Insulin Regulation of Proteostasis and Clinical Implications

Abstract

Maintenance and modification of the cellular proteome are at the core of normal cellular physiology. Although insulin is well known for its control of glucose homeostasis, its critical role in maintaining proteome homeostasis (proteostasis) is less appreciated. Insulin signaling regulates protein synthesis and degradation as well as posttranslational modifications at the tissue level and coordinates proteostasis at the organism level. Here, we review regulation of proteostasis by insulin in postabsorptive, postprandial, and diabetic states. We present the effects of insulin on amino acid flux in skeletal muscle and splanchnic tissues, the regulation of protein quality control, and turnover of mitochondrial protein pools in humans. We also review the current evidence for the mechanistic control of proteostasis by insulin and insulin-like growth factor 1 receptors based on preclinical studies. Finally, we discuss irreversible posttranslational modifications of the proteome in diabetes and how future investigations will provide new insights into mechanisms of diabetic complications.

Introduction

Virtually all cellular functions are determined by cellular proteome abundance and quality. The proteome, made up of all proteins expressed by a cell, tissue, or organism at a single time point, is determined by protein turnover, which is the balance between protein synthesis and degradation. Free amino acids derived from diet, degradation of endogenous proteins (via autophagy and proteasome pathways), and in vivo synthesis (only nonessential amino acids) play critical roles in regulating protein synthesis and degradation. Through highly regulated processes, amino acids may be either incorporated into proteins to become part of the proteome or released back into the free amino acid pool where they may be oxidized, producing carbon dioxide and nitrogen in the process (Figure 1). Since the amino moiety of amino acids is the obligatory component of all proteins, net urinary loss of nitrogen is traditionally used as a measure of net protein degradation. The determinants of which amino acids undergo degradation versus incorporation into the proteome are not fully understood. Irreversibly damaged amino acids derived from degradation of endogenous proteins are likely degraded, while intact amino acids are preferentially acylated to tRNA for peptide synthesis (Goldberg, 2003; Ljungqvist et al., 1997). Meal-derived amino acids are incorporated into proteins, depending on the need and capacity to produce proteins, while the rest are degraded and used as energy sources like other fuels, such as glucose and fatty acids (Walrand et al., 2008). Insulin plays a key regulatory role in amino acid metabolism, and amino acids in turn affect insulin action in a bidirectional way by regulating glucose and protein metabolism (Patti et al., 1998). It is therefore imperative that any discussion of insulin’s impact on proteostasis should include the role of amino acids.

Figure 1. Simplified Whole-Body Model of Proteostasis.

Amino acids, derived from diet, degradation of endogenous proteins, and in vivo synthesis (nonessential amino acids only), are distributed through cellular compartments and the bloodstream. They travel via blood vessels to different tissues, where they may be incorporated into proteins via protein synthesis (PS) and become part of the whole-body proteome. Within the tissues, protein degradation (PD) releases amino acids back into the amino acid pool, where they may be recycled via acylation to tRNA for further PS. The amino acids that are not directed for PS are oxidized, releasing carbon dioxide (CO₂) and nitrogen in the process. The rate of protein turnover, which includes the processes of PS and PD, is tissue-dependent and variably influenced by insulin.

Protein turnover in health is a precisely regulated process that fluctuates in response to many physiologic alterations and becomes altered in different pathological states. This review focuses mainly on human studies in view of the high relevance of proteostasis in human health. We also provide substantial data from both in vitro and in vivo human and animal studies that support insulin’s pivotal role in the regulation of proteostasis via key biological pathways, including protein biogenesis (transcription and translation), folding, trafficking, posttranslational modifications (PTMs), and degradation. Here we show that insulin deficiency, as occurs in type 1diabetes mellitus (T1DM), and reduced insulin sensitivity in type 2 diabetes mellitus (T2DM) and other conditions have significant, detrimental effects on proteostasis. We propose that these alterations in proteostasis may explain many complications that reduce both healthspan and lifespan in diabetes and other conditions with reduced insulin action.

Protein Catabolic State in the Insulin-Deficient State

Insulin is the predominant hormone involved in the regulation of many vital biological functions, including fuel metabolism, mitochondrial biogenesis and physiology, and remodeling of tissues. Increased plasma glucose concentration is the most sensitive and standard biomarker of insulin deficiency, but insulin deficiency in T1DM also causes a catabolic state with profound muscle wasting (Reed, 1954), increased urinary nitrogen excretion (Lukens, 1953), and increased plasma amino acid levels (Felig, 1975). These manifestations indicate depletion of body protein content and are corrected by insulin replacement, demonstrating the key anticatabolic role that insulin plays in the maintenance of the human proteome. The catabolic state of insulin deficiency also causes abnormalities in energy metabolism (Karakelides et al., 2007), with altered mitochondrial physiology and increased reactive oxygen species production, which can lead to oxidative damage of proteins and alterations in the regulation of proteostasis (Zabielski et al., 2016). Understanding insulin’s regulation of proteostasis is essential, not only to define the role of insulin in normal physiological functions, but also to understand the underpinning mechanisms of altered physiology and both acute and chronic complications of diabetes.

Insulin’s Effects on Whole-Body Protein Turnover

Our understanding of insulin’s effects on whole-body proteostasis has evolved over time with the emergence of stable isotope tracer approaches and highly sensitive mass spectrometry techniques. Early nitrogen balance studies showed increased urine nitrogen excretion resulting from amino acid degradation in animals with experimentally induced diabetes and diabetic humans deprived of insulin (Chaikoff and Forker, 1950; Lukens, 1953; Sokhey and Allan, 1924). The increase in urinary nitrogen in individuals with T1DM was specifically due to lack of insulin, since insulin replacement reversed this process (Atchley et al., 1933). In vitro studies also demonstrated increased protein synthesis and decreased degradation in muscles exposed to insulin (Fulks et al., 1975). Together, these findings led to a conclusion that insulin independently prevented protein degradation and stimulated synthesis, but later studies revealed that this conclusion does not completely hold true in human physiology. The use of isotopically labeled amino acid tracers allowed measurement of whole-body protein synthesis and degradation in humans (Bier, 1989; Waterlow, 1984). Studies utilizing stable isotope-labeled amino acid tracer approaches elucidated that, while insulin deficiency does lead to increased protein degradation, it surprisingly also increases protein synthesis, although the increment in protein degradation exceeds that of synthesis (Luzi et al., 1990; Nair et al., 1983). By definition, in a high-protein-turnover state, flux through protein synthesis and degradation cycles is increased. This is a highly energy-consuming process since both synthesis and degradation require ATP as fuel, which therefore contributes to the catabolic state of T1DM. Insulin replacement reversed the changes observed during insulin deficiency (Nair et al., 1987a; Umpleby et al., 1986), and protein degradation measured in nondiabetic fasting men has been shown to decrease in a dose-dependent manner with insulin infusion (Fukagawa et al., 1985). Together, these studies show that insulin’s protein-conserving property in humans is largely a result of its ability to suppress protein degradation, and that protein accretion induced by insulin is in part due to balancing protein turnover by decreasing flux through synthetic and degradation pathways, with greater reduction of degradation at the whole-body level.

It is important to note that the above-mentioned human studies were performed in the postabsorptive condition when the sole source of essential amino acids was degradation of endogenous proteins. Under these experimental conditions, insulin decreases circulating essential amino acid concentrations by its inhibition of protein degradation (Barazzoni et al., 2012; Pozefsky et al., 1969), so results were likely affected by the deficiency of amino acids, which serve as stimulants for signaling cascades that increase protein synthesis. Most notably, the mammalian target of rapamycin (mTOR) pathway is activated by amino acids and serves as a master regulator of protein synthesis as well as cellular, organ, and organismal growth (Bar-Peled and Sabatini, 2014; Laplante and Sabatini, 2012). In normal human physiology, insulin secretion in response to a mixed meal is usually accompanied by an amino acid load. To elucidate the effects of amino acids in addition to insulin on protein homeostasis, subsequent studies revealed that, in both diabetic and nondiabetic persons, whole-body protein synthesis increased and protein degradation decreased significantly when amino acids and insulin were co-infused compared with insulin infusion alone (Luzi et al., 1990). Therefore, amino acids administered with insulin, as occurs in the postprandial state, enhance insulin’s inhibition of protein degradation and induce whole-body protein synthesis, thereby matching nutritional supply with hormonal changes, leading to organismal growth under nutrient-replete conditions.

Insulin Elicits Opposing Effects on Protein Turnover in Different Tissues

The whole-body studies do not indicate whether the regulation of protein turnover (average of all proteins in different tissues) by insulin occurs equally in all tissues. Like its differential effects on glucose metabolism by suppressing hepatic glucose production in liver, while enhancing glucose uptake in skeletal muscle (Rizza et al., 1981), insulin exerts differential effects on protein turnover in different tissues. Isotopically labeled amino acids in combination with blood flow measurements across tissue beds have been utilized to measure incorporation of amino acids into proteins, and appearance of amino acids from degradation of proteins, thereby providing evidence of tissue-specific protein flux in the context of whole-body homeostasis. When regional protein flux was studied in skeletal muscle in the leg and compared with splanchnic tissues (which include liver, gastrointestinal tract, and the other visceral organs) in nondiabetic individuals by using stable isotope tracers of phenylalanine, leucine, and tyrosine, protein degradation in the skeletal muscle exceeded protein synthesis after an overnight fast, leading to net amino acid release from muscle (Figure 2A) (Meek et al., 1998). At the same time, protein synthesis in the splanchnic tissues exceeded degradation, leading to net uptake of amino acids (Meek et al., 1998; Nygren and Nair, 2003). These results indicate that liver, among other splanchnic tissues, increases protein synthesis in response to decreased insulin action (Figure 2A). Insulin infusion alone did not significantly change protein synthesis in skeletal muscle, but it did inhibit protein synthesis across the splanchnic bed. Conversely, insulin strongly inhibited skeletal muscle protein degradation in a dose-dependent manner, but it did not affect protein degradation in the splanchnic region. Insulin infusion caused progressive equalization of protein synthesis and degradation in the skeletal muscle and splanchnic tissues, leading to net amino acid balance in both regions (Meek et al., 1998). These findings and later studies (Nygren and Nair, 2003) indicate that amino acids are preferentially stored within skeletal muscle during times when insulin is present (postprandial), but they are available for release during times of insulin deficiency (fasting) when synthesis of essential plasma proteins by the liver is still required. In this way, skeletal muscle can be viewed as a valuable protein and amino acid reservoir, which allows our bodies to carry out necessary functions via circulating proteins even when we are not actively ingesting amino acids.

Figure 2. Protein Flux between Gut, Splanchnic Tissue, and Skeletal Muscle during Postabsorptive, Postprandial, and Insulin-Deficient States.

(A) In the nondiabetic postabsorptive (fasting) state, protein degradation (PD) exceeds protein synthesis (PS) in the muscle, leading to efflux of amino acids into the systemic pool and uptake by the splanchnic tissue where PS exceeds PD. This allows continued synthesis of necessary proteins, such as clotting factors, in the liver even when amino acids are not actively being ingested.

(B) In the nondiabetic postprandial (mixed-meal fed) state, a high-dose amino acid load from the gut triggers increased PS and decreased PD in the splanchnic bed. Insulin is also secreted and inhibits PD in the muscle. PS increases in muscle owing to the additive effects of insulin and amino acids.

(c)In the insulin-deficient state, muscle PD is greatly increased and PS is not affected. This leads to transfer of a large amount of amino acids from muscle to the splanchnic bed, where both PS and PD are increased, but PS exceeds PD, resulting in net positive protein balance, presumably as a mechanism to allow continued synthesis of necessary proteins and perhaps to deal with the stress incurred by the absence of insulin.

Studies in T1DM further confirmed the profound impact of insulin on body protein turnover. Unlike the relatively low insulin levels in nondiabetic persons in the postabsorptive state compared with the postprandial state, complete insulin deficiency occurs in individuals with T1DM who are not treated with insulin. When regional protein flux was studied in T1DM in the postabsorptive state, splanchnic protein synthesis was substantially higher during insulin deprivation than during insulin treatment (Nair et al., 1995). In contrast, protein synthesis in the skeletal muscle of the leg was not affected by insulin deprivation, while protein degradation was markedly increased, indicating net protein depletion in the skeletal muscle bed results from accelerated protein degradation and not due to reduced protein synthesis (Figure 2C). Protein degradation in both leg and splanchnic tissues was suppressed by insulin treatment. Therefore, insulin’s whole-body protein-conserving property resulted mostly from inhibition of protein degradation in skeletal muscle (Nair et al., 1995). Together, these studies in T1DM demonstrated the pivotal role of insulin, maintaining protein balance at the whole-body level.

Amino acids are integral protein components, which must be present for insulin to act as an anabolic hormone in both the splanchnic tissues and skeletal muscle. Insulin, by its inhibition of protein degradation, reduces the essential amino acid concentrations in the postabsorptive state. Therefore, insulin infusion, with its associated hypoaminoacidemia, does not stimulate protein synthesis in skeletal muscle or splanchnic tissues during the postabsorptive state, but the addition of amino acids leads to dose-dependent increases in protein synthesis and protein accretion in both skeletal muscle and splanchnic tissues (Figures 2B and 3) (Nygren and Nair, 2003). Irrespective of whether insulin is infused intra-arterially (Chow et al., 2006; Gelfand and Barrett, 1987) or intravenously (Barazzoni et al., 2012; Nygren and Nair, 2003), it decreases muscle protein degradation, but does not enhance protein synthesis when administered without amino acids. When a multiple regression analysis was performed to determine the relative impact of insulin and amino acid concentrations on regional protein flux, leg protein synthesis was found to be regulated by both insulin and amino acid concentrations, while leg protein degradation was suppressed primarily by insulin. In contrast, splanchnic protein synthesis was stimulated and degradation was inhibited by amino acids (Nygren and Nair, 2003).

Figure 3. The Effects of Insulin and Amino Acids on Phenylalanine Balance and Protein Dynamics in Skeletal Muscle and Splanchnic Bed.

The top panels show net phenylalanine balance, and the bottom panels show changes in protein synthesis (PS) and protein degradation (PD), with use of labeled phenylalanine and tyrosine as tracers across the leg and splanchnic beds in fasting healthy individuals during infusion of normal saline (NS), insulin alone (Ins), insulin + baseline replacement of amino acids (LoAA/Ins), insulin + high-dose physiologic amino acids (HiAA/Ins), and somatostatin + baseline replacement of insulin, glucagon, and growth hormone + high-dose physiologic amino acids (SRIH/AA) or saline (SRIH/NS). The SRIH/AA treatment shows the effects of high-dose amino acids alone, since insulin, glucagon, and growth hormone are maintained at baseline levels. Phenylalanine is an essential amino acid that cannot be synthesized by humans, and it is disposed of in muscle exclusively by incorporation into proteins (protein synthesis). Therefore, phenylalanine balance represents the difference between PS and PD in muscle. In liver, it is converted to tyrosine, so an independent tyrosine tracer was used along with the phenylalanine tracer to measure splanchnic PS and PD.

(A and B) NS infusion led to a statistically significant negative phenylalanine balance in the leg, while insulin infusion balanced phenylalanine in the leg. Adding amino acids to insulin increased phenylalanine balance in a dose-dependent manner in both the muscle and splanchnic bed.

(C)PD exceeds PS in the leg during NS infusion, but PS increases and exceeds PD when both insulin and high-dose physiologic amino acids (HiAA/Ins) and when high-dose amino acids alone (SRIH/AA) are infused.

(D)In the splanchnic bed, PD and PS are nearly equal during NS and insulin infusion, but PS increases in a dose-dependent manner and exceeds PD when amino acids are added to insulin or when high-dose amino acids alone are infused.

Data are represented as mean ± SEM. *Indicates that phenylalanine balance is different between baseline and intervention (p < 0.05). ^#Indicates that the rate of PD is different than the rate of PS (p < 0.05). Adapted from Nygren and Nair (2003).

Measurement of individual amino acids has provided insight, but also elicited further questions about proteostasis during different metabolic conditions. For example, in a pioneering report, plasma branched-chain amino acids (BCAAs) were shown to be higher in insulin-resistant obese people, and were most responsive to insulin’s ability to lower amino acid concentrations (Felig et al., 1969). Plasma BCAAs have also been found to be increased in type 2 diabetic humans (Wahren et al., 1972), and they rise to a greater degree after protein ingestion in diabetics than in nondiabetic controls (Wahren et al., 1976). Insulin deficiency (Nair et al., 1995) and resistance (Halvatsiotis et al., 2002) increase leucine transamination, which affects BCAA concentrations. BCAA infusion in humans has also been reported to reduce insulin-induced glucose disposal (Nair et al., 1992; Tessari et al., 1985). Recent evidence indicates that plasma BCAA levels can predict the development of T2DM by many years (Wang et al., 2011). Whether these amino acids play a causal role in diabetes in humans or are a result of insulin resistance remains a question, but rodent models show that dietary BCAAs may contribute to obesity-related insulin resistance (Newgard et al., 2009; White et al., 2016). Overall, these studies imply that reduced insulin action increases BCAA concentrations and, in return, BCAAs further inhibit insulin action.

Effects of insulin deficiency in vivo cannot be fully delineated from those of glucagon and cortisol, both of which, but especially glucagon, increase rapidly and consistently during insulin deficiency. Glucagon has been shown to enhance not only protein degradation but also leucine oxidation during insulin deficiency (Nair et al., 1987b), and inhibition of hyperglucagonemia during insulin deficiency normalizes increased leucine oxidation in T1DM (Charlton and Nair, 1998). Hyperglucagonemia also has been shown to inhibit amino acid-induced stimulation of protein synthesis (Charlton et al., 1996). Increased plasma cortisol increases protein degradation (Simmons et al., 1984). It is possible that the effects of insulin deficiency in vivo on protein turnover may be enhanced by the effects of both glucagon and cortisol excess. The cortisol increase following transient insulin deprivation is at best marginal in humans, but glucagon levels consistently increase upon insulin deprivation in T1DM. Glucagon’s effect has been clearly delineated from that of insulin or its deficiency in studies where somatostatin was used to suppress glucagon. These studies clearly demonstrated that glucagon accelerates degradation of proteins and leucine oxidation during insulin deficiency and high glucagon reduces amino acid-induced protein synthesis in humans (Charlton et al., 1996).

In general, amino acids and insulin levels rise together in the postprandial state, but they can diverge in certain clinical situations, such as T1DM with insulin deficiency or in conditions of severe insulin resistance (Henderson et al., 2010) where insulin’s action is reduced but amino acid levels continue to be high. Skeletal muscle protein degradation that exceeds protein synthesis seems to result in net transfer of amino acids to the splanchnic bed (Figure 2C), explaining the muscle wasting observed in persons with chronic uncontrolled diabetes, or what was described as “melting down of flesh into urine” by Aretaeus in AD 50 (Reed, 1954).

While both protein degradation and synthesis in the splanchnic bed are upregulated during insulin deficiency, synthesis exceeds degradation, leading to net positive protein balance (Figure 2C), presumably to allow continued synthesis of plasma proteins, including clotting factors that are essential for survival of species. Among splanchnic tissues, intestinal mucosal protein synthesis is lower during insulin deprivation in T1DM than during insulin treatment or in nondiabetic individuals, indicating that insulin is required for maintenance of intestinal mucosal protein synthesis (Charlton et al., 2000). While liver tissue protein synthesis seems to be unaffected by increasing insulin or amino acids in nondiabetic swine (Ahlman et al., 2001), insulin administered directly to the liver via the portal vein increased liver tissue protein synthesis to a greater degree than when insulin was administered systemically (Freyse et al., 2006). The impact of insulin deficiency on liver tissue protein synthesis in humans remains to be determined. In diabetic rodents, liver protein synthesis is reduced and seems to be related to lower abundance of specific mRNA levels (Jefferson et al., 1983). Because the liver is a major site where proteins are synthesized before being released into systemic circulation, it is important to know whether synthesis rates of these liver-derived plasma proteins are increased during insulin deficiency.

Regulation of Plasma Protein Turnover by Insulin

Plasma proteins, which are largely synthesized and secreted by the liver, are affected by insulin deficiency and replacement in different ways. Early studies reported a differential effect of insulin deficiency in T1DM on albumin and fibrinogen synthesis rates (De Feo et al., 1991), with synthesis rates being decreased for albumin but increased for fibrinogen. Increased fibrinogen synthesis during insulin deficiency may also be contributed to by increased glucagon levels in human (Tessari et al., 1997). In the swine model, it has also been shown that insulin with amino acids maintains synthesis of albumin but inhibits that of fibrinogen (Ahlman et al., 2001). Mouse models of impaired insulin signaling in liver show that expression of albumin is suppressed by the transcription factor, forkhead box O (FoxO) isoform 1, which is inhibited by insulin signaling (Chen et al., 2016). A more comprehensive study evaluating 41 plasma proteins showed that fractional synthesis rates of 24 were altered in patients with T1DM who were deprived of insulin compared with nondiabetic controls (Jaleel et al., 2009). Insulin treatment normalized synthesis of only 13, and actually altered synthesis of 14 additional proteins. Many of the proteins which demonstrated different synthesis rates during insulin deprivation or treatment included apoplipoproteins, mediators of complement pathways, antioxidants, transport proteins, and those involved in coagulation and platelet aggregation, which strongly suggests that cardiovascular complications of DM may also involve pro-coagulant changes in the plasma proteome in addition to atherosclerotic changes in the vessel walls. The mechanisms behind these alterations in turnover of specific plasma proteins are still not known. Since most of these proteins are synthesized and released by the liver, there is evidence that FoxO transcription factors mediate at least some of these effects. Namely, FoxOs control albumin expression in the liver (Chen et al., 2016), modulate hepatic mitochondrial function in response to insulin signaling (Cheng et al., 2009), and are critical mediators of gluconeogenesis (Haeusler et al., 2014; O-Sullivan et al., 2015; Xiong et al., 2013). But determinations of protein synthetic rates or proteomic analyses in cellular or mouse models in which hepatic FoxO signaling is disrupted remain to be carried out.

These observations clearly indicate the variable effects of insulin deficiency on synthesis of individual plasma proteins. Moreover, despite improving glycemic control, normal synthesis is not restored for all proteins, suggesting that the peripheral insulin administration is not equal to portal delivery, consistent with results from the diabetic dog model (Freyse et al., 2006). Many of the proteins that exhibit altered synthesis in T1DM are involved in lipoprotein metabolism, inflammation, and coagulation, all of which play important roles in cardiovascular disease. It remains to be determined whether the above abnormalities will explain the persistently increased rates of cardiovascular disease in patients with T1DM (Almdal et al., 2004; Kannel and McGee, 1979), even when glycemic control is achieved (ACCORD Study Group, 2016), decreasing the risk of microvascular complications (The Diabetes Control and Complications Trial Research Group et al., 1993). Although much more work needs to be done to advance our understanding of plasma protein alterations in diabetes, these findings provide some insight and possible mechanisms underlying the increased cardiovascular disease risk associated with diabetes.

Insulin’s Effects on the Mitochondrial Proteome and Other Protein Pools

As described above, the presence and absence of insulin and amino acids induce changes in protein flux in skeletal muscle and splanchnic tissues. These changes represent the average of each individual protein’s turnover within that tissue, and not all proteins respond the same way to physiologic alterations. For example, the fractional synthesis rate of myosin heavy chain, a major structural protein that acts as part of the contractile apparatus within skeletal muscle, is not affected by acute insulin deficiency or replacement in persons with T1DM (Charlton et al., 1997). In contrast, insulin deprivation in diabetic mice increases skeletal muscle mitochondrial protein degradation and decreases synthesis, resulting in reduced amounts of protein available for mitochondrial respiration and β-oxidation (Robinson et al., 2016; Zabielski et al., 2016). As expected, mitochondrial respiration and ATP production is reduced in diabetic mice (Franko et al., 2012; Zabielski et al., 2016) deprived of insulin, along with coupling and phosphorylation efficiency, while oxidant emission is higher. Disruption of mitochondrial function or fission/fusion can itself induce muscle atrophy via FoxO3 (Romanello et al., 2010), and these changes often accompany insulin deprivation. Proteins involved in cellular uptake of fatty acids are paradoxically upregulated, leading to an accumulation of incomplete fatty acid oxidation products within skeletal muscle (Zabielski et al., 2016).

Mitochondrial protein regulation by insulin has been demonstrated in human muscle as well. Persons with T1DM have reduced muscle mitochondrial ATP production and expression of oxidative phosphorylation (OXPHOS) genes when deprived of insulin compared with when treated with insulin (Karakelides et al., 2007). In nondiabetic persons, skeletal muscle mitochondrial mRNA and protein synthesis increase in response to high-dose insulin provided with amino acids (Robinson et al., 2014; Stump et al., 2003). This is associated with increased activity of mitochondrial oxidative enzymes and ATP production in nondiabetic individuals, but type 2 diabetic persons exposed to the same conditions do not exhibit the same increased ATP production (Stump et al., 2003). Further studies have shown that amino acids are essential components which must be present for skeletal muscle mitochondrial protein synthesis to increase in response to insulin, although insulin alone enhances transcript levels of many genes encoding mitochondrial proteins (Barazzoni et al., 2012; Karakelides et al., 2007). Synthesis of myofibrillar and sarcoplasmic proteins, on the other hand, is not significantly affected by insulin, even when amino acids are co-infused (Robinson et al., 2014).

In addition to the studies in skeletal muscle, several recent studies in adipose tissue reveal the importance of insulin resistance and diabetes in the regulation of both the mitochondrial and cellular proteome in fat. A proteomic analysis of visceral adipose tissue from diet-controlled type 2 diabetic people and lean normal glucose-tolerant controls revealed upregulation of inflammatory pathways and downregulation of metabolic pathways, including mitochondrial proteins involved in β-oxidation, tricarboxylic acid (TCA) cycle, and OXPHOS, although these study participants were not matched for age (Kim et al., 2014). When comparing proteomic analysis of mitochondrial isolates from visceral adipose biopsies of obese with those of non-obese individuals, again TCA cycle and β-oxidation proteins were reduced in this tissue (Lindinger et al., 2015). Another study using age-, gender-, and BMI-matched individuals also revealed downregulation of an OXPHOS subunit in visceral adipose tissue from diabetic patients compared with controls, suggesting that both obesity and diabetes affect the mitochondrial proteome (Fang et al., 2015). Along similar lines as mentioned in muscle, fatty acid transport proteins were increased in adipose tissue of diabetic people relative to controls (Kim et al., 2014).

T2DM and obesity are often accompanied by hyperinsulinemia, and one study aimed to determine the role of hyperinsulinemia on adipocyte proteostasis (Minard et al., 2016). Differentiated 3T3-L1 adipocytes were treated with 10 nM insulin for up to 5 days and proteomic analyses, which included measures of synthesis and degradation rates, were performed. In agreement with insulin’s ability to enhance protein synthesis and decrease degradation, hyperinsulinemia increased the synthesis and stability of 662 proteins, while it decreased only one, and had minimal effect on the long-lived mitochondrial proteins. Several proteins involved in folding and quality control, including chaperones, were increased indicating that the hyperinsulinemia maintained proteostasis. It is important to note that the control cells were treated with 10% serum, which contains growth factors that may already partially or fully activate the phosphoinositide-3 kinase (PI3K)-Akt-FoxO pathway and suppress degradation pathways, although hyperinsulinemia did significantly suppress autophagy flux. Clearly, adipose tissue depends on insulin signaling to maintain its capacity for lipid storage, since loss of insulin signaling by deletion of the insulin receptor, specifically in adipocytes, results in lipodystrophic diabetes and adipocyte apoptosis (Boucher et al., 2016; Sakaguchi et al., 2017).

In conjunction with new discoveries about the effects of diabetes and insulin on mitochondrial protein expression and degradation, high-resolution mass spectrometry has also revealed novel targets for PTMs of mitochondrial proteins in response to insulin. In healthy persons, a phosphoproteomic analysis of mitochondrial isolates from muscle biopsies before and after a 4-hr hyperinsulinemic-euglycemic clamp demonstrated a 2-fold increase in the number of mitochondrial phosphoproteins (Zhao et al., 2014). Nearly half of the phosphorylation sites (94 sites) were exclusively detected after insulin stimulation. The identified phosphoproteins included 22 subunits of the electron transport chain, as well as several proteins in the TCA cycle, β-oxidation, and mitochondrial transports, among others. The functional significance of these phosphorylation events remains to be investigated, but these data along with others showing the significant impact of diabetes on degradation rates of mitochondrial proteins (Zabielski et al., 2016) reveal a global regulation of the mitochondrial proteome by insulin signaling.

More recently, advances in high-resolution mass spectrometry have led to the identification of specific low-molecularweight peptides, which are released after protein degradation (Robinson et al., 2016). Through this method, persons with T1DM have been found to have higher degradation of muscle proteins involved in mitochondrial OXPHOS (e.g., ATP synthase, cytochrome c oxidase), proteome homeostasis (ribosomal proteins), regulation of DNA structure and transcription (histones), muscle contraction (myosin isoforms), glucose metabolism (phosphoglucomutase-1 isoform), oxygen transport (myoglobin), and calcium cycling (sarcoendoplasmic reticulum calcium ATPase isoforms) during insulin deprivation compared with when insulin is replaced to maintain euglycemia. Conversely, some proteins, such as calpain-3 isoform b (proteolytic enzyme) and fibrinogen β chain isoform 1 preproprotein (involved in cell adhesion), were degraded at a faster rate during insulin replacement compared with deprivation (Robinson et al., 2016). These data indicate that although mitochondrial proteins are specific targets of insulin’s regulation of muscle proteostasis, many other protein pools are also altered by insulin signaling.

Taken together, these data illustrate the complex and variable effects of insulin on the mitochondrial proteome and other protein pools within muscle and adipose tissue. The increased degradation of many proteins during insulin deficiency is consistent with the clinical muscle wasting observed in patients with uncontrolled diabetes, but newer techniques and higher-resolution proteomics are beginning to reveal the complexity of protein pools and cellular compartments that are controlled by insulin. The specific proteins degraded during insulin deficiency are integral to numerous biological processes that coordinate energy production and consumption, including protein synthesis, ATP production, and muscle contraction, allowing for coordinated modulation of muscle function in response to nutrient and growth factor availability.

Mechanisms of Insulin’s Effects on Skeletal Muscle Protein Flux

Insulin’s impact on skeletal muscle protein synthesis in vitro has been observed as far back as the 1950s (Manchester and Young, 1958; Sinex et al., 1952), and since the discovery of insulin in 1921, researchers had hypothesized a cellular receptor that transmits insulin signaling. The discovery of the insulin receptor (IR) as a tyrosine kinase receptor began a large body of research to identify the signaling cascade by which insulin influences a multitude of actions on the cell (Kasuga et al., 1982). We now know that IR is ubiquitously expressed and highly homologous to the insulin-like growth factor 1 receptor (IGF-1R), with both receptors transmitting similar intracellular signals to regulate glucose and protein metabolism in skeletal muscle and across tissues. While circulating insulin levels rise and fall dynamically in response to glucose load, IGF-1 levels remain more consistent and are bound to IGF binding proteins which limit their bioavailability. Another stark contrast is that insulin is solely produced by b cells in the pancreas, while IGF-1 can be produced by many tissues, although 75% of circulating IGF-1 is produced by the liver in response to growth hormone. IGF-1 is classically described as a critical hormone for muscle growth, since its expression increases in response to exercise, and infusion of IGF-1 or growth hormone locally in muscle causes hypertrophy (Adams and Haddad, 1996; Adams and McCue, 1998). Indeed, in primary human myoblasts and myotubes, expression of IGF-1R is reported to be 6-fold higher than expression of IR (Palsgaard et al., 2009). However, classic studies show that IR increases in abundance during myoblast differentiation into myotubes (Beguinot et al., 1986), and recent work demonstrates that expression of IR predominates overexpression of IGF-1R in differentiated muscle by nearly 4-fold (O’Neill et al., 2016). Indeed, insulin is functionally more important than IGF-1 in maintaining muscle mass, since knockout of IR, specifically in skeletal muscle, leads to a 20% decrease in muscle size, whereas muscle-specific deletion of IGF-1R displays no change in muscle mass (O’Neill et al., 2015). Nonetheless, IR and IGF-1R play overlapping roles in the maintenance of muscle mass, since muscle-specific knockout of both receptors in MIGIRKO mice leads to a marked 60% muscle atrophy and reduced survival.

In the fed state and under normal growth conditions, insulin and IGF-1 are abundant and engage their respective receptors (Figure 4A). Upon ligand binding to the extracellular surface, IR and IGF-1R autophosphorylate and bind to insulin receptor substrates (IRS) (Taniguchi et al., 2006). The signals are further propagated through PI3K to Akt, which then phosphorylates FoxO transcription factor isoforms 1, 3, and 4, inhibiting FoxO activity and preventing translocation into the nucleus. Akt can also activate the mammalian target of rapamycin complex 1 (mTORC1) by phosphorylating and inhibiting tuberous sclerosis 1 and 2. In the presence of insulin and amino acids, mTORC1 is potently activated (Nobukuni et al., 2005), thereby promoting protein synthesis and muscle hypertrophy. By regulating mTORC1 and FoxO pathways, insulin modulates the two critical regulators of protein synthesis and degradation in muscle (Figure 4).

Figure 4. Mechanisms for the Regulation of Protein Turnover in Muscle by Insulin and Insulin-like Growth Factor 1.

(A) In the fed state, muscle growth and maintenance of muscle mass are stimulated by insulin, amino acids, and, to a lesser extent, insulin-like growth factor 1 (IGF-1). Insulin engages the insulin receptor and signals via the IRS-PI3K-Akt pathway to suppress translocation of forkhead box O (FoxO) isoforms 1, 3, and 4, and inhibit their transcriptional activity. Inhibition of FoxOs suppresses proteasomal and autophagy-lysosomal protein degradation. In addition, mammalian target of rapamycin complex 1 (mTORC1) is activated by amino acids and Akt to enhance protein synthesis, which ultimately leads to net protein gain and muscle growth.

(B) Decreased insulin signaling, as occurs with fasting or in diabetes, increases FoxO isoform translocation and transcription of critical mediators of ubiquitin-proteasome and autophagy-lysosome systems, leading to a marked increase in protein degradation that outweighs protein synthesis, leading to muscle atrophy and a high-protein- turnover state. IGF-1R, insulin-like growth factor-1 receptor; IRS, insulin receptor substrates; PI3K, phosphoinositidase-3 kinase.

Insulin’s Molecular Regulation of Muscle Protein Synthesis

Insulin, via Akt, can activate the mTORC1 complex, which is a nutrient sensor and master regulator of protein synthesis and cellular/organ growth (Bar-Peled and Sabatini, 2014; Laplante and Sabatini, 2012). These pathways are critical for normal muscle development and function, since muscle-specific deletion of mTOR or the mTORC1 regulator RAPTOR in mice results in myopathies leading to muscle atrophy and early death (Bentzinger et al., 2008; Risson et al., 2009). In addition to activation by insulin and other growth factors, amino acids, particularly leucine and arginine, are known to be potent regulators of mTORC1– S6 kinase activity (Hara et al., 1998; Sancak et al., 2008). Recent reports show that these sensing mechanisms are initiated at the lysosomal membrane by modulators of mTORC1 complex activity. Sestrin2 senses leucine and CASTOR proteins sense arginine to influence Rag GTPases, which modulate mTORC1 activity and increase protein synthesis (Chantranupong et al., 2016; Kim et al., 2008; Sancak et al., 2008; Wolfson et al., 2016). Once mTORC1 is activated, protein synthesis is increased by activation of the S6Kinase cascade and inhibition of eukaryotic initiation factor 4E-binding protein 1 (4E-BP1), which together allow for maximal translation of proteins by ribosomes (Ma and Blenis, 2009; Shah et al., 2000). Maximal activation of mTORC1 occurs with both insulin/growth factor stimulation and amino acid supplementation, indicating a duality of regulation of this critical node of growth (Avruch et al., 2006).

Human studies have shown that administration of insulin alone reduces amino acid concentrations, which does not allow for stimulation of protein synthesis, despite increases in transcript levels of many genes (Barazzoni et al., 2012). This lack of enhancement of protein synthesis by insulin alone was attributed to lack of activation of mTOR, p70 ribosomal S6 kinase (p70S6K), and 4E-BP1, although there was enhanced phosphorylation of Akt. As mentioned above, amino acids are potent regulators of mTOR, and thus adding amino acids with insulin increases muscle protein synthesis (Robinson et al., 2014), supporting the important regulation of mTOR signaling by pathways distinct from insulin. Human studies have suggested that in vivo amino acids derived from degradation of proteins are preferentially acylated to tRNA (Ljungqvist et al., 1997), thus ensuring that, even in the fasting state, proteome content is preserved. Of interest, the studies on MIGIRKO mice, in which IR and IGF-1R are deleted, specifically in muscle, demonstrated increased muscle protein synthesis and mTORC1 activation in association with substantially higher muscle protein degradation (O’Neill et al., 2016), which explains why some muscle mass is preserved despite deletion of both IR and IGF-1R. There also was evidence of increased muscle amino acid concentrations consistent with the hypothesis derived from the human studies discussed above showing that amino acids per se can stimulate protein synthesis, but the higher degradation of proteins due to reduced insulin effect results in muscle wasting.

Insulin’s Molecular Regulation of Protein Degradation

Protein degradation in skeletal muscle occurs by two cellular pathways: the ubiquitin-proteasome and autophagy-lysosome pathways, which are controlled by several intracellular signaling mechanisms (Cohen et al., 2015). Classic in vitro studies demonstrate that insulin suppresses muscle protein degradation (Fulks et al., 1975), but the investigation of ubiquitin ligases and use of specific inhibitors of proteasomes and lysosomes have shed light on how insulin and IGF-1 control these pathways of degradation. In insulin-deficient rats, muscle protein degradation was increased within 7 days, primarily via ubiquitination and proteasomal degradation (Lecker et al., 1999), and the proteasome inhibitor MG132 was sufficient to block this increase (Price et al., 1996). Proteasomal degradation was also increased in the insulin-resistant db/db mouse model, and it was partially reversed by treatment of mice with the insulin sensitizer rosiglitazone (Wang et al., 2006). These results are consistent with human studies in T1DM demonstrating increased protein degradation in insulin-deficient states (Nair et al., 1983) corrected by insulin.

Insulin and IGF-1 can suppress the transcriptional increase in E3 ubiquitin ligases, atrogin-1, and MuRF-1, which accompany increased proteolysis in muscle, indicating direct regulation of the ubiquitin-proteasome system by these growth factors (Sacheck et al., 2004). Insulin also suppresses autophagy in muscle more than in liver (Naito et al., 2013). As mentioned earlier, insulin activates mTORC1 to increase protein synthesis, and mTORC1 also suppresses autophagy at least in part by phosphorylation of autophagy-activating kinases ULK1/2 (Jung et al., 2009). There is some debate as to whether mTORC1 activation or inhibition leads to increased proteasomal degradation. Two studies show that genetic or pharmacologic inhibition of mTORC1 enhances proteolysis by proteasomal degradation (Zhao et al., 2015), possibly via activation of Mpk1/ERK5 signaling, leading to upregulation of proteasomal components (Rousseau and Bertolotti, 2016). However, other cellular models of chronic mTORC1 activation also showed increases in protein degradation (Zhang et al., 2014). The latter study indicated that proteasomal subunits were increased by Nrf1 signaling to replenish amino acid recycling. Regardless, animal models show us that both proteasomal and autophagy-lysosomal degradation are increased in muscle from mice with muscle-specific deletion of IR or deletion of both IR and IGF-1R (O’Neill et al., 2016). Insulin deprivation in STZ-induced diabetic mice also increases protein degradation through the autophagy-lysosome pathway (Robinson et al., 2016). Taken together, these studies indicate that insulin regulates muscle proteostasis via suppression of both ubiquitin-proteasome and autophagy-lysosomal degradation.

The downstream targets of insulin and IGF-1 that control muscle protein turnover are FoxO isoforms (O’Neill et al., 2016). Skeletal muscle expresses three isoforms of FoxO: FoxO1, FoxO3, and FoxO4. The suppression of all three isoforms of FoxO by IR and IGF-1R via Akt is critical to the maintenance of muscle mass in the fed state (Figure 4A). With prolonged fasting or in the context of insulin-deficient diabetes, FoxO transcription factors translocate to the nucleus and activate transcription of a number of genes, including E3 ubiquitin ligases and proteasomal subunits as well as autophagy-lysosome mediators (Figure 4B). The upregulation of the proteasome and autophagy-lysosome pathways by FoxOs enhances protein degradation, leading to muscle atrophy and a high-protein-turnover state.

The discovery that FoxO transcription factors are downstream targets of IR/IGF-1R signaling originally occurred in Caenorhabditis elegans. In 1997, daf-16 was cloned and identified as the C. elegans homolog of mammalian FoxOs (Ogg et al., 1997). Daf-16 was known to be downstream of daf-2, the IR/IGF-1R homolog, and to mediate the longevity-promoting effects when daf-2 activity was reduced. FoxOs were soon identified as direct, suppressible targets of insulin-PI3K-Akt signaling in mammalian cells (Guo et al., 1999; Nakae et al., 1999; Rena et al., 1999; Tang et al., 1999). In skeletal muscle, FoxOs were next identified as critical regulators of a transcriptional program for muscle atrophy (Sandri et al., 2004), upregulating both the ubiquitin-proteasomal and autophagy-lysosomal protein degradation pathways in response to starvation or growth factor deprivation (Mammucari et al., 2007; Zhao et al., 2007). While these studies initially implicated FoxO3 as the more important isoform, recent work has shown that deletion of all three FoxO isoforms expressed in muscle (FoxO1, FoxO3, and FoxO4) is required to prevent muscle atrophy from genetic loss of IR and IGF-1R or starvation (Milan et al., 2015; O’Neill et al., 2016). Impressively, the 60% decrease in muscle mass with muscle-specific deletion of IR and IGF-1R is completely rescued with deletion of FoxOs 1, 3, and 4, demonstrating that FoxOs, not mTORC1, are the specific targets of insulin and IGF-1 signaling that control muscle size. Thus, insulin and IGF-1 signaling pathways are critical to the regulation of muscle protein turnover, and this regulation is dependent on suppression of FoxO-regulated protein degradation (Figure 4).

The Importance of Insulin and Diabetes in Protein Quality

Insulin affects not only protein quantity, but also quality, which can have important clinical implications. Genes are expressed through a multi-step process of transcription into mRNA, followed by translation into polypeptides (Figure 5). Polypeptides subsequently undergo highly regulated folding processes in order to become functional proteins, although misfolding may also occur, yielding dysfunctional proteins. The proteome is much larger than the genome it stems from, since various proteins can be made from one gene via the complex processes of mRNA splicing, protein folding, and PTMs, which are covalent additions of functional groups. Some PTMs, such as phosphorylation, glycosylation, methylation, and ubiquitination, occur via reversible targeted enzymatic reactions and have many beneficial functions, including promoting proper protein folding and stability, aiding in cell signaling, and targeting proteins for degradation. Insulin is well known to act by phosphorylation of specific amino acids in signaling proteins, such as IRS-PI3K-Akt and MAPK signaling intermediates, to perform its biological functions (Taniguchi et al., 2006). Other PTMs, such as oxidation, carbonylation, glycation, and deamidation, occur spontaneously and irreversibly as a result of oxidative or other environmental stress. Ideally, these irreversibly damaged proteins are targeted for degradation by cellular defense mechanisms, but high levels of oxidant or other metabolic stress can overwhelm the system, leading to build up of dysfunctional proteins. Accumulation and/or aggregation of these, as well as misfolded proteins, can interfere with normal physiological processes and lead to disease (Figure 5) (McCoy and Nair, 2013).

Figure 5. Protein Biogenesis and Homeostasis.

Protein synthesis begins with transcription of genes into mRNA, which is followed by translation into polypeptides. Insulin facilitates both of these processes, and amino acids are necessary for translation. Polypeptides are folded into specific configurations in order to become functional proteins, but they may become dysfunctional if misfolded. Properly folded proteins may undergo reversible posttranslational modifications (PTMs), which aid in various protein functions, or they may undergo irreversible PTMs (oxidation, glycation, etc.) from oxidative or other environmental stress. The proteins damaged by irreversible PTMs, as well as misfolded ones, are usually targeted for proteasome-mediated degradation or autophagy, and amino acids may be recycled. If excess misfolding occurs or environmental stress overwhelms the system, damaged proteins may accumulate and/or aggregate, leading to disease. LMW, low molecular weight; ROS, reactive oxygen species.

Certainly, the widespread glycation that occurs in uncontrolled diabetes represents a classic example of how irreversible and detrimental PTMs occur in response to environmental stress (hyperglycemia due to insulin deficiency) and lead to disease (Brownlee, 2000). Glycated hemoglobin (hemoglobin A_1c) acts as a biomarker of overall diabetes control, and higher levels are associated with end-organ damage, including nephropathy, neuropathy, retinopathy, and cardiovascular disease (Nathan et al., 2005; The Diabetes Control and Complications Trial Research Group et al., 1993). Glycation is only one PTM affecting protein quality during insulin deficiency, however, and is unlikely to fully explain the myriad complications associated with diabetes.

Diabetic patients have been found to have significantly greater oxidative protein damage compared with nondiabetic individuals (Figure 5) (Martin-Gallan et al., 2003). Importantly, ApoA-1, a major component of high-density lipoprotein (HDL) integral to the transfer of cholesterol from peripheral macrophages to the liver for subsequent excretion or recycling, has been shown to undergo accelerated oxidative PTM in individuals with T1DM who are deprived of insulin (Jaleel et al., 2010). Furthermore, HDL in type 1 diabetic patients with either good or poor glycemic control has significantly decreased cholesterol efflux capacity and antioxidative function compared with HDL in nondiabetic controls (Manjunatha et al., 2016). This occurred in association with more irreversible PTMs affecting functionally important HDL proteins, including ApoA-4, ApoD, and ApoE, in diabetic patients. Given HDL’s important role in maintaining vascular health, this study implicates altered HDL protein quality as a potential mechanism for the higher rates of cardiovascular disease in patients with T1DM, even when their quantity of HDL is normal (Manjunatha et al., 2016).

The study of protein quality during normal and altered physiological states has only begun in recent years, and we have much more to learn. On the basis of the recent findings of damaging PTMs associated with impaired function of HDL-associated proteins in T1DM, we suspect that investigation in this field will continue to grow and hope that it will elucidate more mechanisms responsible for the widespread complications associated with diabetes.

Conclusion

The past several years have generated remarkable advances in our knowledge of the dynamic regulation of cellular physiology by insulin through regulation of protein function, synthesis, and degradation. From classic studies in isolated skeletal muscle showing that insulin can change protein turnover to recent advances in identifying specific protein pools that are degraded in the diabetic state, our understanding of insulin’s effect on proteostasis has clearly progressed. In muscle, we have identified the specific signaling pathways downstream of insulin that control muscle protein degradation and help maintain muscle mass. The coordination of amino acid flux between tissues by insulin helps orchestrate proteostasis at the organismal level to maintain normal growth. Lastly, new discoveries in the effect of diabetes on PTMs of the proteome have enhanced our understanding of this complex disease.

Despite this progress, many questions remain unanswered. Whether FoxO isoforms also mediate insulin’s regulation of proteostasis in tissues other than muscle is not fully known. The specific downstream targets of the insulin-FoxO axis that regulate degradation of specific protein pools, such as mitochondrial or other cellular proteins, are not completely understood. Furthermore, insulin’s effect on proteostasis has been best studied in insulin-sensitive tissues, including skeletal muscle and, to a lesser extent, liver and adipose tissue, but whether these mechanisms hold true in other tissues, especially brain and heart, requires more investigation. Diabetic cardiomyopathy is an entity prompting more studies directly on heart tissue than on vasculature, but human tissue samples are not easy to obtain. Moreover, the link between diabetes and vascular disease is a prime reason for the development of complications and mortality. How insulin affects the proteome in the vasculature is less well explored. In addition, the link between diabetes and cognitive decline, including Alzheimer’s disease, is of great interest, and future studies will need to expand on the knowledge of insulin resistance/diabetes in the regulation of protein aggregates and other aspects of proteostasis in the CNS. As we near the 100-year anniversary of the discovery of insulin, we realize that understanding the biology of this pivotal hormone is more important than ever, with diabetes rates reaching pandemic levels. Fortunately, we live in an exciting time when the discovery of insulin’s regulation of proteostasis is more achievable than ever and will likely lead to novel areas of intervention to slow the development of complications and morbidity from diabetes.