Somatotropic Signaling: Trade-Offs Between Growth, Reproductive Development, and Longevity

Abstract

Growth hormone (GH) is a key determinant of postnatal growth and plays an important role in the control of metabolism and body composition. Surprisingly, deficiency in GH signaling delays aging and remarkably extends longevity in laboratory mice. In GH-deficient and GH-resistant animals, the “healthspan” is also extended with delays in cognitive decline and in the onset of age-related disease. The role of hormones homologous to insulin-like growth factor (IGF, an important mediator of GH actions) in the control of aging and lifespan is evolutionarily conserved from worms to mammals with some homologies extending to unicellular yeast. The combination of reduced GH, IGF-I, and insulin signaling likely contributes to extended longevity in GH or GH receptor-deficient organisms. Diminutive body size and reduced fecundity of GH-deficient and GH-resistant mice can be viewed as trade-offs for extended longevity. Mechanisms responsible for delayed aging of GH-related mutants include enhanced stress resistance and xenobiotic metabolism, reduced inflammation, improved insulin signaling, and various metabolic adjustments. Pathological excess of GH reduces life expectancy in men as well as in mice, and GH resistance or deficiency provides protection from major age-related diseases, including diabetes and cancer, in both species. However, there is yet no evidence of increased longevity in GH-resistant or GH-deficient humans, possibly due to non-age-related deaths. Results obtained in GH-related mutant mice provide striking examples of mutations of a single gene delaying aging, reducing age-related disease, and extending lifespan in a mammal and providing novel experimental systems for the study of mechanisms of aging.

I. INTRODUCTION

Phenotypic characteristics of all living organisms are determined by the interplay of genetic and environmental influences. This applies to the control of aging as well. There is also increasing evidence that the endocrine system plays a central role in mediating the impact of genetic and environmental factors on aging and longevity. Mice with mutations causing a deficiency of growth hormone (GH) or resistance to its actions are remarkably long-lived (50, 70, 100), and humans with similar mutations are protected from major age-related diseases (115, 207). The involvement of somatotropic (GH-related) signaling as well as secondary alterations in insulin signaling in the control of mammalian aging represents a fundamental and evolutionarily conserved mechanism. Insulin-like growth factor I (IGF-I), insulin, and their receptors are homologous to signaling molecules and receptors that have major impact on aging and longevity in yeast, worms, and insects (114, 153, 206, 299). Since life-extending GH and IGF-I-related mutations reduce or block the corresponding endocrine signals, it is pertinent to ask what trade-offs may be involved in balancing the detrimental, “pro-aging” actions of the somatotropic axis with its beneficial effects on other characteristics.

Against this background, the objectives of this article are to present evidence that reduced somatotropic signaling can delay and/or slow down the aging process and promote a remarkable optimization of the healthy lifespan, to discuss the mechanisms believed to be responsible for these effects, and to relate the findings in mammals to the fundamental genetic control of aging in yeast and invertebrates. We will also describe age-related changes in the GH-IGF-I axis in mammals and discuss the controversial topic of potential benefits of GH therapy in the elderly.

II. OVERVIEW OF THE SOMATOTROPIC AXIS

A. Growth Hormone

Growth hormone (GH), also called somatotropin, is a 22-kDa protein hormone composed of 191 amino acid residues forming a single chain with four helical regions and two disulfide bridges. It is produced and secreted by specialized cells in the anterior lobe of the pituitary gland, the somatotrophs (also referred to as somatotropes). Its secretion is controlled primarily by two hypothalamic peptides: growth hormone releasing hormone (GHRH) and somatostatin (SST, also called somatotropin release inhibiting factor, SRIF). The interplay of GHRH and SST actions and GH/IGF-I feedback on their release produce a pulsatile pattern of GH release with brief secretory episodes followed by gradual clearance from the circulation during interpulse intervals (121). In the human, these pulses occur approximately every 2–2.5 h (307).

Release of GH is strongly associated with sleep, stimulated by hypoglycemia and inhibited by overeating, hyperglycemia, and obesity (68, 121). Many aspects of the regulation of GH secretion are species specific. For example, exercise, starvation, and stress stimulate GH release in different mammalian species including humans (128, 318) and ruminants, but some of the same stimuli inhibit GH release in the rat (298) and may have different effects on GH release at different time points. Interestingly, prolonged calorie restriction preserved pulsatile GH release in the latter species, presumably by attenuating the suppressive effects of aging (277). In mice, fasting was reported to both increase and suppress circulating GH levels with discrepancies between results obtained in different laboratories being possibly related to differences in protocols for blood collection and associated handling of the animals or to time of treatment (191, 281). Key gonadal steroid hormones, testosterone and estradiol, stimulate GH release, and these actions likely account for the prepubertal elevation of GH levels and contribute to the well-known growth spurt during this period.

Aging is associated with a pronounced, progressive decline in circulating GH levels which, in the human, is due primarily to reduction in the amount of the hormone released in each secretory burst (309). After age 20, GH secretion declines by approximately one-half every 7–12 yr (107, 215) (FIGURE 1). The age-related decline in GH secretion and average serum levels is accompanied by a less steep but statistically significant decline in serum IGF-I (195) (FIGURE 1). These changes are often referred to as “somatopause,” by analogy with the age-related decline in gonadal function, the menopause and the andropause. Declining GH levels are believed to contribute to concomitant changes in body composition (including increased adiposity and reduced muscle mass) and to other physiological changes that accompany aging. Secretion of GH also declines with age in other species, including the domestic dog (214) and laboratory rats and mice (162, 276). However, recent data from the Jackson Laboratory indicate that in some inbred mouse strains IGF-I levels in old age resemble or exceed those measured in early adulthood (R. Yuan, personal communication). The age-related reduction in GH release led to an interest in using GH to reduce the symptoms of aging. This important and highly controversial issue will be discussed later in this article.

Figure 1.

Age-related decline in circulating levels of human growth hormone (GH) and insulin-like growth factor I (IGF-I).[From Corpas et al. (69) and Zadik et al. (335), with permission from The Endocrine Society.]

Growth hormone acts by binding to specific cytokine family receptors located in the membrane of target cells (FIGURE 2). GH receptors are present in multiple cell types including hepatocytes, adipocytes, muscle fibers, bone cells, and cells of the immune system (156, 208). Each GH molecule has two receptor binding sites allowing it to interact simultaneously with two GH receptors. GH binding leads to dimerization of GH receptors and activation of the associated JAK2, a member of the Janus family of tyrosine kinases (FIGURE 2). JAK2 activation leads to tyrosine phosphorylation of both JAK2 and the GH receptor. This initiates signaling via several pathways and, in particular, includes signaling via phosphorylation of signal transducers and activators of transcription (STATs). Phosphorylated STATs disassociate from the receptor-JAK2 complex, move to the nucleus, and bind to promoters of GH-regulated genes. In addition, this pathway creates a negative-feedback loop that terminates the GH signal by stimulating the expression of suppressors of cytokine signaling (SOCS) proteins that inhibit JAK activity and STAT activation and promote internalization of the GH receptors.

Figure 2.

Schematic representation of key mechanisms of growth hormone (GH) signaling. GH binds to two transmembrane GH receptors and alters expression of numerous genes via activation of STAT and ERK; details and references are in the text.

GH signaling pathways also include mitogen-activated protein kinase (MAPK) as well as protein kinase B (Akt) and mammalian target of rapamycin (mTOR), one of the Akt targets. For more information on GH signaling, the reader is referred to review articles (45, 157, 166, 316).

Mutations of the gene encoding the GH receptor in humans or targeted disruption of the same gene in mice lead to the loss of receptor function and consequently to GH resistance and severe retardation of postnatal growth. This endocrine syndrome was first described by Laron et al. (172) in 1966 in a patient exhibiting dwarfism and elevated GH levels and became known as Laron dwarfism. In the 1990s, it was produced in mice by Zhou et al. in the Kopchick laboratory (340). Studies in Laron dwarfs and in GH receptor-deleted, Ghr−/− (Laron dwarf) mice provided a wealth of information on the role of GH in aging and will be discussed later in this article.

In the circulation, GH is present in a free form or bound to a binding protein. Growth hormone binding protein (GHBP) in the human and some other species consists of cleaved extracellular fragments of the GH receptor and thus provides a clinically useful estimate of GH receptor abundance. In mice, GHBP is produced separately from the GH receptor by alternate splicing of the same gene (269, 339).

In contrast to steroid as well as most polypeptide hormones, the biological activity of GH exhibits peculiar species specificity. Thus fish, amphibians, and reptiles respond to mammalian GHs, and “lower” mammals respond to primate GH while the opposite is generally not the case (108). In rodents, human GH binds to both GH and prolactin (PRL) receptors and produces a full spectrum of both GH and PRL effects (102, 122, 279). Lactogenic (PRL) activity of human GH in rats and mice has been documented for over 40 years (102, 122) but is overlooked in many recent publications. Very often, the authors focus on the somatotropic (GH) effects of human GH administration to rodents, ignoring the possible contribution of the concomitant activation of PRL receptors and eliciting effects unrelated to somatotropic signaling.

B. Insulin-Like Growth Factor

One of the key actions of GH is stimulation of the hepatic expression of the IGF-I gene. IGF-I, formerly known as somatomedin C, mediates many, although not all, GH actions. Most of the circulating IGF-I is of hepatic origin, but some is derived from adipose tissue and from other sources. Since production of circulating (“endocrine”) IGF-I depends on GH, serum (or plasma) IGF-I levels are used as an index of GH secretion and its biological actions in both research and clinical (diagnostic) settings. This allows evaluation of the activity of the somatotropic axis without collecting serial blood samples that would be needed for precise evaluation of GH secretion. Actions of IGF-I include inhibition of GH release, primarily via actions within the hypothalamus, thus creating a classical negative-feedback relationship between these two hormones. As expected from this relationship, GH levels are elevated in animals in which circulating IGF-I is reduced by genetic GH resistance or by deletion of the IGF-I gene in the liver (327, 340).

In contrast to the levels of circulating (endocrine) IGF-I, which are determined primarily by GH stimulation, local expression of IGF-I in different tissues and its local (paracrine and autocrine) actions can be relatively or completely independent of GH. For example, expression of IGF-I in the somatic cells of the ovary and the testis is regulated primarily by pituitary gonadotropins rather than by GH (51, 131). Lupu et al. (190) in the Efstratiadis laboratory reported that blocking GH signaling by deletion of GH receptors led to the expected profound suppression of IGF-I message levels in the liver, while IGF-I expression in the kidneys was only partially suppressed and its expression in the heart and in the brain was not affected. Unaltered expression of IGF-I in the heart of GH-resistant mice may be physiologically important because IGF-I is viewed as a cardioprotective agent (54). In fact, cardiac specific overexpression of IGF-I in male mice increased their median lifespan (176). However, interpretation of this very intriguing finding needs to be cautious, because plasma IGF-I levels in these transgenics were increased and thus GH secretion may have been suppressed. Preserved expression of IGF-I in the hippocampus of mice with complete GH deficiency will be discussed later in this article. Growth hormone-dependent and GH-independent IGF-I biosynthesis and its targets are schematically represented in FIGURE 3.

Figure 3.

The somatotropic axis. This diagram contrasts endocrine, primarily hepatic IGF-I that acts as a mediator of GH actions with partially or completely GH-independent IGF-1 produced by other organs that can act locally in paracrine or autocrine fashion.

IGF-I receptors belong to the tyrosine kinase receptor family, are composed of α and β subunits, and resemble insulin receptors. IGF-I receptors are present in most cell types but, somewhat surprisingly, not in the hepatocytes, which are an important source of IGF-I. Consequently, IGF-I affects virtually every tissue and organ. Multiple actions of IGF-I include stimulation of cell proliferation and growth, inhibition of apoptosis, and mimicking some of the insulin effects. IGF-I receptors are present in the cell membrane and signal primarily via activation (phosphorylation) of insulin receptor substrates (IRS-1 and IRS-2) and Akt leading to maintenance of cytoplasmic localization of a transcription factor, FOXO, and inhibition of its various effects (including effects on antioxidant defenses). Signaling pathways and biological effects of IGF-I are described in far more detail in recent reviews (39, 112, 235).

In the circulation, IGF-I is present mainly as a part of a ternary complex with one of the IGF binding proteins (IGFBPs) and acid-labile subunit (ALS). Biological activity of IGF-I is believed to reflect primarily the amount that is free rather than complexed. There are six different IGFBPs of which IGFBP3, produced by the liver under influence of GH, is quantitatively most important (143). The physiological role of IGFBPs is very complex and includes 1) regulating the amount of free (bioavailable) IGF-I in circulation and locally, within the target tissues; 2) protecting IGF-I from metabolic degradation and clearance and thus maintaining an IGF-I reservoir; and 3) exerting IGF-independent actions mediated mostly by separate receptors (65).

III. SOMATOTROPIC SIGNALING IN THE CENTRAL NERVOUS SYSTEM: ROLE IN AGING

Results obtained in Caenorhabditis elegans and Drosophila indicate that insulin/IGF-I-like signaling in neurons and brain is involved in the control of aging and longevity in these animals (6, 325). The impact of deleting IGF-I receptors or IRS-2 selectively in the brain of mice (297) on age-related decline of cognitive function and on longevity will be discussed later in this article. Long-lived mutant mice with GH resistance or deficiency maintain youthful measures of learning and memory into advanced age (154, 155). Against this background, the role of somatotropic signaling in the mammalian brain in the control of aging and particularly cognitive aging is of considerable interest and will be discussed below.

A. Brain GH

Growth hormone affects multiple targets, but its role in the control of cognitive function and brain development is largely unknown. Although pituitary somatotrophs are the primary site of GH production, extrapituitary tissues have been reported to express GH genes (123, 124) and neural tissues were shown to produce and to respond to GH (123). Immunoreactive GH has been shown to be present in the rat midbrain, cortex, hippocampus, striatum, olfactory bulb, and cerebellum at different concentrations (219, 224), and the level of GH signal density in the various brain regions was found to decline during aging (219, 224). Recent studies (79, 80, 287–289, 291) indicate that in murine GH is produced endogenously within the hippocampal formation, a brain structure associated with learning and aspects of emotional experience. Interestingly, both GH mRNA and protein can be detected in the hippocampus of Ames dwarf mice (288, 289, 291), in which pituitary GH is absent because of the loss-of-function mutation at the Prop-1 locus and the resulting failure of somatotroph differentiation. Since the circulating level of GH is very low in these animals, GH present in the dwarf brain is mostly produced locally within the central nervous system (CNS). Furthermore, these studies suggest that the local synthesis of GH in the hippocampus is Pit1 independent.

Despite controversy in the field, there is growing evidence that GH can cross the blood-brain barrier (BBB) from circulation into the brain (1, 63). For example, Johansson et al. (139) reported increased GH concentration in the cerebrospinal fluid in a GH-deficient patient after recombinant human GH treatment. GH receptors are present in the choroid plexus at much higher levels than in any other brain region and may function as GH transporters across the BBB leading to accumulation in the cerebrospinal fluid (CSF) (164). GH can also cross the BBB into the brain through simple diffusion independent of a specific transport system (232). In certain pathological conditions, including hypoxic-ischemic brain injury, the BBB is often compromised, so increasing access of GH to the CSF (258). Taken together, there is growing support for the passage of GH across the BBB, although the exact mechanisms remain to be further elucidated.

B. Function

Several studies suggest that GH may play a role in influencing aspects of mood and cognition (275). GH-binding sites have been identified in several areas of the brain, suggesting the local physiological function of GH signaling in these tissues. Furthermore, in a transcriptional profiling study aimed at identifying genes that were altered in the course of learning and memory formation, GH expression in the hippocampus was significantly increased with acquisition of a hippocampal-dependent learning task (79). This increase occurred in animals that had learned, whereas naive animals and those exposed to other types of training experiences had very low levels of GH expression. GH has been shown to enhance both α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)- and NMDA receptor-mediated excitatory postsynaptic potentials (EPSPs) partially through the JAK-Stat and MAPK signaling pathways (196). These findings suggest that hippocampal GH acts in an autocrine or paracrine manner to contribute to enhanced synaptic function during memory formation. Moreover, studies in CPEB-1 (cytoplasmic polyadenylation element binding protein 1) knockout mice (336), which have defects in synaptic plasticity and hippocampal-dependent memory, support a role of GH in certain types of synaptic plasticity and in learning and memory. With the use of two-dimensional gel electrophoresis and mass spectrometry, GH was found to be the most dramatically changed of the proteins with altered abundance levels in the hippocampus of CPEB-1 knockout. GH mRNA and premRNA were also reduced, suggesting that GH transcription may be directly regulated by CPEB-1 (336).

C. Brain IGF-I

IGF-I plays an essential role in regulating somatic growth and development and is also important in neuronal function (332). As discussed earlier, many types of cells, including all types of cells in the brain, are capable of IGF-I production (71). IGF-I expression in the CNS is particularly high during fetal development and in the mouse reaches its peak expression during the first 2 wk of postnatal life, predominantly in neurons, but also in glial cells, suggesting that it may play a critical role in the development of the nervous system (71). IGF-I has indeed been shown to act as a prominent neurotrophic factor during development of the nervous system, stimulating differentiation and survival of specific neuronal populations (71, 313, 321). Several studies have demonstrated neurotrophic actions of IGF-I and IGF-II in both neurons and glia including stimulation of DNA and RNA synthesis, induction of neurite outgrowth, regulation of neurotransmitter release and synaptogenesis, and protection against neurotoxic insults (82, 118, 135, 145, 247). In the adult CNS, IGF-I is a neuromodulator involved in regulation of synaptic plasticity (272, 302, 303).

Altered IGF-I signaling in the neuroendocrine hypothalamus may contribute to reproductive aging (210, 294). It was suggested that the decline in circulating IGF-I levels during aging plays a direct role in age-associated brain deterioration (242). Restoring IGF-I levels was reported to enhance neurogenesis and ameliorate the age-related cognitive malfunction in aged brain (197). In transgenic mice with increased expression of IGF-I in the brain, the weight and volume of the brain are increased substantially due to increases in neuron number and total myelin (72, 174). In contrast, transgenic mice with ectopic brain expression of IGFBP-1, an inhibitor of IGF action, and mice with ablated IGF-I gene expression have brain growth retardation with impaired neuronal somatic and dendritic growth (60). Levels of the synaptic proteins were significantly reduced in the brain of IgfI−/− mice, suggesting a reduction in synapses (60, 184). Furthermore, IGF-I gene deletion in humans has been reported to be associated with mental retardation (326).

In the studies of Ames dwarf mice, circulating IGF-I is very low (55) and the IGF-I immunoreactivity detected in the hippocampus of these mice seems unlikely to be from peripheral circulation and transport across the BBB. Therefore, IGF-I is hypothesized to be produced locally in the hippocampus and act directly on the progenitor cells in the hippocampus (291). Phosphorylation of Akt and cAMP responsive element-binding protein (CREB) were increased in the hippocampus of Ames dwarf mice (287) along with activation of the anti-apoptosis signal transduction cascade. These data support the suggestion that increase in hippocampal GH and IGF-I expression and subsequent activation of PI3K/Akt-CREB and anti-apoptotic signals might be responsible for the maintenance of the integrity of neuronal structure and cognitive function (154, 155) in these long-lived mice.

D. Aging, Hippocampal Neurogenesis, and GH-IGF-I

In the adult mammalian brain, neurogenesis occurs in the olfactory bulb and hippocampus (283). Physiological and behavioral events, such as aging, stress, diseases, seizures, learning, and exercise, can modulate hippocampal neurogenesis (103, 283). These newborn neurons go through an extended maturation process and eventually functionally integrate themselves into the existing DG circuitry (103, 283). The significance of the production of new neurons during adulthood is unknown, although studies in rodents suggest that neurogenesis plays an important role in learning and memory (88, 109, 111).

Neurogenesis in the DG decreases with age in rats, mice, monkeys and humans (53, 110, 341). Several studies have suggested that cognitive decline in aging may be attributable to decreased DG neurogenesis (83, 267). Conversely, hippocampal neurogenesis was reported to increase in aged mice living in an enriched environment associated with enhanced synaptic plasticity (by increasing long-term potentiation), dendritic spine density, and expression of synaptic proteins, receptors and neurotrophins (149, 222, 227). Moreover, training in a task that requires hippocampal function stimulates granule cell proliferation in the hippocampus (103, 109, 111).

Aging has a major impact on the activity and differentiation capacity of stem cells within the DG in the adult hippocampus, leading to a marked reduction of the number of newborn neurons within this area in rodents, non-human primates, and potentially in humans (103, 148, 160). The progressive decline of neuronal stem cells during aging indicates that the progenitor cells become unresponsive to environmental cues in the aged hippocampus and may also suggest that the newly born stem cells lose their appropriate signaling mechanism in response to the mitotic stimulus (103, 160). However, there is also evidence suggesting that alteration in the local environment in the aged brain does not provide sufficient support or mitotic stimuli for these newborn cells. Several factors that regulate neuronal birth in the adult DG were recently studied. Blockade of adrenal steroids, blockade of excitatory input into the DG, and NMDA receptor inactivation all lead to an increased birth of granule cells (110, 213).

Some cell proliferation factors, such as fibroblast growth factor-2 (FGF-2), IGF-I, and vascular endothelial growth factor (VEGF), have been shown to be responsible for the regulation of hippocampal neurogenesis (103). Intracerebroventricular (ICV) or subcutaneous injections of FGF-2 enhance dentate neurogenesis in neonatal, adult, and aged brain (138, 314, 337). IGF-I levels have been closely associated with increased neurogenesis in the adult hippocampus under different conditions (61, 185, 241, 304, 338). VEGF is another factor that considerably increases DG neurogenesis in both the intact and the injured adult brain following ICV administration (138). Moreover, it has been hypothesized that age-related decreases in some of these proliferation factors are a consequence of age-related impairments in the neurotrophic factors (103, 271), such as IGF-I.

Interestingly, increased hippocampal neurogenesis has been reported in the adult Ames dwarf mice (292). BrdU labeling studies showed an increase in numbers of newly generated cells (BrdU positive) and newborn neurons (neuronal nuclear antigen and BrdU positive) in the DG of dwarfs compared with normal mice at 3 mo of age (292). Furthermore, there was a dramatic reduction of BrdU-positive cells in the DG during aging in both dwarf and normal mice, indicating an age-related decline of hippocampal neurogenesis. Although there was no significant difference in the total number of BrdU-labeled cells between aged dwarf and normal mice, the total number of newly generated neurons (BrdU and NeuN double positive) in old dwarf mice was significantly greater than in old normal mice. Considering the evidence for a significant increase in hippocampal IGF-I protein expression and activation of an anti-apoptosis signal transduction cascade, it is likely that the increase in the fraction of newborn neurons in aged dwarf mice was due, at least in part, to an effect of local IGF-I on cell survival.

V. EXTENDED LONGEVITY OF MICE WITH CONGENITAL GH DEFICIENCY OR DELETION OF GH RECEPTORS

In 1996, Brown-Borg et al. (50) reported that Ames dwarf mice live ∼50% longer than their normal siblings and suggested that extension of their longevity is most likely due to deficiency of GH. Ames dwarf mice (257) are homozygous for a recessive mutation of the Prophet of pituitary factor-1 (Prop1) gene, which is normally expressed in the anterior pituitary and required for differentiation of cells that express Pituitary factor 1 (Pit1) and develop into somatotrophs, lactototrophs, and thyrotrophs, that is cell producing GH, PRL, and thyroid-stimulating hormone (TSH) (18, 20, 278). The causative role of GH deficiency in the remarkable extension of longevity of these mutants was supported by the demonstration that little mice with isolated GH deficiency (85) and “Laron dwarf” mice with complete GH resistance due to deletion of GH receptors (340) are also long-lived (70, 100). Subsequent work established that the extension of longevity of these mutants is reproducible (23, 24, 233) and that the remarkable longevity of Laron dwarf (Ghr−/−) mice is not limited to a particular laboratory, genetic background, or diet composition (21, 23, 42, 70). The possibility that effects of Ames dwarfism on longevity may not be due to their endocrine phenotype but instead to some other, unknown effect of the Prop1^df mutation was eliminated (or at the very least made exceedingly unlikely) by the report of comparable longevity extension in Snell dwarf mice (100) which have the same endocrine defects as Ames dwarf mice (deficiency of GH, PRL, and TSH) due to mutation of Pit1, a gene distinct from Prop1 and located on a different chromosome (17, 19, 20, 177, 270).

Major longevity benefits in animals with severe endocrine defects, growth retardation, and dwarf phenotype appeared counterintuitive and were initially received with considerable skepticism. However, these findings were not unprecedented. Everitt (91) reported in 1980 that rats deprived of all anterior pituitary hormones by hypophysectomy (surgical removal of the pituitary gland) lived longer than normal animals, as long as they were provided adequate glucocorticoid replacement. Powers et al. (243) demonstrated that hypophysectomy can also extend longevity in mice, although in this species the effect on longevity was dependent on the age of the animals at the time of surgery. In a work that appears to have been overlooked by gerontologists until recently, Silberberg (268) reported histopathological findings in Snell dwarf mice up to the age of 41 mo (mice normally live ∼24–30 mo) and mentioned “unusual longevity of the dwarf animals” in the discussion section of her paper. In further support of the role of GH in the control of aging, Banks et al. (11) recently reported that treatment of short-lived “senescence accelerated” (SAMP8) mice with an antagonist of GHRH receptor increased telomerase activity, improved cognition, reduced tumor incidence, and increased mean (although not maximal) life expectancy. Evidence linking somatotropic signaling and longevity in genetically normal (“wild-type”) rather than mutant or gene knockout mice and in other species will be discussed later in this article.

VI. LONGEVITY OF MICE WITH DELETION OF GENES AFFECTING IGF-I SIGNALING AND PATHWAYS DOWNSTREAM OF THE IGF-I RECEPTOR

One of the phenotypic characteristics shared by Ames dwarf, Snell dwarf, little, and Ghr−/− mice is profound suppression of circulating levels of IGF-I (55, 81, 306, 340). Since IGF-I mediates many of the GH effects, it seems likely that extension of longevity in these mutants may be due to reduced IGF-I signaling. Consistent with this possibility, heterozygous deletion of the IGF-I receptor gene reduces the number of IGF-I receptors by ∼50% and extends longevity of female (although not male) mice (130). Results of subsequent studies indicate that the magnitude of lifespan extension in Igf1r+/− mice with partial IGF-I resistance is strongly dependent on the genetic background (36). Partial inactivation of the IGF-I receptor gene only in the brain impacts development of GH-secreting cells in the pituitary leading to reduced growth and adult body size and significant extension of longevity (144). Interestingly, in amyloid precursor protein plus presenilin transgenic mice with phenotypic characteristics resembling various features of Alzheimer's disease, heterozygous deletion of IGF-I receptor gene resulted in significant improvements of special memory and motor skills and reduced neuroinflammation and neuronal loss (64). Hypomorphic mutation of the IGF-I gene leads to a reduction of circulating levels of IGF-I and extends female longevity similarly to the findings in the Igf1r+/− mice (260).

Deletion of IGF-I gene expression selectively in the liver led to a marked (∼70%) decline in circulating IGF-I levels and a modest decrease in adult body size (328), but median lifespan was not affected in females and somewhat reduced in males (M. Adamo, personal communication). This may have been due to reduced negative feedback of IGF-I on GH secretion and the resulting increase in plasma GH levels (328, 329). Elevated GH levels could impact longevity by several mechanisms including induction of insulin resistance (329), shifting adipocytes secretory profile to proinflammatory, and enhancing mTOR signaling.

Svensson et al. (296) recently described effects of inactivating the IGF-I gene in mice selectively in the liver at 1 mo of age. As expected from studies in hepatic IGF-I deleted (LID) mice (328), IGF-I levels were reduced by ∼80% and body weight was reduced, with the greatest differences in body weight noted in middle-aged (12–24 mo old) males. Mean longevity of females was significantly increased, while a similar trend in males and a suggestive increase of maximal longevity in females were not statistically significant. Fertility did not appear affected in either sex.

The important role of IGF-I in the control of mouse longevity is strongly supported by findings in animals with deletion of pregnancy-associated plasma protein A (PAPP-A), a protease involved in degradation of IGF-I binding proteins and particularly IGFBP4. Deletion of PAPP-A thus leads to increased levels of IGFBP with the consequent reduction of free (bioavailable) IGF-I at the tissue levels. Both average and maximal lifespan are increased in both sexes of PAPP-A-1- mice (66, 67).

Extension of mouse longevity was produced also by deletion of insulin receptor substrate 1 (IRS1) (261) or IRS2 (297) as well as by heterozygous or homozygous deletion of IRS2 selectively in the brain (297). The effect of heterozygous deletion of IRS2 in all tissues on longevity reported by Taguchi et al. (297) were not seen in the studies of Selman and co-workers (261, 262) possibly due to the use of diets with a different percentage of fat. It is difficult to know whether the impact of reducing IRS expression on longevity is due to suppression of IGF-I signaling, insulin signaling, or perhaps both. Although effects of IGF-I and insulin on cell function are distinct with relatively little overlap, signaling by both of these hormones involves activation of IRSs. Alterations in insulin signaling are involved in mediating the effects of GH on aging and longevity (details later in this article), and deletion of insulin receptors in adipose tissue increased longevity in mice (34).

In the context of discussing the effects of somatotropic axis on aging, it is of considerable interest that mice with deletion of S6 kinase (S6K) are long-lived (263). S6K is regulated by the mTOR, and mTor signaling is reduced in long-lived GH-resistant and GH-deficient mice (40, 265). Pharmacological suppression of mTOR increases longevity of mice (120), and involvement of TOR pathway in the control of aging is highly conserved (205).

Key characteristics of long-lived GH and/or IGF-I related mouse mutants are listed in TABLE 1.

Table 1.

Phenotypic characteristics of long-lived GH- and IGF-I-related mouse mutants

Name	Gene	Lifespan (Male)	Lifespan (Female)	Strain	BW	Blood GH	Blood IGF-I	Insulin Sensitivity	Cellular Stress Resistance	In Vivo Stress Resistance
Ames Dwarf [150]	Prop-1	49%	68%	Heterogeneous	↓	↓	↓	↑	↑	↑
Snell Dwarf [100]	Pit-1	>40%	>40%	DW/J × C3H/HeJ	↓	↓	↓	↑	↑	↑
Little [100]	Ghrhr	>20%	>25%	C57BL/6J	↓	↓	↓	↑	?	?
GHR KO [70]	Ghr/bp	>35%	>35%	129Ola × BalbC	↓	↑	↓	↑	↑	?
PAPPA KO [67]	Pappa	>25%	>25%	C57BL6 × 129SV/E	↓	NC	NC	?	?	?
AC5 KO [331]	Ac5	32%	32%	129/SvJ-C57BL/6	NC	↓	?	?	↑	?
MIF KO [119]	Mif	?	16%	C57BL/6J 129/SvJ	↓	?	NC	NC	NC	?
Klotho Tg [161]	Klotho	>20%	>18%	C57BL/6 × C3H	NC	?	NC	↓	↑	?
Surf1 KO [76]	Surf1	18%	18%	129S1/SvJxC57 BL6J × DBA2	NC	?	?	?	↑	?
P66 shc KO [209]	p66 shc	30%	30%	C57Bl/6Sv/129	NC	?	?	↑	↑	↑
Igf-I Tg Heart [176]	IGF-I	18%	?	FVB	NC	?	↑	?	?	?
Irs1 KO [261]	Irs-1	16%	30%	C57BL/6	↓	NC	NC	↓	?	?
Irs2 KO [297]	Irs-2	?	18%	C57BI/6	NC	?	?	↑	?	?
FIR KO [34]	Ir	18%	18%	129Sv × C57BI/6	↓	?	?	NC	?	?
Igf-IR ± KO [130]	Igfr	16%	33%	129/Sv	↓	?	↑	?	↑	↑
Igf-IR ± KO [36]	Igfr	0%	<5%	129/SvJ-C57BL/6	↓	?	NC	↓	?	↑ (female)
S6KO [263]	S6k	NC	19%	C57BL/6	↓	NC	NC	↑	?	?

Significant alterations in the various characteristics of mutant as compared with normal (control) mice are identified by arrow pointing up (increases) or down (decreases) (34, 36, 50, 65, 67, 70, 76, 100, 119, 130, 161, 176, 209, 261, 263, 297, 331). NC, no change.

VII. EXTENSION OF LONGEVITY IN GH-DEFICIENT AND GH-RESISTANT MUTANTS INVOLVES VARIOUS TRADE-OFFS

In organisms ranging from yeast to mice, there are numerous examples of extension of longevity by targeted deletion of individual genes or by loss-of-function or hypomorphic mutations. These findings imply that the normal physiological levels of expression of these genes are not optimal for life expectancy and could be viewed as factors that promote aging and reduce longevity. Failure of natural selection mechanisms to eliminate genes with such detrimental effects fits the concept of antagonistic pleiotropy, first formulated by Williams (323). This concept is based on the appreciation that the force of natural selection declines after the age of maximal reproductive effort and posits that genes with positive effects on attainment and maintenance of reproductive fitness early in life will not be selected against even if they have detrimental effects later on. Genes involved in GH signaling would appear to fit this concept. Growth hormone actions early in life promote somatic growth and sexual maturation and may also enhance gonadal function and fecundity, thus clearly increasing reproductive fitness, while its effects in adult life include increased susceptibility to cancer and promoting insulin resistance and hyperinsulinemia, that is, effects that may accelerate aging and be detrimental to long-term survival. When viewed in light of antagonistic pleiotropy, the anti-aging and life-extending effects of reduced GH signaling in mutant mice would be expected to be associated with trade-offs reflecting loss of beneficial effects of GH early in life. Phenotypes of Ames dwarf, Snell dwarfs, and Ghr−/− mice in fact do provide examples of such trade-offs.

The most obvious phenotypic effect of GH deficiency or resistance is profound suppression of postnatal (and particularly postweaning) growth and reduction of adult body size [in Ames, Snell. and Ghr −/− (Laron) dwarf mice, by >50%]. Although the impact of small body size of GH-related mouse mutants on reproductive success or survival under natural conditions has not been examined, it can be assumed that in competition with normal males, male dwarfs would be unlikely to establish and defend territories and thus would have reduced mating opportunities. It is also likely that in temperate or northern climates, these diminutive animals would have reduced winter survival due to unfavorable body surface-to-mass ratio.

Sexual maturation, as measured by the age of establishment of vaginal introitus (“vaginal opening”) in females and balano-preputial separation in males is delayed in Ghr−/− mice by ∼1 wk (73, 147). Although both sexes are fertile, various parameters related to fecundity including interval from pairing until conception, litter size, interval between litters, and percentage of animals that reproduce are reduced in Ghr−/− mice compared with normal animals from the same strain (56, 58, 334). Humans with Laron dwarfism, i.e., analogous endocrine syndrome of GH resistance, exhibit reproductive phenotype closely resembling Ghr−/− mice. Both sexes are fertile, but puberty is delayed (167). In the author's laboratory, sexual maturation in Ames dwarfs is delayed by 3–4 wk (52 ± 8 versus 23 ± 2 days in females and 48 ± 6 versus 25 ± 2 days in males; P < 0.0001). This characteristic is strongly dependent on genetic background. In some stocks segregating for Ames or Snell dwarfism, sexual maturation of the mutants is delayed much more, and in some cases, indefinitely (17). In our breeding colony of Ames dwarf mice, nearly all males are fertile and we use them routinely for breeding. Female Ames dwarf and Snell dwarf mice can ovulate and mate, but their fertilized eggs fail to implant and develop due to prolactin deficiency and the resulting luteal failure (15). In many mammalian species, including mice, activation of corpora lutea and secretion of progesterone in amounts needed for implantation and maintenance of pregnancy absolutely require the action of PRL, which is lacking in Snell and Ames dwarf mice due to the absence of lactotrophes (18, 177, 278). Female dwarf mice treated with PRL can get pregnant, deliver live young, and raise them to weaning (14).

It is interesting that delayed sexual maturation and reduced fecundity of GH-resistant (and likely also GH-deficient) mice appear to be counterbalanced by a delay in the age-related decline in fertility (10; Westbrook & Bartke, unpublished data). This echoes the findings in genetically normal animals subjected to calorie restriction in which fertility may be partially or completely suppressed, but a potential to reproduce upon shift to ad libitum feeding is maintained beyond the age at which reproductive function normally ceases (129). Moreover, age-related alterations in neuroendocrine function related to male reproduction are attenuated in Ghr −/− mice. Decline in plasma testosterone and increases in prolactin and in luteinizing hormone responses to gonadotropin-releasing hormone stimulation accompanied aging in normal mice but were absent in 2-yr-old Ghr−/− animals (57).

Alterations in immune function may provide another example of trade-offs between longevity and various aspects of physiological functioning in GH-deficient and GH-resistant mutants. Snell dwarf mice were described in 1972 as being grossly immunodeficient (92). These findings are somewhat difficult to interpret, because early mortality and extremely short life expectancy of Snell dwarf mice reported in the same publication (92) were not seen in other laboratories working with these mutants (259, 266, 268), and thus may have been peculiar to environmental conditions and/or husbandry practices in this particular study. However, findings of Fabris et al. (92) in Snell dwarf mice were very important by providing some of the earliest evidence that function of the immune system is subject to endocrine control. More recent work clearly established that under modern husbandry conditions, Snell dwarf mice are long rather than short-lived and provided evidence that aging of their immune systems is significantly delayed as assessed by the ratio of naive to memory T cells (100).

In Ames dwarf mice, activity of the immune system is lower than in normal mice as indicated by reductions in cellularity of the thymus and spleen (90) in the responses of lymphocytes to mitogens (89) and in the activity of natural killer cells (90) However, the ability to produce antibodies in response to a challenge with tetanus toxoid was not compromised in Ames dwarf mice (116), and these animals can thrive under standard conditions, as opposed to specific pathogen-free or “barrier” conditions, which implies resistance to various pathogens. Similarly, we are not aware of any reports of immunodeficiency or increased susceptibility to infectious disease in GH-resistant or GH-deficient humans.

Maintenance of the immune system is energetically costly and activation of a “survival program” which is believed to occur in response to calorie restriction includes partial suppression of immune functions. This does not interfere with health or survival under modern laboratory conditions but can increase the susceptibility of the animals to viral, microbial, and helmitic pathogens or parasites (105, 158, 285). Interesting similarities between the effects of CR and reduced somatotropic signaling on immune function also include a delay in immune aging in animals subjected to CR (231).

Thus, in laboratory rodents, both reproductive and immune functions tend to be reduced by either dietary or genetic interventions that produce a major extension of longevity. However, these suppressive effects are counterbalanced by attenuating the age-related decline in reproductive and immune functions by the same interventions.

VII. GROWTH SIGNALING, CELLULAR PROTECTION, AND AGING IN SIMPLE ORGANISMS

The trade-off between investment in either growth or cellular protection is likely to be a fundamental property of all living organisms. This trade-off also provides the foundation for several evolutionary theories of aging including antagonistic pleiotropy and the disposable soma. Not surprisingly, the great majority of long-lived mutants in organisms ranging from bacteria to mice are stress resistant but also display defects in either growth rate, fertility, or size (187). Notably, the trade-off between cellular protection and longevity and growth is not necessarily observed under standard laboratory conditions, since the trade-off may be subtle but detrimental for fitness. Therefore, to understand somatotropic signaling and its effect on aging and disease in mammals, it is important to introduce the fundamental relationship between growth-stimulating nutrient signaling pathways, cellular protection, and aging in yeast, worms, and flies. In yeast, the Tor/Sch9 (Tor/S6K) pathway is perhaps the most closely associated with the mammalian IGF-I/AKT and Tor/S6K pathways since all three 1) are affected by amino acid/protein levels, 2) play central roles in cellular division and organismal growth, and 3) promote aging and cellular sensitization to stress. In the yeast, Saccharomyces cerevisiae, mutations in either the Tor/Sch9 and glucose-response Ras/adenylate cyclase/PKA (Ras/AC/PKA) pathways extend lifespan and also increase cellular resistance to a variety of stresses including oxidative and heat stress (94, 96, 101, 186). Whereas Tor/S6K-deficient yeast cells display slow growth and generate both cells and colonies of reduced size, cells deficient in Ras/AC/PKA signaling grow normally and generate cells and colonies of normal size. Notably, in many mammalian cell types, Tor, S6K, and Ras can all be activated in response to GH and/or IGF-I signaling, indicating that downregulation of both the Tor/S6K and Ras/AC/PKA pathways in yeast can model many of the molecular changes that occur in GH, GHR, and IGF-I-deficient mice. As discussed elsewhere in this review, it is not clear what portion of the effects of GH and GHR deficiencies on protection and lifespan are mediated by reduced circulating IGF-I levels. In fact, GH/GHR can affect Tor/S6K and AKT signaling independently of its effects on IGF-I levels and also affect insulin levels, raising the possibility that multiple factors are responsible for the effects of GH on longevity (126).

One of the consistent effects of the inactivation of the Tor/Sch9 and Ras/AC/PKA pathways in yeast is a major reduction in age-dependent DNA damage including point mutations, gross chromosomal rearrangements, and small DNA insertions/deletions, all damage important for cancer development in mammals (93, 95, 192, 193). These effects on genomic instability but also those on cellular protection and longevity extension are mediated in large part by stress resistance transcription factors Gis1 and Msn2/Msn4 (94, 96, 101), and require expression of the mitochondrial superoxide scavenger SOD2, which is positively regulated by all three transcription factors (94, 96, 101).

In worms, the relationship between the pro-aging signaling daf-2 insulin/IGF-I-like pathway and body size is not clear, since many long-lived mutant alleles are of normal size. However, both the daf-2 and the Tor pathways converge onto a set of genes that affect growth of the worm into a reproductive adult but also limit its longevity (137), suggesting that the effect of these pathways on growth and longevity is conserved. Similarly to the effects observed in yeast, the downregulation of the PI3K age-1/AKT pathway downstream of the daf-2 insulin/IGF-I-like receptor promotes cellular protection and longevity by activation of stress resistance transcription factor DAF-16 (140, 151).

In fruitflies, the insulin-like receptor (InR) and its downstream substrate chico also regulate the size of the organism as well as metabolism and longevity. Flies deficient in Chico, the insulin receptor substrate, are small, display a smaller cell size (35), and have an increased lifespan (300). Deficiencies in the genes encoding for insulin-like ligands (Drosophila insulin-like peptides; DILP 1–7) also increase lifespan and resistance to oxidative stress and heat (48).

In summary, in yeast, worms, and flies, there is a strong inverse association between lifespan extension and growth, determined by the activities of conserved proteins including AKT, Tor, and S6K, which are regulated by GH, IGF-I, or both.

VIII. GROWTH SIGNALING, ADIPOSITY, AND AGE-RELATED DISEASE: FROM SIMPLE ORGANISMS TO MAMMALS

A. Growth Signaling and Reserve Carbon Sources in Simple Organisms

One of the phenotypes consistently associated with mutations that extend lifespan in organisms ranging from yeast to mice is the increased storage of reserve carbon sources. In yeast, the Ras/AC/PKA and the Tor/Sch9 pathways block the accumulation of glycogen, the major reserve carbon source stored during aging (97, 101, 240). As for stress resistance, the glycogen accumulation effects of reduced activity of the Ras/AC/PKA and the Tor/Sch9 pathways are also mediated in large part by stress resistance kinase Rim15 and the downstream transcription factors Msn2/Msn4 and Gis1 (96, 240, 317). The effect of analogous nutrient signaling pathways on the storage of reserve carbon sources is also observed in worms. For example, worms with reduced activity in components of the insulin-like/AKT and TOR/S6K pathways are long-lived and accumulate both fat and glycogen (151). Analogously to the effects observed in yeast, the stress resistance transcription factor DAF-16 promotes the accumulation of fat in worms (218). Defects in insulin-like signaling also increase fat storage in Drosophila (300).

B. Growth Signaling and Adiposity in Rodents

In agreement with the phenotype of lower eukaryotes with deficiencies in growth pathways, mice with deficiencies in the growth hormone-IGF-I axis, including mice lacking GH and the GH receptor (GHR−/−), are long-lived and obese (19, 20, 179). GHR−/− mice, which lack the GH receptor and binding protein and which, as a result, have severe IGF-I deficiency and growth defects, display a major increase in fat mass (29–31, 41). Surprisingly, these mice are not insulin resistant and, in fact, are insulin sensitive (40, 78, 183). These studies are in contrast to those of mice and humans with mutations that reduce IGF-I gene expression directly, which instead are associated with insulin resistance (52). When placed on a high-fat diet, the GHR−/− mice display an approximately twofold increase in fat mass versus the threefold increase observed in wild-type mice, suggesting that the increased fat deposition on the standard diet is not due to an altered response to high-fat nutrition. When the sites of fat deposition were examined, GHR−/− mice displayed lower levels of intra-abdominal fat and higher levels of subcutaneous fat. Remarkably, the insulin level on the low-fat diet was over 65% lower and that on the high-fat diet 80% lower in GHR−/− compared with that in controls (30). Glucose levels were also more than 40% lower in GHR−/− compared with those in wild-type mice on both the low- and high-fat diets. In contrast, insulin but not glucose levels were elevated in the GH overexpressor mice, particularly on the low-fat diet (30). Not surprisingly, these mice placed on a high-fat diet develop diabetes (228), whereas GHR−/− mice are protected from glomerulosclerosis and other markers for diabetes induced with the drug streptozotocin (28). Notably, mice with excess GH function, which are larger than control mice, also displayed a 2.1-fold increase in fat mass; however, the percent body fat for these mice was greatly reduced compared with wild-type mice and especially compared with GHR−/− mice (body fat: wild-type low-fat diet = 10%, GHR−/− low-fat diet = 20%, GH overexpressor low-fat diet = 6%). This increase in percent body fat in GHR−/− begins at ∼13 wk and continues for at least 2 yr (31). Notably, no significant difference in percent lean mass was observed in either female or male GHR−/− (31). In contrast, in 9- to 13-wk and 1-yr-old GH deficient rats, visceral fat was decreased, although marrow adiposity was generally increased in GH-deficient compared with wild-type rats (74). This is particularly surprising, considering the increased storage of either glycogen or fat in organisms lacking pro-growth genes ranging from yeast to humans (see next section).

These studies indicate that reduced GH and GHR levels promote obesity (FIGURE 4). However, mutations that directly cause reduced expression of IGF-I are associated with insulin resistance, suggesting that GH may act independently of IGF-I to promote insulin resistance (FIGURE 4).

Figure 4.

Key mechanisms linking GH signaling with adiposity, insulin resistance, and cardiovascular disease (CVD). Details and additional mechanisms are in the text.

C. Growth Signaling and Adiposity in Humans

Much of the understanding of the relationship between human GH, IGF-I, and obesity comes from studies of Laron syndrome (LS, or GHR-deficient, GHRD) subjects, who display severe dwarfism, high GH levels, and low serum IGF-I levels (171). A study of 25 GHRD subjects showed that they develop marked subcutaneous and visceral adiposity and tend to have elevated cholesterol (170). Long-term treatment of GHRD adults with IGF-I decreased subcutaneous fat as well as serum cholesterol and triglycerides (170). Because GHRDs cannot respond properly to GH since the receptors are mostly inactive, the authors hypothesized that IGF-I acts directly on adipose tissue storage. However, another preliminary study with just three children suggested that long-term IGF-I treatment may actually increase adiposity, raising the possibility that part of the effect of the GH and GHR on obesity may be independent of IGF-I levels (168). Another adiposity-related characteristic of GHR-deficient subjects is nonalcoholic fatty liver, which was detected in the majority of GHRD males but also in many GHRD females (169).

Remarkably, the obese GHRD subjects display a major reduction in diabetes prevalence. Whereas a few cases of diabetes have been reported for the GHRD subjects in the Middle East (169), diabetes has not been diagnosed in any of the 99 GHR-deficient subjects in Ecuador (115). In 13 relatives and 16 GHRD male and female subjects ranging from 20 to 50 years of age, no significant difference in fasting glucose concentrations was observed, but the insulin concentration in the GHRD group was about one-third compared with that in the relatives (115). Consequently, based on the HOMA-IR (homeostatic model assessment-insulin resistance) index, the GHRD subjects (HOMA-IR = 0.34) appeared to be much more insulin sensitive than their relatives (HOMA-IR = 0.96). These human results are consistent with those for GHRD mice (see above). In agreement with these findings, despite the marked obesity, HOMA-IR was elevated in only one of the GHRD subjects tested in the Middle East by Laron et al. (142). One of the changes that may explain part of the uncoupling between obesity and insulin resistance/diabetes in GHRD subjects is the increase in adiponectin, which is one of the major hormones produced by fat cells and acts to increase insulin sensitivity (142). An increase in adiponectin levels along with improved insulin signaling were also reported in subjects lacking the growth hormone releasing hormone receptor (GHRHR) (225).

D. Mechanisms of GH/IGF-I-Dependent Insulin Resistance

Adiponectin levels have been consistently shown to be reduced in insulin-resistant and diabetic mice and humans and in patients with cardiovascular diseases (141). The well-established insulin-sensitizing effect of adiponectin indicate that its low levels contribute to insulin resistance and the consequent metabolic diseases. Thus the elevated adiponectin in GHRD could explain in part why a syndrome that promotes obesity does not cause diabetes. In fact, replenishment of physiological levels of adiponectin in a diabetic mouse model reduced insulin resistance and the replenishment of the combination of adiponectin and leptin completely reversed the insulin resistance in this diabetic lipoatrophic animal (330). Moreover, transgenic mice overexpressing adiponectin share many phenotypic characteristics with long-lived Ghr−/− and Ames dwarf mice (26, 189) and an increase in median lifespan was associated with hepatic expression of human adiponectin (230). One of the mechanisms by which adiponectin may affect insulin resistance is by decreasing plasma triglycerides through fatty acid oxidation and transport (141). These triglycerides have been reported to interfere with PI3K activation and the subsequent translocation/activation of the glucose transporter 4 (141).

Growth hormone has also been linked directly to insulin resistance using mice overexpressing or lacking GH and 3T3-F442A adipocytes. In mice, elevated GH expression increased the expression of p85a, the regulatory subunit of PI3K, and also caused a major reduction in adiponectin (75), whereas GH deficiency increases adiponectin and reduces p85a protein levels in epididymal fat, changes accompanied by a major decrease in insulin and insulin resistance index (75). The authors proposed a direct effect for GH in the regulation of p85a levels resulting in decreased PI3K activity and increased insulin resistance (FIGURE 4). Because IGF-I levels were reduced and adiponectin levels were increased in the fat cells of GH-deficient mice (75), the regulation of p85a by GH could be mediated indirectly by modulation of IGF-I and/or adiponectin. However, the studies indicating that mutations that directly affect the IGF-I gene increase and not decrease insulin resistance (52, 117, 327) support the direct effect of GH on insulin resistance independently of IGF-I (FIGURE 4). Another mechanism linking the absence of GH signals with improved sensitivity to insulin is reduced inhibitory (serine) phosphorylation of insulin receptor substrate 1 (IRS-1), as discussed later in this article.

E. Growth Factors, Cardiovascular Disease, and Cancer in Humans

Obesity is one of the major risk factors for diabetes but also for coronary artery disease and stroke. Thus the adiposity associated with GH and IGF-I deficiencies, in mice and humans, would be expected to also promote cardiovascular diseases. However, subjects from Brazil with deficiency in the growth hormone releasing hormone receptor (GHRHRD) gene, and consequently in GH, are obese, display higher levels for LDL and an inflammatory marker for CVD, but do not develop premature atherosclerosis (5, 226). Also, prolonged treatment of these GH-deficient subjects with GH caused weight loss but also an increase in the intima-media thickness and atherosclerotic plaques (226) as well as elevated diastolic and systolic blood pressure (226). Results consistent with those from GHRHRD subjects were obtained for the GHRD cohort of Ecuador. About 30% of the deaths for both GHRDs and their normal relatives were from cardiac disease and stroke, providing evidence in support of normal cardiovascular disease incidence in severely IGF-I-deficient subjects (115).

As discussed in the rodent section, both GH and GHRD mice either display a reduced incidence of tumors and an increase in benign tumors or a delay in the age at which tumors develop (133, 134, 312). In GHRD humans, none of the 30 deaths that occurred in the Ecuador cohort were reported to be from cancer despite a 20–22% portion of the deaths of their relatives caused by cancer (115). The exposure of human epithelial cells to IGF-I-deficient serum obtained from GHRD subjects caused a decrease in the DNA damage but also an increase in the apoptosis caused by treatment with hydrogen peroxide, indicating that very low IGF-I may both protect cells from oxidative DNA damage and promote cell death in highly damaged cells (115). These results in GHRD mice and humans are in agreement with the effect of deficiencies in Tor/S6K, Ras/AC/PKA, or both in reducing genomic instability in yeast (see model organisms section) (188). In support of a progenomic instability effect of growth factors during aging conserved from yeast to humans is also the finding that in the human epithelial cells incubated with serum obtained from GHRD subjects the expression of TOR was reduced and both N-Ras and PKA were among the most downregulated genes compared with cells incubated with serum from normal subjects (115). Taken together, these results provide additional evidence for a fundamental and conserved trade-off between cellular protection and growth.

Although a range of phenotypes including increased adiposity, reduced age-dependent cancer, reduced insulin resistance, or diabetes were found in both GHRD mice and humans, survival analyses of both GHRHRD and GHRD subjects revealed an increased mortality in young subjects but normal longevity (115, 225), suggesting that severe IGF-I deficiency in humans promotes some side effects that counterbalance its antiaging effects. Considering the unusually high (50%) portion of deaths from accidents, alcohol-related causes, and unknown causes in GHRDs, an alternative possibility is that the very short stature of these subjects may increase their risk of accidents or cause behavioral changes that promote early deaths (115). Although subjects with GHRD (GHRHR deficiency) or hypopituitarism including GH deficiency (159) have essentially normal life expectancies and can reach very advanced age, isolated GH deficiency was reported to reduce longevity (32).

IX. SOMATOTROPIC AXIS-RELATED GENETIC VARIANTS ASSOCIATED WITH EXCEPTIONAL HUMAN LONGEVITY

As discussed elsewhere in this review, the inhibition of components of the GH/IGF-I axis but particularly very low levels of GH or GHR have a consistent effect on protection against aging and age-related diseases in laboratory mice, resulting in a 40–60% lifespan extension. A number of studies are now suggesting that either mutations or polymorphisms that affect the GH or IGF-I levels or signaling are associated with human longevity. For example, Bonafe et al. (38) found that subjects carrying a specific allele of the IGF-I receptor have low levels of plasma IGF-I and are more represented among long-lived people. Another study of a cohort of Ashkenazi Jewish centenarians, their offspring, and matched controls showed overrepresentation of mutations in the IGF-IR gene among centenarians relative to controls (284). In agreement with the study by Bonafe et al. (38), these mutations were associated with reduced activity of the IGF-IR in lymphocytes, suggesting a state of IGF-I resistance (284). Another study identified additional potentially protective polymorphisms in two additional IGF-I pathway genes: a single nucleotide polymorphism (SNP) in AKT1 was significantly associated with increased longevity across three cohorts including the Ashkenazi Jewish centenarians, and a polymorphism in the FOXO3A gene was associated with longevity in females only (239). FOXO3A and other FOXO forkhead transcription factors are human homologs of the DAF-16 transcription factor associated with longevity extension in worms. Both in worms and mammals, these forkhead transcription factors are phosphorylated and negatively regulated by IGF-I signaling proteins including AKT (112, 152). The link between FOXO3a SNPs and longevity was previously described in several studies including one of a genetic variation within the FOXO3A gene associated with longevity in Japanese (322) and German centenarians (99). As discussed in the previous section, neither GHR- nor GHRHR-deficient subjects appear to be long-lived, as observed consistently for GH- and GHR-deficient mice. However, the major reduction in both diabetes and cancer provides solid evidence to support a conserved role for growth factors in promoting both aging and age-related diseases. It will be important to determine whether other factors such as EGF are also promoters of aging.

In comparisons of middle-aged offspring of exceptionally long-lived families with their partners in the Leiden Longevity Study, no significant differences were detected in the levels of IGF1 or IGFBP3 in the circulation (250). However, the levels of adiponectin that are negatively regulated by GH were higher in the offspring (Westendorp and VanHeemst, personal communication) and the offspring were more insulin sensitive (250, 320), thus resembling the findings in long-lived GH-deficient and GH-resistant mice.

X. MECHANISMS LINKING REDUCED SOMATOTROPIC SIGNALING WITH EXTENDED LONGEVITY

Delayed aging and remarkably increased longevity of GH-deficient and GH-resistant mutant mice prompted a search for mechanisms that may be involved. Studies in several laboratories identified numerous characteristics of these animals that are likely to delay aging and prevent age-related disease. Many of these characteristics have been previously associated with extended longevity in animals subjected to calorie restriction and/or in epidemiological studies of exceptionally long-lived people and their relatives. The list of candidate mechanisms of extended longevity in Ames dwarf, Snell dwarf, Ghr−/−, and/or little (Ghrhr^lit) mice includes increased stress resistance, reduced oxidative damage, enhanced capacity to metabolize xenobiotics, increased ratio of anti- to proinflammatory cytokines, improved insulin signaling combined with reduced insulin levels, reduced hepatic cholesterol biosynthesis and serum lipid levels, metabolic shift toward greater fatty acid utilization, reduced body temperature, reduced depletion of very small embryonic-like stem cells in the bone marrow, and improved genome maintenance. Additional mechanisms of extended longevity of GH-related mutants can be deduced from changes in gene expression in different organs of these animals. The reader is referred to recent reviews (19, 20, 22) and to TABLE 2 for details, references, and additional putative mechanisms of extended longevity in these animals. However, what may best explain the protective changes listed above, and the longevity phenotype of these dwarf mice and also of other dwarf organisms such as the Tor/S6K deficient yeast, is the previously discussed ability of most, if not all, organisms to either remain in a progrowth and proreproductive state associated with a higher rate of aging or in a standby state in which the protection of the germline and soma are the most important tasks.

Table 2.

Characteristics of GH-resistant and/or GH-deficient mice believed to contribute to their extended longevity

Characteristics of GH-Resistant and/or GH-Deficient Mice
Energy and fat metabolism
Reduced hepatic sterol biosynthesis (44)
Reduced lipolysis and serum lipids (315)
Increased expression of PPARγ, PPARα, and PGCIα (198, 199)
Altered mitochondrial function and reduced respiratory quotient (62, 319)
Hypothyroidism (severe in some mutants; mild or absent in others) (20, 319)
Reduced body temperature (in spite of increased mass of brown adipose tissue) (125, 132)
Increased food intake and O2 consumption per unit body weight at standard animal room temperature (319)
Shift of energy utilization from growth and fecundity to maintenance and repair (19)
Insulin signaling and glucose homeostasis
Reduced mass and secretory activity of pancreatic β cells; hypoinsulinemia (43, 70, 183, 237, 340)
Shift from pro- to anti-inflammatory adipokine profile: increased adiponectin, reduced TNF-α and IL-6 (3, 201, 315)
Increased whole animal insulin sensitivity (40, 202)
Mild hypoglycemia (43, 125)
Stress resistance and oxidative damage
Cellular resistance to a variety of oxidative, cytotoxic and metabolic stresses (175, 253)
Improved antioxidant defense mechanisms; reduced ROS production (49, 235)
Greater capacity for xenobiotic metabolism (7)
Reduced oxidative damage to macromolecules (49, 256)
Reduced mutation rate (104)
Growth, tissue maintenance, and cancer
Reduced hepatic IGF-I expression and circulating IGF-I levels (20, 340)
Preserved local IGF-I signaling (including the brain and the heart) (200, 288)
Reduced mTOR-mediated translation and cell size (40, 127, 265)
Delayed onset and reduced severity of neoplastic disease (133, 134, 312)
Delayed maturation; reduced adult body size (20, 147, 334)
Improved maintenance of stem cells (246)
Organ-specific alterations in apoptosis (106, 150)
Cardiovascular function
Reduced blood pressure (179)
Reduced cardiac extracellular collagen (127)
Rheological advantages of reduced body size (254)
Other potential mechanisms
Delayed immune system aging (100)
Altered N-glycosylation (308)
Altered profile of expression of multiple genes and micro RNAs (8, 25, 77, 211, 305)

The picture that emerges from these studies is a network of interacting phenotypic characteristics that collectively protect from age-related disease and improve life expectancy (FIGURE 5). This “longevous” phenotype resembles many characteristics of genetically normal animals subject to calorie restriction (CR) that are also long-lived and partially protected from age-related disease. Thus both dwarf mice and CR animals are small, hypoinsulinemic, and insulin sensitive with reduced body size, body temperature, and fecundity, as well as a reduced incidence of cancer. However, important differences also exist. For example, CR animals are lean but long-lived dwarf mice are not. In fact, Ghr−/− males are clearly obese with percentage of body fat reaching ∼40%, i.e., values roughly double those measured in their normal siblings (31, 41). The paradox of obesity coexisting with improved insulin signaling, delayed onset of cancer, and extended longevity appears to be due to striking differences in the secretory activity of adipose tissue exemplified by increased rather than reduced adiponectin levels in Ghr−/−, Ames, and Snell dwarfs (29, 46, 315). In addition, adipocyte senescence is delayed (J. Kirkland, personal communication), and the accretion of white adipose tissue is predominantly subcutaneous (29, 41) in these mutants.

Figure 5.

Mechanisms believed to be involved in linking GH signaling with healthspan and lifespan. Details and references are in the text and in TABLE 2.

Recent studies involving surgical removal of most of the epididymal and perinephric fat pads provided evidence that visceral adipose tissue of Ghr−/− males is an important source of adiponectin and promotes enhanced insulin sensitivity (201). This is in sharp contrast to the findings in control (normal) animals in which visceral fat secretes less adiponectin and more proinflammatory cytokines and promotes insulin resistance (201).

It is very difficult to know which of the characteristics of dwarf mice believed to represent candidate mechanisms of delayed aging can be directly related to suppression of somatotropic signaling and which are secondary or tertiary. For example, deficiency of GH and IGF-I interferes with development of insulin-producing β cells in the pancreas. However, reduction of insulin levels in dwarf mice can also be a response to the improved insulin sensitivity caused by the absence of GH signaling. Increased accumulation of subcutaneous fat in these mutants likely reflects the absence of lipolytic effects of GH but could also be due to improved sensitivity to insulin. Delayed onset and reduced incidence of age-related disease can be related to increased resistance to oxidative stress, improved insulin signaling, altered stem cell maintenance, delayed senescence, and altered secretory profile of adipocytes and associated macrophages, as well as reduced somatotropic support of cancer growth.

Of the numerous mechanisms likely to contribute to delayed aging of the GH-related mutants, we are particularly interested in improved insulin sensitivity combined with reduced insulin levels because these characteristics are consistent with highly conserved impact of insulin-like and insulin signaling on aging in other organisms and have been specifically associated with human longevity (234, 250, 320). Moreover, opposite phenotypic characteristics, hyperinsulinemia, and insulin resistance figure prominently in the definition of metabolic syndrome which is associated with increased risk of various age-related diseases and thus with reduced human life expectancy (223, 280).

Our studies revealed that improved whole animal insulin sensitivity of Ghr−/− mice is associated, most likely causally, with alterations in the insulin signaling pathway and in acute responses to insulin stimulation in different insulin target tissues (40, 42, 78). In the liver, the levels and activation of insulin receptors were increased in Ghr−/− mice. In the skeletal muscle of these animals, there were increases in the levels and activation of Akt1 and Akt2 leading to increased levels of a glucose transporter, Glut4. Improved muscle insulin sensitivity in Ghr−/− mice was also associated with reduced inhibitory (serine 307) phosphorylation of insulin receptor substrate1 (IRS-1). This, in turn, was likely due to reductions in mTOR signaling (40) and in activation of c-Jun NH₂-terminal kinase, JNK1 (4). The role of inhibitory (serine) phosphorylation of IRS-1 in the control of insulin signaling is well documented in rodents as well as in the human (2, 84). Studies of the interactive effects of GH receptor deletion with 30% calorie restriction resulted in identification of characteristics and responses related to insulin signaling which were consistently associated with alterations of longevity. These were present in animals in which longevity was extended by GH resistance or by CR but, importantly, were not present in animals in which CR failed to produce a further increase in lifespan (16, 40, 202).

XI. STRESS RESISTANCE AND RESPONSES IN THE CELLS AND TISSUES OF MICE WITH ALTERATION IN SOMATOTROPIC SIGNALING

Long-lived yeast with mutations in the Tor/S6K and Ras pathways, both major IGF-I signaling components, are remarkably resistant to a variety of stresses including paraquat, hydrogen peroxide, heat, and chemotherapy drugs (96, 245). Long-lived nematodes C. elegans with mutations in the insulin/IGF-I pathway also show an increased resistance to multiple forms of stress, including oxidative stress, heat, UV stress, and the heavy metals cadmium and copper (12, 173, 180, 216). It has been hypothesized that resistance to multiple forms of stress leads to delayed aging and extended longevity (98). Interestingly, the mean lifespan of worms can also be extended by transient exposure to nonlethal thermal stress, which induces a temporary increase in stress resistance (181). Similar observations have previously been reported in Drosophila after transient exposure to thermal stress (203). These results suggest that the relationship between stress resistance and lifespan found in C. elegans could also extend to mammals and that mutations that extend lifespan in mice may also render them resistant to oxidative damage and death caused by agents that induce oxidative stress. It was found that mouse embryonic fibroblast (MEF) cultures derived from hemizygous IGF-IR^(+/−) mice are significantly more resistant to death induced by hydrogen peroxide than those derived from control mice (130). Female mice carrying a heterozygous deletion of this gene live longer than normal controls (130). Interestingly, hemizygous IGF-IR^(+/−) mice survive significantly longer than controls following in vivo injection of paraquat, which causes massive oxidative damage. The increase in the in vivo stress resistance of these mutants is greater in females than in males, thus corresponding to their longevity (130). Another early example of increased stress resistance of long-lived mice involved animals with targeted mutation of the p66shc gene, a downstream effector of IGF-I receptor signaling, which were reported to live 28% longer than their controls (209). p66^shc (molecular mass 66 kDa) is encoded by the protooncogene SHC locus and is a splice variant of p52^shc and p46^shc. p66shc was found to inhibit c-fos promoter activation, which occurs in response to environmental stress such as UV light and oxidative stress (209). Resembling findings in hemizygous IGF-IR^(+/−) mice, MEFs from the p66shc mutant mice show resistance to apoptosis induced by H₂O₂ and ultraviolet light. Moreover, the p66shc mice also show increased survival after paraquat injections in vivo (209). However, interpretation of these results requires caution because of the relatively short lifespan of control mice in both of these studies. Thus the differences in lifespan, and perhaps also in resistance to the oxidative stress in these animals, may have in part reflected differences that might be unrelated to aging (163). Bokov and colleagues recently reported that in C57BL/6 mice, heterozygous deletion of IGF-I receptor caused a very modest (less than 5%) extension of longevity in females and no extension of longevity in males (36).

A series of in vitro experiments have shown that dermal fibroblast cell lines derived from adult long-lived Ames dwarf, Snell dwarf, and GHRKO mice are resistant, in culture, to the cytotoxic effects of multiple forms of stress including hydrogen peroxide, cadmium, ultraviolet light, paraquat, and heat (175, 204, 217, 253). Stress resistance of these fibroblasts was demonstrated in defined media in the absence of continued hormonal alterations and therefore might be due to epigenetic effects during early postnatal development (217, 253). Intriguingly, GH replacement sufficient to restore normal lifespan in Ames dwarf mice also led to a reversal of the resistance of skin-derived cells to multiple forms of stress (233). Thus cellular stress resistance appears to be determined by GH exposure during critical developmental windows (233). Ames dwarf mice have been shown to be very resistant to the lethal effects of exposure to the oxidative toxins diquat and paraquat in vivo (37, 290). Paraquat induces oxidative stress by catalyzing the generation of the superoxide anion and the oxidation of cellular NADPH. Diquat is a potent redox cycler and is readily converted to a free radical that, in reaction with molecular oxygen, generates superoxide anions and subsequently other redox products. These products can induce lipid peroxidation in cell membranes and potentially cause cell death. As measured by overall survival after a lethal dose of diquat or paraquat, Ames dwarf mice, both males and females, were much more resistant to both stressors than were normal littermates, and this resistance to paraquat toxicity was maintained with age in the Ames dwarf mice (37, 290).

Biochemical studies have begun to delineate stress response pathways that might contribute to the phenotype in long-lived mutant mice. The MAPKs comprise a ubiquitous group of signaling proteins that play a prominent role in regulating cell proliferation, differentiation, and adaptation, emerging as important regulators of cellular responses to various stimuli (59, 146). There are three major MAPK subfamilies, namely, the extracellular signal-regulated protein kinases (ERK), the c-Jun NH₂-terminal kinases (JNK), and p38 MAPK. Each of these sets of kinases has been implicated in cell injury and disease pathogenesis (59, 146). Exposure of primary skin-derived fibroblasts to peroxide, cadmium, or paraquat leads to phosphorylation of the stress-activated kinases ERK1 and ERK2. Surprisingly, the level of ERK phosphorylation induced by these stresses is significantly lower in cells from dwarf mice. This suggests defects in the pathways that sense cellular injury and lead to activation of the kinase MEK, responsible for phosphorylation of ERK (292). Moreover, p38 phosphorylation response to the mitochondrial inhibitor rotenone was also attenuated in fibroblasts from Ames dwarf compared with normal mice (194).

Stress resistance of Snell dwarf mice was also studied by using 3-nitropropionic acid (3-NPA) challenging. 3-NPA induces mitochondrial free radical generation in liver tissue and activates MAPK signaling cascade. After treatment with 3-NPA, phosphorylation of ERK signaling is lower in the liver of Snell dwarf mice, suggesting that this tissue is protected, in vivo, from oxidative damage (194). Analysis of the responses of hepatic tissues to oxidative stress revealed dramatic differences in the pattern of stress-induced MAPK activation in Ames dwarf mice compared with controls (290, 292). While normal animals responded by increases in ERK and p38 phosphorylation, dwarf mice responded by increases in p38 and JNK activation. It is suggested that the lower levels of diquat-induced P-ERK and P-p38 or the higher levels of JNK phosphorylation may be part of the mechanism of resistance of dwarf mice to oxidative stress (290, 292).

XII. SOMATOTROPIC SIGNALING AT DIFFERENT STAGES OF LIFE HISTORY

Studies of GH-deficient, GH-resistant, and IGF-I-resistant mutant mice provided strong evidence that reducing somatotropic signaling can extend longevity. However, results of these studies did not allow for separating the importance of adult as opposed to developmental effects of GH and IGF-I on lifespan. The potential significance of early “developmental” events is emphasized by the recent appreciation of the impact of prenatal and early postnatal events on adult phenotype, presumably via epigenetic mechanisms. This includes rapidly accumulating evidence that prenatal events and birth weight can determine susceptibility to chronic disease in adulthood (9, 249, 286).

Several studies addressed the role of early versus adult actions of hormones of the somatotropic axis on rodent aging. Sonntag et al. (273) utilized genetically dwarf rats in which GH signaling is reduced but longevity is not altered and demonstrated that the treatment of these animals with exogenous GH between the ages of 4 and 14 wk produced a significant extension of longevity. The authors suggested that early GH treatment corrected developmental defects that may have been present in these animals, while adult GH deficiency (after GH treatment was stopped) protected them from cancer (273). In contrast with these observations, we have reported that treatment of Ames dwarf mice with GH limited to the period between 2 and 8 wk of age can be sufficient to drastically shorten longevity of these normally long-lived mutants (233). Further work will be needed to determine whether these results represent species differences or developmental differences between GH-deficient dwarf rats and hypopituitary Ames dwarf mice.

As was previously mentioned, Svensson et al. (296) recently described the effects of suppressing hepatic IGF-I expression in mice at the age of 1 mo using pharmacological induction of Cre-Lox system. This treatment reduced circulating IGF-I levels to approximately one-fifth of the levels measured in control animals, reduced growth, and extended mean lifespan of females (296). Extension of mouse longevity in response to reduced availability of nutrients between birth and weaning (286) likely involves suppression of IGF-I signaling during this period. Collectively, results of these studies suggest that alterations in the somatotropic signaling limited to either “early” or “adult” stages of life history can have significant impact on longevity. Precise delineation of the impact of GH and IGF-I at different stages of ontogency on aging, age-related disease, and lifespan in different species will be important for devising interventions that could produce health benefits.

XIII. IMPACT OF DIFFERENCES IN SOMATOTROPIC SIGNALING ON LONGEVITY OF GENETICALLY NORMAL ANIMALS

There is persuasive, although largely indirect, evidence that differences in somatotropic signaling lead to differences in longevity not only in mutants with complete absence of GH or GH receptors or in transgenic animals with pathologic GH excess, but also in genetically normal individuals. Much of this evidence is derived from studies relating lifespan to body size because somatic growth and adult body size strongly depend on the actions of GH and IGF-I. It has been known for over 50 years that in comparisons of mice from inbred strains, outbred stocks, or selected lines differing in body size, smaller animals live longer (22, 254, 324). Similar negative association of body size and lifespan is very robust and well documented in comparisons of different breeds of domestic dogs and in large cohorts of mixed breed dogs (86, 113, 238). In dogs, these differences have been related to circulating IGF-I levels (295). Negative correlation of body size and lifespan was described also in rats (248) and horses (47) as well as in various human data sets analyzed by Samaras (255). His important work on this relationship was summarized in a recent book (254).

In 2002, Miller et al. (212) reported that in a genetically heterogeneous population of mice derived from crossing four inbred strains, longevity of individual animals was negatively correlated with body weight measured in early adulthood. More recently, Yuan et al. (333) reported a similar negative correlation of adult plasma IGF-I levels and median lifespan in comparison of 31 inbred mouse strains. Sun et al. (286) reported that mice with mild nutritional restriction during lactation (produced by adding pups to a litter) were smaller than mice raised in smaller litters and lived significantly longer.

It should be mentioned that these results contrast sharply with the well-known positive correlation of body size and longevity in comparisons of animals (mammals or birds) from different species. In this context, it is intriguing that circulating IGF-I levels were recently reported to be generally lower in larger species (282). Discussion of the possible physiological significance of differences in body size comparisons “within species” versus “between species” and likely reasons for numerous exceptions from the overall relationships between body size and longevity are outside the scope of this article.

XIV. IS AGE-RELATED DECLINE IN GROWTH HORMONE LEVELS A RESULT OR A CAUSE OF AGING? CAN TREATMENT WITH GH SLOW DOWN OR REVERSE AGING?

As was mentioned earlier in this article, levels of circulating GH reach their maximum around the time of attaining adulthood and thereafter progressively decline. In the rat, age-related decrease of pulsatile GH release has been linked to altered secretion of somatostatin into portal circulation (274). Deconvolution analysis of data derived from frequent blood sampling in the human provided evidence for age-related alterations in the hypothalamic control, of GH secretion, its modulation by gonadal steroids, and in GH autofeedback leading to reduced GH secretion (310, 311). Reduced levels of GH that are a consistent finding in middle-aged and especially in elderly individuals (107, 215, 251, 309) can be viewed as a result (a symptom) and perhaps also as a “biomarker” of aging. Thus alterations in GH levels could be added to a long list of the effects of aging on various physiological parameters including circulating levels of gonadal sex steroids, and, in primates, adrenal dehydroepiandrosterone (DHEA) and its sulfate, which also decline with age.

However, it has been suggested that in addition to being the normal accompaniment of aging, decline in GH levels is also a cause of aging, or more precisely, a cause of many of its symptoms. This interpretation of the functional significance of age-related decline in GH release is based primarily on two sets of observations: 1) some of the symptoms of aging, including loss of muscle mass and bone mineral density (BMD), increased adiposity, reduced energy, decline in sexual activity and increased risk of developing psychological depression resemble symptoms of adult GHD; and 2) in young and middle-aged adults, most of the symptoms of GHD markedly improve in response to GH therapy (27, 264). This line of reasoning gained powerful support from a widely quoted landmark study of Rudman et al. (252) who reported in 1990 that GH treatment of elderly men reduced adiposity, increased lean body mass, increased BMD in some of the examined sites of the skeleton, and tended to increase skin thickness, i.e., succeeded in reversing some of the symptoms of aging. Not surprisingly, this report led to considerable interest in using GH treatment as an antiaging therapy. However, subsequent studies involving larger numbers of subjects and more rigorous design (double blind, placebo controlled rather than open label) were less successful in producing the desired outcomes and identified various side effects including troublesome evidence for producing insulin resistance (33, 165, 236). Meta-analysis of available data confirmed the ability of injected GH to cause beneficial although generally modest changes in body composition (reduced adiposity and increased muscle mass) but reiterated the evidence for side effects such as arthralgias and carpal tunnel syndrome and raised the issue of the potential risk of diabetes and cancer (182). Thus these analyses supported the currently prevailing opinion of clinical endocrinologists and guidelines from the Endocrine Society and the American Association of Clinical Endocrinologists that GH therapy in endocrinologically normal elderly must be considered as experimental treatment and cannot be recommended for routine use (182, 301).

It is interesting that in spite of these publications and recommendations and the fact that use of recombinant human GH as an antiaging treatment is specifically prohibited in the United States, GH along with various “GH-related products” continues to be aggressively promoted. Advertisements promise the users many appealing benefits including improvements in energy level and sex life, reversal of balding, and return of original hair color with assurances of looking and feeling younger. Although many of these claims appear to be grossly exaggerated (to say the least), the beneficial potential of administering GH or increasing its release in selected groups of elderly individuals deserves attention and continues to be explored. Of particular interest in this context is the evidence that treatment with ghrelin analogs can produce near-normal patterns of GH release and thus reduce the incidence and severity of untoward side effects (220, 221). However, ghrelin levels normally increase with age (293), perhaps as a result of reduced GH negative feedback (244), and in mice this has been linked to increased adiposity, reduced thermogenesis and development of insulin resistance (178). Moreover, the ability of GH to increase muscle mass suggests that combining GH treatment with a program of exercise could offer a therapeutic approach to treatment of sarcopenia and frailty, particularly in women in whom androgen therapy would be generally contraindicated. However, the benefits of such interventions remain to be conclusively shown (165, 301), and the risks of cancer and diabetes would have to be considered before recommending GH therapy.

It should be pointed out that the controversial issues concerning the relationship of declining GH levels to aging (whether a symptom or a cause), the possibility of using GH as an anti-aging agent and the risk-benefit considerations of such use mirror the corresponding and generally better understood issues concerning estrogen replacement in peri- and postmenopausal women and androgen therapy in elderly men. It should also be pointed out that while GH therapy may eventually find its place in the management of individual geriatric patients, there is considerable evidence (described earlier in this article) that action of GH during lifespan tends to accelerate rather than to slow or prevent aging and age-related diseases. This would suggest that age-related decline in GH-release can be viewed not only as a symptom or one of the mechanisms of aging but rather as an evolutionarily selected protective mechanism that reduces the risk of age-related disease, primarily cancer and type 2 diabetes. The latter interpretation would seem consistent with reduced lifespan and various indices of premature or accelerated aging in patients with GH secreting tumors (136, 229) and in transgenic mice overexpressing various GH genes (13). Moreover, increased serum IGF-I levels in hepatocyte-specific IGF-transgenic mice are associated with increased mortality (87).

GRANTS

Preparation of this article was supported by The National Institute on Aging via Grants P01 AG031736, R01 AG019899, and R21 AG038850 (to A. Bartke); AG20642, AG025135, P01 AG034906, Ted Bakewell (The Bakewell Foundation), and the V Foundation for Cancer Research (to V. Longo); and a University of Michigan Nathan Shock Center Pilot Grant (to L. Sun).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.