PPAR gamma is activated by diverse ligands and regulates the differentiation of many cell types. Based on evidence that activation of PPAR gamma 2 by rosiglitazone stimulates adipogenesis and inhibits osteoblastogenesis in U-33/gamma 2 cells, a model mesenchymal progenitor of adipocytes and osteoblasts, we postulated that the increase in marrow fat and the decrease in osteoblast number that occur during aging are due to increased PPAR gamma 2 activation. Here, we show that the naturally occurring PPAR gamma ligands 9,10-dihydroxyoctadecenoic acid, and 15-deoxy-Delta(12,14)-PGJ(2), also stimulate adipocytes and inhibit osteoblast differentiation of U-33/gamma 2 cells. Strikingly, 9,10-epoxyoctadecenoic acid and the thiazolidine acetamide ligand GW0072 [(+/-)-(2S,5S)-4-(4-(4-carboxyphenyl)butyl)-2-heptyl-4-oxo-5-thaizolidineN,N-dibenzyl-acetamide] prevent osteoblast differentiation, but do not stimulate adipogenesis, whereas 9-hydroxyoctadecadienoic acid stimulates adipogenesis but does not affect osteoblast differentiation. The divergent effects of PPAR gamma 2 ligands on osteoblast and adipocyte differentiation were confirmed in primary murine bone marrow cultures using rosiglitazone and GW0072. These findings indicate that the proadipogenic and antiosteoblastogenic effects of PPAR gamma 2 are mediated by distinct regulatory pathways that can be differentially modulated depending on the nature of the ligand, and they support the idea that increased fatty acid oxidation during aging may inhibit osteoblast differentiation. Moreover, there may be selective PPAR gamma 2 modulators that block the adverse effects of fatty acid oxidation products while retaining beneficial activities such as insulin sensitization.

Cells of the bone marrow stroma can reversibly convert among different phenotypes. Based on this and on evidence for a reciprocal relationship between osteoblastogenesis and adipogenesis, we have isolated several murine bone marrow-derived clonal cell lines with phenotypic characteristics of osteoblasts or adipocytes, or both. Consistent with a state of plasticity, cell lines with a mixed phenotype synthesized osteoblast markers like type I collagen, alkaline phosphatase, osteocalcin, as well as the adipocyte marker lipoprotein lipase, under basal conditions. In the presence of ascorbic acid and beta-glycerophosphate-agents that promote osteoblast differentiation-they formed a mineralized matrix. In the presence of isobutylmethylxanthine, hydrocortisone, and indomethacin-agents that promote adipocyte differentiation-they accumulated fat droplets, but failed to express adipsin and aP2, markers of terminally differentiated adipocytes. Furthermore, they were converted back to matrix mineralizing cells when the adipogenic stimuli were replaced with the osteoblastogenic ones. A prototypic cell line with mixed phenotype (UAMS-33) expressed Osf2/Cbfa1-a transcription factor required for osteoblast differentiation, but not PPARgamma2-a transcription factor required for terminal adipocyte differentiation. Stable transfection with a PPARgamma2 expression construct and activation with the thiazolidinedione BRL49653 stimulated aP2 and adipsin synthesis and fat accumulation, and simultaneously suppressed Osf2/Cbfa1, alpha1(I) procollagen, and osteocalcin synthesis. Moreover, it rendered the cells incapable of forming a mineralized matrix. These results strongly suggest that PPARgamma2 negatively regulates stromal cell plasticity by suppressing Osf2/Cbfa1 and osteoblast-like biosynthetic activity, while promoting terminal differentiation to adipocytes.

A variety of short-lived mouse strains (SAMP strains) and control strains of less abbreviated life span (SAMR strains) have been proposed as murine models of accelerated senescence. Each SAMP strain, in addition to displaying "progeroid" traits of accelerated aging, exhibits a singular age-related pathology. The application of this animal model to the study of normal aging processes has been and remains controversial. Therefore, we have undertaken a study of dermal fibroblasts derived from the short-lived SAMP6 strain, which shows early-onset and progressive osteopenia. We have investigated cellular and molecular characteristics that are associated with in vitro aging of normal human fibroblasts, and which are exacerbated in fibroblasts from patients with Werner syndrome, a human model of premature senescence. We found that SAMP6 dermal fibroblasts, relative to SAMR1 and C57BL/6 controls, exhibit characteristics of premature or accelerated cellular senescence with regard to in vitro life span, initial growth rate, and patterns of gene expression.

In our efforts to characterize cellular senescence we have shown that the mRNA encoding WS3-10 protein is overexpressed in senescent human diploid fibroblasts (HDF) when compared with their younger counterparts, and that forced expression of the WS3-10 cDNA in young HDF results in suppression of calcium-dependent membrane currents, presumably due in part to the presence of a calcium binding domain within the WS3-10 protein. We have now expressed this protein in E. coli and have obtained affinity purified antibodies. Western blot analysis utilizing these antibodies showed that WS3-10 protein is also overexpressed in senescent HDF when compared to young HDF, and in normal fetal lung HDF when compared to SV40-transformed fetal lung HDF. HeLa cells do not express WS3-10 protein. In addition, we looked for WS3-10-related species in a variety of rat tissues. Analysis of WS3-10 immunologically related proteins in rat tissue extracts revealed two WS3-10 homologs, sized 22 kDa and 20 kDa. The latter presumably result from proteolytic removal of the C-terminal end of the 22 kDa polypeptide. The ratio between these polypeptides varies in a tissue-specific manner. Two proteins immunologically related to WS3-10 with sizes of 39 kDa and 91 kDa were present in rat spleen and skeletal muscle, respectively.

The RNA species encoded by IGFBP-3 (insulin-like growth factor binding protein-3), PAI-1 (plasminogen activator inhibitor-1) and SPARC (secreted protein-acidic and rich in cysteine; a.k.a. osteonectin) are overexpressed in senescent human diploid fibroblasts (HDF). Their extracellular products have the ability to modulate cell growth in culture and have been shown to have inhibitory effects on DNA synthesis and/or cell growth. This overproduction may contribute to a number of features of aging, including osteoporosis, atherosclerosis and diabetes mellitus type II. Based on analysis of steady-state mRNA levels, which showed similar patterns for all three along with overexpression in senescent cells, we further investigated their transcription rates and stability to determine reasons for their overexpression and to determine if coordinate gene regulation was involved. Characterization of the rates of transcription and the levels of message stability of these genes in early passage (young) versus late passage (old) HDF revealed that IGFBP-3, PAI-1 and SPARC are coordinately overexpressed but not regulated by a unique or simple mechanism encompassing all three transcripts. Only PAI-1 shows an increase in the rate of transcription, while all three show evidence that their overexpression is due to an increase in the stability of RNA. Thus, the overexpression of these genes in senescent fibroblasts involves interactions not only at the transcriptional level but also with protein factors involved in determining the stability and the degradation of RNA.

Senescent primary human skin fibroblasts were used as feeder layers for cloning lymphoid cells by limiting dilution. B-cells including mouse hybridoma cells, human multiple myeloma cells C1R and DIG, as well as an immortalized T-cell line (Jurkat cells) were cloned using this approach. From heterogenous populations, homogeneous (clonal) populations were obtained and further analyzed. The major advantages of senescent fibroblasts as feeder cells are (i) the need to establish primary cultures from experimental animals for preparing feeder cells is obviated; (ii) the risk of contamination with infectious agents is diminished; (iii) primary skin fibroblasts are easily maintained in culture, can be kept at confluence for extended periods of time, and do not undergo spontaneous transformation; and (iv) lymphoid cells are readily separated from fibroblast monolayers. The use of senescent fibroblasts overcomes limitations inherent with primary mouse cell cultures and irradiated cells commonly used as feeder cells in cloning techniques.

The phenotype of replicative senescence is a dominant trait in human diploid fibroblasts (HDF), Therefore, we have sought to identify overexpressed and/or newly expressed causal genes by constructing and screening a subtracted cDNA library derived from polyA(+)RNA of prematurely senescent Werner syndrome (WS) HDF, We have identified 15 cDNA clones that are overexpressed in senescent and WS HDF, Among them are six known sequences coding for: acid sphingomyelinase, fibronectin, SPARC, nm23-metastasis suppressor protein, and two translation factors, eIF-2 beta and EF-1 alpha. Among the 10 unknown clones are: S1-5, which encodes a secreted protein containing EGF-like domains and paradoxically stimulates DNA synthesis of young HDF in an autocrine and paracrine manner; S1-3, which encodes a protein containing "zinc finger" domains, suggesting nucleic acid binding properties; S1-15, which shows sequence similarities to human alpha 2-chimerin; and S2-6, which represents a new member of the LIM family of proteins. The other five clones do not have any significant homology to known sequences. Steady-state mRNA levels of all gene sequences thus far studied are elevated in both WS and senescent normal HDF when compared to young HDF,which suggests that senescent and WS HDF enter a final common pathway where multiple gene overexpression may generate diverse antiproliferative mechanisms and pathogenic sequelae.

Previously, we reported that fibronectin (FN) mRNA was overexpressed in normal late-passage (old) and prematurely senescent Werner syndrome (WS) fibroblasts when compared to normal early-passage (young) cells (Muranoet al. Mol. Cell. Biol.11, 3905–3914, 1991). Therefore, we investigated the expression and function of the α5β1 FN receptor (FNR), a member of the integrin family, in young and senescent normal and WS cells. Levels of the α5 polypeptide, a unique subunit of the α5β1 FNR, were reduced in old cells, so that old cells produced fewer α5β1 heterodimers on the plasma membrane. The reduced levels of α5 polypeptide might be due to deficient translation and/or nonfunctional α5 mRNA since increased mRNA levels and unchanged polypeptide turnover were found in these cells. Moreover, the α5 polypeptides on the senescent cell surface were less accessible to monoclonal antibody, suggesting sequestration of this subunit, which might affect receptor–ligand binding. In contrast, β1 subunit, a common subunit for the β1 integrin subfamily, showed relatively stable levels during cellular aging, but underwent slower intracellular processing. Old cells exhibited reduced attachment to FN, which might be in part mediated by the α5β1 FNR. More importantly, old cells were deficient in response to FN-induced DNA synthesis and cell proliferation. This induction was pronounced in young cells, however, and could be completely inhibited by α5-specific monoclonal antibody, indicating mediation by α5β1 FNR. WS cells behaved like normal old cells in the above assays. Our results indicate that reduction of α5β1 FNR expression and its mediated effects are associated with the senescent phenotype of fibroblasts. These findings provide new insight into the mechanism(s) of replicative senescence in human fibroblasts.

We have previously demonstrated that senescent human diploid fibroblasts (HDF) produce large amounts of IGF-binding protein-3 (IGFBP-3) in comparison to early-passage vigorously proliferative HDF. In order to determine whether this excess IGFBP-3 accumulation plays a role in the observed attenuation of DNA synthesis in senescent HDF, we examined the response of these cells to IGF-I and two IGF-I functional analogs, [QAYL]IGF-I and insulin, both of which have extremely low binding affinity for IGFBP-3 but which exert their mitogenic effect via the IGF-I plasma membrane receptor. Senescent HDF showed an increased sensitivity of DNA synthetic response to [QAYL]IGF-I and insulin compared to IGF-I. IGF binding activity was significantly higher in conditioned medium of senescent HDF than the medium of young HDF, and virtually all of this enhanced binding capacity could be accounted for by IGFBP-3. Addition of recombinant IGFBP-3 to young cells at a constant molar ratio of 1:1 with respect to IGF-I significantly attenuated the response to IGF-I and abolished the response at a 2:1 molar ratio. These data indicate that IGFBP-3 accumulated in medium of senescent HDF can bind and sequester IGFs when present in molar excess and thereby account for a significant part of the attenuated response of senescent HDF to IGF-I.

In order to analyze changes in metabolism of insulin-like growth factor binding proteins (IGFBPs) related to cell senescence and cell density, we compared human diploid fibroblasts (HDF) in the proliferatively vigorous first half (young cells) and senescent HDF in the last 10% (old cells) of the replicative lifespan after seeding cells over an eightfold range and proliferation to high density. Increasing the seeding cell density of both young and old HDF led to elevated rates of IGFBP-3 secretion, an increasing ratio of the 42/38 kDa species of IGFBP-3, and degradation of all species of IGFBPs derived from both the fetal bovine serum component of the culture medium and from HDF. At a given seeding density old HDF produced more IGFBP-3 and degraded more IGFBPs than young HDF. IGFBP-4 was degraded by a protease that appeared to be different from the protease(s) involved in degradation of the other IGFBPs. Young HDF at all seeding densities contained a cell-associated 29 kDa IGFBP, whereas this protein could not be detected in old cells. Thus, although certain changes in IGFBP metabolism are similar in young HDF seeded at high densities and in old HDF, young and old phenotypes can be distinguished by characteristic qualitative and quantitative changes in IGFBPs derived from fetal bovine serum and from HDF.

We previously reported that plasminogen activator inhibitor type‐1 (PAI‐1) mRNA was present at higher steady‐state levels in prematurely senescent fibroblasts derived from a subject with Werner syndrome (WS) compared to early passage (EP) fibroblasts from an age‐matched normal subject (Murano et al., 1991, Mol. Cell. Biol. 11:3905–3914). To explore the generality of this phenomenon with respect to chronological age of donor (in vivo aging) and the late‐passage (LP) or senescent phase of the fibroblast replicative lifespan, we assayed PAI‐1 mRNA in cells and PAI‐1 antigen in medium conditioned by 20 normal fibroblast strains at EP and LP and six WS strains during their curtailed replicative lifespans. The lowest accumulations of PAI‐1 were found in medium conditioned by fetal and newborn cells with a shallow but progressive rise seen in postnatal cells from normal donors of increasing chronological age. With few exceptions, normal LP fibroblasts showed increased PAI‐1 accumulations in medium compared to their EP counterparts. Conditioned medium from four of the six WS strains showed PAI‐1 accumulations that were significantly higher than the media of any normal controls at EP and LP. PAI‐1 mRNA levels were generally commensurate with the cumulative amount of PAI‐1 in the medium but the frequent exceptions indicate that translational and post‐translational mechanisms also regulate PAI‐1 output. The augmentation in PAI‐1 output of fibroblasts as a direct function of chronological age and during in vitro senescence suggests that PAI‐1 may play an important role in the reduced capacity for wound healing and the increasing tendency to thrombogenesis and atherogenesis seen during biological aging and in particular in persons with Werner syndrome. © 1994 Wiley‐Liss, Inc.

We have found that insulin-like growth factor binding protein-3 (IGFBP-3) accumulates to higher levels in medium conditioned by a strain of normal fibroblasts at late passage (LP) and a strain derived from subjects with Werner syndrome (WS) of premature aging, compared to medium conditioned by the same normal cells at early passage (EP) (Goldstein et al., Proc. Natl. Acad. Sci. USA, 88:9680-9684, 1991). To explore the generality of this phenomenon with respect to chronological age of donor (in vivo aging) and LP (in vitro senescence) we assayed IGFBP-3 in medium conditioned by 18 normal fibroblast strains at EP and LP and two WS strains at the midpoint of their curtailed replicative lifespans and assessed IGFBP-3 mRNA levels in cells by Northern analysis. The lowest accumulations of IGFBP-3 were found in medium conditioned by fetal cells with progressively increasing amounts postnatally; direct correlations between IGFBP-3 levels and donor age were seen in EP cells 3 days after subculture (during logarithmic growth) r = 0.80, P < 0.001, and 7 days after subculture (at confluence) r = 0.77, P < 0.001. With two exceptions, conditioned medium of cell strains accumulated more IGFBP-3 at LP; IGFBP-3 levels correlated with chronological age after 3 days, r = 0.50, P = 0.05, and after 7 days, r = 0.75, P < 0.001. IGFBP-3 content of WS culture medium fell within the range of LP normal cells. Cumulative IGFBP-3 levels were inversely proportional to the thymidine labeling index, a measure of proliferative vigor. With some exceptions IGFBP-3 mRNA levels were commensurate with the amount of IGFBP-3 accumulated in the medium, suggesting that distal translational and posttranslational mechanisms also regulate IGFBP-3 production in some strains. The trend toward augmented IGFBP-3 output of fibroblasts as a direct function of chronological age and in vitro senescence and as an inverse function of proliferative vigor is consistent with the known inhibitory effect of excess IGFBP-3 on IGF-mediated DNA synthesis and the reduced regenerative potential that is evident during biological aging in vivo. (C) 1993 Wiley-Liss, Inc.

Insulin-like growth factor binding protein 3 (IGFBP-3) mRNA levels were consistently higher in both senescent normal human diploid fibroblasts (HDFs) at late passage (old cells) and prematurely senescent HDFs from a subject with Werner syndrome (WS) during serum depletion and repletion of growth medium and during proliferation from sparse to high-density inhibited cultures, compared to normal early-passage (young) HDFs. However, IGFBP-3 protein accumulated to higher levels in conditioned medium of old cells than in medium of WS and young cells, in that order, under the same conditions. Insulin-like growth factor I (IGF-I) was not detected in naive medium or in any of the media conditioned by these three cell types, whereas IGF-II was detectable in serum-repleted medium and remained relatively constant. Thus, molar ratios of IGFBP-3/IGF-II were consistently higher in old and WS cells and increased substantially as all three cell types became quiescent, due to either serum depletion or high cell density. These data are consistent with either an adaptive or a causal role for IGFBP-3 protein in the senescent and quiescent growth arrest of HDFs.

Genes that play a role in the senescent arrest of cellular replication are likely to be overexpressed in human diploid fibroblasts (HDF) derived from subjects with Werner syndrome (WS) because these cells have a severely curtailed replicative life span. To identify some of these genes, a cDNA library was constructed from WS HDF after they had been serum depleted and repleted (5 days in medium containing 1% serum followed by 24 h in medium containing 20% serum). Differential screening of 7,500 colonies revealed 102 clones that hybridized preferentially with [32P]cDNA derived from RNA of WS cells compared with [32P]cDNA derived from normal HDF. Cross-hybridization and partial DNA sequence determination identified 18 independent gene sequences, 9 of them known and 9 unknown. The known genes included alpha 1(I) procollagen, alpha 2(I) procollagen, fibronectin, ferritin heavy chain, insulinlike growth factor-binding protein-3 (IGFBP-3), osteonectin, human tissue plasminogen activator inhibitor type I, thrombospondin, and alpha B-crystallin. The nine unknown clones included two novel gene sequences and seven additional sequences that contained both novel segments and the Alu class of repetitive short interspersed nuclear elements; five of these seven Alu+ clones also contained the long interpersed nuclear element I (KpnI) family of repetitive elements. Northern (RNA) analysis, using the 18 sequences as probes, showed higher levels of these mRNAs in WS HDF than in normal HDF. Five selected mRNAs studied in greater detail [alpha 1(I) procollagen, fibronectin, insulinlike growth factor-binding protein-3, WS3-10, and WS9-14] showed higher mRNA levels in both WS and late-passage normal HDF than in early-passage normal HDF at various intervals following serum depletion/repletion and after subculture and growth from sparse to high-density confluent arrest. These results indicate that senescence of both WS and normal HDF is accompanied by overexpression of similar sets of diverse genes which may play a role in the senescent arrest of cellular replication and in the genesis of WS, normal biological aging, and attendant diseases.