Keywords

1 Introduction

The growth hormone (GH) and insulin-like growth factor 1 (IGF-1) axis (collectively known as the somatotropic axis) was demonstrated to be a major determinant of mammalian longevity more than 20 years ago [1, 2]. Since then, numerous laboratories have attempted to elucidate mechanisms underlying the role of this axis in longevity, leading to an ever-growing list of possible mechanisms to disentangle [3, 4]. During the same timeframe, the growing obesity epidemic in the developed world has resulted in a dramatic increase in adipose tissue (AT) research. Since GH plays integral roles in the physiology of AT, there has been a surge in research attempting to understand how AT impacts longevity in GH-mutant mice. This review is aimed to give an overview of AT, GH, and how the interplay between the two influences longevity.

2 Adipose Tissue

Traditionally, AT was believed to be metabolically inactive, solely acting as a tissue to store excess calories. However, paradigm-shifting studies demonstrated that AT secretes adiponectin [5], leptin [5, 6] and resistin [7], which paved the way for future work on AT as an endocrine organ. Since then, our understanding of AT has been greatly expanded. We now know that there are at least three distinct types of adipose tissue: white adipose tissue (WAT); brown adipose tissue (BAT); and beige AT. As predicted, each type of AT depot has specific functions. Further differentiating these types of AT is the significant cellular heterogeneity within an AT depot itself [8]. Despite this heterogeneity, AT is largely made-up of postmitotic adipocytes and their replicative precursors, termed preadipocytes. The differentiation of preadipocytes into mature adipocytes is transcriptionally controlled through the coordination of CCAAT/enhancer-binding proteins (C/EBPs) and peroxisome proliferator-activated receptor gamma (PPARγ) [9]. This section is dedicated to defining the similarities and differences between WAT, BAT and beige AT, which are summarized in Table 11.1.

Table 11.1 Different types of adipose tissue. The major similarities and differences between white adipose tissue (WAT), brown adipose tissue (BAT), and beige adipose tissue
Full size table

2.1 WAT

The defining characteristic of WAT in both humans and mice is the storage of excess energy. Morphologically, WAT is characterized by a large, unilocular lipid droplet, with few mitochondria. WAT has both similarities and differences in mice and humans. In mice, WAT is present in superficial subcutaneous depots, mainly in the scapular and inguinal regions. WAT is also present in the intra-abdominal region of mice in the form of perigonadal (epididymal and paraovarian in males and females, respectfully), mesenteric, and retroperitoneal AT. In humans, WAT is more widely distributed and is present subcutaneously in the gluteal, femoral, clavicular, and abdominal regions. WAT in humans is also present intra-abdominally in intraperitoneal, retroperitoneal, mesenteric, and omental AT depots. The major difference in WAT distribution between mice and humans is the large perigonadal depot in mice, and the large omental depot in humans. Most mouse studies in the context of metabolism and aging use inguinal WAT (iWAT) to represent subcutaneous AT, and perigonadal AT to represent the so-called visceral AT. Although this practice is widely used and accepted, it is worth mentioning that some investigators prefer a more stringent use of the term “visceral” to include only AT that directly drains into the portal vein, rather than any intra-abdominal AT depot. By this definition, only mesenteric AT in mice would be considered “visceral” [10].

In obese subjects, there are critical changes in WAT physiology. AT itself is surrounded by a thick extracellular matrix (ECM). During weight gain, the ECM in WAT must expand to accommodate hypertrophic adipocytes. However, this results in poor vascularization [11] and subsequent hypoxia. Hypoxia in the adipocyte is just one of several instances that cause an increased secretion of proinflammatory cytokines to be released from WAT during obesity [12]. Another unfavorable impact of obesity on WAT is ectopic lipid distribution. For example, intra-myocellular lipids [13,14,15,16,17] and intra-hepatic lipids [18,19,20,21] are associated with insulin resistance, while epicardial fat is associated with an increased risk of coronary artery disease [22,23,24].

2.2 Bat

The differences between BAT and WAT begin during development, as BAT comes from a mesoderm lineage that is myogenic factor 5 (MYF5) positive [25]. Unlike WAT, BAT is characterized by multilocular lipid droplets, many mitochondria, and is rich in both innervation and microvasculature. These anatomical traits are important for the main function of BAT, thermogenesis. Sympathetic nerves provide a source of norepinephrine (NE) to stimulate thermogenesis, blood vessels provide nutrients to the tissue as well as aid in heat dissipation, and multilocular lipid droplets have an increased surface area to facilitate an increased rate of lipolysis. The thermogenic circuit relies on several transcriptional inputs including peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) [26] and PR domain containing 16 (PRDM16) [27]. Moreover, cues from other transcriptional machinery such as thyroid hormone receptor [28] and retinoic acid receptor [29] play important roles in the thermogenic circuit. Readers interested in learning more about the transcriptional control of the thermogenic program are directed to the following reviews [30,31,32]. Central to the function of BAT is uncoupling protein 1 (UCP1), which dissociates the electron transport chain to release chemical energy in the form of heat.

Thermogenesis itself begins with the release of NE from the sympathetic nervous system, which acts on β3-adrenergic receptors, which are associated with G-protein coupled receptors (GPCRs) of the Gs subtype [33, 34]. Subsequently, a rise in cytosolic cAMP results in the activation of protein kinase A (PKA) [35], which has several functions including activating mitogen-activated protein kinase (MAPK) p38 [36] and increasing cytosolic free fatty acid (FFA) levels in the cell by phosphorylating perilipin [37]. This, in turn, causes the release of comparative gene identification-58 (CGI-58) to activate adipose triglyceride lipase (ATGL), the major triglyceride lipase in BAT [38,39,40]. The breakdown of triglycerides into FFAs in BAT is critical for two reasons. First, the resulting FFAs can be shuttled into the mitochondria where they can undergo β-oxidation to produce ATP and reduced electron carriers to maintain thermogenesis [30]. Second, the FFAs act as activators of UCP1 [41, 42]. Conversely, purine nucleotides act as inhibitors of UCP1 [41].

It was long believed that BAT in humans was non-existent past adolescence. However, a decade ago, seminal studies that “rediscovered” BAT in adult humans proved otherwise [43,44,45]. Because a few hundred milligrams of BAT can oxidize up to 60% of consumed glucose and lipids in a cold-acclimated mouse [46, 47], investigators became interested in using BAT therapeutically to combat the growing obesity epidemic. Although humans do possess BAT, it differs in several ways from BAT in mice. Mice have several distinct BAT depots including the large interscapular depot, as well as the axillary, cervical, paraaortic, cardiac, and perirenal depots. In humans, brown adipocytes appear interspersed in WAT, mainly in the supraclavicular region, but are also present in the para-aortic, cervical, axillary, perirenal, and paravertebral regions. It is worth noting that recently a supraclavicular BAT depot was discovered in mice [48]. There is an interscapular BAT depot in humans, although it disappears during adolescence. Currently, the gold standard for assessing the location and activity of BAT in humans is positron emission tomography coupled with computed tomography (PET/CT) during the infusion of radiolabeled 18-fluoro-deoxyglucose (18FDG) in a patient either wearing a “cold vest” or receiving a β3-agonist, such as mirabegron [49]. However, this method does not always accurately reflect BAT activity, and has led to vastly different estimates of the total volume of BAT present in humans, ranging over two orders of magnitude from only a few, to a few hundred milliliters [50]. The measurement of BAT in humans, along with a detailed description of its pitfalls has been reviewed elsewhere [49]. Regardless, we know that BAT is present in adult humans and has a significant impact on metabolism. A clear example of this is a study in which type 2 diabetics spent several hours a day over a 10-day period at 15 °C, which resulted in a significant increase in their glucose infusion rate (GIR) during a euglycemic clamp [51].

One of the biggest advances in our understanding of thermogenic AT in the past few years is the presence of UCP1-independent thermogenic mechanisms. For example, both brown and beige AT thermogenesis can occur through a creatine-based substrate cycle [52,53,54]. Moreover, beige AT thermogenesis can be controlled through ATP-dependent calcium cycling [55]. These findings can begin to explain why UCP1 null mice only become obese under thermoneutral temperatures [56, 57], while BAT-deficient mice are obese and insulin resistant at standard room temperature [58, 59]. To-date, however, these UCP1-independent forms of thermogenesis have not been examined in GH mutant mice.

2.3 Beige AT

Beige AT is distinct from both WAT and BAT. Beige adipocytes reside within WAT depots, contain mitochondria expressing UCP1, and are therefore thermogenic. Under basal conditions, thermogenic output of beige adipocytes is relatively low, however, stimulants such as cold exposure, exercise, or treatment with PPARγ agonists significantly increase the expansion and energy expenditure in these cells in a process termed beiging. Although beiging was described more than 30 years ago [60, 61], only recently have the specific lineages and molecular regulators that give rise to beige AT been worked out [62, 63]. It is worth noting that this is an area of ongoing investigation, with no clear consensus. For example, some studies have shown that beige AT derives from a lineage distinct from BAT that is positive for myosin heavy chain 11 (MYH11) and platelet derived growth factor receptor alpha (PDGFRα) in mice [64,65,66,67], while other beige adipocytes have been found to be positive for paired box 3 (PAX3) and MYF5 [68]. Adding further complexity are conflicting studies in which some investigators demonstrate that beige adipocytes are formed de novo in response to external cues [66, 69], while others argue they arise from the transdifferentiation of white adipocytes [70, 71]. Regardless of the developmental origin of beige adipose tissue, during cold-exposure, thermogenic beige adipocytes replace non-thermogenic white adipocytes, which is reversible when cold-acclimated mice are placed at thermoneutrality [72].

3 Properties of Adipose Tissue

3.1 Adipose Tissue Heterogeneity

Beyond the previously discussed inter-depot differences, AT is highly heterogenous within each depot. By volume, the majority of AT is composed of mature adipocytes. These adipocytes have a turnover rate of approximately 10% annually in humans, with a much faster turnover rate of 5% daily in mice [73, 74]. Because of this, preadipocytes, or committed adipocyte progenitors, are critical to the AT niche. Work has already been done to identify markers of white, brown, and beige adipocytes, however, studies identifying novel cell surface markers of preadipocytes that give rise to these adipocytes is mostly lacking, although some markers have been identified [75]. For example, sorting preadipocytes with high CD29 expression enriches for a population of preadipocytes that differentiate into cells with a high expression of UCP1 [75]. Preadipocytes are also a major source of tumor necrosis factor α (TNF-α), suggesting their role in AT extends beyond acting as a precursor cell [76].

Immune cells are another cell type with a large role in AT. It has been appreciated for years that macrophages are present in AT, and that their presence increases with obesity. However, considerable improvements have been made to our understanding of different subpopulations of macrophages in AT [77,78,79]. For example, there is good evidence that both M1 (referred to as classically activated) and M2 (referred to as alternatively activated) macrophages exist within AT [80, 81]. Along with macrophages, natural killer cells are recruited to AT during obesity, causing insulin resistance [82]. Changes in resident immune cells may also play a beneficial role in the physiology of AT. For example, cold-exposure causes an influx of M2 macrophages and eosinophils that aid in thermogenesis [83,84,85].

AT stores energy as triglycerides during caloric excess, and must liberate FFAs during periods of caloric demand through lipolysis. Since lipolysis is stimulated by the sympathetic release of NE, it makes sense that nerves are a core component of AT [86]. To transport nutrients to AT, or dissipate heat and FFAs (in BAT and WAT, respectfully), there must also be the presence of microvasculature, including endothelial cells and smooth muscle cells [8]. Certainly, AT heterogeneity has garnered attention in recent years. With advances in technologies such as single cell RNA-sequencing, the makeup and functional importance of the AT niche is sure to be further developed in the near future.

3.2 Adipose Tissue as a Secretory Organ

Studying WAT as an endocrine organ began 20 years ago with the discovery of leptin, resistin, and adiponectin being secreted from WAT [5,6,7]. WAT has also been known to secrete proinflammatory cytokines such as TNF-α and IL-6, which facilitate the development of insulin resistance during obesity. What has been much less studied, however, is the role of BAT as a secretory organ. Recently, this has changed as secreted factors from BAT (referred to as batokines) have been demonstrated to have both autocrine/paracrine and endocrine effects. For example, the lipokine 12,13-diHOME has been demonstrated to have an autocrine effect on BAT that results in increased lipid uptake [87]. Other factors that have paracrine/autocrine action in BAT are vascular endothelial growth factor A (VEGFa) and nitric oxide (NO), which increase angiogenesis [88, 89]. Moreover, fibroblast growth factor 2 (FGF2) and nerve growth factor (NGF) increase innervation and the recruitment of preadipocytes [90,91,92]. Endocrine factors that are secreted from BAT include insulin-like growth factor-binding protein 2 (IGFBP2) [93], WNT10b [93], and FGF21 [94]. BAT also secretes microRNAs. For example, both mice and humans show an inverse relationship between BAT activity and circulating levels of miR-92a [95]. Readers interested in learning more about the secretory function of BAT are encouraged to read a relevant review [96].

3.3 Alterations in AT during Aging

The main changes in WAT during aging is the gradual decline in tissue mass, the redistribution from subcutaneous to intra-abdominal depots, and the ectopic distribution of lipids in organs such as the liver and muscle [97,98,99,100]. Metabolically, aged WAT has a decline in its sensitivity to insulin and fatty acids [97, 101,102,103]. Moreover, aged WAT has an increased secretion of harmful proinflammatory cytokines such as TNF-α and IL-6 [104, 105]. There does appear to be an increase in macrophage infiltration in subcutaneous WAT, although this does not seem to apply to intra-abdominal WAT [106]. A final means through which WAT changes during aging is through the preadipocyte pool. Preadipocytes from aged tissue have lower levels of the transcription factors PPARγ C/EBPα, and their target genes [107, 108], which may explain the reduction in their capacity to differentiate into mature adipocytes. Moreover, senescent preadipocytes accumulate in aged WAT, which could contribute to metabolic impairment and increased systemic inflammation [97]. Readers interested in learning more about WAT remodeling during aging are directed to a relevant review [97].

Some of the changes mentioned above apply to BAT. For example, BAT mass decreases with age. Although senescence in BAT has been understudied, it is plausible to assume that brown preadipocytes also senesce, and lose the ability to differentiate into mature brown adipocytes with age. Many of the age-related changes in BAT relate to impaired thermogenic capacity. One mechanism for this is through the increased visceral AT expression of forkhead box protein A3 (FOXA3), which impairs BAT mass and function [109]. Interestingly, deletion of FOXA3 increases BAT late into life and extends longevity [109]. Another deleterious change in aged BAT is the presence of sympathetic neuron-associated macrophages, which chelate NE, resulting in decreased thermogenic output [110]. Finally, the well-documented age-dependent mitochondrial dysfunction impairs thermogenesis. The decline in the thermogenic function of BAT likely plays a critical role in the metabolic impairment and obesity observed during middle-age.

4 Growth Hormone

GH is a 22 kDa peptide hormone that is secreted from somatotrophs in the anterior pituitary. Its secretion is induced by the release of growth hormone releasing hormone (GHRH), and inhibited by the release of somatostatin (SST), both of which are released from the hypothalamus. GH has negative feedback on GH release from the pituitary, as well as on GHRH from the hypothalamus. Another level of feedback is through IGF-1 which acts on both the pituitary and hypothalamus [111]. Ghrelin, a “hunger” hormone is another factor that stimulates the release of GH [112]. AT can regulate GH production through FFAs and leptin which inhibit and stimulate GH production, respectively [113, 114].

In circulation, GH acts by binding to growth hormone receptor (GHR) on target tissues. Mainly, GH acts on the liver to stimulate the production of IGF-1, but GH can also act on other tissues such as muscle and AT [115]. Therefore, GH can elicit direct effects, or indirect effects through the action of IGF-1. Once bound to a homodimerized GHR, there is a conformational change in the receptor structure which brings together the associated janus kinase 2 (JAK2) domains together, allowing for transactivation [116] and subsequent phosphorylation of signal transducer and activator of transcription 5 (STAT5). Activated STAT5 can then enter the nucleus and act as a transcription factor [117]. GH has been demonstrated to signal through other non-canonical pathways including mammalian target of rapamycin (mTOR) and extracellular signaling-regulated kinase (ERK) [118].

5 Examples of Altered Growth Hormone Action

5.1 Humans

The two main ways that GH is altered in humans is through its overproduction in acromegaly and GH resistance, or through its under production in GH deficiency. Patients with acromegaly suffer from increased GH secretion, and subsequent increases in IGF-1 production. The increased secretion of GH is oftentimes the result of a pituitary adenoma. Acromegaly patients are more prone to cancer [119,120,121], diabetes [122], and are often short-lived compared to people with normal GH secretion [123, 124]. GH deficiency has multiple etiologies that influence the age of the onset of disease. In children, congenital GH deficiency is usually the result of mutations in genes encoding GH, GHRH, or other pituitary factors involved in the secretion of GH [125]. In adults, acquired GH deficiency is typically the result of hypopituitarism or irradiation of a pituitary adenoma [126]. Beyond deficiency, patients can be resistant to GH through mutations in the gene encoding GHR [127]. This, disease, termed Laron syndrome, causes patients to have low IGF-1, with elevated levels of GH [128]. Patients with both GH deficiency and resistance demonstrate decreased height, increased obesity, decreased bone mineral density, and altered lipid metabolism [3]. Interestingly, patients with Laron syndrome appear to be protected from cancer [129, 130] and diabetes [131], although Laron syndrome patients from cohorts in Israel and Turkey appear to still develop diabetes [132, 133], making the “protected” status from diabetes less clear.

5.2 Mice

To further understand the impact of GH signaling, several transgenic lines over-expressing GH have been created, the most commonly used being the bovine GH (bGH) transgenic line [134, 135]. These mice have a transgene that ectopically expresses GH under a strong promoter such as phosphoenolpyruvate carboxykinase (PEPCK). bGH mice are noticeably larger than their control littermates, and exhibit increased muscle mass. bGH mice demonstrate increased insulin resistance, and severe hyperplasia and hypertrophy of their hepatocytes [136]. Many of these mice die of hepatic cancer. Aging in these mice appears to be accelerated, and lifespan is reduced to around 1 year-of-age [135, 137].

Ames dwarf mice were first described in 1961, and suffer from a spontaneous mutation in the gene encoding prophet of pituitary factor 1 (Prop1) [138]. Snell dwarf mice were first described in the 1929, and suffer from a spontaneous mutation in the gene encoding pituitary factor 1 (Pit1) [139]. Since Prop1 is a transcription factor for the Pit1 gene, the phenotypes of Ames and Snell dwarfs are essentially identical, with both strains of dwarf mice lacking the production of GH, thyroid-stimulating hormone (TSH) and prolactin [140]. The downstream consequences of these mutations are a decreased production of IGF-1 and the thyroid hormones, T3 and T4. Both Ames and Snell dwarf mice are extremely long-lived, with an ~50% increase in longevity in Ames dwarf mice (1) and ~42% increase in Snell dwarf mice [141]. Along with increased longevity, these animals have a decreased incidence of cancer [142], increased insulin sensitivity [143] and increased adiposity, particularly in the subcutaneous depot [144].

Growth hormone receptor knockout (GHRKO) mice were developed to replicate Laron syndrome in mice by disrupting the gene encoding GHR/GH binding protein [145]. As with Laron syndrome patients, GHRKO mice are small, have increased adiposity, and have increased circulating GH with low circulating IGF-1 [145]. GHRKO mice also live ~38% longer than control mice [146]. Studies of cognitive function [147, 148] and tissue histopathology [149] revealed that GHRKO mice have a delay in aging, similar to that of Ames and Snell dwarf mice. To study the effects of GH action in specific tissues, many tissue-specific lines of GHRKO mice have been created [150,151,152,153]. The first line attempting to knockout the GHR gene in AT was done using Cre driven by the adipocyte protein 2 (Ap2) promoter (termed FaGHRKO) [154]. Unlike the global GHRKO mice, these animals had an increase in IGF-1 and had no improvement in insulin sensitivity [154]. It has since been reported that the Ap2 promoter is “leaky” and is expressed in macrophages, endothelial cells, and in the brain [155, 156]. Therefore, an AT-specific GHRKO mouse (termed AdGHRKO) was generated using the more specific adiponectin promoter-driven Cre [157]. These animals show no difference in GH or IGF-1, have increased fat mass, decreased liver triglyceride content, and are insulin sensitive [157].

Other mouse lines such as the GHRH knockout (GHRHKO) mice live ~45% longer than their control littermates, and have decreased IGF-1 production and increased adiposity [158]. “Little” mice have a mutation of the GHRH receptor gene, live ~25% longer than their normal littermates, and develop increased adiposity [141]. Several current reviews discuss mechanisms of altered longevity in all these mouse lines in more detail [4, 159, 160].

6 Adipose Tissue in Mice and Humans with Altered Growth Hormone Action

What is clear is that a defining phenotype of GH-related mutations is alterations in adiposity. For example, patients with acromegaly and bGH mice have decreased adiposity [161,162,163,164], while patients and mice that are GH-deficient or GH-resistant have increased adiposity [144, 165,166,167]. Particularly noteworthy is that the iWAT depot in GHRKO mice is equal in mass to that of their control littermates, despite GHRKO mice weighing approximately 66% less [166, 168]. It has also been demonstrated that dwarf and GHRKO mice maintain a higher extra- to intra-peritoneal distribution of lipids [169], which is the opposite of acromegaly patients which have a higher ectopic distribution of lipids [170]. Beyond alterations in mass and distribution, there appears to be an alteration in the endocrine function of AT in GH-mutant animals. For example, leptin and adiponectin are decreased in bGH mice, and increased in dwarf and GHRKO mice [171, 172]. In fact, altered endocrine function may partially explain an unexpected phenotype, where surgical removal of the epidydimal WAT (eWAT) in dwarf and GHRKO mice results in decreased insulin sensitivity [173, 174]. This finding may be due to the fact that eWAT in long-lived GH-mutant mice secretes less pro-inflammatory cytokines such as TNF-α and IL-6, while secreting more adiponectin [173, 174]. The secretory function of BAT in these animals has not yet been assessed. Lowering senescent cell burden can extend longevity [175]. Therefore, it is important that 18-month-old dwarf and GHRKO mice have a reduced senescent cell burden, and that their preadipocytes demonstrate an increased differentiation capacity, suggesting that their AT has a “younger” phenotype [169]. Senescence in BAT of long-lived GH-mutant mice has not yet been investigated.

Although many studies have been conducted on WAT in GH-mutant mice, far less has been done relating to the BAT of these animals. To date, we know that BAT in GHRKO and Ames dwarf mice has an increased expression of Ucp1, and that BAT in bGH mice has a decreased expression of Ucp1 [176, 177]. This is accompanied by an increase or decrease in BAT mass in Ames dwarf/GHRKO and bGH mice, respectively [176, 177]. The highly active BAT observed in Ames dwarf and GHRKO mice may at least partially explain the increased rate of energy expenditure and reduced respiratory quotient observed in these animals [178]. Particularly curious in Ames dwarf mice is the increased BAT activity despite having a depleted thyroid hormone axis. This suggests that other circulating factors may be responsible for the increased BAT activity, although this hypothesis remains to be tested. It does appear that the increased BAT activity of these animals may play a role in several biomarkers of healthy aging. For example, placing Ames dwarf mice at thermoneutrality (30 °C) eliminates differences in oxygen consumption rate and respiratory quotient, and reduces their enhanced insulin sensitivity [179]. Further testing will need to be done to determine if thermoneutral housing also impacts longevity in these animals.

7 Final Thoughts

GH has a critical role in metabolism due to its profound effects highly metabolic tissues such as the liver, muscle and AT. Increased adiposity is associated with comorbidities ranging from diabetes to Alzheimer’s disease. Therefore, it is of major consequence that the AT in long-lived GH-mutant mice functions in a more metabolically beneficial way. These differences include AT distribution (extra- instead of intraperitoneal), endocrine function (shift from pro- to anti-inflammatory cytokines), and replication and senescence status (differentiates well and has a low senescent cell burden), as well as an increase in thermogenesis and BAT activity. In terms of whole-body physiology, these alterations cannot be understated. For example, both WAT and BAT in these animals most likely significantly contribute to improved glycemic control, which is believed to be a major factor in their improved healthspan and lifespan. There are still areas that should be examined in the context of AT in GH animals. For example, we now know that AT secretes many proteins, metabolites, lipids, miRNAs, and is a rich source of exosomes. Examining how these are alteredin GH-mutant mice is a currently unexplored area. Moreover, extrapolating these findings to human subjects would be of considerable interest.