Fibromodulin gene transcription is induced by ultraviolet irradiation, and its regulation is impaired in senescent human fibroblasts.

Cells undergoing replicative senescence display an altered pattern of gene expression. Senescent fibroblasts show significant changes in the expression of mRNAs encoding extracellular matrix-remodeling proteins; among these mRNAs, the mRNA encoding fibromodulin is highly decreased in these cells. To understand the molecular basis of this phenomenon, we explored the regulatory mechanisms of the human fibromodulin gene. We found that fibromodulin gene promoter contains a cis-element, crucial for its basal expression, that forms a DNA-protein complex when exposed to nuclear extracts from exponentially growing human fibroblasts and not to extracts from cells undergoing senescence by repeated in vitro passages or by mild oxidative stress. The purification of this complex showed that it contains the damage-specific DNA-binding protein DDB-1. The latter is known to be induced by UV irradiation; therefore we checked whether fibromodulin gene promoter is regulated upon the exposure of the cells to UV rays. The results showed that, in exponentially growing fibroblasts, the promoter efficiency is increased by UV irradiation and the DDB-1-containing complex is robustly enriched in cells exposed to UV light. Accordingly, in these experimental conditions the endogenous fibromodulin mRNA accumulates to very high levels. On the contrary, senescent cells did not show any activation of the fibromodulin gene promoter, any induction of the DDB-1-containing complex, or any accumulation of fibromodulin mRNA. These phenomena are accompanied in senescent cells by a decrease of the UV-damaged DNA binding activity.

the case of human fibroblasts, they stop growing and undergo a condition known as replicative senescence (1). The presence and the role of senescent cells in vivo remain to be definitively addressed; however, there is a significant interest in the study of this cellular phenotype for at least two reasons: i) accumulation of senescent cells in the tissues of living organisms could have deleterious effects, actually favoring the development of age-related dysfunctions and diseases, and ii) cellular senescence could represent a strategy adopted by the cells to prevent the development of malignancies (2). Senescent cells have a well defined microscopic and molecular phenotype characterized by enlarged and flattened shape, expression of ␤-galactosidase activity, shortened telomeres, and a specific gene expression profile (3).
The mechanisms underlying the changes of gene expression profile accompanying replicative senescence are not completely understood. Senescent human fibroblasts show significant differences, compared with early passage cells, in the expression of genes encoding matrix-remodeling proteins and growth factors and cytokines involved in wound healing. Among these genes, stromelysin 1 and 2, plasminogen activators 1 and 2 and urokinase plasminogen activator, and metalloelastase (4,5) are overexpressed, whereas elastin, stromelysin 3, and prostaglandin-1 synthase are suppressed (4). Similarly, fibromodulin (FM) 1 mRNA was found to be significantly underexpressed in human fibroblasts undergoing senescence by repeated in vitro passages or mild oxidative stress (6) and in postmitotic fibroblasts (7). Although experimental evidence suggests that the observed up-regulation of genes expression is due to the stabilization of APA-1 transcription factor in senescent cells (5), the mechanisms underlying the observed gene down-regulation are not understood.
FM belongs to the family of leucine-rich proteoglycans, which regulates the assembly of extracellular matrix proteins. FM binds to collagen and is involved in the generation of collagen fibrils (8). FM gene knock-out mice show decreased tendon stiffness; the tendons contain irregular collagen fibrils that often show a reduced diameter (9). Lumican is another protein belonging to the same procollagen family. Lumican/fibromodulin double knock-out mice show more severe defects, including skin fragility, reduced corneal trans-parency, and age-dependent osteoarthritis (10). Human FM is encoded by one gene located on chromosome 1q32 (11), and the regulation of its expression is still not understood.
DDB-1 is a 127-kDa protein originally identified as a component of the damage-specific DNA-binding heterodimeric complex DDB (12)(13)(14). It is involved in nucleotide excision repair of UV-induced DNA damage through the interaction with DDB-2 or CSA proteins (15). The DDB-1/DDB-2-containing complex contributes to the so-called global genome repair, and the DDB-1⅐CSA complex acts in the transcription-coupled-repair pathway (15). In addition to these DNA-damage repair functions, DDB-1 and DDB-2 have been proposed to play a role in transcription regulation. In fact, the DDB dimer was demonstrated to act as a co-factor of the E2F1 transcription factor (16), as well as being associated with the histone acetyltransferase SPT3-TAF(II)31-GCNSL acetylase complex (17), and to be part of the machinery responsible for the basal transcription of the RNA-dependent protein kinase gene promoter (18). DDB-1 also interacts with virus-encoded transcription factors, such as the X protein of the hepatitis B virus (19). Furthermore, the Drosophila homolog of DDB-1 is not only a DNA-damage repair factor but is also necessary for development of the fly (20).
Here we report that the transcription of the human FM gene is induced by UV irradiation. Experimental evidence indicates that this regulatory mechanism is dependent on DDB-1 activity and is abolished in senescent cells. Accordingly, although the amount of DDB-1 is not modified, the UV-damaged DNA binding activity is clearly impaired in senescent fibroblasts.

MATERIALS AND METHODS
Cell Cultures-IMR-90 human primary fibroblasts and HeLa cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% (v/v) fetal bovine serum (Invitrogen), 1% penicillin/ streptomycin (Invitrogen). Cultures were maintained at 37°C in a 5% CO 2 -humidified atmosphere. The population doubling level (PDL) of IMR-90 cells was calculated by using the following equation: ⌬PDL ϭ log(n h /n i )/log2, where n i is the initial number of cells and n h is the final number of cells at each passage (21). The cells used were at 28 PDL (HF-28 cells) or at 60 PDL (HF-60 cells). More than 50% of HF-60 were positive for senescence-associated ␤-galactosidase, which was assayed according to Dimri et al. (22).
IMR-90 were treated, when indicated, with diethylmaleate (DEM) at a final concentration of 100 M in complete medium. This DEM-containing medium was prepared by dissolving 3.3 l of a 0.3 M concentrated solution of DEM in 10 ml of freshly prepared medium. After 3 days of culture in the presence of DEM, the cells stopped growing and expressed the phenotype of senescent cells, including the positivity for senescence-associated-␤-galactosidase (6).
UV-irradiated cells were obtained by growing IMR-90 cells to subconfluence in 100-mm plates in complete medium. The latter was removed and replaced with 1 ml of cold phosphate-buffered saline. The cells were UV-irradiated (254 nm) with 15 J/m 2 . Phosphate-buffered saline was then replaced with 10 ml of fresh medium. Mock-irradiated cells were treated following the same procedure but were not exposed to UV light.
Reporter Vectors and Transfection Experiments-The region of the human FM gene from nucleotide Ϫ2032 to nucleotide ϩ20 (ϩ1 corresponds to the transcription start site) was obtained by PCR using as template the DNA prepared from blood donor leukocytes. This fragment was cleaved with SacI/XhoI and cloned in the pCAT basic vector, giving rise to the FP-CAT1 vector. The DNA fragments used to generate the vectors FP-CAT2 (Ϫ833 to ϩ20), FP-CAT3 (Ϫ122 to ϩ20), FP-CAT4 (Ϫ80 to ϩ20), FP-CAT5 (Ϫ122 to Ϫ20), and FP-CAT6 (Ϫ122 to Ϫ42) were obtained by PCR using the FP-CAT1 as template. In FP-CAT5 and FP-CAT6, a region of 50 bp located upstream from the transcription start site of the herpesvirus thymidine kinase gene precedes the CAT gene. FP-CAT6mut vector was generated by PCR using the same forward oligonucleotide as for FP-CAT6 and a reverse primer containing four mismatches corresponding to the mutations present in the oligonucleotides m4 and m8 (see below).
The cDNA encoding human DDB-1 was obtained by retrotranscribing total RNA from IMR-90 cells. It was then cloned into the pRC-CMV vector. Cloned cDNA was completely sequenced. The RNA interference experiments were performed by using two dsRNA: iDDB-1, 5Ј-CUCCUUGGAGAGACCUCUA-3Ј, and iGFP, 5Ј-GCAAGCUGACCC-UGAAGUUCAU-3Ј. Conditions of the siRNA experiments were as described previously (23).
Cell transfections were carried out with Lipofectamine-plus (Invitrogen) in 60-mm dishes, according to the manufacturer's instructions. Each transfection was performed in quadruplicate with 2 g of plasmid DNA plus 0.2 g of CMV-luciferase vector, to correct for variations in DNA uptake and transfection efficiency. CAT gene expression was assessed by measuring the CAT protein levels by enzyme-linked immunosorbent assay (Roche Applied Science).
The EMSA experiments with the UV-irradiated probe were performed as described in by Martinez et al. (17). The sequence of the double-stranded oligonucleotides used for this experiment were: oligonucleotide 1, 5Ј-TACCCACGCTCACGGCAGCCGCA-3Ј, and oligonucleotide 2, 5Ј-TACGGACGAACACGGCGCCCGCT-3Ј. UV irradiation (254 nm) was done with 15,000 J/m 2 using a UV-Stratalinker apparatus.
The proteins binding to the B4 oligonucleotide were purified from HeLa cell extracts in 5 ml of buffer A (50 mM Hepes, pH 7.9, 12.5 mM MgCl 2 , 10 mM KCl, 1 mM EDTA, 17% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) as described previously (26). Briefly, 50 mg of HeLa nuclear extracts in buffer A were loaded on a DEAE column (2-ml bed volume) (Amersham Biosciences). Retained proteins were eluted by step-increasing the concentration of KCl from 100 to 500 mM. The 200 mM KCl active fraction (7.6 mg) was dialyzed against buffer Z (25 mM Hepes, pH 7.6, 100 mM KCl, 12.5 mM MgCl 2 , 1 mM dithiothreitol, 20% (v/v) glycerol, 0.1% (v/v) Nonidet P-40), challenged with 400 g of competitor (poly(dI-dC), Sigma), and loaded on a DNA affinity column constructed with the multimeric B4 oligonucleotide covalently linked to CNBr-activated Sepharose. The retained sample was eluted with increasing KCl concentration (from 250 mM to 1.5 M). The active fraction was concentrated on Centricon filters (3-kDa cutoff), resolved by SDS/PAGE (12% (w/v) gel), and silver-stained with a commercially available kit (Sigma). The bands were excised from the gel and digested with trypsin, and the peptide mixture was analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry as reported previously (27).
For Western blot analyses, total extracts from IMR-90 cells were challenged with antibodies against DDB-1, DDB-2, CSA, and tubulin (Santa Cruz Biotechnology), and the signals were detected by using the ECL kit (Amersham Biosciences).
UV Treatment and Real Time PCR-1.0 ϫ 10 6 fibroblasts were seeded onto 100-mm culture plates in 10 ml of Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. After incubation for 4 days at 37°C in 5% CO 2 , the medium was replaced with 1 ml of cold phosphate-buffered saline, and the monolayer of subconfluent cells was irradiated with 15 J/m 2 of UV light (254 nm). The phosphate-buffered saline was then replaced with 10 ml of fresh culture medium, and the cells were allowed to recover for 12-72 h. Total RNA was prepared from IMR-90 cells by using the RNeasy mini kit (Qiagen) and subjected to cDNA synthesis with random hexanucleotide primers and MultiScribe reverse transcriptase (Invitrogen) at 48°C for 1 h. The cDNA was then amplified in an iCycler iQ real time PCR detection system (Bio-Rad) using iQ TM SYBR Green Supermix in triplicate in 25-l reaction volumes. Relative quantification of gene expression was performed using the 2 T Ϫ⌬⌬C method (28). Reference mRNA was that of c-Abl. The ratios between 2 T Ϫ⌬⌬C before the UV treatment and those calculated for the samples at various times after the exposure to UV light are expressed as fold changes. The primer sequences were as follows: c-Abl forward, 5Ј-TGGAGATAACACTCTAAGCATAACTAAA-GGT-3Ј; c-Abl reverse, 5Ј-GATGTAGTTGCTTGGGACCCA-3Ј; Fmod forward, 5Ј-CTACCTCCAAGGCAATAGGATCA-3Ј; Fmod reverse, 5Ј-TGGCGCTGCGCTTGA-3Ј; p21 waf1 forward, 5Ј-CTGGAGACTCTCAG-GGTCGAA-3Ј; p21 waf1 reverse, 5Ј-CGGCGTTTGGAGTGGTAGAA-3Ј; GADD45 forward, 5Ј-AGACCCCGGACCTGCACT-3Ј; and GADD45 reverse, 5Ј-CCGGCAAAAACAAATAAGTTGACT-3Ј.

RESULTS
Transcription from Fibromodulin Gene Promoter Is Decreased in Senescent Fibroblasts-We have observed previously that human senescent fibroblasts express low levels of FM mRNA (6). To ascertain whether this phenomenon is due to a decreased transcription of the FM gene, we generated a reporter construct (FP-CAT1) in which the CAT gene is under the control of about 2,000 bp flanking the transcription start site of FIG. 1. Identification of a cis-element that is crucial for the basal expression of the human fibromodulin gene promoter in exponentially growing fibroblasts. A, schematic representation of the reporter vectors generated to examine the genomic region upstream of the transcription start site of the human FM gene. Vectors from FP-CAT1 to FP-CAT6 contain the indicated fragments of the FM gene cloned upstream from the CAT gene. In FP-CAT5 and FP-CAT6 the CAT gene is preceded by the transcription start site of the human thymidine kinase gene (mesh pattern). The FP-CAT6mut vector is identical to FP-CAT6 but for four nucleotide changes contained in the oligonucleotides m4 and m8 reported in Fig. 2 Fibromodulin Expression in UV-irradiated and Senescent Cells the human FM gene (see Fig. 1A). As shown in Fig. 1B, the reporter gene in this construct is efficiently transcribed in exponentially growing IMR-90 fibroblasts at 28 PDL (HF-28 cells). The same construct was transfected in IMR-90 fibroblasts grown in culture for several passages thus reaching 60 PDL (HF-60 cells) and showing the phenotype of senescent cells, including the expression of ␤-galactosidase in more than 50% of the cells. As shown in Fig. 1B, the transcription of the FP-CAT1 vector transfected in HF-60 cells was dramatically suppressed, compared with the transcription efficiency observed in HF-28 cells, whereas that of CMV-CAT was very similar to that observed in HF-28 cells.
Another method to induce the senescence in cultured cells is based on the exposure of cells to a mild oxidative stress (29). Therefore, HF-28 cells were treated for 3 days with 100 M DEM (HF-DEM cells), and this treatment resulted in the arrest of the growth and the appearance of ␤-galactosidase staining. Also in this case, although the efficiency of the CMV promoter was the same in HF-28 and HF-DEM cells, the results reported in Fig. 2B clearly show that the transcription from the promoter in the FP-CAT1 vector is suppressed in both the HF-28 and the HF-DEM cells.
To identify the minimal promoter region still able to drive an efficient transcription of the reporter gene in HF-28 cells and not in HF-60 and HF-DEM cells, we generated a collection of 5Ј-deletion mutants of the promoter region (see Fig. 1A). As shown in Fig. 1B, the deletion of a great part of the 2,000 bp flanking the transcription start site at the 5Ј terminus, down to position Ϫ122 (FP-CAT2 and FP-CAT3), did not modify the promoter efficiency, whereas a further deletion down to nucleotide Ϫ54 (FP-CAT4) resulted in a promoter that drives a weak transcription to the same extent in HF-28, HF-60, and HF-DEM cells. Similarly, we have deleted the region encompassing the transcription start site, and the obtained fragments were cloned upstream from the transcription start site of herpesvirus thymidine kinase gene (FP-CAT5 and FP-CAT6). Fig. 1B shows that the fragment from Ϫ122 to Ϫ42 contains a promoter element that supports the transcription in HF-28 and not in HF-60 or HF-DEM cells.
DDB-1 Protein Interacts with the Fibromodulin Gene Promoter-To identify the cis-element(s) that is contained in the Ϫ122 to Ϫ42 region of the FM gene promoter and that is possibly responsible for the observed decline of transcription efficiency in senescent cells, five overlapping double-stranded oligonucleotides covering this region were tested by EMSA for their ability to bind proteins present in the nuclear extracts from HF-28 and HF-60 cells. As shown in Fig. 2B, three of these oligonucleotides (B1, B3, and B5) form specific shifted complexes with either HF-28 or HF-60 nuclear extracts. The oligonucleotide B4, which partly overlaps B3 and B5 sequences, forms a DNA-protein complex when challenged with HF-28 nuclear extracts. On the contrary, this complex was barely visible when the nuclear extracts from HF-60 cells were used.
To better characterize the cis-element involved in the formation of the complex, we generated a collection of mutant B4 oligonucleotides. Fig. 2C shows that two of these mutant oligonucleotides (m4 and m8) were unable to compete for the formation of the complex, thus suggesting a sequence-specific binding. On this basis, the FP-CAT6 construct (see Fig. 1A) was mutagenized through the insertion of the nucleotide changes present in the m4 and m8 oligonucleotides. This vector was transfected in HF-28, HF-60, and HF-DEM cells, and, as shown in Fig. 2D, the mutant vector drives low levels of transcription both in HF-28 and in HF-60 and HF-DEM cells. These results support the possibility that the region of the FM gene promoter covered by the B4 oligonucleotide contains a specific cis-ele- Fibromodulin Expression in UV-irradiated and Senescent Cells ment that interacts with the transcription factor(s) responsible for the different efficiency of the promoter in exponentially growing cells compared with senescent cells.
Considering that a robust shifted complex was observed when the B4 oligonucleotide was challenged with HeLa cell nuclear extracts (see Fig. 2A), we decided to use the latter as a source to purify the protein(s) present in the shifted complex. Nuclear extracts from HeLa cells were chromatographed on a DEAE column, as described under "Materials and Methods." The fraction eluted with 0.2 M KCl was able to shift the B4 oligonucleotide (see Fig. 3A). A further purification was obtained by affinity chromatography on a Sepharose column bearing the multimerized B4 oligonucleotide. As shown in Fig.  3B, the factor was eluted from the column with 0.5 M KCl. This fraction was analyzed by SDS-PAGE, the silver staining of which is reported in Fig. 3C. The bands were excised from the gel, digested with trypsin, and analyzed by mass spectrometry. Peptide mass fingerprint analysis with 12 of 15 measured tryptic fragments allowed us to identify one band of about 120 kDa as the DDB-1. We were unable to identify the other bands present in the 0.5 M KCl fraction; mass spectrometry analysis suggests that some of them could have been DDB-1 degradation products.
To confirm that DDB-1 is present in the protein-DNA complex responsible for the electrophoretic shift of the oligonucleotide B4, we did the supershift experiments reported in Fig. 3D. The results show that antibodies directed against DDB-1 or against the two known partners of this protein, i.e. DDB-2 and CSA, did not induce any evident supershift. On the contrary, when we challenged the nuclear extract with a combination of anti-DDB-1 and anti-DDB-2 antibodies, a clear supershift was observed, whereas no supershift appeared with the combination of anti-DDB-1 and anti-CSA antibodies. No specific band shift was observed in control experiments in which the oligonucleotide B4 was challenged with the above mentioned antibodies without nuclear extracts (see supplemental Fig. 1).

DDB-1 Is Necessary for the Basal Transcription from the FM Gene
Promoter-To ascertain whether DDB-1 is involved in the regulation of the FM gene promoter, we overexpressed this protein in IMR-90 cells. As shown in Fig. 4A, this overexpression significantly activated the transcription of the FP-CAT3 vector and not of the FP-CAT4 vector.
We then tested the effect of the suppression of the endogenous DDB-1 by using an siRNA targeting the DDB-1 mRNA. We observed that a 50% decrease of the DDB-1 protein concentration in transfected cells was accompanied by a significant reduction of the transcriptional efficiency of the FM gene promoter in the FP-CAT3 vector and not of the CMV-CAT vector. A control siRNA, which suppresses the expression of enhanced GFP had no effect on the FP-CAT3 transcription (see Fig. 4B). Nuclear extracts from cells overexpressing DDB-1 or transfected with siRNA targeting the DDB-1 mRNA were analyzed by EMSA. As shown in Fig. 4C, DDB-1 overexpression resulted in an increased intensity of the shifted band, whereas DDB-1

Fibromodulin Expression in UV-irradiated and Senescent Cells
suppression caused a decrease of the shifted band. These results demonstrated that the basal levels of transcription driven by the FM gene promoter are, at least in part, dependent on DDB-1 concentration.
Fibromodulin Gene Expression Is Activated by UV Irradiation-It has been demonstrated that DDB-1 is translocated into the nucleus upon UV irradiation (30); therefore we asked whether UV irradiation is associated with a change in the transcription from the FM gene promoter. To this aim, HF-28 cells were transfected with the FP-CAT6 vector (see Fig. 1A) and exposed to UV irradiation (15 J/m 2 ). CAT assay allowed us to observe that UV irradiation activates FM gene promoter efficiency, leading to a 3-fold increase 72 h after the irradiation (Fig. 5A).
To ascertain whether endogenous fibromodulin gene is similarly regulated, we extracted RNA from HF-28 cells before and at various times after the exposure to UV irradiation and analyzed the RNA by real time PCR. As shown in Fig. 5B, a dramatic accumulation of FM mRNA was observed, reaching the maximal expression 72 h after irradiation. These results support the hypothesis that the binding of DDB-1 to the FM gene promoter not only affects the basal levels of transcription but also plays a role in the regulation of this gene upon UV irradiation.
Fibromodulin Gene Does Not Respond to UV Irradiation in Senescent Human Fibroblasts-Considering that the FM gene transcription is repressed in senescent cells, we explored its response to UV irradiation in these cells. To this aim, FM mRNA was measured in HF-60 cells at various times after the exposure to UV, and, in contrast to that observed with HF-28 cells, no significant increase of FM mRNA was observed (Fig. 5B). Another two mRNAs, encoding GADD45 and p21 waf1 , showed a robust accumulation upon UV irradiation in both HF-28 and HF-60 cells (Fig. 5C). Therefore, we first addressed whether the interaction between the B4 element of the FM gene promoter and the DDB-1 protein is modified upon UV treatment. To do this, we used EMSA with the B4 oligonucleotide to analyze nuclear extracts from HF-28 cells exposed to UV irradiation. Fig. 6 shows that the intensity of the shifted band of the B4 oligonucleotide is significantly increased in the extracts from UVtreated cells. Then, we addressed whether the enhancement of the B4 oligonucleotide shift upon UV irradiation is abolished in senescent cells. As shown in Fig. 5, no effects on the intensity of this band were seen by using extracts from HF-60 cells exposed to UV rays, thus suggesting an impairment of DDB-1 function in these cells.
DDB Activity Is Impaired in Senescent Fibroblasts-One possible explanation of the results presented above is that DDB-1 expression is suppressed in senescent cells. To address this point, we measured by Western blot the levels of DDB-1, DDB-2, and CSA in exponentially growing and in senescent IMR-90 fibroblasts. Fig. 7A shows that these three proteins are similarly expressed in HF-28 and HF-60 cells.
Given that we demonstrated that the DDB-1-dependent transcription of the FM gene, both in basal conditions and after UV irradiation, is impaired in senescent cells, it could be hypothesized that the well known ability of DDB-1 to bind UVdamaged DNA, regardless of its sequence, is also impaired in senescent cells. To test this possibility we used an enzymelinked immunosorbent assay-based method, which measures the binding of nuclear proteins to DNA fragments exposed to UV rays (17). An oligonucleotide witha sequence that is not related to that of the B4 oligonucleotide was UV-irradiated (15,000 J/m 2 ) and challenged with nuclear extracts from HF-28 and HF-60 cells. Nuclear extracts from HF-28 cells have the capacity to bind this irradiated oligonucleotide, which is decreased in cells transfected with the siRNA for DDB-1 (see supplemental Fig. 2). As shown in Fig. 7B, although the irradiated cold oligonucleotide efficiently competed for this binding, a 50-fold excess of the unirradiated cold probe was unable to compete. On the contrary, HF-60 extracts, when challenged with the irradiated probe, led to a very faint band, thus suggesting that the binding activity of the damaged DNA in these cells was greatly reduced. DISCUSSION In this study, we have explored the regulation of FM gene expression in normal human fibroblasts, either exponentially

Fibromodulin Expression in UV-irradiated and Senescent Cells
growing or arrested by repeated in vitro passages. These experiments support several conclusions on the regulation of this gene and the phenotype of fibroblasts undergoing replicative senescence.
First, we have found that the FM gene is regulated by UV irradiation and that this leads to a dramatic (50-fold) accumulation of FM mRNA. Our knowledge of the functions of this protein is mostly based on the study of the phenotype of the FM gene knock-out in the mouse (9,10). These studies show several abnormalities mostly concerning defects in collagen fibrillogenesis. A significant amount of experimental data indicate that UV irradiation provokes skin damage localized in the dermis, where collagen fibrils are deposited (31). Chronic exposure to UV rays provokes the so-called photoaging of the skin, which is characterized by disorganization of collagen fibrils, reduced levels of procollagens I and III due to increased degradation by collagenase and metalloproteinases, and increased levels of elastin. Therefore, it is absolutely not surprising to find that UV treatment of cultured fibroblasts induces the expression of the FM gene, the product of which plays a crucial role in collagen fibrillogenesis. Thus, it could be speculated that FM overexpression is aimed at correcting, or at counteracting, the connective tissue trouble induced by exposure to UV light by assisting in the assembly of new collagen fibrils. On the other hand, chronic exposure to UV light could induce sustained overexpression of FM that could contribute to the development of the photoaging phenotype.
The second point that deserves comment concerns the find- FIG. 5. FM gene expression is induced by UV irradiation in exponentially growing human fibroblasts and not in senescent cells. A, FP-CAT6 vector (white bars) and CMV-CAT (black bars) were transfected in HF-28 cells exposed or not to UV rays (15 J/m 2 ). CAT expression was measured 48 h after the UV irradiation or mock treatment. The efficiencies of the promoters are indicated as percentages of the mean values observed with the two vectors in the mocktreated cells. B, FM mRNA was measured by real time PCR in total RNA preparations from HF-28 or HF-60 cells before and at various times after the exposure to UV irradiation. White bars indicate the relative abundance of the mRNA, considering as ϩ1 the abundance of the FM mRNA in HF-28 cells not exposed to UV light. The values of black bars have been calculated by considering ϩ1 the FM mRNA concentration in HF-60 cells not exposed to UV light. All of the values represent the mean Ϯ S.D. of triplicate experiments on two distinct total RNA preparations. C, the same total RNAs analyzed in the experiments of B were used to calculate the changes in the concentrations of GADD45 and p21 waf1 mRNAs. The bars indicate the relative abundance of the two mRNAs, considering as ϩ1 their abundance in HF-28 cells not exposed to UV light.
FIG. 6. The interaction of DDB-1 with the FM gene promoter is induced upon UV irradiation only in HF-28 cells. Nuclear extracts from UV-irradiated or unirradiated HF-28 or HF-60 cells were used for EMSAs to test the formation of the complex with the oligonucleotide B4. The last two lanes indicate the competition experiments performed on the unirradiated HF-28 nuclear extracts, by using a 50-fold molar excess of unlabeled B4 oligonucleotide (s.c.) or of Sp1 site-containing oligonucleotide (ns.c.).

Fibromodulin Expression in UV-irradiated and Senescent Cells
ing that FM gene expression is regulated by a cis-element, present in the promoter region in the proximity of the transcription start site, that interacts with the DDB-1 protein in EMSA experiments. Numerous experimental results indicate that this factor is involved in the cellular responses to UV irradiation (12)(13)(14)(15). It forms two complexes with DDB-2 and CSA, respectively, which play an important role in the molecular machinery for repairing UV-induced DNA damage (15). Other observations, however, suggest the possibility that DDB-1 is involved in the regulation of gene transcription. Our results support this possibility and also suggest that DDB-1 coordinately contributes to the assembly of DNA repair mechanisms and, at the same time, to the regulation of the transcription of UV-induced genes. Additional studies are needed to address whether DDB-1 involvement in gene regulation upon UV irradiation is only targeted at the FM gene or has a more general effect.
Third, FM gene expression is impaired in senescent human fibroblasts, and this appears to be dependent on the suppression of the DDB-1 functions. In fact, although DDB-1 is expressed in senescent fibroblasts at levels comparable with those observed in exponentially growing cells, the B4 cis-element of the FM gene promoter fails to form any detectable complex when challenged with HF-60 nuclear extracts. On the other hand, UV-damaged DNA binding activity of senescent cells is severalfold lower than that present in the extracts from growing HF-28 cells. These phenomena are apparently due to the failure of the mechanisms governing the intracellular trafficking of DDB-1. It has been demonstrated that DDB-1 is mostly cytosolic in unirradiated cells, whereas it translocates into the nucleus after UV irradiation (30). The mechanism regulating this trafficking and its relationship with UV irradiation are not known. However, our results, indicating that this mechanism is suppressed in senescent cells, suggest a possible explanation; the molecular machinery driving DDB-1 into the nucleus could be related to the cell cycle progression. This is obviously supported by the main peculiarity of senescent cells compared with exponentially growing cells, that is, the close association of replicative senescence with the arrest of the cell cycle. Experimental evidence further supports this hypothesis. DDB-1 nuclear translocation does not take place at the same time in all of the cells exposed to UV; 24 h after UV irradiation DDB-1 is localized in the nucleus of about 40% of the cells, and this percentage increases to 70% after 48 h and to 100% only after 72 h (30). This suggests that the DDB-1-nuclear translocation mechanism, to sense the DNA damage-induced signaling, should pass through a specific cell cycle step. In agreement with this hypothesis, there are several results indicating that DDB-1-containing complexes functionally or even physically interact with cell cycle regulatory proteins (16).
The altered regulation of the FM gene expression in senescent cells could have significant effects. Because the number of senescent fibroblasts is high in the elderly, decreased basal levels of FM expression could alter the extracellular matrix assembly in these individuals, leading to a condition similar to that observed in FM knock-out mice that is characterized by skin fragility, reduced corneal transparency, and age-dependent osteoarthritis. In addition, the lack of response of the FM gene to UV irradiation could have deleterious effects on the sensitivity of aged tissue exposed to sun rays.
Solid experimental evidence indicates that the efficiency of repair of DNA double strand breaks is decreased in senescent cells (probably because of a decline of end-joining activity (32)), and an accumulation of unrepairable double strand breaks is suggested to have a causative role in aging (33). On the contrary, it is not completely clear whether nucleotide excision repair is altered in senescent cells. Repair of UV-induced DNA damage on nontranscribed strands (global genomic repair) is reduced in terminally differentiated cells (34), and the number of UV photoproducts found in normal skin increases with age (35). This is in agreement with the dramatic decrease we observed in the UV-irradiated DNA binding activity in nuclear extracts from HF-60 cells. It is conceivable that replicative senescence could be associated with a decreased efficiency of the DDB-1/DDB-2-dependent global genomic repair. This is not expected to provoke significant alterations because senescent cells do not replicate their DNA; thus the accumulation of mutations in nontranscribed strands is not harmful. On the contrary, if DDB-1 is involved in gene regulation, as suggested by our results, the effects on the functions of senescent cells and on the tissues in which they are present should be relevant. The understanding of the complete array of genes regulated by DDB-1 is expected to make a significant contribution to address this point.

FIG. 7. UV-damaged DNA binding activity of senescent fibroblasts is decreased compared with exponentially growing cells.
A, Western blot analysis of DDB-1, DDB-2, and CSA in HF-28 and HF-60 cells. Tubulin was used as a control of loading. B, the oligonucleotide 1 was UV-irradiated (15,000 J/m 2 ) and incubated with HF-28 or HF-60 nuclear extracts. DNA-protein complexes were separated on nondenaturing gel as described under "Materials and Methods." The shifted complex is efficiently competed by unlabeled UV-irradiated oligonucleotide 1 (Oligo 1 UV) and not by an unrelated double-stranded DNA fragment (Oligo 2) or by unirradiated oligo 1 (Oligo 2).