zou and the accompanying deterioration of organ function, et al., 2000; Kyng et al., 2001, 2003b; Weindruch et al., slowing of metabolism, and increased carcinogenesis. 2001; Welle et al., 2001), but little is known of A role of genome instability in the aging process is transcriptional regulation after DNA damage in aged supported by the existence of segmental progeroid individuals. Descriptions of coordination of gene expression in response to diverse pathogenic microbes *Correspondence: VA Bohr; E-mail: [email protected] have revealed valuable information on how immune Received 15 December 2004; revised 8 March2005; accepted 8 March cells respond to stress (Huang et al., 2001). In yeast, 2005; published online 16 May 2005 the stress response involves both damage-specific Transcription after DNA damage in aging and WS KJ Kyng et al 5027 transcriptional changes and a common set of genes tion by multiple repair systems, and 4NQO is unique referred to as the environmental stress response (ESR), because WS cells have a known sensitivity to this agent. which are regulated in response to several types of stress (reviewed in Gaschand Werner-Washburne,2002). Sucha coordination of gene expression in response to Results multiple DNA-damaging agents has not been reported in humans, and the pattern of stress-specific and Overview common pathways in human aging is not understood. It is possible that specific shortcomings in stress-induced To investigate age-associated changes in stress-induced gene expression are essential to the development of the gene regulation, we analysed data from approximately aging phenotype. This notion prompted us to explore 7000 human genes. Genes were characterized as stress age-related transcription changes after stress, thus responsive if they changed their expression by threefold mimicking the challenges faced during in vivo aging. or more in cells from young donors (n ¼ 6) in at least one Werner syndrome (WS) is a human segmental time point upon exposure to 4NQO, g-irradiation, or progeria syndrome withmany features of thenormal UV-irradiation. Of 2868 genes withgene ontology t aging process. WS cells display various forms of (GO) annotation, 449 (15.7%) were included in the genomic instability and while they are not hypersensitive further analysis (see Supplementary Table 1). Many to most DNA-damaging agents including UV irradia- more genes changed significantly but failed to make the tion (Poot et al., 2001), they are particularly sensitive to threefold cutoff, which is very restrictive relative to the carcinogen 4-nitroquinoline-1-oxide (4NQO). The comparable studies (Ly et al., 2000; Nantel et al., 2002; mechanisms underlying this particular hypersensitivity Lu et al., 2004). Thus, working from this set of genes are not understood, but are widely considered as an should be considered a conservative representation of important feature of WS at the cellular level. the actual changes. Figure 1 presents a graphical A recent review underlines the value of accumulated overview of stress-induced transcriptional reprogram- gene expression data in the understanding of complex ming in young-, old-, and WS donors. Analysis of the biological processes (Smithand Greenfield, 2003). This transcriptional response to individual DNA-damaging would be particularly relevant in the aging process, agents in young donors revealed that a common set of where several different mechanisms have been thought 85 genes was regulated after all of the stress types tested to be causative, and thus it would be valuable to use (Figure 2). For consistency withtheterminology global genome expression as a mirror reflecting the established in yeast, we refer to these genes as the diversity of cellular processes affected. human environmental stress response (H-ESR). In We have previously characterized gene expression addition, eachDNA-damaging agent was able to differences in untreated primary fibroblasts (Kyng et al., exclusively alter the expression of specific subsets of 2003b) and found that the steady-state transcription genes (Figure 2). Table 1 shows the number of stress patterns in old age and WS were very similar. In responsive genes in cell lines from aged and WS donors accordance with this finding it has been suggested that relative to young donors. We observed that the response several phenotyical aspects WS are secondary conse- in WS was particularly defective after 4NQO, and rather quences of aberrant gene expression (Nakura et al., proficient after UV. To further explore the age- 2000). Gene expression is changed to adapt to cellular associated decline in stress response capacity, we stress (Lee et al., 1995; Moradas-Ferreira and Costa, proceeded to investigate the degree to which different 2000; Volkert and Landini, 2001; Kyng et al., 2003a; pathways were affected. Moggs and Orphanides, 2003), and this adaptation is likely to be affected in aging because of constant Common stress response genes exposure to cellular stress. Here, we have explored how gene expression is Common stress response genes are listed in Table 2. In coordinated in response to multiple types of DNA young donors 59 of the common genes were induced and damage, and how specific stress response pathways are 29 repressed with an overlap of three genes that were aberrant in aging and WS. We were particularly induced or repressed depending on the stress type. The interested in resolving whether transcription patterns direction of change for each gene was generally the same reflect the relative sensitivities to DNA-damaging agents in young, old, and WS donors (Table 2). The common in WS and old donors. To answer these questions, we stress response (H-ESR) comprised 3% of 2868 GO- used cDNA microarrays to study expression changes of annotated genes analysed. In old and WS, a large part of 6912 RNA polymerase II transcribed genes, generating the H-ESR genes were only regulated after certain types approximately one million data points. A total of 15 of damage, or not at all. There was a tendency towards primary human fibroblast cell lines derived from normal stronger regulatory signals causing the highest fold young donors, normal old donors, and individual with changes (5.2–40.9-fold) to be conserved, whereas genes WS were exposed to 4NQO, g-irradiation, and UV- withlower fold changes(3.4–15.2-fold) were not irradiation, and analysed at 0, 1, 6, and 24 hafter consistently regulated in old and WS cells. Analysing treatment. From a DNA repair perspective, these DNA- exposure to eachdamage type separately, thecommon damaging agents were interesting because UV is stress response was on average 66% conserved in old age repaired by nucleotide excision repair (NER), g-irradia- and WS compared to young donors (Table 1). Examin-

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5028

Figure 1 Stress response in cells from young, old-, and WS donors. Time-course expression proﬁles, in hours, from 0 to 24 h after exposure to DNA-damaging stress for cells from young, old, and WS donors. The expression proﬁles of 449 genes with GO annotation whose expression levels changed X3-fold in at least one of the time-courses in the young donor cell line are represented. Each line represents a single gene. See Supplementary Table 1 for gene list

ing regulation of the common stress response genes by Functional categories in the common stress response eachDNA-damaging agent (Table 1), we analysed up- Categorization according to GO annotation (Figure 4) and downregulation separately and plotted the values in revealed that many common stress response genes Figure 3. Figure 3 is a central figure, demonstrating how represented general responses to cell and tissue injury. age-associated deficiencies vary withthestress type. Stress response-, cell growth-, and cell death-related Induction of the 85 common stress response genes was genes accounted for 41% of the genes in Table 2 and relatively conserved in bothold age and WS, whereas displayed the greatest fold changes. The detection of stress-induced repression was more variable. Supple- genes previously associated withstress activation con- mentary Table 3 shows the genes from Figure 3a with firmed that the microarray system was capable of defective regulation after stress in old or WS cells. reproducing existing findings. Half the genes in the To explore age-associated changes in the common stress response category were differently expressed in old stress response as a function of time, we examined the and WS donors compared to young. For example, time dependency of the common stress response superoxide dismutase (SOD1) was uniformly upregu- (Table 3). Of the 85 H-ESR genes, 51 were consistently lated in young and old, but not in WS where only 4NQO changed at 1 h after all damage types in young donors induced upregulation. In the cell death category, and were marked as IE (immediate-early) genes in BNIP3L, a BCL2 interacting protein involved in the Table 2. IE genes did not belong to one particular negative regulation of survival was uniformly upregu- functional category. For all three donor groups, similar lated in young, but not after UV in old or after numbers applied at 1, 6, and 24 h, and thus we detected g-irradiation in WS. Thus, individual stress response no significant time-related difference in the extent of pathways were affected differently with aging and age-associated changes. in WS.

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5029

Figure 2 Young donors: common stress response genes (H-ESR). Stress-regulated gene expression in cells from young healthy donors. In all, 449 GO-annotated genes showing more than threefold change were included. (a) Venn diagram of overlapping sets of 4NQO-, g-, and UV-regulated genes. (b–h) Graphs correspond to regions of the Venn diagram. X-axis is the time after damage from 0 to 24 h. Each line represents a single gene responding to the different stress agents. Y-axis is gene expression change from þ 50- to À50- fold (linear scale). (b) Common genes responding to all stress types. (c) Genes responding specifically to 4NQO only. (d) Genes responding specifically to g-irradiation only. (e) Genes responding specifically to UV only. (f) Genes responding to 4NQO and g-irradiation but not UV. (g) Genes responding to 4NQO and UV but not g.(h) Genes responding to g-irradiation and UV but not 4NQO. Lists of genes in panels b–hare in Supplementary Table 4

Table 1 Overview of stress-induced gene expression changes Damage 4NQO g-irradiation UV-irradiation type X3-fold, Common Stress X3-fold, Common Stress X3-fold, Common Stress any stress stress specific any stress stress specific any stress stress specific response response response

Young 100% (204) 100% (85) 100% (38) 100% (279) 100% (85) 100% (86) 100% (310) 100% (85) 100% (66)

Old 78% (160) 82% (70) 68% (26) 47% (131) 60% (51) 33% (28) 50% (154) 65% (55) 32% (21)

WS 26% (54) 46% (39) 16% (6) 29% (80) 56% (48) 8% (7) 66% (205) 86% (73) 47% (31)

Numbers in parantheses indicate absolute number of genes. Average of common stress response genes conserved in aging (av. 69%) and WS (av. 63%): 66%. Average of stress speciﬁc genes conserved in aging (44%) and WS (24%): 34%

Stress induction of cell cycle-related genes was negative regulator of the cell cycle and a negative RNA signiﬁcantly reduced in old and WS donors. Cyclin L2, pol II transcription regulator. Thus, no cell cycle-related displayed in Figure 5, was perhaps the single gene genes from the H-ESR group were consistently induced displaying the clearest age- and WS-associated changes, in old and WS (Table 2). and it seems that the pathway resulting in Cyclin L2 Remaining genes in the common stress response coded induction in young donors was defective in bothaging for proteins involved in nucleic acid binding, enzymes, and WS. Cells from old and WS donors also failed to signal transduction proteins, transport proteins, structur- induce the cell cycle-related kinase LCK and CTCF, a al proteins, chaperones and others, are given in Table 2.

Oncogene Oncogene 5030

Table 2 Common stress response genes (H-ESR genes) GenBank Da Da Old Da WS Oldb WSb Gene name Synonym Young 4NQO g UV 4NQO g UV

Cancer NM_001219 9.7 IE 6.8 IE 7.9 IE mmmmmmCalumenin CALU AA723035 4.7 — — m — mm— m Zinc-ﬁnger protein 36, C3H type-like 1 ZFP36L1

Cell cycle WS and aging in damage DNA after Transcription T95823 15.2 IE — — — — — m — m Cyclin L2 CCNL2 NM_006565 8.5 — — m — m ——m CCCTC-binding factor (zinc-ﬁnger protein) CTCF NM_005356 4.5 IE — — — — m — — — Lymphocyte-speciﬁc protein tyrosine kinase LCK

Cell death NM_001350 17.7 IE 24.4 IE 13.5 IE mmmmmmDeath-associated protein 6 DAXX AA156940 9.3 — — m — m ——m Programmed cell death5 PDCD5 AL132665 4.4 IE — — mm— m — m BCL2/adenovirus E1B 19 kDa interacting protein 3-like BNIP3L

Cell growth NM_000598 40.9 IE 46.1 IE 30.6 IE mmmmmmInsulin-like growthfactor-binding protein 3 IGFBP3 NM_002026 35.4 IE 34.7 IE 25.5 IE mmmmmmFibronectin 1 FN1 Kyng KJ NM_002026 26.4 IE 27.7 IE 22.1 IE mmmmmmFibronectin 1 FN1 NM_002011 21.2 IE 21.1 IE 11.6 mmmmmmFibroblast growthfactor receptor 4 FGFR4 NM_001901 7.5 IE 8.9 IE — mmm— mmConnective tissue growthfactor CTGF al et NM_002506 4.9 IE 5.5 — mmmm— m Nerve growthfactor, beta polypeptide NGFB X16323 3.8 — — — — — — m — Hepatocyte growthfactor (hepapoietin A; scatter factor) HGF

Chaperone NM_003380 19.6 IE 22.5 IE 15.2 IE mmmmmmVimentin VIM N93021 5.1 IE 5.4 IE — mmmm— m Tubulin-speciﬁc chaperone a TBCA

Enzyme AA443630 6.5 IE 7.8 IE 5.6 IE mmmmmmAldehyde dehydrogenase 3 family, member B2 ALDH3B2 R45108 6.3 IE 4.4 IE — mmm— mmThioredoxin reductase 2 TXNRD2 AA456051 5.7 IE 5.4 IE — mmm— mmUDP-glucose ceramide glucosyltransferase UGCG T98195 4.8 — — — — — — mmTAF10 RNA polymerase II, TATA box-binding protein TAF10 AA458661 4.4/ 5.3/ — mkm——m Hydroxyacyl-Coenzyme A dehydrogenase, type II HADH2 À4.4 À4.3 R00496 3.9 — — mm—— mmUDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acetylga- GALNT2 lactosaminyltransferase 2 R76394 3.8 — — m —— m — m Protease, serine, 23 SPUVE N32919 À3.4 — — — — — — — k Glucosamine-phosphate N-acetyltransferase 1 GNPNAT1 R28660 À4.0 IE — — k ————k Phosphodiesterase 3A, cGMP-inhibited PDE3A N32019 À4.8 IE — — k —— — kkCholine/ethanolaminephosphotransferase CEPT1 AA459381 À5.0 IE — — k — k — kkSphingosine-1-phosphate lyase 1 SGPL1 NM_003368 À5.9 IE À4.5 IE — kkk— kkUbiquitin-speciﬁc protease 1 USP1

GABA signalling N80593 À5.0 IE — — k — k ——k Gamma-aminobutyric acid (GABA) B receptor, 1 GABBR1

Glutamate signalling NM_002064 6.4 — — m —— m — m Glutaredoxin (thioltransferase) GLRX Table 2 (Continued ) GenBank Da Da Old Da WS Oldb WSb Gene name Synonym Young 4NQO g UV 4NQO g UV

G-protein signalling NM_005100 5.2 9.4 IE 6.1 IE mmmmmmA kinase (PRKA) anchor protein (gravin) 12 AKAP12

GTPase-mediated signal transduction NM_014578 14 IE — 7.6 IE m — mmmmRas homolog gene family, member D ARHD

Integrin receptor signalling W56754 9.9/ ——m — m ——m Integrin, beta 8 ITGB8 À3.0

Nucleic acid binding NM_001968 7.5 IE 10.1 IE 6 IE mmmmmmEucaryotic translation initiation factor 4E EIF4E W02101 6.5 5.3 — mmm— mmHeterogeneous nuclear ribonucleoprotein A2/B1 HNRPA2B1 H63976 5.8 IE — 5.1 IE — — — mmmEucaryotic translation initiation factor 3, subunit 6 interacting EIF3S6IP AA188179 5.1 5.2 — mmm— — — Actin-related protein 2/3 complex, subunit 1B, 41 kDa ARPC1B NM_005679 4.4 — — — — — — — — TATA box-binding protein (TBP)-associated factor, RNA TAF1C polymerase I, C AA455111 À6.3 IE À6.3 — kkk— kkHeterogeneous nuclear ribonucleoprotein C (C1/C2) HNRPC

Pathogenesis

NM_002089 11.8 IE 16.5 IE 10.4 IE mmmmmmChemokine (C-X-C motif) ligand 2 CXCL2 WS and aging Kyng in KJ damage DNA after Transcription

Proliferation NM_003177 À11.1 — — — kk ——k Spleen tyrosine kinase SYK al et

Signal transducer (other) R52654 6.6 IE 4.8 3.7 mmmmmmCytochrome c, somatic CYCS AA428778 4.4 IE 3.7 — mmmm— m Ephrin-B1 EFNB1

Stress response BM903615c 8.9 IE 5.3 9.9 mmmmmmOrnithine decarboxylase antizyme 1 OAZ1 NM_000700c 8.4 IE 8.2 IE 6.9 IE mmmmmmAnnexin A1 ANXA1 NM_000972c 7.6 IE 6.7 IE 4.2 mmmmmmRibosomal protein L7a RPL7A N80129c 6.1 9.4 IE 7.6 mmmmmmMetallothionein 1X MT1X R67915 5.8 6.7 IE 4.6 mmmmmmChromosome 13 open reading frame 12 HSPC014 NM_006096c 5.3 IE — — — — m — — — N-myc downstream-regulated gene 1 NDRG1 NM_002122c 4.9 IE — — m — — — — — Major histocompatibility complex, class II, DQ alpha 1 HLA-DQA1 NM_000454 4.3 4.5 — mmmm— — Superoxide dismutase 1 SOD1 NM_004040 4.1 IE — — m —— — m — Ras homolog gene family, member B ARHB NM_006406c 4.0 4.6 — mmm— mmPeroxiredoxin 4 PRDX4

Structural protein NM_000090 8.9 IE 11.4 IE 11.2 IE mmmmmmCollagen, type III, alpha 1 COL3A1 NM_000090 5.2 IE 8.5 IE 6.0 IE mmmmmmCollagen, type III, alpha 1 COL3A1 NM_006314 À4.3 — — k —— — kk CNK1 Oncogene 5031 Transcription after DNA damage in aging and WS KJ Kyng et al 5032 Gene name Synonym Connector enhancer of KSR-likesor (Drosophila of kinase ras) suppres- Translocated promoter region (toATP activated synthase, MET H+ oncogene) transporting, mitochondrial F0 complex TPR ATP5I Adaptor-related protein complex 3, beta 2 subunitSolute carrier family 18 (vesicular monoamine), member 2 SLC18A2 AP3B2 Catenin (cadherin-associated protein), delta 1 CTNND1 ) m k UV Classiﬁcation based on other sources. IE indicates immediate-early genes whose expression c b g Highest fold change after 4NQO, gamma-irradiation or UV-irradiation at 1 h, 6 h, or 24 h after a Continued WS

Figure 3 Age-associated aberrations in common and stress- mmm mmm kkk —— —— specific genetic reprogramming in cells from old and WS donors. 3-fold change. Table 2 ( Black bars indicate genes regulated in young donors and are X defined as 100%. Dark-gray bars indicate relative number of genes k — — regulated in old donors. Light-gray bars indicate relative number of UV 4NQO genes regulated in WS donors. See Supplementary Table 3 for gene lists of damage-specific deficiencies in H-ESR gene regulation in old b g k —— — — — — — Solute carrier family 30 (zinc transporter), member 5 SLC30A5 — and WS corresponding to panel a. Selected genes from panel b are listed in Table 4 Arrows indicate b mm mmm m k k 4NQO Table 3 Time dependency of common stress response Any time 1 h 6 h 24 h WS Old 4.2 — a À

D Young 85 51 59 54 Old 40 24 26 21 WS 29 18 23 18 Old a Number of genes consistently changed X3-fold after all three damage D types at the indicated time a 5.3 — 6.3 — — D 4.2 IE — — À À Young

À Stress-specific responses In aging and WS, an average of 34% of stress-specific responses were conserved, vs an average of 66% of the common response genes (Table 1). The age- or 3 fold at 1 hafter damage. See Supplementary material for complete gene list WS-related deficiency was dependent on the damage X type. Figure 2 shows how stress-specific genes were identified and select genes are listed in Table 4. The NM_003292AA431433NM_004644 8.5 IE 5.6 IE 4.5 IE — 4.5 IE — 4.6 IE — 4.2 AA134753 N74229 AA025276 GenBank changed damage, shown only when? 3 fold change after all damage types. Transport Other A representative selection of the 85 genes in the H-ESR categorized by GO annotation is shown. complete lists of genes corresponding to eacharea of the

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5033 old donors was in the 4NQO response, whereas WS cells reacted relatively more like normal aging after g- irradiation and UV. We found several known stress response genes that responded only to 4NQO, including MT1H, oxidative stress responsive gene ATOX1, protein kinase H11, which has heat–shock protein activity, and PDCD2 involved in apoptosis. None of these were regulated in WS. Most genes responding specifically to 4NQO only (Table 4) were IE genes like the growth-related PDGRFRL, IFNGR2, and NET1. NET1 contains a DH domain, which suggests that it interacts with members of the RAS superfamily of GTP-binding proteins. RAS-related genes RAB11B and RIS1 were Figure 4 Functional classification of common stress response also induced. Figure 2g reveals the interesting result that genes in Table 2 according to GO annotation. Categories with p2 pathways overlapping between UV and 4NQO are genes were omitted induced 1 hafter 4NQO but 24 hafter UV. Figure 6 exemplifies gene expression profiles regulated via pathways that appear to be altered in WS after 4NQO. Figure 6a shows BTG3 that is upregulated 25– 50-fold after 4NQO and UV, except in WS donors, where the pathway activating BTG3 after 4NQO is shut down. A similar expression pattern is seen in Figure 6b for MYDD88, INPP1, and PPP2R2B, three genes regulated so closely that they are likely to belong to a common pathway, which in WS cells is activated after UV but not after 4NQO. Figure 6c shows a group of closely regulated genes, where once again the pathway governing their repression after 4NQO in WS cells is defective. Our results suggest that WRN is needed for the 4NQO-induced regulation of these genes.

g-irradiation Conservation of g-induced common stress response gene regulation was almost similar in old and WS (Figure 3a). In young donors, 86 g-irradiation- specific responding genes included the DNA repair protein ATRX, the MAPK activator CHRNA7, heat– shock transcription factor 2 (HSF2), and other transcription regulators (ARNT, HCLS1, STAT5B, HOXD9, MADH4), see Table 4. RAB1B, RAB2, RAB6A, and RIT2 were only changed in young donors suggesting that RAS-related gene expression pathways are disrupted in old age and in WS after g-irradiation, Figure 5 Cyclin L2 expression profile. Time-course expression but not after 4NQO. Preservation of g-specific gene profile, in hours, from 0 to 24 h after exposure to DNA-damaging regulation in WS was small compared to old stress for cells from young, old, and WS donors. Y-axis is gene (Figure 3b). expression change UV exposure Characteristic for the UV response was a large number of intermediate-early genes down-regu- Venn diagram in Figure 2 can be found in Supplemen- lated in all three donor groups (Figure 2e and Table 4). tary Table 4. In young donors, 66 genes responded to UV only. Figure 2g and hreveals a delay in theUV-induced regulation of genes that are also induced by 4NQO or 4NQO Examining common and stress-specific genes g-irradiation. PDCD10 (programmed cell death10) separately, 63% of H-ESR genes induced by 4NQO in responded to UV only, where PDCD2 responded to young donors were induced in WS cells, while only 8% 4NQO only, indicating that separate but similar of repression was conserved (Figure 3a). For 4NQO- mechanisms were involved in the damage-specific specific regulation, 23% of the induction and 0% of responses. Known stress response genes reacting to repression was maintained in WS cells (Figure 3b). As UV only were CD69 and TNFRSF10B/KILLER. Three seen in Table 1, the primary difference between WS and human ERCC genes, which are known to be involved in

Oncogene Oncogene 5034

Table 4 Stress-speciﬁc genes GenBank Da Young Da Old Da WS Gene name Synonym

4NQO speciﬁc genes NM_006207 5.92 IE 5.76 IE — Platelet-derived growthfactor receptor-like PDGFRL NM_002213 5.91 5.78 — Integrin, beta 5 ITGB5 AA127069 4.91 4.88 4.24 Ras-induced senescence 1 RIS1 NM_004218 4.28 IE 8.13 IE 5.54 IE RAB11B, member RAS oncogene family RAB11B NM_005534 3.79 IE — — Interferon gamma receptor 2 (interferon gamma transducer 1) IFNGR2 WS and aging in damage DNA after Transcription NM_001998 3.64 IE 3.98 IE 3.19 IE Fibulin 2 FBLN2 AA418694 3.47 3.12 — ATX1 antioxidant protein 1 homolog (yeast) ATOX1 NM_005863 3.31 IE 6.33 IE 7.55 IE Neuroepithelial cell-transforming gene 1 NET1 AA456028 3.22 IE 3.12 — Rab geranylgeranyltransferase, beta subunit RABGGTB H69049 3.19 3.06 IE 3.28 IE CD36 antigen (collagen type I receptor, thrombospondin receptor) CD36 NM_000786 3.15 — — Cytochrome P450, family 51, subfamily A, polypeptide 1 CYP51 NM_005719 3.05 3.30 IE — Actin-related protein 2/3 complex, subunit 3, 21 kDa ARPC3 NM_005951 3.01 — — Metallothionein 1H MT1H NM_014365 3.01 — — Protein kinase H11 H11 AA418689 3.00 4.49 — Polymerase (RNA) II (DNA directed) polypeptide F POLR2F NM_004652 À3.13 — — Ubiquitin-speciﬁc protease 9, X chromosome USP9X N54788 À3.23 IE — — Cytochrome b reducatse 1 CYBRD1 Kyng KJ R91950 À3.33 IE À3.33 — Cytochrome b-5 CYB5 AK055180 À3.33 IE À3.13 — Programmed cell death2 PDCD2 NM_004996 À3.45 IE À3.57 IE — ATP-binding cassette, subfamily C (CFTR/MRP), member 1 ABCC1 al et AL833554 À3.45 IE — — Interleukin 16 (lymphocyte chemoattractant factor) IL16 R74171 À3.45 IE À3.33 IE — Rap2-interacting protein x RIPX NM_006856 À3.57 IE À3.45 IE — Activating transcription factor 7 ATF7 NM_001932 À3.70 IE À3.85 IE — Membrane protein, palmitoylated 3 (MAGUK p55 subfamily member 3) MPP3

g-speciﬁc genes H73714 3.39 IE — — Replication factor C (activator 1) 1 (145 kDa) RFC1 NM_002865 3.38 IE — — RAB2, member RAS oncogene family RAB2 AA453467 3.37 — — Lactate dehydrogenase C LDHC W45572 3.20 IE — — ADP-ribosylation factor 1 ARF1 NM_138271 3.08 IE 3.28 IE — Alpha thalassemia/mental retardation syndrome X-linked (RAD54 homolog, Sachar- ATRX romyces cerevisiae) NM_002053 3.08 IE 3.10 3.08 IE Guanylate-binding protein 1, interferon-inducible, 67 kDa GBP1 NM_002715 3.03 4.34 IE — Protein phosphatase 2 (formerly 2A), catalytic subunit, alpha isoform PPP2CA NM_000245 À3.03 — — Met proto-oncogene (hepatocyte growth factor receptor) MET NM_000746 À3.03 IE — — Cholinergic receptor, nicotinic, alpha polypeptide 7 CHRNA7 AA460302 À3.03 IE — — Acetyl-Coenzyme A acyltransferase 2 (mitochondrial 3-oxoacyl-Coenzyme A thiolase) MYO5B X68560 À3.03 — — Sp3 transcription factor SP3 T95553 À3.13 — — Helicase withzinc-ﬁnger domain HELZ NM_030981 À3.13 IE — — RAB1B, member RAS oncogene family RAB1B N72976 À3.13 IE — — General transcription factor IIIC, polypeptide 4, 90 kDa GTF3C5 NM_004506 À3.33 IE — — Heat–shock transcription factor 2 HSF2 R31512 À3.33 — — Activating transcription factor 7-interacting protein ATF7IP NM_005424 À3.33 IE — — Tyrosine kinase withimmunoglobulin and epidermal growthfactor homology domains TIE BC024739 À3.45 À4.55 IE À3.03 DEAD (Asp-Glu-Ala-Asp) box polypeptide 18 DDX18 NM_002869 À3.45 IE — — RAB6A, member RAS oncogene family RAB6A NM_002930 À3.45 — — Ras-like without CAAX 2 RIT2 NM_001827 À3.45 IE — — CDC28 protein kinase regulatory subunit 2 CKS2 Table 4 (Continued ) GenBank Da Young Da Old Da WS Gene name Synonym

NM_000284 À3.70 À3.13 IE À3.23 Pyruvate dehydrogenase (lipoamide) alpha 1 PDHA1 NM_145803 À3.70 IE À3.57 IE — TNF receptor-associated factor 6 TRAF6 NM_012448 À3.70 IE À5.00 IE À3.45 IE Signal transducer and activator of transcription 5B STAT5B NM_000236 À3.85 À3.33 IE — Lipase, hepatic LIPC R68464 À4.17 IE À3.23 IE À3.23 IE Junctional adhesion molecule 2 JAM2 NM_000397 À4.17 À6.67 IE À3.23 IE Cytochrome b-245, beta polypeptide (chronic granulomatous disease) CYBB H50677 À4.35 IE À3.33 IE — RNA-binding motif protein 6 RBM6

UV-speciﬁc genes N57750 4.25 IE 6.54 IE 5.17 IE Protein kinase, cAMP-dependent, catalytic, alpha PRKACA NM_001781 3.90 — — CD69 antigen (p60, early T-cell activation antigen) CD69 AA485272 3.77 — — Hexokinase 1 HK1 NM_005194 3.65 IE — — CCAAT/enhancer-binding protein (C/EBP), beta CEBPB NM_000395 3.51 — — Colony-stimulating factor 2 receptor, beta, low-afﬁnity (granulocyte-macrophage) CSF2RB NM_001885 3.46 IE — 3.18 IE Crystallin, alpha B CRYAB W68559 3.35 IE — 3.40 IE Angiotensin I-converting enzyme (peptidyl-dipeptidase A) 1 ACE NM_001299 3.33 — — Calponin 1, basic, smoothmuscle CNN1 NM_003842 3.28 — — Tumor necrosis factor receptor superfamily, member 10b TNFRSF10B NM_005399 3.26 IE — — Protein kinase, AMP-activated, beta 2 noncatalytic subunit PRKAB2 AA169176 3.13 IE — 3.35 IE Glycerol-3-phosphate dehydrogenase 2 (mitochondrial) GPD3 N62248 À3.03 IE — À3.45 IE Eukaryotic translation initiation factor 4A, isoform 2 EIF4A3 W49583 À3.13 À4.55 IE — Growthdifferentiation factor 11 GDF11 NM_000846 À3.13 IE — — Glutathione S-transferase A2 GSTA2 H56918 À3.13 IE — — Eucaryotic translation initiation factor 4A, isoform 1 EIF4A2

NM_006286 À3.13 — — Transcription factor Dp-2 (E2F dimerization partner 2) TFDP2 WS and aging Kyng in KJ damage DNA after Transcription AA236141 À3.33 — — ATPase, Cu2+ transporting, alpha polypeptide (Menkes syndrome) ATP7A N51858 À3.45 — — Ubiquitin ligase E3 alpha-II KIAA0349 AA057313 À3.57 — — MYST histone acetyltransferase (monocytic leukemia) 4 MYST4 al et NM_007217 À3.57 — À4.00 IE Programmed cell death10 PDCD10 NM_003474 À3.57 IE — À3.45 IE A disintegrin and metalloproteinase domain 12 (meltrin alpha) ADAM12 AA398492 À4.17 À3.23 À3.23 A disintegrin-like and metalloprotease (reprolysin type) withthrombospondin type 1 ADAMTS3 motif, 3 NM_006166 À6.25 IE À7.69 IE À10.00 IE Nuclear transcription factor Y, beta NFYB NM_139276 À8.33 IE À8.33 IE À7.14 IE Signal transducer and activator of transcription 3 (acute-phase response factor) STAT3 T58773 À10.00 IE À5.88 IE À10.00 IE Inositol polyphosphate-5-phosphatase, 40 kDa INPP5A NM_002312 À12.50 IE À7.69 IE À16.67 IE Ligase IV, DNA, ATP-dependent LIG4 NM_004637 À16.67 IE À20.00 IE À25.00 IE RAB7, member RAS oncogene family RAB7 N70701 À20.00 IE À12.50 IE À25.00 IE Aldehyde dehydrogenase 8 family, member A1 ALDH8A1 AA481547 À25.00 IE À25.00 IE À50.00 IE Protein tyrosine phosphatase, receptor type, C-associated protein PTPRCAP

A representative selection of stress-specifc genes is shown. aMaximum fold change, shown only when X3-fold change. IE indicates immediate-early genes whose expression changed X3-fold at 1 h after damage. See Supplementary material for complete gene list Oncogene 5035 Transcription after DNA damage in aging and WS KJ Kyng et al 5036 UV repair, were present on our array. Consistent with the notion that repair protein-encoding genes are not extensively regulated at the transcription step, we found that ERCC3/XPB was 2.1-fold upregulated after UV, while to our surprise ERCC4/XPF and ERCC5/XPG were bothdownregulated 2–3-fold (see Supplementary material). Gene expression pathways induced by UV were better conserved in WS cells than in cells from old donors (Figure 3). Notably, the downregulation of common stress response genes after UV was 89% conserved in WS vs 44% in old donors.

Discussion

The results presented here establish that there is an age- associated alteration of stress-induced gene regulation. We speculate that aberrant gene expression may under- lie the increased susceptibility to various forms of stress seen with aging, and thus contribute to the aging process itself. We have interpreted the data carefully, keeping in mind the limitations of studies on cultured cells. Since WS cells may generate higher levels of reactive oxygen species (ROS), it is possible that at the same level of oxygen they have more responses to oxidative damage as compared to normal cells. However, the resemblance between old and WS suggests that this phenomenon also occurs in old cells. There have been reports that high oxygen levels in culture may contribute to the accumulation of mutations in cultured cells (Busuttil et al., 2003; Parrinello et al., 2003). This could mean that unchallenged cells in culture are in a perpetual state of stress, masking some of the gene expression changes elicited by induced DNA damage. Thus, it is possible that the common stress response could involve more genes than the 85 identified here. Gene expression patterns in senescent WS cells overlap with those in normal strains (Choi et al., 2001), and senescence appears to be a p53-dependent event mediated via pathways identical to those in normal cells (Davis et al., 2003). We saw no significant induction of p53 in neither old nor WS cells (see Supplementary Table 6), indicating that the gene profiles reflected early passage rather than cellular senescence. Also, gene profiles of replicative senescence (early passage o25 PDL, late passage >60 PDL) has previously been shown by Park et al. (2001) to be different from progeria or elderly donor, suggesting that the gene profiles presented here are specific to normal aging and WS. In addition, we routinely examined the cell cultures and did not see a preponderance of spindle- shaped cells. We have also tried b-gal staining, and after

Figure 6 Selected profiles of genes whose expression was deficient after 4NQO in WS cells. Time-course expression profiles, in hours, from 0 to 24 hafter exposure to DNA-damaging stress for cells from young, old, and WS donors. Y-axis is gene expression change. Gene names are printed on graph

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5037 Table 5 Cell lines used in this study Coriell repository Genotype Donor phenotype PDL Age number (years)

AG11747 Normal young Not clinically affected 14–15 22 AG10803 Normal young Not clinically affected 10 22 GM03440 Normal young Not clinically affected ? 20 GM02937 Normal young Not clinically affected ? 22 GM01891 Normal young Not clinically affected ? 24 AG09975 Normal young Not clinically affected 15–16 25 AG10884 Normal old Not clinically affected 11–12 87 AG13208 Normal old Not clinically affected 12–13 89 AG13129 Normal old Not clinically affected 11–12 89 AG07725 Normal old Not clinically affected 15 91 AG08433 Normal old Not clinically affected 21 94 AG12795 WS (mutation Short stature, bird-like appearance, gray hair, juvenile bilateral 14–18 19 not identified) cataracts, atrophic skin, and hypogonadism AG12797 WS (mutation Short stature, bird-like appearance, gray hair, skin hyperpigmentation, 10–11 36 not identified) juvenile bilateral cataracts, atrophic skin, diabetes, and hypogonadism AG06300 WS (WRN Gray hair, muscle wasting, wrinkling of skin, dystrophic nails, high- 32–33 37 protein F1074L pitched voice, hypogonadism, and a general aged appearance mutation) AG12799 WS (mutation Short stature, gray hair, hyperpigmentation of skin, juvenile bilateral ?25 not identified) cataracts, atrophic skin, and hypogonadism

exposure to oxidative stress, we found less staining in have long been known to be activated following DNA WS cells (von et al., 2004) than in normal. damage inducing stress (Friedberg et al., 1995). Upre- Overall, the common stress response was better gulation of stress response genes has also been conserved than specific stress responses (Figure 3), and associated with aging in unchallenged cells, which could there was a tendency for strong regulatory signals (high mean that stress gene pathways are primed and there- fold change) to be better conserved in aging and WS, see fore easily induced. In unchallenged cells from old- and Tables 4 and 5. Thus, it appears that the weaker and WS donors, we have previously found various stress- specific stress response pathways are most severely related genes to be bothup- or downregulated (Kyng affected in aging. The common stress response genes et al., 2003b), whereas other groups have found were consistently regulated in response to eachdamage predominant upregulation of stress genes in aging type, suggesting that they are components of the same (Weindruch et al., 2001; Lu et al., 2004). The cellular response. The existence of the common response characteristic age-influenced expression profile of Cyclin suggests a convergence of pathways from upstream L2 in Figure 5 deserves mention since overexpression of cellular events activated by eachDNA damage type. cyclin L2 can induce apoptosis and suppress tumor The generality of the common stress response could growth(Yang et al., 2004). These functions may be reflect that it is controlled by one common regulatory mediated via Cyclin L2’s interaction withRNA poly- system; however, the different expression patterns with- merase II and its role as a transcription regulator. Our in the common stress response (Figures 2b, 5 and 6) results suggest a new role for Cyclin L2 where universal suggest that different regulatory mechanisms are at play. induction after stress could result in apoptosis via a Moreover, damage type-specific expression profiles pathway that is defective in old and WS donors. indicate that different signalling pathways are involved Reduced expression of cell cycle-related genes in aging depending on the damage type, and corresponding (Ly et al., 2000) has previously been demonstrated, and findings in yeast support this conclusion (Gasch and our findings extend that result by showing that stress- Werner-Washburne, 2002). It thus appears that a variety induced regulation of cell cycle-related genes is deficient of stress-specific upstream signals can activate the in old and WS donors. Genes encoding proteins common stress response program as part of the involved in nucleic acid binding also showed age- condition-specific response. This also explains how associated changes in expression. We have previously select parts of the common stress response can be found that WS and aging are associated with altered defective in aging and WS. expression of genes in DNA and RNA metabolism Major functional categories did not include DNA (Kyng et al., 2003b). We have also demonstrated that repair genes, which is in accordance with the existing DNA and RNA metabolism-related genes are over- consensus that DNA damage-activated genes include expressed after stress, and that this overexpression is only a few bona fide DNA repair enzymes (Friedberg deficient in cells from patients withthepremature aging et al., 1995; Jelinsky et al., 2000), as DNA repair in syndrome Cockayne Syndrome (Kyng et al., 2003a). mammalian cells is generally based on constitutive levels Thus, there is an accumulation of evidence connecting of repair enzymes. Growthregulation, however, is defects in expression of genes in DNA and RNA associated withtheaging process and growthfactors metabolism withaging.

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5038 Age-associated changes varied with the stress type in a several pathways. Upon closer experimental investiga- reproducible manner, suggesting that altered gene tion, the aberrant regulation in WS of these genes could expression involves specific pathways. The same path- provide clues to the pathology of WS and aging. ways were to a large extent altered in WS patients, which Regulation of genes in Figure 6a–c was deficient in places WRN in a position central to the stress response WS cells after 4NQO but not after UV. Our results deficits seen in aging. WRN is involved in DNA repair suggest that WRN is needed for the 4NQO-induced and transcription (Opresko et al., 2003), WS cells regulation of these genes. There can be several accumulate DNA damage (von et al., 2004) and it is consequences of the altered gene expression in WS. plausible that lack of WRN function could affect BTG3 overexpression (Figure 6a) is involved in the specific stress-induced gene regulation pathways. WS negative control of cell cycle progression from the G0/ cells were particularly nonresponsive to 4NQO in a way G1 to S phase (Guehenneux et al., 1997). Thus, WS cells not observed in aged donors, while other pathways, like may progress through the cell cycle in spite of persisting those activated after UV-irradiation, were somewhat DNA damage. This is supported by our recent work more changed in aging than WS. This correlates with the (von et al., 2004). BAP1 repression (Figure 6c) was hypersensitivity of WS cells to 4NQO. It also supports deficient in WS after 4NQO but not after UV. BAP1 is the idea that malfunction of key pathways can have thought to be a tumor suppressor gene that functions in detrimental effects on gene expression regulation, and the BRCA1 growth control pathway, indicating that WS offers an explanation for some of the differences seen cells are deficient in the BRCA1 pathway after 4NQO. between the premature aging of WS and normal aging. This is particularly interesting since we have preliminary Thus, the stress response deficits seen in aging may be data suggesting that WRN interacts physically with explained by the malfunction of WRN and/or other key BRCA1 (Cheng et al., our unpublished data). DUSP5 proteins not yet recognized acting in the same pathways (Figure 6c) is a tyrosine phosphatase involved in the as WRN. negative regulation of the MAPK superfamily, particu- WS cells have a selective survival sensitivity to 4NQO larly MAPK3/ERK1. MAPK3 is involved in boththe which is intriguing, as they are not sensitive to UV or initiation and regulation of mitosis and postmitotic most other DNA-damaging agents (Ogburn et al., 1997; functions in differentiated cells, by phosphorylating a Prince et al., 1999). The mechanism underlying the number of transcription factors suchas ELK-1. 4NQO sensitivity of WS cell lines is not known. 4NQO PPP2CB is also involved in MAPK modulation, produces bulky DNA adducts, oxidative damage, and suggesting that regulation of the MAPK pathway is strand breaks (Gebhart et al., 1988), and the lesions are defective in WS cells after 4NQO. Thus, WS cells are generally thought to be repaired by NER (Jones et al., deficient in the 4NQO-induced downregulation of genes 1989). 4NQO has often been referred to as a UV in multiple pathways where they respond normally to mimetic, and 66 genes overlapped between the 4NQO UV. This fits with our observations that WRN has many and UV response vs 15 between 4NQO and g-irradiation protein interaction partners (Opresko et al., 2003). The (see Supplementary Table 4). However, 4NQO adducts, observation that g-irradiation-specific regulation is very unlike UV-induced cyclobutane pyrimidine dimers impaired in WS cells (7 and 8% conserved) without a (CPDs), are not preferentially repaired in transcription- corresponding significant sensitivity to g-irradiation ally active genes, and in mitochondrial DNA, which has suggests that the 4NQO sensitivity of WS cells is no NER, 4NQO adducts are removed (Snyderwine and connected to the missing repression of common stress Bohr, 1992). We found several known stress response response genes. genes that responded only to 4NQO, including MT1H, g-induced regulation of common stress response genes oxidative stress responsive gene ATOX1, protein kinase was almost similar in old and WS cells (Figure 3a), H11, which has heat–shock protein activity, and which is in agreement with the observation that WS cells PDCD2 involved in apoptosis. Only a subset of path- do not display significant g-irradiation hypersensitivity ways is shared between 4NQO and UV, and they appear (Prince et al., 1999). g-Irradiation can affect all cellular to be regulated differently. This correlates with existing components and induces several types of damage: AP results that UV repair-deficient Xeroderma Pigmento- sites, base modifications, DNA–DNA and DNA– sum group A cells retain 40–60% of 4NQO excision protein crosslinks and strand breaks, withthemost repair (Jones et al., 1989). We have preliminary data deleterious effect being DNA double-strand breaks. g- showing that WRN leaves the nucleoli after 4NQO and Specific gene regulation was more reduced in WS than in X-rays, but not after UV (Karmakar et al., our old, further supporting that the stress-specific responses unpublished data), which further supports that WS cells are not essential to survival. are deficient in 4NQO response pathways, but not after Characteristic for the UV response was a large UV. Thus, our results support the idea that the 4NQO number of intermediate-early genes downregulated in response is different from the UV response. This could all three donor groups, perhaps as a result of UV- also suggest that repair pathways other than NER, for induced dimers blocking transcription (Figure 2e). example base excision repair (BER) of oxidative Figure 2g and hreveals a delay in theUV-induced damage, are needed for repair of 4NQO-induced regulation of genes that are also induced by 4NQO or damage. g-irradiation, suggesting that certain UV pathways react The inability of WS cells to regulate transcription slower to DNA damage signals. UV irradiation causes after 4NQO was striking and appeared to involve crosslinking and transcription blocking DNA damage,

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5039 the most prominent lesion being cyclobutane pyrimidine tion and survival. This implies that various defects may dimers (CPDs) (Friedberg et al., 1995). UV-induced have similar end-results on a cellular level, and it will be DNA damage is repaired by NER. WS cells may possess interesting to identify proteins other than WRN, whose more proficient UV repair capacity than cells from old absence or malfunction can result in an age-like stress donors, bothin telomeric regions (Kruk et al., 1995) and response profiles. Multiple defects leading to a similar overall genomic repair (Weirich-Schwaiger et al., 1994). end point could also explain how heterogeneous WS cells have normal recovery of RNA synthesis after mutations accumulated withage can result in theaging UV irradiation and in primary WS fibroblasts gene- phenotype. This theory would also support the existence specific repair of UV-induced damage is normal (Webb of a relatively homogeneous gene expression profile in et al., 1996). Consistent with this, we found that gene unchallenged cells from aged donors, which is very expression pathways induced by UV were better similar to that of WS cells (Kyng et al., 2003b). Our conserved in WS cells than in cells from old donors hypothesis is supplemented by recent speculations that (Figure 3). genome damage may result in heterogeneous gene By investigating age-associated deficiencies in gene expression changes in middle age that leads towards a expression changes induced by different DNA damage homogeneous aged state (Lu et al., 2004). Clarification agents, we have taken a novel approach to deciphering of the molecular mechanisms underlying transcriptional the genetic basis of human aging. How processes of regulation of stress response genes could be used in DNA repair and transcription change in relation to diagnostics to identify drug targets, and to modify the aging is a complex question, and we have shown that role of cellular stress response mechanisms in age- there is an age-associated aberration of gene expression associated diseases suchas cancer, heartdisease responses to DNA damage, something which has not (reperfusion injury), and diabetes. Preventive interven- previously been clear. This understanding opens new tions against diseases of old age could then be tested for possibilities for exploring the mechanisms leading to their ability to reverse or prevent the age-associated DNA damage accumulation and aging. alteration of stress-induced gene regulation. A central question is whether transcription changes reflect causative mechanisms of other origins, or whether they themselves have a causative effect on Materials and methods aging. Transcriptional changes after DNA damage have recently been speculated to play a role in DNA-damage Cell lines, culture conditions, and damaging agent exposure surveillance and in determining cellular fate after A total of 15 primary human skin fibroblast cell lines were damage, thus making transcription a potential thera- obtained from the Coriell Cell Repositories (Camden, NJ, peutic target for anticancer strategies (Ljungman and USA), and classified into three groups based on genotype as Lane, 2004). It is possible that gene expression changes listed in Table 5: normal young (av. 22.5 years, n ¼ 6), normal are downstream to other ‘causative’ events not identified old (av. 90 years, n ¼ 5), and WS (av. 29 years, n ¼ 4). The cells here, such as stalled replication forks, or primary DNA were cultured in minimal essential media supplemented with repair defects. The widespread effects of lacking a single 15% fetal bovine serum, vitamins, essential and nonessential protein like WRN involved in DNA and RNA amino acids, penicillin/streptomycin, and 2 mML-glutamine metabolism support that this could be the case. (all supplied by Invitrogen, Carlsbad, CA, USA). Before damage induction, cells were grown as monolayers until 75% However, our data indicate that transcription changes confluency. For g- and UV irradiation similar protocols were mediate the process of aging with its diverse manifesta- applied: cells were washed with phosphate-buffered saline tions, becoming a cause of aging by themselves, and thus (PBS), then irradiated with 10 Gy of g-irradiation in a representing a valid goal for intervention. This is Gammacell 40 Exactor 137Cs g source (Nordian International supported by reports that age-related changes in gene Inc., Kanata, Ontario, Canada) or placed under a UV lamp expression are reversed by caloric restriction (Wein- (254 nm) and exposed to 10 J/m2. Cells were allowed to repair druch et al., 2001), indicating that transcription is a in fresh media for the times indicated before the RNA was means of directing cellular fate. extracted. Cells exposed to 4NQO were washed with PBS, then Our results show that examination of how cells from incubated for 1 hat 37 1 in serum-free media containing 4NQO aged donors behave under stress contributes informa- at a final concentration of 0.4 mg/ml. They were washed again withPBS and allowed to repair in freshmedia for thetime tion that would have been missed by only examining indicated. unchallenged cells. We found that cells from patients suffering from the premature aging of WS show stress Microarray hybridization response deficiencies similar to those of normal aging. Interestingly, after 4NQO, cells lacking WRN are RNA for expression analysis was extracted before exposure markedly defective in transcription regulation. In and 1, 6, and 24 hafter exposure to eachof theDNA- relation to this finding, we have shown that 4NQO, damaging agents. RNA isolation was performed using normally referred to as a UV mimetic, induced very QIAGEN RNeasy mini kit withan added DNase step (QIAGEN, Valencia, CA, USA). Single channel labeling 33P different gene expression changes compared to those nylon membrane based cDNA microarrays containing 6912 seen after UV. The essential role of the common stress genes and expressed sequence tags (ESTs) were provided by response genes suggests that, under different conditions, the Gene Expression and Genomics Unit (GEGU), National multiple damage-specific signals are integrated into one Institute on Aging, NIH. Array hybridization and data common regulatory response needed for cellular func- analysis were supervised by the GEGU and carried out as

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5040 previously described (Fan et al., 2002; Kyng et al., 2003a). scans were inspected visually to confirm calculated ratios. Protocols on array printing, labelling, and hybridization as CV calculations for all genes are given in Supplementary well as information on software packages are available at Tables 7–16. the GEGU website (www.grc.nia.nih.gov/branches/rrb/dna/ dna.htm). Hybridization intensities were quantitated using Data analysis Array-Pro analysis software (Media Cybernetics, CA, USA) and then stored as Excel spreadsheets. Normalization of the data was performed in Excel using standardized spreadsheets. To eliminate noise from low level expression, spots quantified at 10 (arbitrary unit) were Experimental design o replaced by the value 10 (less than 20 data points). Nonspecific We followed MIAME (Minimum Information About a uniform background across entire arrays due to experimental Microarray Experiment) guidelines for the presentation of variation was normalized in Excel using global normalization. our results (Brazma et al., 2001). Study design and data The data value for each spot on each membrane was analysis were in cooperation withtheStatistics and Experi- divided by the median intensity value of that membrane to mental Design Section (SEDS) and the Gene Expression and obtain a normalized intensity value. CVs for replicate Genomics Unit (GEGU), bothpart of theResearchResources measurements were low (0.11), and thus further normalization Branchat NIA, NIH. Thebasic design was based on was deemed unnecessary. Changes in gene expression previously published studies (Chen et al., 2000; Bakay et al., after DNA damage were then calculated by dividing the 2002; Ostermeier et al., 2002), and has been described in a median of four replicate microarray measurements after DNA recent paper (Kyng et al., 2003b). First, to minimize bias damage by the median of four replicate microarray measure- due to individual genetic background differences, age groups ments prior to DNA damage. The resulting value, referred to and WS patients were represented by an average of five as the fold difference, was tested for significance following the cell lines. Grouping WS patients for gene expression analysis is guidelines suggested by Miller et al. (2001) and others consistent with the fact that different mutations found in (Lawrance et al., 2001). Based on the CV between replicates WS all prevent nuclear localization of the protein that in and on the number of replicates and genes analysed, we turn looses its function (Matsumoto et al., 1997). Second, to determined the expected number of false positives for any reduce experimental biological (growthconditions), and given fold difference. Withidentical experimental para- technical (labelling and microarray hybridization) variability, meters and a CV of 0.16, we have previously demonstrated quadruple replicates were performed for eachcondition. that expression changes down to 1.5-fold difference are We generated approximately one million data points and a statistically significant (Kyng et al., 2003b). Following DNA study of this magnitude was only possible because RNA damage, many genes were differentially expressed by 30–50- samples were pooled within each donor group prior to fold, and to focus on changes that were biologically as well as microarray hybridization. Analysis of individual cell lines statistically significant we chose a threefold cutoff. At a rather than pooling within donor groups would have increased threefold difference, the two-tailed P-value was 1.7E-10, thus the amount of information available, but the goal of 1.7E-8% of genes were expected to show a fold difference our approach was to isolate changes specific to the group greater than 3 by chance. Assaying 6912 genes that corre- rather than individual donors. Pre-profile mixing of RNA lates to (6912*1.7E-10)p1 false positives. Thus, the use effectively normalizes bothintra- and inter-patient sources of of a threefold cutoff was very stringent and ensured that the variation (Hilsenbeck et al., 1999; Lawrance et al., 2001; genes identified were indeed differentially expressed. Our Ostermeier et al., 2002; Welle, 2002; Kyng et al., 2003b) results were further solidified by use of multiple time points, and results in relatively high sensitivity and specificity for multiple stress types, and different cell lines acting as internal gene expression changes that would be detected by many controls, in effect constituting 27 repeats that all shared individual expression profiles. It has thus been demonstrated some elements. To concentrate on changes that could be that stringent yet robust data can be generated by mixing a interpreted in a context of gene function, we included only small number of individuals witha defined condition ( n ¼ 5) genes classified in GO in the further analysis (http:// (Chen et al., 2000; Bakay et al., 2002). We have previously www.geneontology.org). This was the case for 2868 (41%) of published an evaluation of the effects of pre-profile mixing of the 6912 genes on the microarrays. GO is a structured, RNA in the particular cell lines used in this study, concluding hierarchical vocabulary to describe gene functions in areas of that there is a good correlation between pooled and indivi- biological processes (why), molecular function (what), and dual gene expression and that multiple donors are needed to cellular components (where). Using the standardized GO avoid bias from differences in genetic background (Kyng et al., annotation ensured that all gene products have a consistent 2003b). description that can be employed when comparing other sources of data withthisstudy. GO annotation assignment was an integrated function of GeneSpring software (Silicon Reproducibility statistics Genetics, Redwood City, CA, USA), which was also used To verify that growth conditions were reproducible, cells from for generating tables of differentially expressed genes and self- eachdonor were independently grown twice for eachexperi- organizing map (SOM) clustering. mental condition and RNA was isolated separately (Biological replicates). The average coefficient of variation (CV ¼ s.d./ Verification of transcriptional profiling by cDNA microarrays mean) between biological replicates was 0.12. Variation due to technical issues was assessed by generating two independent Gene expression levels of untreated cells in this experiment probe preparations from eachof thesetwo RNA samples and were verified by relative RT–PCR (Kyng et al., 2003b). Also, hybridizing them to separate arrays, resulting in a CV of 0.11 we have previously used the same cDNA microarray system to (Array replicates). In summary, for eachexperimental condi- characterize the transcriptional response to hydrogen perox- tion, an age group was analysed by four replicate microarray ide-induced DNA damage in human fibroblasts and have hybridizations and comparing all four replicates, the average verified the results by Northern blotting (Kyng et al., 2003a). CV was 0.16, demonstrating very high reproducibility. Array For more references on the cDNA microarray system

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5041 developed by the Gene expression and Genomics Unit at the H-ESR, human environmental stress response ; PDL, popula- NIH, see the website www.grc.nia.nih.gov/branches/rrb/dna/ tion doubling; IE, immediate-early gene; GO, gene ontologyt. dna.htm. Acknowledgements This work was supported by the Danish Center for Molecular Gerontology, Aarhus Universitets Forskningsfond (E-2002- Abbreviations SUN-3-11), Lundbeckfonden (117/02), and The Danish WS, Werner syndrome; CV, coefﬁcient of variation; EST, Cancer Society (DS 04 024). We thank members of the expressed sequence Tag; 4-NQO, 4-nitroquinoline-1-oxide; laboratory for reading and comments.

References

Bakay M, Chen YW, Borup R, Zhao P, Nagaraju K and Lawrance IC, Fiocchi C and Chakravarti S. (2001). Hum. Mol. Hoffman EP. (2002). BMC Bioinformatics, 3, 4. Genet., 10, 445–456. Brazma A, Hingamp P, QuackenbushJ, SherlockG, Lee CK, Klopp RG, WeindruchR and Prolla TA. (1999). Spellman P, Stoeckert C, AachJ, Ansorge W, Ball CA, Science, 285, 1390–1393. Causton HC, Gaasterland T, Glenisson P, Holstege Lee CK, WeindruchR and Prolla TA. (2000). Nat. Genet., 25, FC, Kim IF, Markowitz V, Matese JC, Parkinson H, 294–297. Robinson A, Sarkans U, Schulze-Kremer S, Stewart J, Lee J, Dawes IW and Roe JH. (1995). Microbiology, 141 Taylor R, Vilo J and Vingron M. (2001). Nat. Genet., 29, (Part 12), 3127–3132. 365–371. Ljungman M and Lane DP. (2004). Nat. Rev. Cancer, 4, Busuttil RA, Rubio M, Dolle ME, Campisi J and Vijg J. 727–737. (2003). Aging Cell, 2, 287–294. Lu T, Pan Y, Kao SY, Li C, Kohane I, Chan J and Yankner Chen YW, Zhao P, Borup R and Hoffman EP. (2000). J. Cell BA. (2004). Nature, 429, 883–891. Biol., 151, 1321–1336. Ly DH, Lockhart DJ, Lerner RA and Schultz PG. (2000). Choi D, Whittier PS, Oshima J and Funk WD. (2001). FASEB Science, 287, 2486–2492. J., 15, 1014–1020. Matsumoto T, Shimamoto A, Goto M and Furuichi Y. (1997). Collier IE, Popp DM, Lee WH and Regan JD. (1982). Mech. Nat. Genet., 16, 335–336. Ageing Dev., 19, 141–146. Miller RA, Galecki A and Shmookler-Reis RJ. (2001). Davis T, Singhrao SK, Wyllie FS, Haughton MF, Smith PJ, J. Gerontol. A Biol. Sci. Med. Sci., 56, B52–B57. Wiltshire M, Wynford-Thomas D, Jones CJ, Faragher RG Moggs JG and Orphanides G. (2003). Toxicol. Lett., 140–141, and Kipling D. (2003). J. Cell Sci., 116, 1349–1357. 149–153. Fan J, Yang X, Wang W, Wood III WH, Becker KG Moradas-Ferreira P and Costa V. (2000). Redox. Rep., 5, and Gorospe M. (2002). Proc. Natl. Acad. Sci. USA, 99, 277–285. 10611–10616. Murano S, Thweatt R, Shmookler Reis RJ, Jones RA, Friedberg EC, Walker GC and Siede W. (1995). DNA Repair Moerman EJ and Goldstein S. (1991). Mol. Cell. Biol., 11, and Mutagenesis. ASM Press: Washington, DC. 3905–3914. GaschAP and Werner-WashburneM. (2002). Funct. Integr. Nakura J, Ye L, Morishima A, Kohara K and Miki T. (2000). Genom., 2, 181–192. Cell. Mol. Life Sci., 57, 716–730. Gebhart E, Bauer R, Raub U, Schinzel M, Ruprecht KW and Nantel A, Dignard D, Bachewich C, Harcus D, Marcil A, Jonas JB. (1988). Hum. Genet., 80, 135–139. Bouin AP, Sensen CW, Hogues H, van het HM, Gordon P, Guehenneux F, Duret L, Callanan MB, Bouhas R, Hayette S, Rigby T, Benoit F, Tessier DC, Thomas DY and Whiteway Berthet C, Samarut C, Rimokh R, Birot AM, Wang Q, M. (2002). Mol. Biol. Cell, 13, 3452–3465. Magaud JP and Rouault JP. (1997). Leukemia, 11, 370–375. Niedermuller H, Hofecker G and Skalicky M. (1985). Mech. Hasty P, Campisi J, Hoeijmakers J, van SH and Vijg J. (2003). Ageing Dev., 29, 221–238. Science, 299, 1355–1359. Ogburn CE, Oshima J, Poot M, Chen R, Hunt KE, Gollahon Hilsenbeck SG, Friedrichs WE, Schiff R, O’Connell P, Hansen KA, RabinovitchPS and Martin GM. (1997). Hum. Genet., RK, Osborne CK and Fuqua SA. (1999). J. Natl. Cancer 101, 121–125. Inst., 91, 453–459. Opresko PL, Cheng WH, von Kobbe C, Harrigan JA and Huang Q, Liu D, Majewski P, Schulte LC, Korn JM, Young Bohr VA. (2003). Carcinogenesis, 24, 791–802. RA, Lander ES and Hacohen N. (2001). Science, 294, Ostermeier GC, Dix DJ, Miller D, Khatri P and Krawetz SA. 870–875. (2002). Lancet, 360, 772–777. Jelinsky SA, Estep P, Church GM and Samson LD. (2000). Park WY, Hwang CI, Kang MJ, Seo JY, Chung JH, Kim YS, Mol. Cell. Biol., 20, 8157–8167. Lee JH, Kim H, Kim KA, Yoo HJ and Seo JS. (2001). Jones CJ, Edwards SM and Waters R. (1989). Carcinogenesis, Biochem. Biophys. Res. Commun., 282, 934–939. 10, 1197–1201. Parrinello S, Samper E, Krtolica A, Goldstein J, Melov S and Kayo T, Allison DB, WeindruchR and Prolla TA. (2001). Campisi J. (2003). Nat. Cell Biol., 5, 741–747. Proc. Natl. Acad. Sci. USA, 98, 5093–5098. Poot M, Yom JS, Whang SH, Kato JT, Gollahon KA and Kruk PA, Rampino NJ and Bohr VA. (1995). Proc. Natl. RabinovitchPS. (2001). FASEB J., 15, 1224–1226. Acad. Sci. USA, 92, 258–262. Prince PR, Ogburn CE, Moser MJ, Emond MJ, Martin GM Kyng KJ, May A, Brosh Jr RM, Cheng WH, Chen C, Becker and Monnat Jr RJ. (1999). Hum. Genet., 105, 132–138. KG and Bohr VA. (2003a). Oncogene, 22, 1135–1149. Seluanov A, Mittelman D, Pereira-SmithOM, Wilson JH and Kyng KJ, May A, Kolvraa S and Bohr VA. (2003b). Proc. Gorbunova V. (2004). Proc. Natl. Acad. Sci. USA, 101, Natl. Acad. Sci. USA, 100, 12259–12264. 7624–7629. Lambert B, Ringborg U and Skoog L. (1979). Cancer Res., 39, SmithL and Greenﬁeld A. (2003). Hum. Mol. Genet., 12 2792–2795. (Spec. No. 1), R1–R8.

Oncogene Transcription after DNA damage in aging and WS KJ Kyng et al 5042 Snyderwine EG and Bohr VA. (1992). Cancer Res., 52, Welle S. (2002). Exp. Gerontol., 37, 583–590. 4183–4189. Welle S, Brooks A and Thornton CA. (2001). Physiol. Genom, Volkert MR and Landini P. (2001). Curr. Opin. Microbiol., 4, 5, 67–73. 178–185. Xu G, Snellman E, Bykov VJ, Jansen CT and Hemminki K. von KC, May A, Grandori C and Bohr VA. (2004). FASEB J., (2000). Mutat. Res., 459, 195–202. 18, 1970–1972. Yang L, Li N, Wang C, Yu Y, Yuan L, Zhang M and Cao X. Webb DK, Evans MK and Bohr VA. (1996). Exp. Cell Res., (2004). J. Biol. Chem., 279, 11639–11648. 224, 272–278. Zahn RK, Zahn-Daimler G, Ax S, Reifferscheid G, Wald- WeindruchR, Kayo T, Lee CK and Prolla TA. (2001). mann P, Fujisawa H and Hosokawa M. (2000). Mech. J. Nutr., 131, 918S–923S. Ageing Dev., 119, 101–112. Weirich-Schwaiger H, Weirich HG, Gruber B, Schweiger M Zou S, Meadows S, Sharp L, Jan LY and Jan YN. (2000). and Hirsch-Kauffmann M. (1994). Mutat. Res., 316, 37–48. Proc. Natl. Acad. Sci. USA, 97, 13726–13731.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc)

Oncogene