Mitotic Misregulation and Human Aging

See allHide authors and affiliations

Science  31 Mar 2000:
Vol. 287, Issue 5462, pp. 2486-2492
DOI: 10.1126/science.287.5462.2486


Messenger RNA levels were measured in actively dividing fibroblasts isolated from young, middle-age, and old-age humans and humans with progeria, a rare genetic disorder characterized by accelerated aging. Genes whose expression is associated with age-related phenotypes and diseases were identified. The data also suggest that an underlying mechanism of the aging process involves increasing errors in the mitotic machinery of dividing cells in the postreproductive stage of life. We propose that this dysfunction leads to chromosomal pathologies that result in misregulation of genes involved in the aging process.

The question of why we age has intrigued mankind since the beginning of time. Extensive studies of model systems including yeast, Caenorhabditis elegans,Drosophila, and mice as well as studies of human progerias and cellular senescence have identified a number of processes thought to contribute to the aging phenotype (1). These include the effects of oxidative damage associated with cellular metabolism and genome instabilities such as telomere shortening, mitochondrial mutations, and chromosomal pathologies. To gain greater insights into the mechanisms that control life-span and age-related phenotypes, we have studied gene regulation of normal and premature aging in actively dividing cells. Studies of fibroblasts derived from young, middle-age, and old-age humans and from humans with Hutchinson-Gilford progeria revealed sets of genes that correlate with, and hence likely contribute to, age-related phenotypes and diseases. The results also suggest that mitotic errors in dividing cells may lead to the altered expression of this collection of genes.

Ten closely matched dermal fibroblast cell lines were classified into four categories based on their chronological and diagnostic similarities: normal young (NY), normal middle (NM), normal old (NO), and Hutchinson-Gilford progeria (P) (2). Collectively, NY, NM, and NO samples allow examination of the expression levels of various genes throughout the natural aging process. Hutchinson-Gilford progeria, on the other hand, is a rare genetic disease in which those affected display at a very early age features typically associated with natural old age, including loss of or graying of hair, diminished subcutaneous fat, cardiovascular disease, and skeletal abnormalities (3). Actively dividing early passage fibroblasts from each age group were initially examined by phase contrast and fluorescence microscopy to characterize age-related morphological changes (2). For each group, the elliptical morphology characteristic of fibroblast nuclei was observed in the majority of cells. However, in contrast to fibroblasts from NY and NM individuals, whose nuclei appeared normal, the NO and P groups had a significant proportion of cells exhibiting aberrant nuclear morphology including multilobed nuclei and irregular nuclear boundaries. The NO and P populations also had a higher proportion of cells with multiple nuclei, consistent with reports of age-dependent increases in micronucleation in human lymphocytes (4). Flow-activated cell sorting (FACS) analysis of the same fibroblasts from the NY and NM groups revealed very similar populations of cells with 2N, S, and 4N DNA content. However, NO and P fibroblasts showed higher percentages of 4N DNA content (Fig. 1), consistent with the larger number of binucleated cells observed by microscopy.

Figure 1

Structural composition and experimental electron density of E. coli DnaG-RNAP. (A) Schematic diagram illustrating the domain boundaries of full-lengthE. coli DnaG (9, 10). Gold, blue, and purple color coding on the DnaG-RNAP domain correspond to NH2-terminal (residues 115 to 240), central (residues 241 to 367), and COOH-terminal (residues 368 to 428) subdomains in the structure, respectively. (B) Representative, solvent-flattened experimental electron density of three β strands within the toprim region of DnaG-RNAP. Contouring is at 1.4σ above the mean. The refined model is shown as a ball-and-stick representation. The figure was generated by Ribbons (27). (C) Secondary structure and conservation in DnaG-RNAP. The DnaG-RNAP sequence is highlighted to illustrate the positions of invariant (green boxes) and highly conserved (yellow boxes) residues as well as the locations of sequence motifs II to VI (19) (gray boxes). Sequence conservation was determined by comparing 28 bacterial primase proteins with ClustalX (28); “invariant” and “highly conserved” residues are defined by the Gonnet 250 weighting scheme (29). Secondary structure content of each subdomain is indicated below the sequence as cylinders (α helices), arrows (β strands), or lines (coil) and is color-coded as in (A). Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

To examine the transcriptional profiles (mRNA abundance) of these fibroblasts, asynchronous and actively dividing cells were cultured in vitro to about 60% confluency (2,5). Messenger RNA levels were analyzed with high-density oligonucleotide arrays containing probes for more than 6000 known human genes. Expression patterns for each age group (NM, NO, and P) were compared with NY fibroblasts (baseline). Only those changes that were reproducible across all comparisons and all independent replicates were considered further (Fig. 2). On the basis of these conservative criteria, we found that 61 genes (∼1% of the genes monitored) showed consistent expression level changes more than twofold between young and middle age. More than half of these 61 genes can be grouped into two functional classes, (i) genes whose products are involved in cell cycle progression (25%) (Table 1) and (ii) genes involved in maintenance and remodeling of the extracellular matrix (ECM) (31%) (Table 2). The first group of genes are all involved in mitosis and are down-regulated between 2.6- and 12.5-fold. These include cyclins A, B, and F, which are typically up-regulated at the G2-M cell cycle phase transition and control mitotic progression (6); polo kinase (pLK), which regulates both spindle assembly and cyclin-dependent kinase 1–cyclin B (Cdk1-cyclin B) activation (7,8); and p55CDC, which participates in the mitotic checkpoint (9). In addition, a group of proteins involved in spindle assembly and chromosome segregation are down-regulated, including the centromere-associated proteins CENP-A (10), mitosin (CENP-F) (11), and histone H2A.X (12), as well as the kinesin-related proteins, mitotic centromere-associated kinesin (MCAK) (13), mitotic kinesin-like protein-1 (14), and kinesin-like spindle protein (HKSP) (15). Transcription factors associated with the G2-M transition, including B-myb (16, 17) and hepatocyte nuclear factor–3/fork head homolog (HFH-11A) (18), are also down-regulated. Myb is known to regulate cellular proliferation and differentiation, and HFH-11A is a homolog of daf-16, a protein that has been shown to regulate key metabolic and developmental genes and plays a role in regulating life-span in C. elegans(17).

Figure 2

Structure of DnaG-RNAP. (A) Ribbon diagram of the DnaG-RNAP crystal structure, color-coded as inFig. 1. The toprim region resides in the central (blue) subdomain. Secondary structural elements are labeled as in Fig. 1C and were determined according to standard parameters (30). The figure was generated by Ribbons (27). (B) View of (A) showing the surface potential of DnaG-RNAP. Positive charge potential (+7 k B T/e) is shown in blue and negative potential (−7k B T/e) is shown in red. Acidic, metal-binding residues from the toprim domain and a basic ridge presented by the NH2-terminal subdomain form the mouth of the putative nucleic acid–binding cleft. The conserved basic depression is also indicated. The figure was generated by GRASP (31). (Inset) Side view of (B) rotated by 90°.

Table 1

A comparison of changes in gene expression in middle age, old age, and progeria versus young. The magnitude of the changes reported was computed as the average values over the set of comparisons, where the highest and lowest values were excluded. Colors indicate genes that are common across age groups. Red, genes common across all three age groups; brown, genes common in middle and old age; blue, genes common in old age and progeria; black, genes unique to an age group.

View this table:
Table 2

A comparison of changes in gene expression in middle age, old age, and progeria versus young. Methods and colors are as in Table 1.

View this table:

A significant number of genes involved in remodeling of the ECM also show altered expression. For example, human macrophage metalloproteinase (HME) and stromelysin 2 are increased 24- and 14-fold, respectively, whereas the protease inhibitors urokinase inhibitor (PAI-2) and cystatin M increase five- to ninefold. A number of proteoglycan cell adhesion proteins are also up-regulated, including dermatopontin, fibromodulin, and thrombospondin, as well as collagens VI and XV and cartilage oligomeric matrix protein (COMP). The preponderance of differentially expressed mitotic and ECM genes compared with genes involved in other cellular processes likely reflects some level of coordinated regulation between these classes of genes (19). The remainder of genes whose expression levels differ consistently by twofold or more include key enzymes involved in the conversion of arachidonic acids to the prostaglandins and thromboxane, such as prostaglandin endoperoxidase synthase 2, endoperoxide synthase type II, cyclooxygenase 2, and endoperoxidase synthase (5- to 30-fold changes). Increasing levels of these enzymes with age may affect a large number of systemic physiological processes, including platelet aggregation, muscle and kidney function, bone formation, and various inflammatory processes.

Analysis of gene expression in fibroblasts isolated from old versus young individuals again revealed a down-regulation of genes involved in the G2-M phase of the cell cycle (Table 1). Some of these same cell cycle genes that were down-regulated in middle age versus young cells are more significantly down-regulated in old-age individuals. In addition, other cell cycle genes are affected. For example, the cell cycle proteins FRAP-related protein [FRP1, a homolog of ataxia-telangiectasia mutated (ATM) and its related protein (ATR)], CKS1, and Myt1 are down-regulated three- to ninefold; all are involved in the G2-M DNA dependent checkpoint pathway, which ensures fidelity before entry into mitosis (20). CDC25B, a key protein triggering entry into mitosis, is also down-regulated. Expression of some of these genes may be controlled by common cell cycle transcriptional control elements (21). There is also an increase in the number of down-regulated genes involved in spindle assembly and segregation, including CENP-E, HMG-2, Ran BP1 (also involved in nuclear transport and regulation of polyadenylation), and scaffold attachment factor. DNA synthesis and repair genes {e.g., thymidylate synthase, minichromosome maintenance protein, Rad 2, poly(adenosine diphosphate–ribose) [poly(ADP-ribose)] synthetase, and proliferating cell nuclear antigen} and RNA synthesis and processing genes (RNA helicases) are also down-regulated. The importance of the down-regulation of some of these repair genes has been clearly demonstrated. For example, in the case of poly(ADP-ribose) synthetase (also known as a polymerase, or PARP), partial inhibition by antisense agents or complete removal by gene knockout results in increased DNA strand breaks, recombination, gene amplification, and aneuploidy (22). Decreased expression levels are also found for proteasomal proteins, which are a core component of the anaphase-promoting complex (APC) and are responsible for the timing of the metaphase to anaphase transition (the final stage of cell division) and other key events in the cell cycle (Table 1). A significant number of genes whose products affect the ECM are again differentially expressed in fibroblasts from old-age individuals; however, the expression levels are less pronounced when compared with that of NM and NY fibroblasts. Genes involved in arachidonic acid oxidation continue to be affected, as do genes involved in lipid transport and metabolism.

Other changes in gene expression occur in old age that are not seen in middle age. For example, genes involved in the stress response and heat shock proteins are up-regulated, including αβ-crystallin, which is thought to be involved in maintenance of the cellular cytoskeleton (23), and a heat shock serine protease human HtrA, which is overexpressed in osteoarthritic cartilage. Mitochondrial genes are down-regulated, which is consistent with age-related mitochondrial dysfunction (1). Similar changes in a number of stress-response and metabolic genes were observed in expression profiles of aged versus young skeletal muscle from mouse (24). Transcript levels of genes involved in oncogenesis, signal transduction, and the cellular immune response also change and may be linked with age-related diseases.

A comparison of mRNA transcript profiles between progeria and normal young cells reveals a consistent and significant change in the expression of 76 genes. Many of these genes overlap with those found in the comparison of young and old-age samples (Tables 1 and 2). In particular, genes involved in cell division and DNA or RNA synthesis and processing are commonly down-regulated in NO and P. ECM-related genes continue to represent a significant fraction of those genes whose expression levels change. Particularly in progeria, one observes changes in expression of the caldesmons, which are actin-binding proteins involved in cell cycle–dependent reorganization of the cytoskeleton (25); desmoplakin I, which plays a role in intracellular adhesive junctions (26); and autotaxins, which are extracellular phosphodiesterases involved in cellular chemotaxis. Transforming growth factor–β (TGF-β) expression also increases 12-fold, consistent with changes in expression of the large number of genes that affect the ECM (27). Genes involved in fatty acid transport and oxidation also vary in similar ways, as with old age and progeria.

Comparison of the genes whose transcript levels change as a function of natural and premature aging reveals classes of genes that can be linked to aging-related phenotypes and diseases. For example, osteoblast (OB)-specific factor 2 (osteoprotegerin) (28) and OB-cadherin (29) play key roles as transcriptional activator and adhesion molecules in bone formation, respectively. Their down-regulation in old age may be linked to bone diseases such as osteoporosis. HME (30) is up-regulated with age and has been shown to be associated with joint destruction in rheumatoid arthritis. Down-regulation of the hyaluronic acid (HA) synthase may also contribute to joint disease; depletion of HA is observed in early experimental osteoarthritis in dogs (31). Cartilage oligomeric matrix protein is significantly up-regulated in old age and has been shown to correlate with disease activity in rheumatoid arthritis and is also arthritogenic when expressed in rat cartilage (32). Cathepsin C, an oligomeric lysosomal protease whose loss of function results in periodontal disease and palmoplantar keratosis, is down-regulated 10-fold in old age (33). In general, the altered expression of a large number of genes that influence the ECM may contribute to age-related changes in the derma.

Expression of other genes possibly linked to age-related diseases is also observed. The breast cancer susceptibility protein–1 (BRCA-1) associated Ring domain protein (BARD1), which binds the NH2-terminus of BRCA-1, a protein implicated in DNA repair and cell cycle checkpoint regulation, is down-regulated in old-age and progeria cells (34). Down-regulation of BARD1 may deleteriously affect BRCA-1 function and may be linked with age-related sporadic breast cancer; mutations of BARD1 have been identified in breast, ovarian, and uterine cancers (35). The hFRP-1 gene, which is also down-regulated in old age, is a homolog of the geneATM mutated in ataxia-telangiecstasia (A-T). A-T is an autosomal recessive disorder characterized by progressive neurodegeneration, immune deficiencies, premature aging, chromosomal instability, and radiation sensitivity (36). ATM, like BRCA-1, plays a key role in the cellular response to DNA breaks, including activation of cell cycle checkpoints and DNA repair. Down-regulation of genes involved in these pathways would result in an increase in genetic instability and sensitivity to reactive oxygen species. αβ-crystallin expression increases with old age (37). It is also overexpressed in a number of neurological disorders such as Alzheimer's disease, Diffuse Lewy Body disease, and Alexander's disease, suggesting a possible link between this protein and age-related neurologic disease. αβ-Crystallin has also been implicated in the formation of age-related nuclear cataracts (38). The amyloid precursor protein (APP)–binding protein, which binds at the COOH-terminal of APP, the proteolysis site in the generation of β amyloid peptides, is down-regulated with age (39). TGF-β, a key growth factor that regulates tissue homeostasis and whose sustained expression is responsible for tissue fibrosis, is highly up-regulated in progeria, consistent with the biopsies from the progeria patients. COX-2 expression is down-regulated in NM, NO, and P samples. COX-2 knockout mice exhibit abnormalities in the kidney, heart, and ovaries that result in renal dysplasia, cardiac fibrosis, cancer, gastric insufficiency, and female infertility—all of which are related to aging (40). Thus, analysis of gene expression across a larger collection of human genes and in a number of different tissues specifically affected by these and other age-related diseases may allow the identification of key genes associated with diseases of aging, which may provide potential points for therapeutic intervention. Clearly, it will be of interest to carry out similar studies with cell types associated with specific diseases such as breast epithelial cells.

In addition to identifying genes that contribute to the aging phenotype, the analysis of gene expression in natural aging and progeria may also provide insights into the underlying mechanisms of the aging process. A comparison of cells from middle- and old-age humans reveals a common set of genes with altered expression levels. These genes are principally involved in the G2-M phase of the cell cycle and in remodeling the ECM; they are likely linked through changes in the cellular cytoskeleton that occur during cell division. A comparison of gene expression in old age and progeria also reveals disregulation of many of the same genes, as well as additional genes involved in DNA or RNA synthesis and processing. The large number of mitosis-related genes that are down-regulated in middle-age, old-age, and progeria fibroblasts does not simply reflect altered cell cycle populations resulting from differential growth rates. This is supported by the fact that the G0-G1 and G2-M populations in young and middle age are virtually identical and that the cells also grow at the same rate. In old-age and progeria cells, the 4N DNA content is even higher, reflecting either a larger number of cells in G2-M or an increased incidence of tetraploidy. The genes whose expression is altered in middle-age, old-age, and progeria also do not correspond with those observed in cellular senescence (41). Indeed, a recent re-examination of fibroblast culture replicative life-span does not show a correlation with donor age (42). In addition, the transcript profiles described here do not resemble those of quiescent (19), contact-inhibited (43), or G1-arrested cell populations (44), nor do the changes in gene expression observed here correspond to those observed in the aged hypothalamus (43) or skeletal muscle (24). In the latter case, a marked stress response was observed along with a lower expression of metabolic and biosynthesis genes. Although a number of the same changes were observed here, the majority of changes we observe in cell cycle, ECM, fatty acid oxidation, and disease-related genes were not observed in muscle or hypothalamus. The altered expression of these genes observed at middle age and elaborated in old age and progeria are likely specific to mitotic versus postmitotic cells.

We suggest that an altered expression of genes involved in cell division occurs with age. These changes result in increased rates of somatic mutation, leading to numerical and structural chromosome aberrations and mutations that manifest themselves as an aging phenotype. Previous studies have demonstrated an increase in aneuploidy with increased age (45), and down-regulation of mitotic genes has been shown to lead to aneuploidy in experimental models. For example, both a motorless mitotic centromere-associated kinesin (MCAK) and antisense inhibition of MCAK lead to chromosome lagging during anaphase (13). It has also been argued that mutations in presenilin 1 and 2, which are associated with both the interphase kinetochore and centrosome and account for most early onset familial Alzheimer's disease, may result in chromosome pathologies (46). Aneuploidy associated with chromosome 21 is involved in Down syndrome, a disease characterized by some features of premature aging. Misregulation of genes involved in cell division may be the result of an intrinsic lack of fidelity that arises in the absence of selection in the postreproductive stage. Alternatively, the growing loss of fidelity may result from the cumulative effects of oxidative damage associated with metabolism, which are slowed by caloric restriction (24). In fact, there may be multiple entry points into this process. For example, Werner syndrome, which is characterized by the premature appearance of aging in young adults (47), shows an increased rate of chromosomal abnormalities caused by mutations in a DNA helicase or exonuclease enzyme known as WRN.

Chromosome pathologies that begin to occur in dividing cells relatively early in life (postreproductive stage) may then lead to misregulation of key structural, signaling, and metabolic genes associated with the aging phenotype, such as osteoporosis, Alzheimer's disease, arthritis, and so forth. Misregulation of this sort is expected to increase in each round of cell division. It may be propagated to other normal mitotic (e.g., leukocytes, epithelial cells, glial cells, and so forth) and postmitotic (e.g., neurons, muscles, and so forth) cells through changes in the ECM and oxidized fatty acid derivatives that affect signaling pathways. Aging, therefore, may occur gradually and in mosaic patterns, rather than as a uniform phenomenon as in cancerous growth, which is clonal. Additional studies are required before we can understand the aging process in complex organisms, both in mitotic and postmitotic tissue, but the studies reported here highlight important mechanisms that may contribute to aging and age-related problems.


View Abstract

Navigate This Article