Biological research on aging is relatively a new field, at least in comparison to research on development. Nevertheless, its progress has been rapid, particularly during the last two decades, due to the utilization of sophisticated techniques of biochemistry, biophysics, molecular biology, and genetic engineering. However, aging is an enormously complex problem, as one has to examine a complex organization to find a few basic changes that cause aging. It is like looking all over a forest to locate a few worms that have started destroying the trees. In order to understand development, one proceeds from a relatively simple organization, step by step, to a complex organization. Development is like planting trees, both in space and time, to grow a forest. One can at least follow the steps. Though all the key steps of development are not yet deciphered, a few steps are beginning to be understood at the molecular level in certain organisms, such as how homeotic genes control segmentation in animals, how myoD gene regulates differentiation of skeletal muscle cell in vertebrates and how ced-3 and ced-4 genes cause the death of certain predetermined cells during the development of Caenorhabditis elegans.

During the past decade we have learned that the expression of several genes decreases after the attainment of reproductive maturity. The cause of such changes in expression of genes is being studied by analyzing specific sequences in their promoter regions, such as the DH-sites, methylation of cytosine and cis-acting elements, and the trans-acting factors that bind to these elements and modulate their expression.

In multicellular organisms, the problems of transmitting information stored in the DNA from cell to cell and from generation to generation are enormous. A diploid (2N) human cell has about 7 picograms (pg) of DNA which is distributed among 23 pairs of chromosomes. The total length of the DNA of 23 pairs of chromosomes is about 2 meters. Therefore, it has to be highly compacted in order to be accommodated within a nucleus whose diameter may range from 5 to 8 μ. In the highly condensed metaphase chromosomes of a liver cell, the DNA is compacted about 10,000-fold (Pienta, Getzenberg & Coffey, 1991). Approximately 3 × 109 base pairs (bp) of DNA are distributed among 23 chromosomes. Only 50,000 to 100,000 genes that are coded by nearly 5 × 107 bp of DNA or 3%–5% of the total DNA are estimated to be functional. So nearly 95%–97% of the DNA in a human cell has no known function.

The DNA in eukaryote chromosomes exists as a complex, unlike that in prokaryotes, mitochondria, and plastids, with two types of chromosomal proteins, histones and nonhistone chromosomal (NHC) proteins, forming a supramolecular complex called chromatin, the genetic apparatus of eukaryotes. The three components – DNA, histones, and NHC proteins – are present approximately in equal proportions. Various types of RNA are also found in the chromatin, but they occur as transcription products of DNA and are not structural components of the chromatin.

The genome in a eukaryote houses far more DNA and genes than it needs for its various functions. This is the inference one draws if one counts the various types of enzymes, structural proteins, mRNAs, tRNAs, and rRNAs that are coded by the genome. Only 3%–5% of the genome accounts for these molecules. Of the remaining, nearly 40% comprise repetitive DNA whose exact function is not known. What the rest of the DNA does is not known. Assuming that it is only the expressed fraction of the DNA or genes that plays a role in the aging process or holds the key to the aging process of an organism, one may look for changes that occur in this fraction during its life span, and hope to find a common pattern of changes in specific sets of genes. Such a finding may give an insight into the possible causes of aging since genes have been implicated in the following fixed-time life processes of mammals: gestation, attainment of maturity and growth completion, duration of reproduction and fertility, life span, and body size. These times may vary with the species of mammals. The variations seen in these characteristics among individuals within a species have been attributed to various extrinsic factors such as nutrition, temperature, radiation, and stress.

Biochemical research on aging commenced in the early 1950s. Until about 1970 the research was confined to mostly studies on changes in enzymes and structural proteins. It had two purposes: First, since enzymes catalyze all functions of the body, an understanding of their changes during the life span may throw light on the aging process.

The shortening of the three types of transcripts is necessary because their corresponding genes contain sequences that are not required in their final products. The RNA polymerase transcribes the entire gene into a pre-RNA, which is as long as the gene. The sequences that are not required in the final product alternate with those that are required.

Aging as a phenomenon in the life spans of organisms has intrigued mankind from time immemorial. Why and how is it that having attained a vigorous adulthood, all functions should undergo decay? The duration of this phase, which is referred to as aging or senescence, varies with species. It can be as short as a few days as in the female octopus which lays eggs only once, broods them, reduces its food intake, and dies soon after her young hatch. Among the so-called marsupial mice of Australia, the males live for only about a year. When they approach the end of their lives, they stop eating and engage in a competitive, brief but frantic mating. All males die shortly thereafter, perhaps due to hormonally induced stress. The females live long enough to suckle their young and wean them. Very few females live long enough to breed again. The female Pacific salmon also lays eggs once, and then dies soon after. These are sort of “sudden death” phenomena that occur soon after one-time reproduction, and the period of aging is too brief to be perceptible in these organisms.

In most species, however, such a phenomenon is not seen even when a large number of eggs or young are produced. Mice and rats give birth to a large number of young, take care of them during their weaning period, and are ready to breed again soon after. In higher mammals such as humans and elephants, only a few young are produced during the entire life span with long gaps, sufficient to take care of the young during the crucial early developmental period.

All multicellular organisms have two striking characteristics: (1) they show a gradual decline in their adaptability to the normal environment after attaining reproductive maturity, and (2) all members of a species have a more or less fixed life span. These two characteristics are inherited, and hence genetically controlled. Rats, mice, and Drosophila, kept under controlled environmental conditions, live for periods characteristic of the species and strain.

Besides these characteristics, it is also known that the time required for all members of a species to reach reproductive maturity is the same, as for example, 10 weeks for rats and 12 years for humans. Likewise, the reproductive period of all individuals of a species is more or less the same. The various stages of development leading to reproductive maturity are precisely timed. Thus, it is likely that developmental changes up to the attainment of reproductive maturity are genetically controlled and occur according to a genetic program.

Medvedev (1990) has brought together all theories on aging, classified them, and made an analysis of each group of theories. We shall consider here only those theories that attempt to explain aging at the genetic level. These theories are based on genes as the primary sites at which changes occur to initiate the process of aging. Factors such as food, temperature, humidity, radiation, pollution, and various stresses, however, influence the rate of aging.

Before discussing these theories, certain characteristics of aging that point to genetic involvement need to be considered.

Some of our basic ideas about genes and their transmission have been established for nearly a century. One of these is Mendel's Law of Segregation, from which we derived the transmission probabilities (τs) given in Chapter 5. However, as noted earlier in this book, some investigators estimate these probabilities from the general likelihood in families rather than specifying them, and the estimates are sometimes significantly different from the mendelian values of 0, ½ and 1. A second basic principle of classical genetics is that genes are faithfully transmitted during mitosis, so that all of our somatic cells have the original inherited genotype. The process of somatic rearrangement in immunoglobulin and TcR genes shows that this postulate, also, is not always correct.

This chapter shows various other ways in which violations of these classical assumptions have important effects on the pattern of genotypes in families and the relationships between genotype and phenotype. The term non-traditional inheritance is often used for such phenomena {Hall, 1990a,b; Holliday, 1987, 1989; Solter, 1988}, but the mechanisms have always been there whether part of our ‘traditions’ or not!

Contrary to the verse by the great population geneticist J. B. S. Haldane, written just after being treated for the cancer that was to take his life, his was among the first generations in which a lot of chaps were bumped off by cancer. Previously most people died from other causes, such as infectious diseases. Although cancers are genetic diseases, selection has prevented most cancers from being directly heritable, but cancer demonstrates the important effects that somatic mutations can have. Cancer reflects evolution in a microcosm, occurring among our own cells during our own lifetime rather than among the individuals in a population over evolutionary time.

Cancer age patterns reflect the time required for somatic mutations to produce their complex cellular phenotypes. This leads naturally to a general consideration of the age patterns of chronic disease and, consequently, of aging in general. These are the subjects of this chapter.

Somatic mutation: the genotype changes with age

An individual begins life as a single cell, but an adult organism is the product of a very large number of subsequent cell divisions that occur as the zygote grows, develops, and renews its tissues. Somatic mutations occur in these somatic cells, and are inherited by their mitotic descendents during the life of the individual.

How can the almost unimaginable amount of genetic variation, arising in an almost unimaginably variable environmental context, be related to specific phenotypes? This chapter extends the concepts of frequency and association developed in the previous chapter, to provide the basic concepts needed for a genetic model of how genes may affect a trait and how they act. Such models take advantage of the special constraints that billions of years of evolution have placed on traits controlled by genes.

Frequency concepts for genetic traits

The prevalence of a genetically related trait in a population depends on the amount of genetic variation that affects it.

Allele and genotype frequencies

The most fundamental quantitative variable in population genetics is the allele frequency (often carelessly called the ‘gene’ frequency), a prevalence measure. If, among the 2N copies of a given gene in a population of N diploid individuals, ni are allele i, then the frequency of that allele is defined as pi = ni/2N. There is no theoretical restriction on the number of alleles that can exist at a locus, but their frequencies must sum to 1. Thus, if one allele is very common, others must be correspondingly rare.

Diverse forms of allometry

Allometry is a term that refers to the quantitative scaling relationships between biological traits, such as between metabolic rate and body size among species. The term can be extended to refer to systematic dose – response relationships between genes, environments, and phenotypes – that is, to risk factors (covariates or concomitants) that determine the penetrance (G → P) function.

A challenge to genetic inference is to identify those situations that magnify or clarify the phenotypic differences among genotypes. Exposure to environmental risk factors often leads to phenotype amplification; that is, to phenotypes that become increasingly divergent, and easier to identify, with exposure dose and duration (or age). Other factors, however, may obscure the effects of different genotypes. We saw in Chapter 6 how to include unmeasured risk factors, such as the ‘environmental’ variance, in terms of their aggregate effect on phenotypic variance. This chapter is concerned with the effects of those risk factors that we can identify and measure on individuals.

We usually have no specific biological model for the dose–response relationships of such variables, and usually parameterize those relationships with general epidemiological models such as those given in Chapter 3.

The complex map from genotype to phenotype

The complexities and levels at which genetic and environmental factors can affect a subsequent outcome phenotype are illustrated schematically in Figure 12.1 for the case of CHD; for example, cholesterol is one component of the atherosclerotic plaques that may lead to clotting reactions producing heart attacks.

Segregation analysis provides evidence that a disease has a genetic causation, but this is only the beginning, since our real goal is to find the genes involved. An important objective is to map the genes to their chromosomal location.

Modern molecular biology provides a rapidly increasing repertoire of methods for the physical mapping of genes. Genes can be isolated and cloned using many different techniques (e.g. see Appendix 1.2), and variants can be related to function and physiology both in experimental systems and in natural settings. Physical mapping will probably become the overwhelming method of choice for identifying human genes.

That day is not yet here, however, and we currently still rely on statistical epidemiological methods. In some cases simple association studies will suffice, but usually we rely on linkage mapping methods. These take advantage of the fact that evolution has produced a genome consisting of genes concatenated on chromosomes. Probably we owe this arrangement to evolution by gene duplication and to the need to coordinate regulation {De Duve, 1991}. In any case, linkage mapping methods search for association relationships that are due to linkage, between genes that have already been mapped and genetically controlled phenotypes whose genes we have not yet identified. The bible on linkage mapping methods is the text by Ott {1991} {Conneally and Rivas, 1980; Lathrop and Lalouel, 1984; White and Lalouel, 1987}. These methods are so powerful that scarcely a week passes without their being used to map some new gene.

It may seem peculiar that genetic diseases, being deleterious and presumably selected against, exist at all. Yet, a number of genetic diseases, especially severely deleterious recessive conditions, have substantial incidence at birth. This chapter and the next illustrate the impact of evolution on the frequency and distribution of genetic disease.

What is responsible for the observed frequency of disease?

How often do mutations introduce new copies of diseaseproducing alleles?

A classical problem in human genetics has been to estimate how often new mutants leading to a given disease are produced. Interestingly, although the loci responsible vary tremendously in length, physiology, etc., a large number of studies, using very different methods, have reached similar estimates. For alleles related to both recessive and dominant qualitative traits, new mutations occur about every 100 000 meioses (i.e., μ ≈ 10-5) {Vogel and Motulsky, 1986}. The range of these estimates is about 100-fold, which roughly encompasses the range of size of coding regions of the genes involved, so that overall the rate per nucleotide per generation is consistently around 10-7 to 10-9.

What maintains the frequency of disease-related alleles?

Other important questions, perhaps not asked as often as they should be, are: what factors maintain the observed frequency of alleles that are deleterious? is it selection? do the alleles affect fitness?

How common is a phenotype in a population? How often does it arise? This chapter introduces some basic concepts of the impact or ‘frequency’ of phenotypes and of the relationship between that and the frequency of genes which may help to produce them.

Basic concepts of trait frequency and risk

Static measures: prevalence and its use as a probability

We can denote by φ the phenotype that we observe on an individual. The prevalence, or frequency, Pr(φ), of a specific phenotype at any given time is the fraction of individuals in a specified population who have the trait, equal to the probability that a randomly sampled individual has that phenotype. A phenotype can usually take on a range of possible values, and over this range Pr(φ) gives the phenotype distribution. Since every person must have exactly one phenotype, the prevalences over the acceptable range must sum to unity.

A phenotype can be a continuous measure such as the level of blood glucose, for which we can write the prevalence of a given glucose level, for example, as Pr(100 mg/dl). Or, a phenotype can be a qualitative trait such as the presence or absence of a disease, which we can denote φ = 1 (presence) and φ = 0 (absence). The term prevalence usually refers to the fraction of individuals who have the trait and is denoted Pr(φ = 1), or when the context is unambiguous just by Prev.