Inflammation and Oxidation in Aging and Chronic Diseases
Publisher Summary
This chapter provides an overview of experimental models for human aging and age-related diseases. In most chronic diseases of aging, oxidative stress and inflammation are prominent. Moreover, many tissues without specific pathology that show modest inflammatory changes during aging share major subsets of those in chronic diseases of aging. An overview of inflammation and oxidant stress in host defense suggests a classification of bystander damage. Mechanisms of inflammation are outlined, including the energy costs. This chapter also reviews in more detail inflammation in arterial and Alzheimer disease. These details are critical to understanding human aging and the role of insulin-like metabolic pathways. Many inflammatory processes emerge during “normal” or usual aging, but in the absence of specific diseases. The slow creep of inflammation from early years may drive the accelerating incidence of chronic diseases.
1.1 OVERVIEW
Human life spans may have evolved in two stages (Fig.1.1A). In the distant past, the life expectancy doubled from the 20 years of the great ape-human ancestor during the evolution of Homo sapiens to about 40 years. Then, since the 18th century, life expectancy has doubled again to 80 years in health-rich modern populations, with major increases in the post-reproductive ages (Fig. 1.1B) and decreases in early mortality (Fig. 1.1C). During these huge demographic shifts, human ancestors made two other major transitions. The diet changed from the plant-based diets of great apes to the high-level meat-eating and omnivory that characterizes humans. Moreover, exposure to infections increased. The great apes abandon their night nests each day and rarely congregate closely for very long in large groups. As group density and sedentism increased in our ancestors, so would their burden of infections and inflammation have increased from exposure to pathogens in raw animal tissues and from human excreta.
I propose that the growth of meat-eating and sedentism selected for gene variants adaptive in host defense and adaptive for high fat intake. Some of these genes may have favored the increased survival to later ages that enables the uniquely human multi-generational caregiving and mentoring. Many such genetic changes had probably evolved by the time of the Venus of Willendorf (cover photograph), 21,000 years ago in the Upper Paleolithic. Her manifest obesity may be viewed as adaptive in times of fluctuating food, with few ill consequences during the short lifespans of the pre-modern era, at the least, fewer than in the modern era of rampant chronic obesity. However, the most recent and rapid increases in life span cannot be due to the natural selection of genes for greater longevity.
I emphasize the plural lifespans, because many concurrent human life history schedules can be recognized in the world today that differ by the rate of growth, age of puberty and sexual maturation, the schedule of reproduction, and life expectancy. Evolutionary biologists recognize the huge plasticity of life history schedules, which vary between populations and respond rapidly to natural or artificial selection (Section 1.2.8). In the not too distant past, human life histories and lifespans may have been outcomes of natural selection, whereas changes in the last 200 years are clearly driven by culture and technology.
I propose that the evolution of the human life span depended on the genetic modulation of synergies between inflammation and nutrition. These dyadic synergies are both substrates and drivers of specific chronic diseases and dysfunctions (Fig. 1.2A). Many aspects of aging are accelerated by infections and inflammation, while drugs and nutritional interventions that slow aging may act by attenuating inflammation and oxidative damage. The current lab models selected for fecundity in atypically clean environments with unlimited food and no stress from predators may not represent aging processes in the bloody, dirty, invasive, and stingy environment of natural selection. Host defense and somatic repair processes are evolved to survive the relentless assaults by microorganisms, parasites, and other predators that are omnipresent in the natural environment. Understanding gene-environment (G x E) interactions in the inflammation-nutrition synergies is fundamental to human aging, past, present, and future. No single gene or mechanism is likely to explain human aging and its evolution, because natural selection acts mainly through successions of small quantitative gene effects. Many gene variants show trade-offs in balancing selection, epitomized by the sickle-cell gene in resistance. A broad theory of aging may emerge by mapping the nutrition-inflammation synergies of pathological aging changes (Fig. 1.2) and their role in oxidative damage.
Because host defense and repair require energy, homeostatic energy allocation strategies were evolved for eco-specific contingencies. High infectious burdens and poor nutrition attenuate somatic repair and growth (Fig. 1.2B). Homeostatic resource allocation involves insulin-like metabolic pathways that operate throughout development and adult life (Fig. 1.3). Insulin-like signaling pathways were recently shown to influence aging in many species. Many aspects of aging at the molecular and cell level can be attributed to ‘bystander’ damage from locally generated free radicals in the immediate microenvironment. DNA, lipids, and proteins are vulnerable to bystander oxidative damage from ROS produced by activated macrophages and from spontaneous reactions with glucose and other sugars. In turn, oxidatively damaged molecules interact with, and can stimulate, inflammatory processes.
This analysis of the complex interactions in the aging of humans and animal models is guided by three “Queries” about inflammation, nutrition, and oxidant stress during aging.
(QI) Does bystander damage from oxidative stress stimulate inflammatory processes?
These Queries are not posed as testable hypotheses because in each domain, multiple outcomes are expected from the trade-offs present throughout natural selection. Consequently, many exceptions are expected in the direction and degree of these associations.
The evidence shows that inflammatory and oxidant damage accumulated by long-lived molecules and cells promote the major dysfunctions of aging that, in turn, drive the acceleration of mortality during aging. Later life dysfunctions of the vasculature, brain, and cell growth may be traced to prodromal (subclinical) inflammatory changes from early in life. These processes are examined across the stages of life history, from oogenesis, fetal and postnatal development, and adult stages into senescence.
This inquiry considers aging as a process that is event-related, rather than time-related, from fertilization to later ages (Finch, 1988; Finch, 1990, p.6). Degenerative changes that eventually lead to increased mortality risk can be analyzed as bystander events from agents acting ‘without and within.’ External agents include infections and physical trauma. Internal agents include free radicals produced by macrophages in host defense and subcellularly by mitochondria through normal metabolism. Most long-lived molecules inevitably accumulate oxidant damage during aging. Arterial aging demonstrates many bystander processes which stimulate inflammatory pleiotropies (multiple targets of a process) and which are major risk factors for mortality from heart attack and stroke. Diabetes and infections cause oxidative damage and accelerate arterial changes through complex recursive pathways (Fig. 1.2A).
The theory of inflammation and oxidative stress in aging draws from the free radical immunological and inflammatory theories of aging and the Barker theory of fetal origins of adult disease (Chapter 4). Free-radical causes of cancer and of aging ‘itself’ were hypothesized by Denham Harman in 1956 to involve genetic damage (Harman, 1956, 2003). Then, in the next decade, Roy Walford’s immunological theory of aging extended the importance of somatic cell variation from mutations and other autogenous aging changes to autoimmune reactions, in which somatic cell neoantigens caused pathological aging (Walford, 1969). Since then, the free radical hypothesis was extended to many aspects of aging through mechanisms that involve oxidant stress (Bokov et al, 2004; Harper et al, 2004; Schriner et al, 2005; Stadtman and Levine, 2003; Beckman and Ames, 1998). Damage from inflammation is now well recognized in aging processes and chronic diseases and is mediated by free radicals and many specific inflammatory peptides (Beckman and Ames, 1998; Ershler and Keller, 2000; Finch and Longo, 2001; Finch, 2005; Franceschi et al, 2005; Wilson et al., 2002). Inflammation was already recognized in arterial disease a century ago by Rudolf Virchow (Bokov et al, 2004) (Section 1.5.3). One aspect of immunosenescence in Walford’s theory has been recognized as the depletion of naive T-cells and the acquisition of memory T-cells that are present in unstable arterial plaques. Most recently, Barker’s ‘fetal origins of adult disease’ identifies maternal nutritional influences on adult vascular and metabolic diseases (Chapter 4). I will argue that exposure to infection and inflammation during development also have major importance to outcomes of aging.
While the ‘aging’ risk factor in the chronic diseases is well recognized demographically, aging changes are often neglected in the disease mechanisms. There are many disconnects between the field of ‘basic’ aging and the biomedical fields of chronic diseases (Alzheimer, cancer, diabetes, vascular disease, etc.). I argue most age-associated diseases interact throughout with ‘normal aging processes.’ Many of the same molecules, cells, and gene systems that are altered during aging are considered separately by research in Alzheimer, cancer, and arterial diseases. Major shared mechanisms in aging and disease may be found to stem from roots in the common soil of aging. Shared mechanisms in aging are emerging in the genetics of aging, the insulin-like signaling pathways of metabolism in yeast, flies, worms, and mammals that influence longevity (Fig. 1.3A). Insulin signaling also operates in human arterial disease (Fig. 1.3B). These convergences of aging processes imply ancient genomic universals in life span evolution. It is time to reach for a more general theory that encompasses ‘normal aging’ and ‘diseases of aging’ in the context of evolution and development. However, we should not expect that gene regulation of longevity and senescence will operate by the strict gene regulatory circuits that govern early development (Davidson, 2006; Howard and Davidson, 2004).
Next is an overview of this book. Chapter 1 has two parts: Part 1 reviews human aging and age-related diseases for a diverse readership, emphasizing inflammation and major experimental models. An overview of inflammation and oxidant stress in host defense suggests a classification of bystander damage. Mechanisms of inflammation are outlined, including the energy costs. Part 2 reviews in more detail inflammation in arterial and Alzheimer disease. These details are critical to understanding human aging and the role of insulin-like metabolic pathways. Many inflammatory processes emerge during ‘normal’ or usual aging, but in the absence of specific diseases. The slow creep of inflammation from early years may drive the accelerating incidence of chronic diseases. This hypothesis is supported by evidence that many diseases benefit from drugs with anti-inflammatory and anti-coagulant activities (Chapter 2) and by energy (diet) restriction, which can have anti-inflammatory effects (Chapter 3).
Chapter 2 examines environmental inflammatory factors in vascular disease and dementia with a focus on infections, environmental inflammogens, and drugs that modulate both vascular disease and dementia. Infections and blood levels of inflammatory proteins are risk factors for future coronary events and possibly for dementia. When early age mortality is high, the survivors carry long-term infections that impair growth and accelerate mortality at later ages (‘cohort morbidity phenotype’). Chronic infections, which are endured by most of the world’s human and animal populations, cause energy reallocation for host defense. Infections and inflammation may impair stem cell generation, with consequences to arterial and brain aging. Diet may introduce glycotoxins that stimulate inflammation. Some anti-inflammatory and anti-coagulant drugs may protect against coronary artery disease and certain cancers, and possibly also for Alzheimer disease. These ‘pharmacopleiotropies’ implicate shared mechanisms in diverse diseases of aging.
Energy balance, inflammation, and exercise are addressed in Chapter 3. Diet restriction, which can slow aging and increase life span, also alters insulin-like signaling. Moreover, diet restriction attenuates vascular disease and Alzheimer disease in animal models, again suggesting common pathways. Diet restriction in some conditions has anti-inflammatory effects and may attenuate infections. Conversely, hyperglycemia is proinflammatory in obesity and diabetes. Exercise and energy balance influence molecular and cellular repair, in accord with evolutionary principles.
Chapter 4 considers developmental influences of infections, inflammation, and nutrition on aging and adult diseases. Birth size, overly small or excessively large, can adversely affect later health through complex pathways. Developmental influences attributed to maternal malnutrition in the Barker hypothesis are extended here to infections. Fogel’s emphasis on malnutrition as a factor in poor health can also be extended to include consequences of infection and inflammation. I argue that infection and inflammation compromise fetal development by diverting maternal nutrients to host defense, with consequences to development that influence adult health and longevity.
Chapter 5 reviews genetic influences on inflammation, metabolism, and longevity in animal models and humans. Mutations in insulin-like metabolic pathways shared broadly by eukaryotes can also influence longevity. These metabolic pathways (Fig.1.3A) also interface with inflammation. Certain mutations of insulin signaling that increase the worm life span also increase resistance to infections. The human apoE alleles influence many aspects of aging and disease; the apoE4 allele shows population differences in frequency and effects that may prove to be exemplars of gene-environment interactions during aging.
The last chapter considers the evolution of human life span from shorter-lived great ape ancestors that ate much less meat and lived in low density populations. Human longevity may have evolved through ‘meat-adaptive genes’ that allowed major increases of animal fat consumption and increased exposure to infection and inflammation not experienced by the great apes. The book closes by discussing environmental trends and obesity, which may influence future longevity.
1.2 EXPERIMENTAL MODELS FOR AGING
Aging and senescence in yeast, fly, worm, rodent, monkey, and human are reviewed with details referred to in later chapters. Lab models are referred to by their common names: fly (Drosophila melanogaster); monkey (rhesus, Macaca mulatta); mouse (Mus musculus); rat (Rattus norvegicus); worm (roundworm, nematode Caenorhabditis elegans); yeast (baker’s yeast, Saccharomyces cerevisiae). Other related species may have different life histories (Finch, 1990) and are identified by full name where discussed.
At the population level, humans and these models share the characteristics of finite life spans determined by accelerating mortality. These species share the characteristic of female reproductive decline and oxidant damage in many cells and tissues during aging. Each species has a canonical pattern of aging that persists in diverse environments (Table 1.1) (Finch, 1990). Insulin-like metabolic signaling influences life span, as shown by mutants (Fig. 1.3A) (Chapter 5), suggesting a core of shared mechanisms in aging. However, lab flies and worms differ importantly from mammals by the absence of tumors during aging. The recent discovery that the adult fly gut has replicating stem cells that replace the epithelium with 1 week turnover (Ohlstein and Spradling, 2006) could give a basis for tumor formation in longer-lived species, such as honeybee queens.
TABLE 1.1
General Characteristics of Aging (Canonical Patterns of Aging)
| Mortality Accelerationa | Reproductive Declineb | Slowed Movementc | Cardiac/Vascular Dysfunctionsd | Abnormal Growthse | Oxidative Damagef | Brain Neuron Loss During Agingg | |
| yeast | + | + | not relevant | not relevant | 0 | + | not relevant |
| fly | + | + | + | + | 0 | + | yes |
| worm | + | + | + | not relevant | 0 | + | not likely |
| mouse, rat | + | + | + | + | + | + | sporadic except in disease |
| monkey | + | + | + | + | + | + | “ |
| human | + | + | + | + | + | + | “ |
byeast, budding diminishes; fly and worm, egg production diminishes before death; mammalian ovary becomes depleted (Finch, 1990).
cspontanteous locomotion: fly (Finch, 1990, p. 65); worm, idem, p. 560; mammals (Slonaker, 1912) and common knowledge.
dfly, slowed pulse and lower threshold for fibrillation (Section 5.6.3, Fig. 5.7) and vascular changes in other insects ibid p.65; rodent, loss of arterial elasticity, myocardial fibrosis, and atheroma (Sections 1.2.2 and 2.5); monkey, coronary artery disease induced by fat (Clarkson, 1998); human, Sections 1.2.2 and 1.5, Fig. 1.4 and 1.6.
efly and worm, no tumor observed in wild-type. The presence of dividing stem cells in the adult fly gut (Ohlstein and Spradling, 2006) might lead to tumors in long-lived fly species that over-winter.
fworm, fly, Section 1.2.4; rodent and human, Sections 1.2.4 and 1.6.2.
1.2.1 Mortality Rate Accelerations
All individual organisms have finite life spans, it is simple to say. The core issue in aging is to resolve environmental effects on endogenous aging processes. The hugely complex gene x environment interactions collectively result in mortality risks that define the statistical life span. Here, we face the immense challenge of moving the level of causal analysis from populations to the individual. Time (age) is the best predictor of future longevity in populations. However, the multifarious aging changes that can be identified in individuals are much weaker predictors of longevity risk, the elusive ‘biomarkers of aging’ discussed below.
Senescence in populations of humans and many other species can be compared by the rate of mortality acceleration during aging (Fig. 1.1C) (Finch, 1990, pp. 13–16; Finch et al, 1990; Johnson et al, 2001; Nusbaum et al, 1996; Pletcher et al, 2000; Sacher, 1977). In humans and rodents, mortality accelerations arise soon after puberty (Fig. 1.1C). The lowest values of mortality, which occur in mammals at about puberty, are designated as initial mortality rates (IMRs) (Finch et al., 1990; Finch, 1990, pp. 13–16). The main phase of mortality acceleration is described by the exponential coefficient of the Gompertz equation (Table 1.2). In humans, flies, and worms, mortality rates decelerate at later ages (Finch, 1990, p. 15) and (Carey et al, 1992; Johnson et al, 2001; Vaupel et al, 1998). Mortality deceleration at later ages is less definitive in lab rodents (Finch and Pike, 1996). These complex curves may also be fitted by multi-stage Gompertz (Johnson et al, 2001) or Weibull equations (Pletcher, 2000; Ricklefs and Scheuerlein, 2002). The mortality acceleration in both equations is the strongest determinant of life span in most populations.
TABLE 1.2
Comparative Demography of Aging
| Initial Mortality Rate (IMR) | Mortality Rate Doubling Time (MRDT) | Maximum Life Span | |
| yeasta | 0.2/d | 10 d | >20 d |
| flyb | 0.1/d | 5 d | >60 d |
| mouse, ratc | 0.1/mo | 4 m | > 48 mo |
| human Sweden)d | |||
| 1751 | 0.0090/y | 7–9 y | <100 |
| 1931 | 0.0008/y | >110 y |
These organisms show exponential accelerations of mortality, approximating a straight line on a semi-logarithmic plot of mortality rates against age (Fig. 1.1B), as described by the Gompertz equation for mortality rates: m(x) = Aexp(αx), where α is the Gompertz coefficient, x is age, and A is the initial mortality rate, IMR. Mortality rate doubling time is calculated as ln 2/α (Finch et al, 1990). Rodents fed ad libitum.
ayeast (Finch, 1990, p. 105) from data of (Fabrizio et al, 2004), chronologic model (non-dividing)
bfly B stock (Nusbaum et al, 1996)
crepresentative rodent strains (Finch and Pike, 1996)
dSwedish historical populations (Finch and Crimmins, 2004, 2005; Crimmins and Finch, 2006a, b), and unpublished. IMR is calculated differently by species according to conventions. For rodents and human, IMR is calculated at the age of sexual maturation (puberty), its lowest value. For worm and fly, IMR is calculated at age 0 (hatching). Also see TABLE 5.1 and Finch (1990), pp. 663–666.
The Gompertz exponential coefficient is conveniently expressed as the ‘mortality rate doubling time’ (MRDT), which ranges 1000-fold between yeast and long-lived mammals (Table 1.2) (Finch, 1990, pp. 662–666). Yeast, worm, and fly show the most rapid senescence, while birds and mammals show gradual senescence. At the other extreme is the theoretical limit of ‘negligible senescence’, with MRDTs of >100 years (Finch, 1990, pp. 206–247; Finch, 1998; Vaupel et al., 2004). Species of long-lived fish (Cailliet et al, 2001; De Bruin et al, 2004; Geuerin, 2004), turtles (Congdon, 2003; Henry, 2003; Swartz, 2003), and conifers (Lanner and Connor, 2001) have not shown reproductive aging and are candidates for negligible senescence; however, data are lacking to evaluate mortality rates. MRTDs within a species vary less than the 10-fold or more variations in IMR. Human populations show a remarkable 10-fold range of IMR variations (Table 1.2), which reflect the level of health allowed by nutrition, infections, and other environmental factors (Chapters 2, 3, 4). Experimental variations of MRDT include 2-fold difference by diet (diet restriction in rodents, Chapter 3) and genotype (Age-1 worm mutant, Chapter 5). Curiously, rodent MRTDs do not vary much by genotype, despite quite different diseases of aging (Finch and Pike, 1996). Human MRTDs are fairly similar across populations, despite major differences in diseases and overall mortality (Finch, 1990; Gurven and Kaplan, 2007), e.g., Sweden (Table 1.2; Fig. 1.1C and Fig. 2.7). Male mortality is generally higher throughout life (Section 5.3).
1.2.2 Mammals
Mammalian aging follows canonical patterns that gradually emerge after maturation and progress across the life span in proportion to the species life span (Finch, 1990). The seeds of aging are found before birth in many tissues, e.g., arteries and ovaries, as discussed below. The occurrence of these aging patterns in at least 5 of the 28 orders of placental mammals implies shared gene regulatory systems evolved hundreds of million years ago that determine the level of molecular and cell turnover and repair in specific tissues. The canonical patterns of aging thus can be considered as genetically programmed aging.
The increasing incidence of diseases of aging corresponds to the acceleration of mortality during aging, as known in detail for humans and rodents. Arterial disease (heart attack, stroke) and cancer are the main causes of death across aging human populations (Fig. 1.4). Vascular deaths increase more or less exponentially after age 40, whereas breast cancer incidence plateaus after menopause. By age 65, vascular deaths exceed those from cancer in most populations (Horiuchi et al, 2003). In 1985, in Japan, Sweden, and the United States, for example, the total male deaths recorded for heart attack and stroke were 2-fold or more than for cancers, 3-fold more than respiratory conditions, and 30-fold more than for infectious diseases (Fig. 1.4) (Aronow, 2003; Himes, 1994; Horiuchi et al, 2003). The relative proportion of heart attacks (ischemic heart disease) and stroke (cerebrovascular disease) vary between populations. However, by 2002 in the United States, cancer mortality appears to have overtaken vascular-related mortality for age 85 and younger, where about 5% more died of cancer than from heart disease (476,009 vs. 450,637) (American Cancer Society, 2006). The campaigns on prevention and intervention of vascular disease are having remarkable impact on vascular changes.
In rodents, the incidence of new pathologic lesions also increases exponentially (Bronson, 1990; Simms and Berg, 1957; Turturro et al, 2002), and roughly paralleling the acceleration in mortality rates (Fig. 1.5). Diet restriction shifts the incidence of lesions to later ages and slows the acceleration of mortality (Chapter 3). The causes of death are often unresolvable, because multiple lesions are common at later ages (Fig. 1.5 and legend). The Berg-Simms colony founded in 1945 gives an unsurpassed documentation of age-related degenerative disease and mortality (Berg, 1976).
Despite the relatively primitive husbandry and hygiene, life span was in the current range. The pathology of aging (specific organ lesions and age incidence) (Simms and Berg, 1957; Simms and Berg, 1962) has been confirmed in modern colonies (Bronson, 1990). Kidney lesions preceded tumors and cardiomyopathy; arterial calcification was occasional. In current colonies, kidney lesions and tumors also predominate, occurring in 80% of aging rodents across genotypes (Bronson, 1990; Turturro et al, 2002). Myocardial lesions are less common than in the Berg-Simms era and may vary; e.g., in aging C57BL/6 mice, myocardial degeneration ranged from 8% (Bronson, 1990) to 40% (Turturro et al, 2002), always less than tumors and kidney lesions. The arterial and myocardial pathology in early colonies is discussed in Section 2.5.2, together with improvements in hygiene and husbandry that increased life span with some parallels to the recent human improvements. Rodents in modern colonies on standard diets are not thought to die from arterial degeneration or thrombosis. This may be incorrect.
Rodent models for aging have been borrowed from existing lines that were originally developed for genetic studies of cancer and other chronic diseases, and of transplantation (immunogenetics). Rodents with delayed incidence of pathology until after 18 m were used as controls for early onset tumors, e.g., the relatively long-lived C57BL/6J and DBA/2J mice. All baseline stocks were selected for traits of fast growth and high fecundity, which is the rule for domestication for animals and plants. Infectious diseases were gradually minimized. The resulting models differ importantly from their feral origins, i.e., the true ‘wild-types.’ For example, wild-caught mice are smaller, mature later, and live longer than lab mice (Miller et al, 2002). Moreover, diet restriction has much less effect on life spans of wild-caught mice (Harper et al, 2006) (Chapter 3, Fig. 3.3). Immune functions also differ in ‘unhygienic’ feral mice and rats, with much higher levels of autoreactive IgG (Devalapalli et al, 2006). The modern lab rodents with unlimited access to food, low physical activity, and tendencies to obesity may thus be fine models for contemporary lifestyles. However, the limited exposure to infections is unlike the real world. It may be necessary to incorporate antigenic challenges in our aging animal colonies to understand the aging mechanisms at work in human populations, past, present, and future.
In humans, arterial degenerative aging changes result from two long-term processes: the inexorable progressive accumulation of arterial wall lipids (Fig. 1.6A) and arterial rigidity, both from starting early in life (Sections 1.2.6 and 1.6). The loss of elasticity increases blood pressure (Fig. 1.6B), independent of clinical hypertension syndromes. The atherosclerotic lesions can lead to clots (thromboses) that block blood flow with catastrophic effects. Mortality from ischemic heart disease and stroke increases exponentially with adult age (Fig. 1.6C). Systolic pressure elevations are major risk factors in heart attack and stroke and are as universal to human aging as menopause and bone thinning.
The loss of arterial elasticity and artery wall thickening (arteriosclerosis) are ubiquitous in mammals, while focal atherosclerosis is more prominent in humans and primates than rodents (Tables 1.1 and 1.4). The aorta and other central arteries become progressively thicker. The accumulation of oxidized lipids begins before birth in microscopic cell clusters (Section 1.5.1). The numerous inflammatory changes include increased macrophages, free radical producing enzymes (NADPH oxidase), cell adhesion molecules (ICAM), cytokines (TGF-β1), and matrix metaloproteinases (MMP-2 and -9). These diffuse changes are generally independent of focal atheromas. Thus, oxidative damage (oxidized lipids) and inflammation are at work from the beginning in arterial aging (Queries I and II).
TABLE 1.4
Comparisons of Arterial Aging Changes in Humans and Mammalian Models with Atherosclerosis and Hypertension
| Arterial Parameter | Human | Monkey | Rat | Rabbit | Atherosclerosis | Hypertension |
| Diffuse intimal thickening | + | + | + | + | + | + |
| Lipid deposits | − | − | − | − | + | +/– |
| Macrophages | + | − | − | − | + | + |
| T cells | + | − | − | + | + | + |
| Matrix ↑ | 0 | + | + | + | + | + |
| Ang II-ACE ↑ | + | + | + | + | + | + |
| Endothelial dysfunction | + | + | + | + | + | + |
| Extra cell. Matrix | + | + | + | ? | + | + |
| ICAM ↑ | ? | ? | + | ? | + | + |
| MCP-1/CCR2 ↑ | + | + | + | + | + | + |
| NADPH oxidase ↑ | ? | ? | + | ? | + | + |
| TGF-β1 | ? | + | + | ? | + | + |
| VEGF ↑ | + | ? | ? | + | + | + |
| Lumenal dilation | + | + | + | + | ? | +/− |
| Wall stiffness ↑ | + | + | + | + | ? | + |
| Collagen ↑ | + | + | + | + | ? | +/− |
| Elastin degeneration | + | + | + | + | 0 | 0 |
| Telomere shortening | + | + | + | ? | + | ? |
Adapted from Lakatta (2003), Lakatta and Levy (2003a,b), Wang and Lakatta (2006), Najjar et al (2005).
Arterial elasticity decreases progressively from alterations in collagen and elastin by inter-molecular AGE adducts (advanced glycation and glyco-oxidation end products) derived from glucose and other reducing sugars (Fig. 1.6D) (Section 1.4.4). AGE adducts contribute to arterial rigidity by intermolecular cross-links between collagen and other proteins. In turn, AGE may kindle local inflammation by activating scavenger receptors. Arterial elastin is very-long lived, as shown by accumulations of racemized D-aspartate (Fig. 1.6D). Racemization spontaneously converts normal L-amino acids to the D-isomers. In long-lived proteins, the accumulation of ‘racemers’ is a direct marker of age (Bada et al, 1974; Helfman and Bada, 1975). Because veins undergo less wall thickening, arterial aging is hypothesized to be driven by the repeated pressure waves at each pulse (Section 1.5.3.2). Blood flow patterns modify gene expression in atheroprone arterial regions.
New macroscopic atheromas appear throughout life. Lipid oxidation may be a key cause of atheroma initiation and progression (Queries 1 and 2). Inflammatory processes are active throughout atherogenesis and are intensified in atheroprone arterial zones. The developing atheromas are described as a complex wounding response with cell growth and cell recruitment; oxidation of lipids and proteins; cell death; and eventual calcification. Environmental influences from infections, diabetes, and stress can accelerate atheroma formation, whereas statins may facilitate atheroma regression (Chapter 2). The insulin/IGF-1 system that modulates life span in flies and worms is also at work in many aspects of atherogenesis (Fig. 1.3B). Animal models vary in susceptibility to arterial lesions. Macaques, chimpanzees (Finch and Stanford, 2004; Wagner and Clarkson, 2005), and rabbits (Yanni, 2004) are more vulnerable to atheroma induction by diet and stress than lab rodents (Moghadasian, 2002). The apoE-knockout mouse has extreme susceptibility to atheromas, in association with its extreme hypercholesterolemia (Rauscher et al, 2003).
The myocardium is altered during aging through inflammatory processes that can interact with arterial changes. Left ventricular stiffness increases progressively with aging (decreased ‘compliance’) and slows the diastolic of filling rate by up to 50% by age 80 (Brooks and Conrad, 2000; Lakatta and Levy, 2003a,b; Meyer et al., 2006). The stiffness is due to ventricular wall thickening and interstitial myocardial fibrosis, and possibly collagen cross-linking through nonenzymatic glycation. Fibrosis is very common during mammalian aging and deeply linked, if not intrinsic, to general inflammatory processes in aging (Thomas et al, 1992). TGF-β1 signaling pathways that regulate collagen synthesis are implicated in myocardial fibrosis. TGF-β1 deficiency (+/– heterozygote knockout) attenuated the age-related increase of left ventricular fibrosis, improved cardiac performance, and possibly increased life span (Brooks and Conrad, 2000). Myocardial stiffness is attenuated in humans during diet restriction in at least one study (Section 3.4.1). Conversely, transgenic mice with increased systemic TGF-β1 developed premature left ventricular fibrosis with increased levels of TIMPS (tissue inhibitor of metaloproteinase, also implicated in arterial aging) (Seeland et al, 2002).
Mitochondrial DNA changes in the myocardium also merit mention because of their interactions with ischemia and oxidative stress. Additionally, myocardial mitochondrial DNA deletions (mtDNA4977, nt 8469–13,447) increase modestly after age 60 (up to 7 per 10,000 mitochondria). Ischemic hearts can have >200-fold more mtDNA deletions (Botto et al, 2005; Corral-Debrinski et al, 1992), which is attributed to the oxidative stress from ischemia. Because the DNA deletion impairs mitochondrial function and increases respiratory chain stress, a vicious cycle is hypothesized to cause further mitochondrial damage. Single base changes (point mutations) also increase with aging in a mutational hotspot (nt 16,025–16,055, control region) in cardiomyocytes, but not buccal epithelial cells, with indications of clonal expansion (Nekhaeva et al, 2002).
Besides these aspects of myocardial aging, there are many other aging changes, as well as compensatory mechanisms that go beyond this discussion. At the behavioral level, and of great importance to human aging, are complex social and psychological links to vascular disease and hypertension (‘social etiology’) (Berkman, 2005; Marmot, 2006; Sapolsky, 2005). Social stress also accelerates vascular changes in primates and rodents (Andrews et al, 2003; Henry et al, 1993). Complex social interaction during aging has not been defined in animal models.
Immunity declines in complex ways during aging: instructive immunity weakens concurrently with increased inflammation in most tissues and chronic diseases. Both changes may contribute to the decreased resistance to opportunistic infections, incidence and severity (Akbar et al, 2004; Miller, 2005; Pawelec et al, 2005; Weksler and Goodhardt, 2002). As examples, the elderly suffer 90% of the influenza deaths, while HIV has a shorter latency in the elderly, reviewed in (Olsson et al, 2000). The decreased resistance is associated with various dysfunctions of systemic and tissue immune mechanisms: the attenuation of adaptive (instructive) immunity and the hyperactivity of acute phase host defense processes. Aging of the adaptive immune responses may be very gradual in populations of humans and lab animals with low burdens of infection and inflammation, and good nutrition. Nonetheless, naive T cells progressively decrease at the apparent expense of memory T cells (CD4 and CD8) (Haynes, 2005; Linton and Dorshkind, 2004; Miller, 2005; Pawelec et al, 2002).
At birth, nearly all T cells express CD28, a major T cell-specific co-stimulator that binds to sites on antigen-presenting cells and activates IL-2 transcription, cell adhesion, and other critical T-cell functions. CD28 is progressively lost during aging (Merino et al, 1998; Pawelec et al, 2005; Trzonkowski et al, 2003). The loss of CD28+ T cells is attributed to chronic antigenic stimulation over the life span. Accelerated loss of CD28 T cells is observed in young HIV patients and is modeled in cultured T cells (Posnett et al, 1999). The CD8+ CD28– T cells are resistant to apoptosis and are considered ‘fully differentiated.’ During influenza inflections, the elderly have decreased cytotoxic T-cell activity in association with shifted cytokine profiles (T-helper type 2 dominance) (McElhaney, 2005).
Declines of thymus function begin before maturation (Krumbahr, 1939; Min et al, 2005; Steinmann, 1986). Infections and malnutrition in the early years can impair thymus development with later consequences to immunity (Moore et al, 2006; Savion, 2006). Striking examples come from West Africa. In rural Gambia, seasonal infections during childhood alter T-cell functions with correspondingly increased adult mortality (Moore et al, 2006). In Guinea-Bissau, a low small thymus is associated with increased mortality from infections (Aaby et al, 2002). These and other environmental effects on immunity during development are discussed in Section 4.6.2.
Between puberty and mid-life, adipocytes gradually replace the lymphocytic perivascular space. These gross changes are preceded by regression of the thymic epithelium and can be delayed by castration before puberty (Chiodi, 1940; Min et al, 2006). After maturation, the thymus continues to generate T cells throughout life, although at lower levels (Douek et al, 2000; Hakim et al, 2005). Immune homeostatic mechanisms decline during aging; e.g., T-cell recovery after chemotherapy is greatly reduced by age 50 (Hakim et al, 2005; Hakim et al, 2005). Extra-thymic aging changes include lower bone marrow production of lymphopoietic progenitor cells, possibly due to decreased growth hormone and IGF-1 (Hirokawa et al, 1986; Linton and Dorshkind, 2004). Most immunologists agree that thymic involution is multi-factorial and that immune aging is not reversed by simply restoring GH, IGF-1, or other hormones that change with aging (Chen et al, 2003; Min et al, 2006). The adverse effects of infections and malnutrition on thymic development may extend to other aspects of immunity.
The major shifts from virgin T cells to memory T cells during the life span are attributed to exposure to common infections, environmental antigens, and auto-antigens. Cytomegalovirus (CMV), an endemic β-herpes virus that is a common infection in childhood, may be a general factor in the clonal depletion of CD28+ T cells (Koch et al, 2006; Pawalec et al, 2005). Up to 25% of the CD8 T cells in older healthy humans are CMV-specific (Khan et al, 2002) and are approaching replicative senescence (Fletcher et al, 2005). A proposed ‘immune risk phenotype’ of aging is characterized by (1) CMV-seropositivity; (2) inverted ratios of CD4:CD8 <1 (unlike the normal CD4 excess in healthy young adults); and (3) increases in ‘fully differentiated’ CD8+ CD28– effector T cells, which have shortened telomeres and limited proliferation (Olsson et al, 2000; Pawelec et al, 2005). Elderly with ratios of CD4:CD8 <1 have 50% higher mortality in two populations: the Healthy Ageing Study (Cambridge UK) (Huppert et al, 2003) and the OCTO and NONA Longitudinal Studies (Jönköping, Sweden) (Wikby et al, 2005). These T-cell shifts decrease resistance to new infections. The greater vulnerability of elderly to influenza may be attributed to imbalances of central memory T cells over the effector memory T cells that mediate virus-specific IFN production (Kang et al, 2004). CMV-seropositive elderly who responded poorly to influenza vaccine also had more CD28- lymphocytes (Effros, 2004; Trzonkowski et al, 2003) and 2-fold higher IL-6 and TNFa (Trzonkowski et al, 2003). The higher cytokine production during aging in immune responses may extend to other classes of T-cells (O’Mahony et al, 1998) and may be a factor in the strong age trend for elevated cytokines (Section 1.8.1).
Besides CMV, many other infections influence the ‘immune aging phenotypes.’ Chronic immune activation can accelerate ‘aging’ of T-cell functions, as observed in infections by HIV (van Baarle et al, 2005) and nematode parasites (Borkow et al, 2000) (Section 2.7.1). As noted previously, childhood infections affect the thymus and impair immunity and increase mortality (Section 4.6). As another example, mice with genetically determined elevations of memory T-cells have shorter life spans and higher prevalence of tumors at middle age (Miller, 2005). Chronic immune activation also increases ‘bystander’ damage (Section 1.4.3) (Query II). We may anticipate that outcome of immune aging depends on gene-environment interactions with inflammatory gene variants, particularly the proinflammatory IL-6 and the antiinflammatory IL-10 (Caruso et al, 2004) (Section 1.3.2). The strong role of the antigenic environment on immune aging is included in the framework of Fig. 1.2A and extends to direct involvement of T-cells with unstable atheromas (Section 2.2.2).
Telomere erosion is implicated in immune aging in association with the reduced proliferation of T cells (Effros, 2004). In peripheral lymphocytes, telomeres shorten by 50 base pairs per year across the life span against initial telomere lengths at birth of about 15,000 base pairs (Hathcock et al, 2005). Telomeres are shorter in primed T-cell subsets, especially the ‘effector memory’ T cells (Akbar et al, 2004). Telomere loss may eventually activate gene regulatory programs leading to cell death (apoptosis) or a post-mitotic state (clonal senescence; considered equivalent to the Hayflick limit; see below). Telomerase reactivation is thought to be adaptive for clonal expansion without rapid clonal senescence. However, T cell proliferation does not always cause telomere erosion, because immune stimulation of B and T cell proliferation can induce telomerases (Akbar et al, 2004; Hathcock et al, 2005; van Baarle et al, 2005). Other evidence argues against telomere erosion as a general mechanism in immune aging (Miller et al, 2000); e.g., although mice have much longer telomeres than humans, mouse T cell proliferative aging is faster. Much is unknown about enzymes that mediate telomere replication, which differs between immune cell types and animal species.
In contrast to the decline of antigen-driven immunity, inflammatory processes in many tissues progressively increase during aging, e.g., muscle, fat, brain (Section 1.8.1). Inflammatory gene expression increases in these and other tissues. Blood IL-6 and C-reactive protein generally increase during aging in human populations, although much of the increase is associated with vascular disease. Tissue-specific macrophages are prominent in atheromas (‘foam cells’), Alzheimer disease (microglia), and bone (osteoclasts). Apart from these degenerative diseases, studies of circulating macrophages from aging humans and rodents are puzzlingly inconsistent about the direction and type of aging changes (Finch and Longo, 2001; Pawelec et al, 2002; Wu and Meydani, 2004). Despite blood IL-6 elevations, induction of IL-6 in response to LPS (gram-negative bacterial endotoxin) decreases with age in peritoneal macrophages (Stout and Suttles, 2005), but increases with age in brain microglia (Xie et al, 2003; Ye and Johnson, 1999).
Lastly, we should be mindful that decreased system-level and integrative functions (‘organ reserves’) contribute to mortality independently of specific immune subsystems. The declining ‘vital capacity’ of lungs (Janssens and Krause, 2004; Meyer, 2005) is strongly associated with survival in general and resistance to respiratory infections. In the Framingham Study, mortality risk at age 50–59 varied inversely with the lung vital capacity (Ashley et al, 1975; Finch, 1990, p. 563). Smoking, which decreases pulmonary volume and respiratory functions, increases vulnerability to influenza and pneumonia (Murin and Bilello, 2005). In the Cardiovascular Health Study of persons 65 years and older, smokers had a 50% higher risk of hospitalization for pneumonia and a 28% higher mortality in the 2.4 years after discharge (O’Meara et al, 2005). Moreover, CMV and other chronic infections may deplete the bone-marrow-derived endothelial progenitor cells that mediate vascular repair (Section 2.7.3). Thus, the decreased resistance to infectious disease should be analyzed in terms of systemic physiology and the ecological life history of exposure in infections and inflammogens (Chapters 2–4), which are subject to gene-environment interactions throughout the life history (Chapters 4 and 5).
Female reproductive senescence is due to the exhaustion of ovarian oocytes in all mammals examined (Finch, 1990, pp. 165–167; Gosden, 1984; vom Saal et al., 1994; Wise et al, 1999). Oocyte numbers are fixed during development by the cessation of primordial germ cell proliferation. Oocyte loss begins before birth and continues exponentially, like radioactive decay (Faddy et al, 1992). Recent evidence refutes the possibility of continuing de novo oogenesis from circulating stem cells (Eggan et al, 2006). Less than half of the original stock remains by puberty. The rate of oocyte loss is slowed by diet restriction, which alters hypothalamic controls of the gonadotrophins (Chapter 3, Fig. 3.17). Fecundity declines long before the failure of ovulation due to oocyte depletion, with marked reduction by age 35 years in women; lab rodents aged 8–12 months are culled as ‘retired breeders’ by production colonies. With the loss of ovarian follicles, the production of estrogens and progestins decreases sharply in human menopause, causing hot flushes, as also observed in macaques (Appt, 2004; Nichols and O’Rourke, 2005). Ovarian steroid loss is implicated in the post-menopausal increase of vascular disease and may interact with vascular inflammatory processes. In males, androgen levels show trends for decline, but more sporadically than in females. Estrogen replacement (hormone therapy), while controversial, appears to be health protective for some women (Section 2.9.4). Androgen replacements may benefit arterial disease, cognition, and glycemic control (Harman, 2005; Jones et al, 2005; Liu et al, 2004; Morley et al, 2005). The sharp rise of vascular disease during middle age is an example for the declining strength of natural selection during aging (evolutionary perspectives, below).
Bones and joints degenerate broadly during aging in mammals in processes that involve inflammatory regulators (Section 1.8). Osteoporosis (bone mineral resorption) occurs through an imbalance of production by osteoblasts versus resorption by osteoclasts. The inflammatory system is involved in bone resorption. First, osteoclasts are of macrophage/monocyte lineage. Then, bone resorption is stimulated by inflammatory cytokines (IL-1, TNFa) (Clowes et al, 2005; Tanaka et al, 2005). Osteoporotic bone loss accelerates after menopause and can be attenuated by estrogen replacement. In some contexts, estrogen has anti-inflammatory activities (Amantea et al, 2005; Thomas et al, 2003) (Section 2.10.4). Osteoarthritis is a focal, age-related inflammatory lesion in the joints that can be painful (Section 1.7). Mechanical pressures activate inflammatory cells and catabolic responses of the articular chondrocytes that cause matrix loss and accumulation of AGEs.
Brain-aging atrophic changes are manifest soon after maturation, in healthy humans by age 30 y and rodents aged 10 m (Finch et al, 1993; Teter and Finch, 2004). The volume of the brain as a whole shrinks by about 0.5%/year in normal humans across the adult age range, 20–98, and is accelerated by Alzheimer disease to about 1%/y (longitudinal MRI) (Burns et al, 2005; Fotenos et al, 2005). The volume of the hippocampus, which is critical to declarative memory, also shrank linearly in healthy elderly observed over 6 y (Cohen et al, 2006).
Neuron loss during aging is more limited than once widely presumed (‘neuromythology’) (Finch, 1976; Gallagher et al, 1996; Rasmussen et al, 1996; Teter et al, 2004; Tomasch, 1971). During normal human aging, the total number of cortical neurons does not change, but neuronal size shrinks. Small cortical neurons increase, while the numbers of large neurons decreases reciprocally (Terry et al, 1987) (Fig. 1.7A).
Synaptic loss parallels brain atrophy and neuron shrinkage, with progressive decreases in the presynaptic protein synaptophysin in the normal aging cerebral cortex (Fig. 1.7B) and dopamine D2 receptors in the cortex and striatum (Fig. 1.7C, D) (Morgan et al, 1987; Reeves et al, 2002; Suhara et al, 2002; Wong et al, 1997). Other receptors, however, may increase—e.g., dopamine D1 receptor (Morgan et al, 1987)—possibly as a compensatory response. The extent of synaptic loss approximates 1% loss per year after age 20. Rodent brains show similar changes in dopamine receptors, scaled to their shorter life span. By the mean life span, synaptic atrophy reaches 30–50%, independent of Alzheimer, stroke, or other clinical conditions in lab rodents and humans. Nonetheless, neuron cell death is minimal in cortex and possibly other brain regions to advanced ages, absent Alzheimer changes. Even at later ages, neuron loss in aging memory-impaired rats is modest or sporadic in brain regions afflicted by Alzheimer disease (Rasmussen et al, 1996; Rapp and Gallagher, 1996). However, sporadic neuron loss may arise from various stressors (Landfield et al, 1977; Meaney et al, 1988) or toxins (Section 3.2.3; Finch, 2004b). In rodents, D2 receptor loss in aging is attenuated by diet restriction (Chapter 3, Fig 3.17) while neuronal atrophy is reversed by nerve growth factors (Smith et al, 1999). The generality of synaptic atrophy during middle age in the absence of neuron loss distinguishes these changes from the subgroups at later ages that develop aggressive neurodegeneration during Alzheimer disease.
Age-related synaptic atrophy (15–30%), while modest relative to Alzheimer disease, plausibly contributes to declines in complex brain functions. Memory capacity declines progressively in humans and rodents, with no evident pathology (Albert, 2002; Rajah and D’Esposito, 2005; Rosenzweig and Barnes, 2003; Woodruff-Pak, 2001). In the hippocampus, a seat of declarative memory, synapse loss is extensive. The hippocampus receives input from the cerebral cortex from the perforant pathway, which is much more damaged during Alzheimer disease than normal aging (Nicholson et al, 2004; Rosenzweig and Barnes, 2003). Some functional deficits may be linked to synaptic atrophy. The D2 receptor loss during aging (Fig. 1.7D) correlated with performance on tasks dependent on the frontal cortex, e.g., the Stroop Color-Word Test interference score (Volkow et al, 2000).
Other deficits may be due to the deterioration of myelinated pathways (white matter), which mediate high-speed exchanges between brain regions. Microglial activation may be a factor in white matter changes during middle age as seen by brain imaging (Bartzokis, 2004; Bartzokis et al., 2003, 2004, 2006; Burns et al., 2005) (Fig. 1.8B). ApoE4 carriers have accelerated myelin deterioration (Bartzokis et al., 2006), consistent with the ‘proinflammatory’ associations of the E4 allele (Section 1.3). White matter inflammatory changes may contribute to the usual slowing of information processing and the decreased multi-tasking by middle age (Bashore, 1994; Madden, 2001; Verhaeghen and Cerella, 2002). Multi-tasking becomes progressively impaired during aging. A striking example is the impairments by middle age in memorizing words while walking an irregular course (Li et al, 2001). Multi-tasking depends on high-speed processing across multiple circuits, which slows progressively during normal aging (Bashore and Ridderinkoff, 2002; Maeshima et al, 2003; Ylikoski et al, 1993). These processes suggest why few professional athletes remain competitive over the age of 40 and why driving errors increase with aging (Campagne et al, 2004).
). Determined from the transverse MRI relaxation rate signal (R2), a measure of myelin integrity (Bartzokis et al., 2004).Skeletal muscle atrophy during aging (‘sarcopenia’), a major factor in frailty (Dow et al, 2005), may be due to motor neuron aging. The atrophy of aging muscle resembles denervation atrophy and is partly reversed by electrical stimulation (Dow et al, 2005). The major role of motor neuron age in skeletal muscle aging was shown by a powerful transplantation experiment: when reinnervated in young rat hosts, the 32-month-old muscle grafts regained full contractile strength (Carlson et al, 2001). Aging muscle also accumulates mitochondrial DNA mutations in association with muscle fibers deficient in cytochrome oxidase (COX) (Brierley et al, 1998; He et al, 2002; Kopsidas et al, 2002). However, COX-deficient fibers are relatively rare (0.1–5%), and the deficiency does not extend through the entire fiber (Brierley et al, 1998; Frahm et al, 2005). Thus, muscle mtDNA mutations may contribute less to muscle aging than motor neuron impairments. Impaired axoplasmic flow by motor neurons in the sciatic nerve (Goemaere-Vanneste et al, 1988) and spinal projections (Frolkis et al, 1985; McQuarrie et al, 1989) could be major causes of the reduced neurotrophic support. Axoplasmic flow also decreases during aging in the central projections (De Lacalle et al, 1996; Geinisman et al, 1977). Cell-level gene expression may identify global or cell-specific impairments in biosynthesis that could cause diverse manifestation of synaptic atrophy.
What may cause the atrophy of neurons during aging? My lab is studying the increase of glial inflammatory changes that are concurrent with neuronal atrophy, in healthy humans, rodents, and monkeys (Morgan et al., 1999; Finch et al., 2002). Astrocytes and brain macrophages (microglia, of bone marrow lineage) are activated by middle age (Fig. 1.8A). The increased volume of astrocytes mainly represents cell hypertrophy. The total number of astrocytes does not increase during normal aging (Bjorklund et al, 1985; Finch et al., 2002; Long et al, 1998), although there are more and larger fibrous astrocytes with thick, GFAP containing processes (Hansen et al., 1987). GFAP is a cytoskeletal protein (intermediate filament) that increases with aging (Nichols et al., 1993) in parallel with increased astrocyte volume (Fig. 1.7E). The age-related increase of GFAP expression can be considered as an inflammatory response (Morgan et al, 1999). GFAP transcription increases in response to oxidative stress and inflammatory stimuli, possibly through redox sensitive elements (NF-1/NF-kB) in the upstream promoter (Morgan et al, 1997b, 1999). Diet restriction attenuates the increase of GFAP transcription and protein levels with aging (Morgan et al, 1999) in parallel with attenuating synaptic atrophy during aging (Chapter 3). Many other brain changes during aging are associated with inflammatory processes (Section 1.8.1).
We hypothesize that glial activation during aging is an inflammatory response to oxidative damage that, in turn, causes synaptic atrophy (Rozovsky et al., 2005). Microglia activation during aging is attenuated by diet restriction (Chapter 3, Fig. 3.19). We are testing these concepts by growing cultures of astrocytes from aging rats, which show age deficiencies in support of neuronal outgrowth (Fig. 1.9). The age deficiencies are associated with increased expression of GFAP and are inversely associated with secretion of laminin and other extracellular substrates. Conversely, we can induce an age-like phenotype in young astrocytes by increasing the cell levels of GFAP, with correspondingly less support of neurite ourtgrowth. These results show the close links of GFAP expression to astrocytic support of neurite outgrowth over a 2-fold range (Rozovsky et al, 2005).
Alzheimer disease (AD) differs from these normal brain aging changes (Section 1.6) by severe neurodegeneration in memory circuits with remarkable selectivity. The hippocampal pyramidal neurons of the CA1 field are devastated, while nearby granule neurons are relatively unscathed. Neuronal vulnerability to endogenous or exogenous insults must ultimately depend on gene expression patterns acquired during cell differentiation.
AD is rare before age 60 and increases exponentially thereafter with a doubling rate of about 5 years (Fig. 1.10) (Kawas and Katzman, 1999; Mayeux, 2003). By age 80, the AD prevalence approaches 50% in some populations but may be less than 15% in others. These huge differences are unexplained. A characteristic of AD is extensive deposits of extracellular senile plaques containing fibrillar Aβ1-42 (Fig. 1.10B) and intraneuronal aggregates of neurofibrillary tangles containing hyperphosphorylated tau. A major hypothesis is that AD is driven by excess production of the 42 amino acid long amyloid β-peptide (Aβ1-42) (Section 1.6). The diagnosis of Alzheimer disease is made by a threshold density of plaques, tangles, and neuron loss in the cerebral cortex and hippocampus. The diagnosis may be adjusted for age because of the increase of neurofibrillary tangles during aging. Non-demented elderly show accelerating increases of tangles to levels that overlap with early clinical AD (Braak and Braak, 1991; Price and Morris, 1999) and also show modest cortical atrophy with aging (Anderton, 1997; Launer et al, 1995). At very advanced ages, plaque and tangle accumulations may overlap with criteria for AD.
Amyloid accumulation in animal models is slowed by anti-inflammatory drugs (Chapter 2) and diet restriction (Chapter 3). Rodents are valuable experimental models to study human AD transgenes because they do not accumulate brain amyloid deposits during aging. Amino acid substitutions in the rodent Aβ sequence (Johnstone et al, 1991) render it less able to aggregate into fibrils and possibly less toxic (Boyd-Kimball et al, 2004). In other vertebrates, the Aβ sequence is remarkably conserved—fish to primates to humans. Aging dogs and macaques accumulate Aβ deposits that resemble senile plaques of AD. However, aging chimpanzees, our closest ancestor, have negligible AD-like changes during aging (Section 6.3) (Finch and Stanford, 2004). The chimpanzee does not have a common risk factor in AD, the apoE4 allele, which evolved in humans.
1.2.3 Cultured Cell Models and Replicative Senescence
Unlike cancer cells, diploid somatic cells typically have a finite capacity for propagation during serial culture, shown first for skin fibroblasts (Hayflick and Moorhead, 1961). After a finite number of subcultures (population doublings), cell division ceases and cultures are considered senescent. The ‘Hayflick’ phenomenon extends to many cell types, including vascular endothelia and lymphocytes (Campisi, 2005; Cristofalo et al, 2004; Hayflick, 2000). Although the end-phase cultures are considered ‘senescent,’ post-replicative cells survive many months if media are refreshed (Matsumura et al, 1979). In fact, senescent cultures are highly resistant to apoptosis. Resistance to apoptosis may link back to the insulin pathways that modulate life span (Fig. 1.3), because senescent cell cultures have decreased endocytic uptake of the IGF-binding protein IGFBP-3 (Hampel et al, 2005). It is cogent to Query I that senescent cultures of fibroblasts and other cell types show increased inflammatory factors including COX-2, IL-1, MMP-3, collagenase, TIMP-1 (tissue inhibitor of metaloproteases) (Han et al, 2004; Parrinello et al, 2005; West et al, 1989; Zeng et al, 1996). These same changes arise in atheromas and, moreover, are blocked in senescing cultures by the COX-2 inhibitor NS398 (Han et al, 2004) (Chapter 2). Inflammatory factors secreted by replicatively senescent cells are implicated in focal tissue remodeling in the progression of pre-malignant cells (Campisi, 2005).
Individual cell variations in replicative potential are associated with variable telomere length (Martin-Ruiz et al, 2004). It is not known if telomere heterogeneity causes the daughter cell differences in proliferative potential, which range from 0 (growth arrest) to 15 or more replications (Matsumura et al, 1985). Somatic cell replicative heterogeneity may contribute to the remarkable differences of individual life span in twins and in highly inbred worms (Finch and Kirkwood, 2000) (Section 5.2).
Contrary to earlier conclusions, adult age up to 92 years does not change the Hayflick limit of skin fibroblasts (Cristofalo et al, 2004; Goldstein et al, 1978; Smith et al, 2002). However, cells from embryos or children have 2-fold higher Hayflick limits (Martin, 1970). Thus, the major effect of age on the cell senescence model is before maturation. In vivo exposure to oxidative stress may be a factor in the reduced proliferation of cells from diabetics (Goldstein et al, 1979). Moreover, the standard protocol of culturing cells for aging studies in ambient air (20% oxygen) is grossly unphysiological. When mouse cells were ‘aged’ at 3% oxygen, closer to the tissue levels, their replicative potential was greatly increased (Parrinello et al, 2003). Nonethless, even under the standard culture conditions, in species comparisons, resistance of cultured fibroblasts to oxidative stress correlates with life span (Kapahi et al., 1999). We may anticipate fruitful further analysis of species differences in resistance to oxidative stress that might, in turn, inform about in vivo vulnerability to bystander effects from inflammatory processes.
1.2.4 Invertebrate Models
The fly and worm models are enabling highly successful studies of genetic influences because their gene regulatory systems of early development are known in detail (Davidson, 2006; Giudice, 2001; Grant and Wilkinson, 2003). Mutations modify longevity in association with altered mortality rate accelerations (Chapter 5). Some mutations that modify aging involve insulin-like signaling pathways and fat depots (Fig. 1.3A). These convergences suggest the importance of energy regulation to aging, as well as to development. The energy-regulating gene circuits have persisted during descent from shared ancestors more than 650 million years ago. Although the causes of death in fly and worm are not well defined, the causes do not include tumors or other abnormal growth during aging.
The lab worm C. elegans naturally grows among the roots of plants. Propagation by self-fertilization eliminates more genetic variation than is possible with inbred laboratory mice (Johnson et al, 2005). Free-living larva hatch about 24 h after fertilization, followed by rapid development through larval stages (L1-L4) and maturation by 72 h. If food is limited, or population density is high, the larval development may be arrested for up to 2 m in the dauer larval stages. Dauer larvae cease feeding and utilize fat depots; body movements decrease, but stress resistance increases (Kimura et al, 1997). With improved conditions, dauer larvae complete maturation and proceed to normal life spans.
Worm life history has four stages lasting 2–3 w (Huang et al, 2004): I, active egg production by self-fertilization in the first 4 d (Bolanowski et al, 1983; Herndon et al, 2002), followed by several post-reproductive stages: II, postreproductive, with vigorous movements; III, dwindling movement leading to the cessation of feeding; and IV, morbidity with little movement and accelerating mortality. Most eggs are produced during the first 4 days (Croll et al, 1977; Johnson, 1987; Klass, 1977).
While life spans in different genotypes and environments are well documented, less is known about the cellular changes and the pathology of aging. Cell death is not obvious during aging, despite the lack of somatic cell replacement. C. elegans is famous for its almost invariant cell number. Neurons look normal in ultrastructure studies throughout life, including neurons of slowed and decrepit worms (Herndon et al, 2002). However, muscle cells deteriorate in the body wall and in the pharynx, which grinds up bacteria that are the diet (Herndon et al., 2002). Lipids, lipofuscins (aging pigments), and lysosomal hydrolases accumulate in muscle and intestine cells (Bolanowski et al, 1983; Epstein et al, 1972; Garigan et al, 2002; Herndon et al, 2002), implying defects in catabolic pathways. Old worms are less resistant to pathogenic bacteria and show shorter latent period after infection (Kurz et al, 2003; Laws et al, 2004). Moreover, aging worms become constipated from bacterial packing in the intestine, which may induce oxidative damage. The usual diet of the bacteria Escherichia coli strain OP50 is considered by some to be mildly toxic; life spans are longer on heat-killed bacteria or other media (Section 2.3.2, Section 5.5.2). Long-lived mutants in insulin-like signaling (age-1) have delayed constipation (Section 5.5.2).
Although these worms are isogenic, constitutive variations in the levels of gene expression arise during development that influence later outcomes of aging (Finch and Kirkwood, 2000). Individual worms vary in the duration of these stages and in rates of aging. This extensive variability may be considered to extend variations present at younger ages in egg laying, feeding, and spontaneous movements (Finch, 1990, p. 560; Finch and Kirkwood, 2000). Individual declines of pharyngeal pumping and body movement were strongly correlated with life span in wild-type and longevity mutants (Chapter 5). For example, when fast pumping is maintained one day longer, the odds ratio for death by or later than a specified date is 1.7-fold greater (Huang et al, 2004a). The levels of expression of a stress-protective gene (hsp-16.2) in young worms predicted future life span, over a 2-fold range (Rea et al, 2005). This first example of individual difference in gene expression in the worm model supports the role of epigenetic variations arising during development that may ultimately represent chance variations in the assembly of the multiple proteins present in transcription complexes (Finch and Kirkwood, 2000). In another model, cultured mammalian cells with a reporter gene did not respond synchronously or to the same level to a diffusible inducer (Zlokarnik et al, 1998).
The fly is a more complex animal with a beating heart. At 25 °C, early development takes 24 h to larval hatching. Three mobile feeding larval stages (LILIII) take 7 more d to pupation. During the 4-day pupal stage, without feeding or movement, the adult body is formed from the imaginal disks (metamorphosis). Adult life spans are about 40 d. Adult flies can over-winter, with extended life span from cool temperature and shorter photoperiods (Flatt et al, 2005; Finch, 1990, p.313; Schmidt et al, 2005). Unlike nematodes, the fly does not have alternate larval stages equivalent to the non-feeding dauer. Juvenile hormone (JH), a series of steroid-like molecules, influences or regulates growth of all developmental stages, particularly the timing of molts and metamorphosis, and also adult diapause (Flatt et al, 2005). JH synthesis is regulated by insulin-like peptides secreted by neurons (Section 5.4). JH also regulates stress resistance and immune responses that are like innate immunity of vertebrates.
Female egg-laying declines exponentially after a fairly stable phase, also observed in the medfly (Ceratitis capitata) (Novoseltsev et al, 2004). Both species have post-reproductive phases that only weakly correlated among individuals with the cessation of egg-laying. As with the worm, somatic cells are not replaced. Major damage is accumulated to the brittle exoskeleton from wear-and-tear (Finch et al, 1990). Unlike the worm, the aging fly shows some indication of neuron loss, in the mushroom body (Technau, 1984). The fat body, a key organ of energy reserves and immune function, gradually atrophies (Finch, 1990, p. 63). Apoptosis with DNA fragmentation increases in flight muscles and fat body (Zheng et al, 2005). The heart rate slows during aging and arrests more easily under the stress of electrical pacing, aging sharply (Wessells et al., 2004) (Section 5.6.3, Fig. 5.7). Insulin-signaling mutants with increased life span have delayed cardiac aging (Chapter 5). Little is known about vasculature of aging flies; other aging insects show indications of circulatory blockage (Arnold, 1961, 1964; Finch, 1990, p. 65).
1.2.5 Yeast
Yeast cells are similar to animal cells in their core biochemistry and organelles. We should not be surprised that the 6000 yeast genes include orthologues of insulin-like signaling and other genes in tissue-grade animals (Fig. 1.3A). Fungal genomes diverged from the animals about 1500 million years ago (Cai, 2006). Aging and life span in yeast are studied with two very different experimental models: replicative life span (Piper, 2006; Sinclair et al, 1998) and chronological life span (Fabrizio and Longo, 2003; Fabrizio et al., 2004).
The yeast replicative life span is defined by asexual reproduction through the formation of smaller buds on the surface of the mother cell. The intervals between budding lengthen as the replicative life span is approached, at about 20 cell divisions. Oxidatively damaged proteins are retained asymmetrically by the mother cell (Aguilaniu et al, 2003), which may be how the detached buds start the replicative clock at zero, independent of mother cell age. The replicative life span model resembles the Hayflick model in that both show limited cloning. The sterile postreplicative cells may have considerable remaining life span (V. Longo, personal comm). Mechanisms in replicative aging include a unique genomic instability in ribosomal DNA (rDNA) cistrons, through aberrant recombination that causes the accumulation of extra-chromosomal rDNA circles. The rDNA instability is modulated by chromatin condensation under the control of Sir2 (silent information regulator), a NAD-dependent histone deacetylase. Increased Sir2 inhibits the aberrant recombination and extends the replicative life span. Sirtuins and their orthologues have many other metabolic activities in animals (Chapter 3 and 5); e.g., diet restriction activates Sirt1 and modulates lipolysis in mammalian fat (Wolf, 2006).
The chronological life span is defined as the cell viability during prolonged periods with limited external nutrients. Yeast and other autotrophic fungi have evolved adaptive mechanisms in their natural habitats for surviving extended periods of starvation, pending episodes of surfeit. When switched from growth media to water, yeast cells become hypometabolic, extending their life span several fold to 15–20 days. Mutations in the kinase Sch9 increase life span by increasing stress resistance and glycogen reserves. Sch9, a functional homologue of Akt/PBK, which modulates life span in animals (Fig. 1.3A), also synergizes with Sir2 (Longo and Kennedy, 2006).
Ongoing studies point to the convergence of mechanisms in these seemingly different models, by the shared dependence of replicative and chronological life spans on Sch9, Ras/cyr/PKA, and Tor pathways (Longo and Kennedy, 2006). The formation of rDNA circles may be regarded as a ‘yeast disease of aging’ specific to the replicative senescence mode. Besides these single cell models, yeast can also grow as filaments (pseudohyphae) (Gognies et al, 2006), which enable the invasion of ripe fruit. Dense fungal mats can form, possibly including domains with metabolic gradients. These alternate life history modes with complex morphology have not been studied for aging processes.
1.2.6 The Biochemistry of Aging
Aging increases the load of oxidative damage in DNA, lipids, and proteins, yeast to humans (Sohal and Wendruch, 1996; Finch, 1990). Free radicals (ROS, reactive oxygen species) generated by mitochondria are a major source of oxidative ‘bystander’ damage. Other damage comes from extracellular ROS generated by macrophages, as is prominent in atheromas. Additionally, DNA, lipids, and proteins become glycated in an oxidizing process that is chemically driven by glucose and other sugars in tissue fluids. These advanced glycation endproducts (AGEs), while not initiated by free radicals, can generate ROS in further complex reactions and by activating macrophages (RAGE pathway, receptor of AGE), discussed below. In the following discussions of adverse effects of ROS, we must be mindful that ROS are essential in functions of the brain, heart, and many other organs that employ ROS in signaling processes. As examples from this large field, in the brain superoxide modulates synaptic plasticity (Hu et al, 2006), whereas in the heart nitric oxide modulates contractility (Massion et al, 2005).
Intracellular ROS is mainly derived from normal mitochondrial respiration (Barja, 2004; Wallace, 2005). The respiratory chain releases electrons that form the superoxide anion (O2.–) by single electron reduction of O2 (Fig. 1.11). Enzymes of free radical homeostasis include catalase; two types of supraoxide dismutase (SOD)—Cu/ZnSOD and MnSOD; and glutathione peroxidase. Superoxide is enzymatically converted by superoxide dismutase (SOD) into H2O2, which is then catalytically degraded by transition metals to the highly reactive hydroxyl radicals. These reactions are limited by the enzymatic degradation of H2O2 by catalase, or by glutathione peroxidase. H2O2 diffuses freely across cell membranes, unlike superoxide.
Pathways of hydrogen peroxide metabolism. Molecular oxygen (O2) is reduced by loss of electrons to form superoxide (O2–, 1e–) or hydrogen peroxide (H2O2, 2e–). Superoxide spontaneously reacts with nitric oxide (NO) to form peroxynitrite radicals (ONOO–). Most H2O2 forms spontaneously, or from the dismutation of O2– by SOD and is used in cell signaling. H2O2 is degraded by intracellular catalase (CAT), extracellular glutathione peroxidase (Gpx), or thiols. (Adapted from Cai, 2005.)
ROS are strongly associated with mitochondrial DNA damage (deletions, rearrangements, and point mutations). The age-related increase of damaged mitochondria DNA (Wallace, 2005; Chomyn and Attardi, 2003) has become a centerpiece in the molecular pathophysiology of aging (Brookes et al, 1998; deGrey, 2005; Harper et al, 2004; Van Remmen and Richardson, 2001). Mitochondrial dysfunctions are found in many disorders of aging, e.g., Alzheimer disease, atherosclerosis, atrial fibrillation, diabetes, deafness, muscle atrophy, retinal degeneration. However, cause and effect are not well resolved in these long-term processes of cell degeneration. Mitochondrial production of ROS increases with age in rat liver and muscle (Bevilacqua et al, 2005; Hagopian et al, 2005; Harper et al, 2004). ‘Proton leak’ across the inner mitochondrial membrane regulates mitochondrial ROS production with high sensitivity and increases during aging (Brookes et al, 1998; Brookes, 2004; Hagopian et al, 2005; Harper et al, 2004). Mitochondrial oxidative damage to DNA and proteins is often attributed to endogenously generated mitochondrial ROS. Because proton leak increases with oxidative damage, progressive mitochondrial impairments of various types may arise during aging through subcellular bystander damage, which propagates cell oxidative damage (Brookes et al, 1998; deGrey, 2005; Harper et al, 2004).
According to the oxidant stress theory of aging, life span should be influenced by levels of enzymes or anti-oxidants that produce or remove free radicals (ROS, NOS) (Bokov et al, 2004; Sohal and Weindruch, 1996; Stuart and Brown, 2005). The role of ROS is being tested in transgenic flies and mice by varying the levels of catalase and SOD that remove ROS (Landis and Tower, 2005; Mele et al, 2006). In flies, transgenic overexpression of mitochondrial Cu/ZnSOD increased life span by >35%, while catalase overexpression did not increase life span, reviewed by Landis and Tower (2005). Mice with partial deficits of MnSOD (heterozygote knockout, Sod2+/–) lived slightly longer (Van Remmen et al, 2003). Although the 2.5% difference was not statistically significant, the survival curves show little overlap. This careful study also showed that SOD2 deficiency increased DNA oxidative damage (8-OH dG) and tumor incidence several-fold—e.g., lymphomas 61% versus 22%. The lack of SOD2 deficiency on skin collagen glycol-oxidation is discussed below and in Section 1.4.2.
From these results and more systematic species comparisons (Kapahi et al, 1991), I suggest that anti-oxidant mechanisms may be related to the levels of molecular turnover and repair. Flies may show these stronger effects on the life span than rodents because adult flies have no somatic cell replacement and, probably, less protein turnover, which, in mammals, removes oxidative damaged molecules. Long-lived organisms may have needed to evolve more effective repair processes (see Section 1.2.8).
Transgenic overexpression of catalase in mitochondria (mCAT) in mice increases life span by 20% (5 months) and delays important pathology (Schriner et al, 2005). This study is exemplary for its genetic design, detailed histopathology, and animal care (husbandry), even reporting the infection rate in sentinel mice. Mortality accelerations were right-shifted by increased mCAT, but without change in slope, implying that aging was delayed. Tissue changes are consistent with the Gompertz interpretation that aging is delayed. At middle age, cardiac pathology was decreased (fibrosis, calcification, arteriolosclerosis), which are common causes of congestive heart failure in human aging. In skeletal muscle, DNA oxidation (8-OHdG) and mitochondrial deletions were decreased. These findings directly link decreased mitochondrial ROS to heart pathology, which is recognized as of inflammatory origin (Query II). In a mouse model of accelerated atherosclerosis (apoE-knockout, apoE–/– with extreme hypercholesterolemia on standard diets), the systemic overexpression of catalase decreased aortic atherosclerosis (lesion area) by 66% and decreased F2-isoprostanes (lipid oxidation product) in plasma by 45% (Yang et al, 2004). Aortic lesion size correlated strongly with aortic isoprostane levels, again consistent with the importance of oxidized damage in inflammation. Cu/Zn-SOD had smaller effects on aortic lesions or lipid oxidation, specifically implicating hydrogen peroxide.
The hyperglycemia of diabetes is associated with another source of oxidative damage through glycation, which has not been well integrated into the free radical theory of aging. Glucose and other reducing sugars can oxidize and cross-link proteins by spontaneous and complex chemical reactions with lysine and arginine sidegroups yielding ‘advanced glycation endproducts’ (AGEs) that include highly reactive carbonyls (ketones and aldehydes) (Biemel et al, 2002; Monnier et al, 2005; Stadtman and Levine, 2003). Carbonyls also form by many other free radical reactions (Stadtman and Levine, 2003). Other targets of glycation are lipids (ALE, advanced lipid endproducts) (Baynes, 2003) and DNA (Bucala et al, 1984).
AGE adducts accumulate progressively during aging in extracellular matrix proteins as cross-links that reduce vascular and skin elasticity (Hamlin and Kohn, 1971; Monnier et al, 2005). Aortic stiffening causes progressive increases in systolic blood pressure (Fig. 1.6B) and pulse wave velocity (De Angelis et al, 2004) that are underway early in adult life. The formation of atheromas is superimposed on these slow arterial aging processes. Diabetes accelerates these vascular and lens changes, implying the importance of glucose and other blood sugars in damage to long-lived proteins during aging (Cerami, 1985). Conversely, AGE formation is slowed by diet restriction, which lowers blood glucose (Chapter 3). The lack of SOD2 deficiency on skin collagen glyco-oxidation (carboxymethyl lysine and pentosidine) in the study of van Remmen et al. (2003), discussed previously, points to the role of blood glucose, rather than extracellular ROS in glyco-oxidation (see below). As discussed below, AGE adducts participate in oxidative stress and inflammation.
Pentosidine was the first chemically characterized AGE cross-link identified in tissues (Sell and Monnier, 1989). Skin collagen pentosidine accumulates progressively during aging in many species, and the accumulations are accelerated by hyperglycemia and diabetes (Sell and Monnier, 1990). However, pentosidine accumulations are dwarfed by glucosepane, a recently characterized AGE that is 50-fold higher than pentosidine in skin collagen (Biemel et al, 2002; Monnier et al, 2005; Sell et al, 2005). Over the life span, glucosepane is added to about 1% of skin collagen arginine and lysine residues. This is equivalent to cross-linking of every five collagen molecules of normal individuals and every other collagen molecule in diabetics. Lens proteins accumulate far less glucosepane than skin collagen (Biemel et al, 2002). We do not yet know the specific contribution of glucosepane and diverse minor glycation products to cross-linking in skin and vascular stiffening.
Besides pentosidine and glucosepane, more than 20 other adducts derive from glucose, pentose, and ascorbate. AGEs form readily in test-tube reactions of proteins with glucose or other reducing sugars through Amadori and Maillard chemistry. The resulting brownish, autofluorescent mixtures are models for brunescent cataracts and other in vivo sites of AGEs (Cerami, 1985; Monnier et al, 2005). The aorta also accumulates fluorescent AGEs (Fig. 1.6D). Oxygen levels are critical to AGE chemistry and may degrade Amadori products (Ahmed, 1986). Glucosepane formation, however, forms directly from reactions that do not depend on oxygen, and is influenced by competing reactions; e.g., in the lens, the lower glucosepane may be due to high levels of methylglyoxal (Sell et al, 2005). Tissues also differ in enzymatic removal of AGEs (deglycating amidoriases) (Brown et al, 2005).
Of critical importance to inflammation, AGE adducts activate scavenger receptors ‘RAGE’ (receptors for AGE) on macrophages and many other cells that stimulate the production of ROS via NAD(P)H oxidases (gp91phox et al.) and electron transport (Fan and Watanabe, 2003; Schmitt et al, 2006). RAGEs are also activated by the amyloid β-peptide of Alzheimer disease and by AGEs present in cooked foods (Lin, 2003; Uribarri et al, 2003) (Chapter 2). AGEs and RAGEs appear to mediate feed-forward loops of oxidative stress and inflammation that increase bystander molecular damage in atherosclerosis, Alzheimer, and other chronic inflammatory diseases (Lu et al, 2004; Ramasamy et al, 2005) (Queries II and III).
RAGE activation also releases cytokines (e.g., IL-6) and leukocyte adhesion factors (e.g., MCP-1 and VCAM-1). Feedback loops induce RAGE by TNFa through production of ROS, mediated by NFkappaB (Mukherjee et al, 2005). RAGE signaling pathways utilize familiar workhorses in inflammation and oxidative stress, including the transcription factor NFkappaB and PI3K (Dukic-Stefanovic et al, 2003; Xu and Kyriakis, 2003). Moreover, PI3K interfaces with other signaling systems implicated in longevity (Fig. 1.3A). Lastly, RAGE activation may stimulate feed-forward ‘vicious cycles’ by autoinduction in the same cell (Basta et al, 2005; Feng et al, 2005; Wautier et al, 2001). RAGE-dependent processes are a major focus in atherosclerosis, particularly inflammation of arterial endothelia by AGE during diabetes (Feng et al, 2005; Naka et al, 2004; Ramasamy et al, 2005) (Section 1.5.1). RAGE-dependent processes are also implicated in Alzheimer disease and cancer. These observations are consistent with Query II that inflammation causes further bystander damage and Query III that nutrition influences bystander damage by AGE production from hyperglycemia and by AGE present in cooked food.
The molecular life span (turnover or half-life, t1/2) is a major determinant of accumulated damage, as exemplified by AGE accumulation in arterial elastin (Fig. 1.6D). In arteries and lungs, elastin may be almost as old as the individual, as evaluated by two independent measures: D-aspartate (Powell et al, 1992; Shapiro et al, 1991), which accumulates linearly through spontaneous racemization (Fig. 1.6D) (Helfman and Bada, 1975) and by ‘bomb-pulse’ 14C radiolabeling1 (Shapiro et al, 1991). Human aortic elastin and cartilage collagen have t1/2 >100 y, while skin collagen is 15 y. With lab tracer labeling, rodent elastin has t1/2 of months to years (references in Martyn et al, 1995; Shapiro et al, 1991). Elastin progressively accumulates glyco-oxidation (AGE) (Fig. 1.6D), at the same rate as collagen, when corrected for turnover (Verzijl et al, 2000). Damage to arterial elastin and collagen contributes to the loss of elasticity and stiffening that cause the increase of blood pressure during aging (Fig. 1.6B) (Section 1.6.3, below). In Alzheimer disease, senile plaque amyloid and neurofibrillary tangles also include very long-lived proteins (also bomb-pulse 14C) (Lovell et al, 2002). Other very long-lived proteins accumulate D-aspartate in tooth dentine, eye lens, and in brain white matter myelin. The accumulating oxidative damage to these life-long molecules is associated with creeping dysfunctions in arteries, skin, and eye lens; the role in myelin dysfunction is not known.
In contrast, molecules with short life spans of days to weeks have less oxidative damage. Diabetics accumulate glycated hemoglobin A1c, for example, which turns over at the erythrocyte t1/2 of about 120 d. Erythrocyte turnover scales with body size across species (M0.18) (Finch, 1990, p. 289). The t1/2 of many proteins is allometrically related to body size and may be a crucial determinant of the rates of damage accumulated during aging across species. Moreover, the rates of basal metabolism correlate with molecular turnover in species comparisons of mammals. The insulin-like signaling pathways (Fig. 1.3) may mediate many of these fundamental energy relationships.
Aging slows the turnover of many shorter-lived proteins (Finch, 1990, pp. 370–373; Goto et al, 2001)—e.g., bulk proteins in worm (Reznick and Gershor, 1979) and in mouse liver (Lavie et al, 1982; Reznick et al, 1981). The causes of slowed turnover during aging are not known and could include impaired proteasomal degradation, as in aging rodents (Goto et al, 2001). These metabolic level aging processes thus tend to accelerate the accumulation of oxidized damage. The effectiveness of diet restriction in slowing aging may be due in part to the accelerated protein turnover and decreased oxidative load (Chapter 3).
The balance of reduction: oxidation (‘redox’) in glutathione and other key homeostatic regulators (Fig. 1.11) is shifted to a more oxidized state (GSSG and protein-SSG) in blood, liver, and other tissues (Lang et al, 1989; Lang et al, 1990; Rebrin et al, 2003; Rebrin and Sohal, 2004), and in whole aging flies (Rebrin et al, 2004). Glutathione, the major redox couple, is at much higher levels than other redox links involving cysteine, thioredoxin, NAD, etc. (Sies, 1999). In healthy aging humans, blood GSH remains relatively stable; however, with cardiovascular disease, diabetes, or kidney disorders, blood GSH tends to drop below the normal range (Lang et al, 2000). Similar redox shifts occur during chronic infections; e.g., in HIV patients, blood GSH decreases in proportion to the viral load (Sbrana et al, 2004). Conversely, redox shifts are opposed by diet restriction (Chapter 3, Fig. 3.14). A component of the GSH shifts of aging could also be a response to low-grade infections, which would also be consistent with the increase of CRP, IL-6, and other acute phase reactants in aging populations (see below and Chapter 2). It is important to resolve the contributions to the oxidized load of aging from three sources: (I) endogenous mitochondrial free radicals and other tissue processes; (II) interactions with the commensal gut and skin flora; and (III) specific infections.
The naked mole rat (Heterocephalus glaber) is adding surprising findings to these debates. H. glaber is the most longevous rodent (at least 28 y), yet has similar body weights of lab mice (30–80 g) (Andziak et al., 2006; Andziak and Buffenstein, 2006). In comparison with lab mice (CB6F1) at 10% of the lifespan (24 m vs. 4 m), H. glaber had indicators of greater oxidative stress than in lab mice, e.g., 10-fold more urinary isoprostanes, 35% higher myocardial isoprostanes, and 25% lower hepatic GSH:GSSG ratios. While it might be concluded that sustained oxidative stress is not incompatable with extraordinary longevity, much remains to be learned about other aspects of metabolism in these remarkable animals. Its membrane lipids differ from other mammals by much lower levels of unsaturated fatty acids in muscle and brain, particularly docosahexaenoic acid (DHA or 22:6 n-3), which is highly susceptible to peroxidation (Hurlbert et al., 2006). The oxidizability of membranes (peroxidation index) fits well with the inverse allometry on lifespan.
These varying results across species suggest that anti-oxidant mechanisms vary within and between phyla. Besides differences in membrane composition and antioxidants, I suggest the importance of molecular turnover and repair. Flies may show these stronger oxidative effects on the life span than rodents because adult flies have no somatic cell replacement, except in the gut, and, probably, less protein turnover, while mammals have extensive molecular turnover, which removes oxidative damaged molecules. Long-lived organisms may have needed to evolve more effective repair processes (see Section 1.2.8).
1.2.7 Biomarkers of Aging and Mortality Risk Markers
For populations, the life span is expressed statistically, often as the life expectancy. However, no measurement has been found that accurately predicts the individual life span from the genotype, or from any ‘biomarker of aging.’ Clearly, the traditional aging changes of gray hair and menopause do not assess an individual’s current health or future longevity. Jeanne Calment lived 70 years after menopause to achieve her longevity record of 122 years.
The N.I.A. has supported an extensive search for biomarkers that predict remaining life span (Biomarkers of Aging Program, begun in 1982) (Reff and Schneider, 1982). Biomarkers have been evaluated in biochemical, cellular, genetics, molecular, and physiological characteristics, and in behavior and cognition. Emphasis has been given to changes of ‘non-pathological aging’ that are distinct from disease. Two decades later, no single biomarker or combination has been found to predict longevity better than the individual age in fly, worm, rodent, or human (Finch, 1990, pp. 558–564; Warner, 2004).
Consider the limits of biomarkers in aging worms, which seem an optimal model for actuarial questions by their minimal genetic, environmental, and social heterogeneity from dominance hierarchies and social interactions (worms are self-fertilizing). Pharyngeal pumping, by which food is ingested, declines progressively during the first week and then nearly ceases a few days before death. Body movements closely parallel the pumping rates, not surprisingly because pumping provides the food needed for energy. The duration of fast pharyngeal pumping and body movements shows strong correlation with individual life spans (Huang and Kaley, 2004). When fast pumping is maintained one day longer, the odds ratio for death by, or later than, a specified date is 1.7-fold greater. The Spearman rank correlation coefficient for the duration of fast pumping and remaining life span was highly significant (r 0.49, P< 0.0001). Despite this statistic, the fraction of life span variance explained by pumping or movement was only 24%. Thus, other variables besides pumping account for about 75% of the individual differences in life span. Mutants with greater longevity have longer phases of active body movement and pharyngeal pumping. (Section 5.5.2).
Even at hatching, worms differ hugely in movements, which Kirkwood and I attributed to chance variations in cell organization and gene expression during development (Finch and Kirkwood, 2000, pp. 58–65). A concrete example is Tom Johnson’s elegant study of worm-to-worm variations in expression of a stress-protective gene (hsp-l6.2) in young worms, which correlated strongly with individual life spans (Rea et al, 2005). Again, the Spearman coefficient of 0.48 accounts for only 25% of the variance in life span. These sobering examples from rigorous studies suggest limits in the predictability of life spans despite strictly controlled genetics and environment. In worms, as in flies, mice, and humans, the heritability of the life span is also about 25% (Finch and Tanzi, 1997; Finch and Ruvkun, 2000) (Section 5.2).
As noted above, partial deficits of MnSOD did not alter mouse life span (Van Remmen et al, 2003). Several biomarkers of aging that respond to diet restriction were the same in Sod2+/– as in control aging mice (ad libitum fed in this study): both genotypes had identical age-related decreases of spleenocyte proliferation and increases in skin collagen of pentosidine and carboxymethyllysine. The lack of effects of MnSOD deficiency on these AGEs indicates that blood glucose was not altered. However, Sod2+/– mice had greater accumulations of oxidized nuclear DNA (8oxodG) in brain, heart, and liver, and 2-fold more lymphomas (83% vs. 41%). This study with its careful analytical chemistry and histopathology shows the uncertainty of connections between robust biomarkers of aging, tumor prevalence, and the life span.
There is reason to consider the individual disease load as more informative than tissue-level aging changes in predicting mortality risk (Karasik et al, 2005). During these same decades of the Biomarkers Program, vast clinical research has developed risk indicators of mortality for the major diseases. From the clinical perspective, disease, not aging, is the cause of mortality, as shown in the exponential increase of tissue lesions in rodent models (Fig. 1.5) and heart attack and stroke in human populations (Fig. 1.6C). Systolic blood pressure may be the most robust overall indicator of human mortality risk, with exponential age-related increases of future heart attack and stroke at all levels of systolic pressure.
The next phase of the biomarker debate may redefine the often vaguely used term ‘disease.’ For example, the clinical threshold of hypertension as a target for intervention is being expanded to include the ‘high normal’ range. A few decades ago, an informal guideline of the expected systolic pressure was “add 100 to the person’s age”! Combinations of risk indicators and disease load are also being intensively studied of the inflammatory marker, C-reactive protein (CRP). In the Women’s Health Study, future heart attacks occurred most frequently in those with both elevated CRP and LDL (Ridker, 2002). New models are needed to resolve the links between the diverse subtle subclinical aging changes that interact to cause circulatory failure on the background of declining organ reserves. It is shocking that 30% of diet-restricted old rats had no gross lesions at necropsy and cause of death was unknown (Shimokawa et al, 1993). Declining homeostasis of glucose and electrolytes, for example, might allow transient disturbances that would arrest a fibrotic heart, even in the absence of thrombotic blockade. The combinatorics of various mild dysfunctions gives rise to a huge number of individual pathophenotypes that may be each estimated as relative risk of mortality, but may never account for more of the variance in life span than in the worm model.
1.2.8 Evolutionary Theories of Aging
The demographics of natural populations are the basis for evolutionary theories of aging. In humans, like most other animals, the major phase of reproduction is accomplished by the young adults. Mortality from arterial and malignant diseases is low until after age 35, which approximates the life expectancy in most human populations before the 19th century (Fig. 1.1A). High levels of extrinsic mortality allow only a minority of humans to survive to older ages, until very recently. The major early causes of mortality are due to extrinsic risks, including infections, malnutrition, and trauma. A demographic pyramid with young adult ages as the largest is widely observed in natural populations. Sufficient reproduction to maintain the population can be distributed in many combinations of age groups, all governed by the level of mortality that allows a sufficient number to survive to adult ages and to reproduce sufficient numbers of offspring, which themselves survive to become reproductive. The reproductive schedule includes the duration of maturation, when individuals are at risk for dying before reproducing, as well as the frequency of reproduction and duration of the reproductive phase. The duration of the reproductive schedule is the prime determinant of lifespan (Hamilton, 1966; Austad, 1993, 1997; Rose, 1991, 2004).
Populations may be described by the Euler-Lotka equation, which is based on life table calculations as the net reproductive rate averaged over age classes, x (Hamilton, 1966; Rose, 1991; Stearns and Koella, 1986). The rate of population growth (r) is calculated at each age class (x) from the sum of the products of the mortality rate m(x) and the reproductive rate b(x).
The net population growth can not be negative for very long, or extinction results. Thus, increases in mortality m(x) must be compensated by commensurate increases in reproduction in one or another age b(x). Note that this equation is the sum of terms that are commutative products in each age, e.g., the number 8 is the product of (2 × 4) or (4 × 2) etc. Thus, an infinite number of different combinations of m(x) and b(x) can satisfy this equation. Because life expectancies are determined by mortality rates, we may speak naturally of the lifespans in plural that characterize a species. Lifespans are highly plastic and changeable in relation to the reproductive schedule and reproductive output. Experimental tests of these relationships are described below.
The duration of lifespans can vary widely in conjunction with the reproductive schedule. At one extreme are species that die after producing vast numbers of fertilized eggs, of which few survive to adulthood, such as in the five species of Pacific salmon (Finch, 1990, pp. 83–95). At the other extreme are species like the great apes that typically have one child at prolonged intervals, of up to ten years in orangutans (Wich et al, 2004).
Differences in extrinsic mortality between populations may lead to different reproductive schedules. In the common opossum (Didelphis virginiana), populations on isolated islands, which are exposed to low predation, are observed to mature more slowly and have fewer offspring per litter (Austad, 1993, 1997). Moreover, aging seems to be delayed, with slower mortality accelerations, slower collagen aging and longer lifespans than mainland opossums, which are under higher predation. These observations are consistent with the hypothesis that senescence is delayed if predation levels are lower in species comparisons (Edney and Gill, 1968), and Steve Austad’s corollary that aging should also be modified commensurate with the reproductive schedule as external predation varies.
Human groups also vary in schedules of maturation and reproduction (Chapter 4, Fig. 4.5C). For example, in pre-industrial foragers, menarche occurs in the Ache by 14 y and first child at 17.7 y, while the Jo/’hoansi menarche is at 16.6 and first child at 18.8 (Walker et al, 2006). The precocious extreme is found in privileged populations with excellent health from abundant food and low loads of infections, where girls have much earlier menarche 12 years or less, closer to menarche in the great apes (Fig. 6.2). While we do not know the relative contribution of environmental factors and genetics in these populations, twin studies show that growth rates, age at menarche, and age at menopause have significant heritability (Chapter 5.2).
The force of natural selection is considered to decline during aging (Fig. 1.12A) because the majority of reproduction is achieved by the younger adults, which are the dominant age group in most natural populations due to external causes of mortality distinct from aging (Hamilton, 1966; Rose, 1991). The declining force of natural selection during aging is then considered permissive for the accumulation over time of mutations in the germ-line with delayed adverse effects emerging in adults. Examples include rare dominant familial genes for Alzheimer disease, breast cancer, or hyperlipidemia, which have little impact on young adults. Such delayed consequences of adverse mutations are not strongly selected against when effects are delayed to later ages that contribute less to reproduction. If a dominant gene caused earlier dysfunctions, its carriers would have difficulty competing for mates, and the early age phenotype would soon disappear. These concepts were introduced by J.B.S. Haldane (1941) and Peter Medawar (1952), and developed into rigorous theory by George Williams (1957) and William Hamilton (1966). In another hypothetical case, genes with adverse later effects might be selected by their benefit during development or in early ages—Williams’ ‘antagonistic pleiotropy’ hypothesis (Williams, 1957; Williams and Nesse, 1998). Thus, the schedule of reproduction for any species to survive (maintain Darwinian fitness) sets a lower age limit in the development of adverse phenotypes.
Many studies show the power of experimentally manipulating the reproductive schedule. In a now classic paradigm in the experimental evolution of aging, flies were selected for reproduction at later ages (Rose and Charlesworth, 1980; Rose, 1991). Within 15 generations, this selection regime yielded flies that lived about 20% longer and had more gradual mortality acceleration (Luckinbill et al, 1984; Rose, 1984; Rose et al, 2004) (Fig. 1.12B). Conversely, selection for early age reproduction shortened life span (Fig. 1.12B). These powerful effects require outbred populations and do not depend on the spontanteous emergence of new mutations and their fixation. Rather, shifts in genotype are attributed to frequency change of existing alleles, which occurs at a vastly greater rate than the mutation frequency. Moreover, the reproductive schedule and lifespan can be shifted back toward initial values in a “reverse evolution” paradigm, by switching the ages of selection (Teotonio et al, 2002; Rose et al, 2004).
These studies also show important gene-environment interactions in the density-dependence of lifespans, which were sensitive to population density in both larval and adult stages (Mueller et al, 1993; Rose et al, 2004). Densitydependent effects could involve microbial flora which grow on excreta and dead larvae. Bacteria (Hurst et al, 2000) and fungi (Rohlfs, 2005) are well known to influence larval growth and even the age of phenotype expression (Hurst et al, 2005). The microbial environment could also be important in aging fly populations that have greatly increased microbial load (Renn et al, 2007; Section 5.6.4).
However, unexpected results have come from studies of the guppy (Poecilia reticulata), which experimentally varied predation pressure (Reznick et al, 2001; Reznick et al, 2006). In some localities with intense predation and with high extrinsic mortality, fish mature earlier and have higher early fecundity. These different reproductive schedules show heritability that persists in the laboratory. Just as in the fly experiments, altering the extrinsic mortality from predation pressure in wild populations caused corresponding advance or delay of reproduction. When guppies were allowed to live out their days in the lab, the early maturing fish reproduce for an additional 200 days longer than the later maturing fish, contrary to the natural populations of opossums or the lab fly studies above. Moreover, the post-reproductive life spans did not differ between the early and late reproducing lines. These intriguing results challenge the evolutionary theory of senescence that was comfortably accepted for several decades.
Moreover, not all animals show reproductive senescence at advanced ages. Two species of turtles (painted, Blanding’s turtles) have not shown decline in fecundity at ages over 60 y in longitudinal field studies (Congdon et al, 2003). Some older individuals continue to be successful in laying and protecting their egg clutches to hatching, as much as the average young adult. Similarly, the very long-lived rockfish (Sebastes) maintain seasonal cycles of de novo oogenesis at least up to age 80, and does not show evidence for declining fecundity (Finch, 1990, pp. 216–219; Finch, 1998; De Bruin et al., 2004).
These and many other examples support the possibility of ‘negligible senescence,’ or negligible decline in reproduction and other functions during the natural life span (Finch, 1990, pp. 207–247; Finch, 1998). ‘Negligible senescence’ as a potential life history type was considered theoretically untenable by most biogerontologists when I first proposed it, at least in part because of Hamilton’s highly regarded mathematical model, which concluded “… even under utopian conditions [of exponential fertility increase and immortality], given genetic variation… phenomena of senescence will tend to creep in” (Hamilton, 1966, p. 25). However, further theoretical developments by James Vaupel and colleagues (2004) show that the level of senescence is highly sensitive to assumptions and parameter values and that there may even be ‘negative senescence’ under some conditions.
Another challenge to evolutionary theory of senescence is the stable characteristics of senescence in related species, whereas evolutionary theory of the weak selection against later deleterious phenotypes should yield a great variety of different aging genes and great individual variability in aging. Indeed, there is great variability in life span, and specific aging phenotypes is certainly manifest between individuals. And, as well studied in twins, life span has low heritability (Section 5.2). Nonetheless, at the species and population level, some aging changes are so generally found that they may be described as canonical patterns (Table 1.1). In all well-studied human populations, arterial elasticity decreases and systolic blood pressure increases with adult age; menopause occurs by age 55 due to depletion of ovarian oocytes; reproductive tract tumors become increasingly prevalent; and bone density decreases, while the risk of disabling fractures increases. These canonical changes of human aging also characterize aging in other mammals (Finch, 1990, pp. 154–155), including details of arterial aging (Table 1.4, Section 1.5.3.3).
The basis of the canonical patterns in aging may be sought in development: Mammals share a basic body plan and program of embyronic gene regulation that determines the patterns of gene expression in differentiated cells. The human genome regulatory machinery was established at least 150 million years ago in mammalian ancestors and is apparently resistant to evolutionary change because of the densely connected transcriptional circuits required for embryonic development (regulatory kernels) (Davidson, 2006). Future studies may show the level of gene circuit regulatory lock-in that determines the replacement of molecules and cells in adults. As discussed above (Section 1.2.6), arterial elastin and other long-lived molecules inevitably accumulate molecular damage (Fig. 1.6D), whereas in circulating erythrocytes, the genomically programmed cell replacement with 120 day lifespans allows much less persistence and impact of oxidative damage. Thus, we must look to genomic regulation during development to understand the cell phenotypes that underly tissue aging processes and that may also be the basis for species differences of lifespan.
Molecular and cell repair, replacement, and organ regeneration require metabolic energy. Tom Kirkwood’s ‘disposable soma’ theory of aging recognizes that energy resources are finite (Kirkwood and Holliday, 1979; Drenos and Kirkwood, 2005). Molecular and cell repair and regeneration require metabolic energy. At each moment, the individual organism assesses its homeostatic condition and allocates energy accordingly. Insufficient food intake attenuates immunity (Chapter 3) and growth (Chapter 4). We will examine this in the impact of diet restriction, which can attenuate immune defenses (Chapter 3), while infections attenuate growth (Section 4.6) and reproduction (Section 5.4.2). Each species has evolved within a relatively predictable balance of energy availability and energy that must be allocated for reproductive success. As one of many examples, when worker bees leave the hive to forage, a highly dangerous activity that accelerates mortality (Finch, 1990, p. 70), they decrease their immune defenses (Amdam et al, 2005). This trade-off of foraging energy against immune defenses is statistically favored because of the ultra-short life span of field bees. Theoretical modeling of trade-offs between investment in somatic maintenance and fecundity (Darwinian fitness) shows very broad curves of fitness, with optima that allow considerable variation in somatic investment (Drenos and Kirkwood, 2005). These results are consistent with observations that many biological functions do not decline appreciably and that major species differences are to be expected. A future goal of aging theory is to understand the physiological trade-offs of immunity, repair, and regeneration that define the species reaction norm of mortality rates to environmental fluctuations (Stearns and Koella, 1986).
1.3 OUTLINE OF INFLAMMATION
Inflammatory processes of innate immunity are evident participants in many tissue changes of normal aging and most of the chronic degenerative diseases of aging, as outlined in Fig. 1.2A and discussed throughout this book. Innate immune responses are the standing initial defense system against invading pathogens. The acute phase of innate immunity is mediated by secretion of systemic ‘acute phase’ proteins and the local activation of macrophages with rapid production of free radicals (Fig. 1.11). Acute phase responses do not depend on prior immune experience of adaptive (instructive) immunity, mediated by B- and T cells. However, exposure to new antigens during the acute phase response can activate the targeted immune responses to pathogen antigens by B- and T cells. Tissue injury from trauma or toxins can also induce inflammatory processes that go far beyond free radical production in tissue matrix remodeling and repair. Host defenses require energy, which is allocated in trade-offs, as just discussed, that determine the duration and type of inflammatory responses and the level of repair and regeneration.
The inflammatory processes at work in atherosclerosis, Alzheimer disease, diabetes, and obesity include a shared set of acute phase responses (Sections 1.5, 1.6, 1.7). Some of the same processes that macrophages employ to remove bacterial invaders are also implicated in arterial disease through the uptake of lipids by macrophages. Blood lipid responses to infections (lipid oxidation, elevated acute phase proteins, triglyceridemia) are also an atherogenic profile. Moreover, emerging evidence indicates roles for T cells in atherosclerosis (Section 1.5.3). Other tissue lesions of aging that involve chronic inflammation may also arise from an early seeding injury. Gastro-intestinal cancer, for example, is associated with local inflammation in response to H. pylori infections (Section 2.8.1). Remarkably, many of the acute phase genes are also upregulated in normal aging, in the absence of these specific lesions (Section 1.8). Cause and effect are unresolved in these complex, long-term processes.
1.3.1 Innate Defense Mechanisms
The acute phase of the innate immunity (‘emergency line’ ring-ups’) include the ancient cardinal signs of inflammation in a localized injury: heat (calor), redness (rubor), swelling (tumor), and pain (dolor).2 ‘Inflammation’ is now understood to include the vastly complex, multi-organ system of defense and repair, in which free radicals have major roles (Fig. 1.11) in directed cytotoxicity and in normal cell signaling.
Acute phase reactions can be stimulated by invading pathogens within one hour and, depending on the level of activation and efficacy of initial defenses, may continue for days or longer. These critical initial defenses rapidly enhance the removal of pathogens by phagocytosis, e.g., by induction and secretion of C-reactive protein (CRP), which is bacteriocidal by binding to Gram-negative bacteria and enhancing their removal by macrophages through phagocytosis. Blood clotting is enhanced by the increase of fibrinogen and other prothrombotic changes. Of major importance, energy resources are mobilized for increased cell activities and systemic responses such as fever. The liver rapidly secretes inflammatory proteins designed to neutralize invading organisms and to mobilize the needed energy. The acute phase inflammatory responses are coded by ancient genes with equivalents (orthologues) in invertebrates (Section 5.4, Fig. 5.4) as well as in lower vertebrates (Azumi et al., 2003) and plants (Ezekowitz and Hoffman, 2003). Transcriptional regulation is at the core of inflammatory responses, often mediated by NF-kB and PPAR.
The acute phase of inflammatory responses may be followed within days by the adaptive immune responses of lymphocyte B-cells and T-cells that recognize new antigens on an infectious invader. Antigen-stimulated immune responses involving somatic gene recombination occur in vertebrates, from bony fish to mammals, but were not evolved in worms, flies, and other invertebrates (Azumi et al, 2003; Marchalonis et al, 2002).
Inflammation in ‘normal’ aging involves the same cells and molecules found in the pathology of arterial atheromas and senile plaques (Table 1.3). I focus on CRP and certain interleukins (IL-1α, IL-6, IL-8, IL-10). These and other acute phase proteins are secreted by the liver (Bowman, 1993), but also by macrophages and other cells elsewhere. The amyloid precursor protein (APP) may also be an acute phase response in the brain (Section 1.6.2). These examples illustrate, but cannot fully represent the remarkable pleiotropies observed in the hundreds of inflammatory mediators.
TABLE 1.3
Inflammatory Components of Vascular Atheromas and Senile Plaque
| Atheroma | Senile Plaque | |
| cells | ||
| astrocytes | 0 | + + |
| mononuclear cells | ||
| macrophage | + + + (foam cell, macrophage; CD68) | + + (microglia, CD68) |
| T-cell | + + (CD3 CD4/Th1) | 0 |
| mast cells | + + | 0 |
| platelets | + + | 0 |
| neovascularization | + + | 0 |
| proteins | ||
| amyloids | ||
| Aβ | ? (macrophages with ingested platelets) | + + |
| CRP | + + | + (neurites) |
| SAP | + | + |
| clotting factors | + + | 0 |
| complement C5b-9 | + (fibrinogen) | + |
| cytokines: IL-1, -6 | + (associated with CRP) | + |
| metals: | Fe | Cu, Fe, Zn |
abbreviations: 0, absent; +/–, weak; +, definitive; + +, moderate; + + +, extensive. amyloids, Aβ: atheroma (De Meyer et al, 2002); senile plaque (Glenner and Wong, 1984; Hardy and Selkoe, 2002; Klein et al, 2001); CRP: atheroma (Rolph et al, 2002; Torzewski et al, 1998); senile plaque (Akiyama et al, 2000; Veerhuis et al, 2003); SAP: atheroma (Li et al, 1995; Meek et al, 1994); senile plaque, (Coria et al, 1988; Veerhuis et al, 2003). Aβ is detected in carotid artery plaque macrophages (De Meyer et al, 2002) and could be derived from platelets adherent to plaques, which contain the amyloid precursor protein (APP) and, when activated, release APP and Aβ-containing peptides (Jans et al, 2004); platelet APP-derived protease nexin 2 (PN-2) is an anti-coagulant (Van Nostrand, 1992). C5b-9 (membrane attack complex); SAA, serum amyloid A (Akiyama et al, 2000; Finch, 2002); B cells: atheromas (Millonig et al, 2002); T-cells: atheromas (Benagiano et al, 2003; de Boer et al, 2006; Häkkinen et al, 2000; (Millonig et al, 2002); mast cells: atheroma (Kelley et al, 2000; Millonig et al, 2002); senile plaque platelets: atheroma (De Meyer et al, 2002; Nassar et al, 2003; von Hundelshausen et al, 2001); complement: atheroma (Torzewski et al, 1998); senile plaque (Akiyama et al, 2000; Eikelenboom and Stam, 1982; McGeer et al, 1989); cytokines: atheroma (Rus et al, 1996); senile plaque (Akiyama et al, 2000). metals: atheromas with intraplaque hemorrhages that deposit extracellular iron; also iron in macrophage from erythrocyte debris (Kolodgie et al, 2003); senile plaques accumulate metals in plaque core and periphery, and colocalized with amyloid fibrils (Lovell et al, 1998; Miller et al, 2006).
When pathogens enter an organ, host defense systems respond immediately to isolate or destroy the invader and to close wounds to prevent further invasion from the skin, airways, or digestive tract. Pathogens are removed by ‘professional’ phagocytes, the ancient macrophage cells residing in every tissue and in the circulation. Acute phase responses are stimulated by receptors that recognize ‘pathogen-associated molecular patterns’ (PAMPS) of common invaders hypothesized by Janeway (Dalpe and Heeg, 2002; Medzhitov and Janeway, 2002; Netea et al, 2004). Bacterial PAMPs include their outer coat components, known as the ‘endotoxins’: lipopolysaccharide (LPS) of gram-negative bacteria (e.g., Escherichia coli) and the lipotechoic acid (LTA) of gram-positive bacteria (e.g., Staphylococcus aureus). LPS is recognized by specialized receptors in many cell types. In liver cells, LPS activates the Toll-4 receptors, leading to rapid secretion of CRP, IL-6, and TNFa. Bacteria are also inactivated by binding to blood lipoproteins (low- and high-density lipoproteins, LDL and HDL), which have specificities for LPS and other bacterial coat components (Khovidhunkit et al, 2004). HDL and LDL, for example, bind LPS. Some viruses are also inactivated by lipoproteins, e.g., Epstein-Barr, Herpes simplex virus. These and other pathogens are also implicated in vascular disease (Chapter 2).
Foci of chronic inflammation are broadly associated with cell proliferation. Hyperproliferation is hypothesized to be the cause of mutations arising through errors in DNA synthesis that increase cancer risk (bystander damage, Section 1.4). This hypothesis is well developed for the role of Helicobacter pylori in intestinal cancer (Section 2.9) and is being considered for other epithelial cancers, e.g., bladder and endometrium (Dobrovolskaia and Kozlov, 2005; Modugno et al, 2005).
Tissue fibrosis is also broadly associated with inflammation and may be considered as generalized wound healing response, which increases fibroblast proliferation and deposition of extracellular collagen and other matrix material. For example, myocarditis from Coxsackie virus infections developed focal scars with thickened collagen networks around surviving myocytes (Leslie et al, 1990). Interstitial myocardial fibrosis also arises during aging in the absence of defined infections in rodents and humans (discussed in Section 1.2.2), and is associated with increased myocardial stiffness during aging (decreased ‘compliance’) (Brooks and Conrad, 2000; Lakatta and Levy, 2003a,b; Meyer et al., 2006). Fibrosis is very common during mammalian aging and deeply linked, if not intrinsic, to general inflammatory processes in aging (Thomas et al, 1992). TGF-β1 signaling pathways regulate collagen synthesis and are implicated in fibrosis of liver (Lieber, 2004), lung (Chapman, 2004), and myocardium (Brooks and Conrad, 2000; Sun and Weber, 2005).
Many inflammatory mediators also have normal functions. (Gene classification systems that may be helpful in organizing data on expression should not be regarded as exclusionary of functions.) IL-6 illustrates these pleiotropies, which range from local cell effects to behaviors of the whole organism. While most circulating IL-6 is secreted by the liver, IL-6 is also made by adipocytes, neurons, and many other cells. IL-6 expression is regulated by transcription factors of convergent pathways that integrate local tissue and systemic signals (Kubaszek et al, 2003; Trevilatto et al, 2003). Moreover, IL-6 and CRP mutually stimulate transcription of the other genes (Arnaud et al, 2005). At a local site of injury, IL-6 regulates cell adhesion molecules that capture circulating neutrophils and macrophages (Kaplanski et al, 2003; Nathan, 2002). IL-6 induction can protect local cells from free-radical-induced death (Waxman et al, 2003). IL-6 is a stimulator (growth and differentiation factor) of B- and T-lymphocytes (Ishihara and Hirano, 2002; Zou and Tam, 2002). IL-6 also influences metabolism and stimulates the resting metabolic rate (BMR); the linear dose response to IL-6 increases BMR by up to 25%, as observed during fever (Tsigos et al, 1997) (Fig. 1.2B). Sustained elevations of IL-6 induce fever-related behaviors of lethargy and poor appetite. Thus, IL-6 sits in a highly integrated network sensitive to the energy state of the individual.
After the acute phase is initiated, a slower phase of adaptive immune responses may be initiated in which lymphocytes are mobilized to recognize new antigens. The differentiation and proliferation of new lymphocyte clones is regulated by IL-1, IL-6, and other interleukins (historically known as lymphokines).
The course of inflammatory responses is regulated by many checks and balances (Li et al, 2005; Nathan et al, 2002; Tracey, 2002) besides the anti-oxidant systems (Fig. 1.11). Local activation of the complement system (C-system) cascade is checked by inhibitors and short molecular life spans (‘tick-over’) of activated C-complexes (Morgan, 1990; Rother and Till, 1988). The complement system is integrated with inflammatory responses and is regulated by cytokines. Among many examples, TGF-β1 represses the first component of the classical C-pathway, C1q mRNA (Morgan et al, 2000).
To counter the damage from the ongoing production of free radicals, body fluids and cells have strong anti-oxidant mechanisms. The redox recycling of glutathione in cells and in the blood is very important. Other regulators also inhibit or modulate inflammatory responses; e.g., local TNFa release may be inhibited by so-called anti-inflammatory cytokines, which include TGF-β1 and IL-10 (Elenkov and Chrousos, 2002).
Free radicals generated by inflammation also induce DNA repair mechanisms. For example, gastric infections of Helicobacter pylori cause infiltrations of immune cells that generate free radicals, in turn, increasing oxidative DNA damage and apoptosis in the gastric mucosa. Experimentally, H. pylori and H2O2 induce enzymes that repair oxidative DNA damage—e.g., the redox-sensitive transcription of APE-1/Ref-1 (apurinic/apyrimidinic endonuclease-1/redox factor-1), a multi-functional enzyme that mediates base excision repair of oxidatively damaged DNA (apurinic sites) (Ding et al, 2004; Lam et al, 2006). This adaptive response (one of many) increases cell resistance to genotoxic free radicals (Ramana et al, 1998).
The immune system is hormonally integrated. Pituitary growth hormone (GH) induces hepatic production of insulin-like growth factor (IGF), which is a mediator of immune and inflammatory functions (Denley et al, 2005; Russo et al, 2005). IGF-1 also feeds back to inhibit pituitary GH secretion. IGF-1 signaling utilizes the PI3 kinases (phosphatidylinositol 3-kinase isoforms) that are general workhorses in signaling (Fig. 1.3A, B).
Macrophages have high-affinity IGF-1 receptors and also secrete IGF-1 (Bayes-Genis et al, 2000), which may be important to both atherogenesis and Alzheimer disease (Section 1.6.4). IGF-1, but not GH, stimulates macrophage secretion of TNFa (Renier et al, 1996). Sex steroids influence immune functions response to trauma (Angele et al, 2000) and pathogens (Soucy et al, 2005). The deep associations of reproduction and immunity enable the evolutionary optimization in the face of the endless assault by infectious pathogenic organisms. Evidence that this selection is ongoing is the extensive genetic variations found in inflammatory responses.
1.3.2 Genetic Variations of Inflammatory Responses
Many genes that influence inflammatory responses may influence the risk of and course of vascular and Alzheimer disease (Chapter 5). Identical twins show extensive heritability in cytokine responses to LPS, accounting for a striking 50% or more of the variance in IL-1β, IL-6, IL-10, and TNFa (de Craen et al, 2005). The IL-6, IL-10, CRP, and apoE gene variants in regulatory and coding elements influence responses to infections (Bennermo et al, 2004; Kelberman et al, 2004).
In bacterial meningococcal meningitis, children carrying an IL-6 promoter variant with lower IL-6 production had 2-fold better survival (Balding et al, 2003). The lower production of IL-6 by the meningococcal LPS may reduce local bleeding (microvascular thromboses). Similarly, gene variants of IL-1β influence inflammatory responses to infections by Helicobacter pylori, a common pathogen, discussed above, which causes peptic ulcers and cancer (Chapter 2) (Blanchard et al, 2004; Rad et al, 2004). H. pylori will often enter discussions of chronic disease. IL-10 (‘anti-inflammatory cytokine’) has a promoter polymorphism that influences secretion by several-fold (Yilmaz et al, 2005) and is associated with different outcomes of hepatitis, meningitis, periodontal disease, and H. pylori (Chapters 2, 4, and 5).
CRP variants may interact with IL-6 variants. Four sites in the CRP gene are associated with plasma CRP levels: upstream promoter (Brull et al, 2003; Kovacs et al, 2005); exon 2 (Zee and Ridker, 2002); the intron (Szalai et al, 2002); and the 3’-untranslated region of the mRNA (3’-UTR) (Brull et al, 2003). Moreover, plasma CRP levels are influenced by alleles of IL-1 and IL-6 (Ferrari et al, 2003; Latkovskis et al, 2004) and of apoE (Section 5.7.4) (Austin et al, 2004; Rontu et al, 2006).
The apolipoprotein E allele apoE4, which is infamous for increasing the risk of heart attack and dementia (Section 5.7), also influences inflammatory responses. In some contexts, apoE4 protein is proinflammatory relative to apoE3. In cultured glia, exogenous apoE4, but not apoE3, induced IL-1 secretion (Chen et al, 2005; Guo et al, 2004). After surgery, apoE4 carriers had higher blood TNFa than the apoE3 (Drabe et al, 2001; Grunenfelder et al, 2004). Transgenic models confirm these effects. IL-6, TNFa, and nitric oxide (NO) production by transgenic apoE4 mice (targeted gene replacement) was greater than with apoE3 (Colton et al, 2004; Lynch et al, 2003). In its newest function, ApoE also mediates lipid antigen presentation to T-cells through CD1 mechanisms that are independent of the MHC system (Hava, 2005; van den Elzen et al, 2005). Other lipoproteins are also important in inflammation (Section 1.4.2.2).
The major histocompatibility complex (MHC) is a cluster of hundreds of genes that modulate instructive immunity and inflammation (Finch and Rose, 1995; Klein, 1986; Price et al, 1999). The MHC has many variant alleles in each of its genes and may be the most polymorphic complex locus in the human genome. The different combinations of alleles across the MHC gene complex (haplotypes) differ in proportion between human populations. MHC genes encode proteins used in antigen presentation, but also a wealth of acute phase and other inflammatory mediators including complement factors (Bf, C2, C4), HSP70, and TNFa (Finch and Rose, 1995). The MHC haplotypes are thought to represent selection for resistance to specific pathogens and toxins (Borghans et al, 2004; Klein, 1986; Wegner et al, 2004). Like IL-6, the MHC is physiologically integrated. MHC allelic variants influence insulin and glucose signaling (Assa-Kunik et al, 2003; Lerner and Finch, 1991; Napolitano et al, 2002; Ramalingam et al, 1997) and reproductive cycles (Lerner et al., 1998, 1992; Lerner and Finch 1991). Because of effects on life history traits that balance trade-offs of immunity, reproduction, and metabolism across the life span, the MHC may be considered a ‘life history gene complex’ (Finch and Rose, 1995).
The emerging genetics of inflammation may also involve haplotypes of the MHC. The MHC haplotypes involving TNF alleles show tentative associations with ischemic heart disease (Porto et al, 2005), whereas TNF isoform variants influence expression of the neighboring complement C4 gene (serum C4a levels) (Vatay et al, 2003). Besides the MHC gene cluster on chromosome 6 (Ch 6) with complement C2, C4, TNFa etc., other chromosomes include clusters of inflammatory and host defense genes: Ch 1, Regulator of Complement (RCA) cluster (C4bp, CR1 and CR2, DAF, factor H, MCP); Ch 2, interleukin-1 cluster (IL-1a, IL-1β, IL-1RN); Ch 11, apo A-I/C-III/A-IV gene cluster; Ch 16, CD11 cluster (integrins CD11a,b,c; adhesion receptors LFA-1, Mac-1, p150,95). Many other inflammatory genes for interleukins, chemokines, etc., are scattered across the chromosomes.
The many inflammatory mediators with genetic variants could contribute to the multi-factorial variability of many diseases. For example, the risk of gastric ulcer in H. pylori infections is associated with particular polymorphisms in both the CD11 cluster (Hellmig et al, 2005) and the IL-1 cluster (Hellmig et al, 2005). The number of possible combinations among these gene variants is very large. Among the 10 or so recognized inflammatory risk factors of vascular disease, each has at least two genetic variants. Thus, the number of possible interactions approximates 2 raised to the 10th power, or 1024. This calculation gives an overestimate because some inflammatory genes are clustered on the same chromosome and hence not randomly assorted. The number of combinations grows faster for genes with more than two variants, as is the case for CRP and IL-6. These infrequent combinations could underlie sporadic cases of the major diseases that do not show obvious heritability. As human genome variations become mapped in populations in greater detail, it may be possible to find inflammatory gene haplotypes of polymorphic loci on different chromosomes that were selected by infectious disease. As a precedent, combinations of alleles in eight inflammatory genes and apoE were very recently found that discriminate Alzheimer disease risk groups (Licastro et al, 2006). Variants in inflammatory genes and insulin/IGF-1 are also implicated as regulators of longevity (Franceschi et al, 2005; van Heemst et al, 2005) (Section 5.7.2).
1.3.3 Inflammation and Energy
The acute phase response to infections rapidly mobilizes host energy needed to sustain fever and acute phase protein synthesis (Fig. 1.2B). White adipose tissue and liver have major roles in the energetics of host defenses (Khovidhunkit et al, 2004; Pond, 2003; Trayhurn and Wood, 2004). Plasma triglycerides increase within 2 h of infection from lipolysis in fat cells and by hepatic synthesis of fatty acids and triglycerides. Triglyceridemia may be sustained for a day or more. These generalized changes are induced by bacterial endotoxins and are mediated by IL-1, IL-6, and TNFa and other cytokines, which have direct metabolic effects. For example, TNFa acts directly on adipocytes to increase lipolysis and lower insulin sensitivity. IL-6 also stimulates lipolysis and in addition hepatic triglyceride synthesis.
Infections cause immediate and chronic energy deficits. The energy consumed by fever comes from 25–100% increases in basal metabolism (Lochmiller and Deerenberg, 2000; Waterlow, 1984). Protein, glycogen, and fat are mobilized; e.g., human energy debts in septic infections are 5,000 kJ/d (Plank and Hill, 2003), which approximates 50% of the normal daily food intake (10,000 kJ/d or 2392 kcal/d). In patients with active infections, white blood cell oxygen consumption increased by 50%, mostly from ATP turnover (Fig. 1.2B) (Kuhnke et al, 2003). The acute phase responses induce ‘sickness behaviors’ through hypothalamic mechanisms that decrease appetite and induce lethargy. T cell activation is closely coupled to the uptake of glucose and other extra-cellular nutrients (Fox et al, 2005). While energy partitioning to various physiological and cellular processes is understood in broad outline (Buttgereit and Brand, 1995; Buttgereit et al, 2000; Rolfe et al, 1999), we do not know how much of the energy consumption associated with fever is due to activation of immune cells and the increased production of CRP and other acute phase proteins. The demands of immunity are considerable, because diet restriction attenuates the primary and secondary immune responses (Chapter 3) (Martin et al, 2007). Growth in children is also attenuated by infections which impair ingestion even if food is not limited (Chapter 4). Malnourished children with infections are in energy debts of 285 kcal per day of infection (McDade, 2003) . The progressively reduced load of infections and inflammation in the 19th and 20th centuries is linked to the increased growth of children (Crimmins and Finch, 2006a) (Chapters 2 and 4).
White adipose tissue is considered an endocrine organ because of its many secreted peptides, particularly leptin and adiponectin (‘adipokines’), which regulate metabolism and eating behavior (Ronti et al, 2006; Trayhurn and Wood, 2004) . Leptin modulates appetite by binding to neurons in hypothalamic centers that regulate energy, body temperature, and reproduction; leptin also influences on insulin sensitivity and stimulates lipid β-oxidation in skeletal muscle. Adiponectin has some overlapping activities by stimulating muscle lipid oxidation by skeletal, and also inhibiting hepatic gluconeogenesis. Both leptin and adiponectin activate the critical AMP-activated protein kinase (AMPK) pathway that increases glucose uptake and lipid oxidation (Chapter 3, Fig. 3.11). Leptin is highly conserved in vertebrates, with homology to IL-6 and TNFa, and binds to class I cytokine receptors (Boulay et al, 2003). Moreover, leptin directly binds CRP in human serum and blocks binding to leptin receptors (Chen et al., 2006). These highly pleiotropic activities of leptin epitomize the nexus of immunity and energy regulation.
Because leptin is secreted in proportion to white fat mass, blood leptin levels are an index of energy reserves (Ronti et al, 2006; Trayhurn and Wood, 2004). Leptin is also a major immunomodulator with complex roles that are still emerging (Loffreda et al, 1998; Matarese et al, 2005; Steinman et al, 2003). Endotoxin induces a leptin surge, which may be an important coordinator of the acute and chronic phases of immune responses. In macrophages, leptin stimulates phagocytosis and increases secretion of IL-6 and TNFa. Adiponectin may have opposing actions (Kougias et al, 2005). In adaptive immunity, leptin is a co-stimulant of T-cell subsets, e.g., naive CD4+CD45RA+ T cells. Exogenous leptin can override energy decisions that attenuate immunosuppression, e.g., in starved mice infected with Streptococcus (Klebsiella pneumonia), leptin injection rapidly corrected the impaired phagocytosis (Mancuso et al, 2006). Reciprocally, diet (energy) restriction may be an intervention for the immunological abnormalities of obesity (Lamas et al, 2004), while fasting may benefit autoimmune disorders because of decreased leptin (Sanna et al, 2003).
Another relationship of fat to immunity is found in lymph nodes, where specialized adipocytes appear to supply follicular dendritic cells with fatty acids (Pond, 2003, 2005). Perinodal adipose tissue differs from adipocytes in other locations by a greater content of polyunsaturated fatty acids and by resistance to atrophy during starvation or hibernation. Intramuscular adipocytes may also be specialized.
Adipocytes also secrete acute phase proteins including CRP, IL-6, and TNFα; complement factors and serum amyloid A3 (SAA3); and prothrombotic factors plasminogen activator inhibitor-1 (PAI-1) (Lyon et al, 2003; Trayhurn and Wood, 2004). SAA-3 and PAH-1 are induced by endotoxin (LPS) and hyperglycemia (Lin et al, 2001; Lin et al, 2005). Moreover, white fat depots have numerous macrophages in proportion to obesity (body mass index) (Bruun et al, 2006; Trayhurn and Wood, 2004; Weisberg et al, 2003). The levels of inflammatory gene expression differ between white fat pad adipocytes and embedded macrophages: IL-6 is expressed in both, with TNFα more prevalent in these macrophages (Weisberg et al, 2003). The gene expression profile of white adipose tissue gives a good account for the association of obesity with the proinflammatory, prothrombotic blood profile in vascular events (Weisberg et al, 2003). Exercise and diet restriction can reduce the macrophage content and inflammatory expression profile of white fat (Bruun et al, 2006) (Chapter 3).
The brain receives information about the inflammatory status by detecting changes in blood sugar, cortisol, and other hormones, but also directly by the vagus and other nerves going from the gut to the brain stem and hypothalamus (Besedovsky and Del Rey, 1996; Black, 2002; Tracey, 2002). These neuronal inputs are part of inflammatory reflexes, which regulate hormonal secretions by the hypothalamus and pituitary and secretion of cytokines by cells throughout the body (Black, 2002; Tracey, 2002). Inflammation activates the hypothalamicpituitary-adrenal axis and increased blood levels of the adrenal steroid, cortisol. Cortisol is called a ‘glucocorticoid’ because it stimulates the biosynthesis of glucose (‘gluconeogenesis’) (Riad et al, 2002), using carbon fragments derived from stored protein and fat. Cortisol elevations can attenuate induction of some cytokines (Franchimont et al, 2003; Nadeau and Rivest, 2003; Refojo et al, 2003), which is part of the basis for steroidal anti-inflammatory drugs (SAIDs) such as prednisone and dexamethasone.
Besides their key roles in disease, many inflammatory mediators also have normal physiological roles that are independent of the acute phase responses. During exercise, IL-6 is released by skeletal muscle in proportion to the intensity of muscular challenge (Helge et al, 2003). In adipocytes, IL-6 directly regulates insulin responsiveness (Bastard et al., 2002). Eating foods rich in fat, or induced hyperglycemia, increases IL-6 and TNFa by 50% within 4 hours (Esposito et al, 2002; Nappo et al, 2002). Could these proinflammatory responses to ingestion have been evolved as a protection against infectious pathogens that were very common in food, until recently? Chapter 6 discusses the dangers of eating raw meat in relation to our evolution. The dense synergies among the hormonal regulatory systems of growth, metabolism, and immune functions may be critical to diseases that limit human life spans.
1.3.4 Amyloids and Inflammation
Amyloid proteins are closely associated with acute and chronic inflammation, and may be universally accumulated during aging.3 Amyloid fibrils are aggregates of 10 nm filaments with repetitive β-sheets in parallel to the fibril axis (Kisilevsky, 2000; Pepys, 2005; Sipe and Cohen, 2000). The cross-β structure of amyloids is detected by binding Congo red, a birefringent dye (‘congophilic amyloids’); however, cross-β structures are also formed by many peptides not yet associated with amyloidosis (Carulla et al, 2005). About 20 peptides encoded by separate genes can form amyloids. Extracellular amyloid fibrils accumulate in diseases of brain (Alzheimer, Creutzfeldt-Jakob) and heart (cardiomyopathy, atheromas). During aging, nearly all tissues accumulate some amyloid (Schwartz, 1970; Tan and Pepys, 1994; Walford and Sjaarda, 1964; Wright et al, 1969). Chronic inflammation often induces amyloids, some of which are acute phase proteins (CRP, SAA, SAP) with anti-microbial activities, discussed below. Bystander effects are also shown, with the accumulation of AGEs and other oxidative damage that attract macrophages and activate scavenger receptors (Section 1.4.4).
Three amyloids are notoriously associated with tissue damage: Aβ (amyloid β-peptide) of Alzheimer disease, amylin of diabetes, and transthyretin in cardiomyopathy. Amyloid deposits are usually embedded with acute phase proteins and sulfated proteoglycans (heparin). The amyloid bulk disrupts functions in the hereditary transthyretin amyloidoses, which damage the myocardium and peripheral nerves (Buxbaum and Tagoe, 2000; Morner et al, 2005). Oligomeric amyloid aggregates are also cytotoxic (Bucciantini et al, 2002; Kayed et al, 2003; Reixach et al, 2004) and important in Alzheimer disease (Section 1.6).
Other amyloidogenic proteins are acute phase responses and some have antimicrobial activities. C-reactive protein (CRP), serum amyloid A (SAA), and serum amyloid P (SAP) bind microbial pathogens. These ancient proteins form pentameric fibrils (pentraxins) that have orthologues throughout vertebrates and invertebrates (Finch and Marchalonis, 1996; Shrive et al, 1999; Ying et al, 1992). The case for anti-microbial functions is clearest for CRP, which enhances the phagocytosis (opsonization) by binding lipopolysaccharide (LPS) and other components of gram-negative bacteria (Bodman-Smith et al, 2002; Ng et al, 2004).
Infections can induce amyloids throughout the body, including the brain, as observed in HIV (Section 2.7.2). In a transgenic model of transthyretin, myocardial amyloid was observed in one mouse colony considered ‘dirty,’ but not in a specific-pathogen free (SPF) colony (Noguchi et al, 2002). Moreover, tissue amyloids in DBA/2 mice were observed in earlier dirty colonies, but not in the cleaner later SPF colonies (Lipman et al., 1993). These responses to the microbial environment suggest that tissue amyloid deposits may be part of host defense innate immunity.
On the other side, some infectious bacteria and fungi have evolved cell-wall amyloid proteins to penetrate host defenses (Gebbink et al, 2005). The microcin amyloid of Klebsiella pneumonia forms cytotoxic pores, as well as non-toxic fibrils (Bieler et al, 2005), whereas the Tafi peptide of Salmonella typhimurium forms amyloid fibrils that enhance intestinal colonization (Sukupolvi et al, 1997). Endless host-pathogen “arms races” that select for specialized host responses may give rise to population-specific genetic risk factors in Alzheimer and other amyloidotic diseases.