Yale University
EliScholar – A Digital Platform for Scholarly Publishing at Yale
Yale Medicine Thesis Digital Library
School of Medicine
January 2019
Insulin And Non-Insulin Dependent Glut4
Trafficking: Regulation By The Tug Protein
Stephen Devries
Follow this and additional works at: https://elischolar.library.yale.edu/ymtdl
This Open Access Thesis is brought to you for free and open access by the School of Medicine at EliScholar – A Digital Platform for Scholarly
Publishing at Yale. It has been accepted for inclusion in Yale Medicine Thesis Digital Library by an authorized administrator of EliScholar – A Digital
Platform for Scholarly Publishing at Yale. For more information, please contact elischolar@yale.edu.
Recommended Citation
Devries, Stephen, “Insulin And Non-Insulin Dependent Glut4 Trafficking: Regulation By The Tug Protein” (2019). Yale Medicine
Thesis Digital Library. 3489.
https://elischolar.library.yale.edu/ymtdl/3489
Insulin and non-insulin dependent GLUT4 trafficking:
regulation by the TUG protein
A Thesis Submitted to
the Yale University School of Medicine
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Medicine
by
Stephen Graham DeVries
2019
Abstract
The body tightly regulates glucose production and disposal despite changing
metabolic demands, including large post-prandial and fasting fluctuations. Specifically,
under the action of insulin, muscle contraction, ischemia, and poor nutrient availability,
cells increase the amount of the glucose transporter type 4 (GLUT4) at the plasma
membrane by mobilizing a sequestered pool of transporters. In this work, we demonstrate
that the TUG (tether containing a UBX domain for GLUT4) protein mediates both
insulin-dependent and insulin-independent pathways to increase GLUT4 at the plasma
membrane. In mice fed a high fat diet to induce insulin resistance, the regulation of the
endoproteolytic cleavage of the TUG protein was disrupted. We also present evidence
that helps to identify the key protease, Usp25m, that cleaves the tethering protein TUG in
both an insulin-dependent and insulin-independent manner, releasing GLUT4 from its
storage location in the basal state to the plasma membrane in an activated state. Finally,
our results also suggest that in the adipocytes and myocytes, activated AMPK leads to
cleavage of the TUG protein.
Acknowledgements
I would like to thank Dr. Estifanos Habtemichael, who patiently taught me the
techniques necessary to work in a cell biology laboratory at the beginning of medical
school. His constant guidance and feedback were invaluable to the work that led to this
thesis. I would also like to thank Don Li, who helped me adapt and optimize my planned
projects. His mentoring both in research and in medical training has been a central part of
my training as a physician and as a scientist. Finally, I would like to thank my advisor,
Dr. Jonathan Bogan, whose constant enthusiasm for science, optimism, and support made
working in his lab the highlight of my time in medical school.
Table of contents
1. Introduction
1
1.1. Insulin resistance
1
1.2. Macronutrient contributions to obesity
5
1.3. GLUT4 transporters and their regulation by membrane trafficking
7
1.4. Thyroid hormone agonists
15
1.5. Exercise-induced glucose uptake in muscle and the role of AMPK
16
1.6. Mouse model of type 2 diabetes mellitus
21
2. Statement of purpose, specific hypothesis, and specific aims
22
3. Methods
23
3.1. Reagents and cell culture
23
3.2. Mice
25
4. Results
26
4.1. Usp25m interacts with TUG
26
4.2. TUG cleavage differences in HFD and RC fed mice
27
4.3. Activated AMPK
27
5. Discussion
35
5.1. Usp25m interacts with TUG
35
5.2. HFD inhibits TUG cleavage
36
5.3. AMPK
36
References
39
1
1. Introduction
1.1. Insulin resistance
The human body, in terms of its nutritional requirements, is optimized for a time
when food was scarce and unpredictable (1). However, over the last half-century in the
developed world and now increasingly in developing countries, dense caloric foods have
become easily available. Obesity, as well as comorbid conditions like non-alcoholic fatty
liver disease (NAFLD) and atherosclerosis, are on the rise. Type 2 diabetes in particular
is implicated in significant and growing morbidity and mortality in the United States and
worldwide (2). The burdens of this disease include kidney failure, retinopathy, and
neuropathy (3). While type 2 diabetes has been a major health problem in developed
countries for decades, the populations with the largest rates of increase in the disease are
in Asia and the Indian subcontinent. In these countries, the number of people living with
type 2 diabetes is projected to increase by over 75% by the year 2034.
The rising number of cases of type 2 diabetes has been apparent in the United
States for decades. The prevalence of diabetes in the United States increased from 0.9%
in 1958 to 4.4% in 2000, and 90%-95% of these diagnoses were type 2 diabetes. Notably,
the increase in cases over this forty year period was not uniform across age groups: in the
18-29 year-old group, the increase in diabetes diagnoses went up by 40%, while in
persons aged 30-39, the increase was 95%. For older patients aged 40-49, the number of
diagnoses rose by 83%, and in patients aged 50-59, 49%. The rates were slowest to rise in
the oldest age groups; in patients 60-69, diagnoses rose by 40%, while in patients over
70, 33% (4).
2
Over the last few decades, both epidemiological and laboratory studies have shed
light on the pathophysiology of type 2 diabetes. At a basic level, insulin resistance leads
to the disease, and beta-cell dysfunction follows (5).
However, it remains unclear if hyperinsulinemia is the primary cause of insulin
resistance, or if it is entirely secondary (6). The former model is supported by evidence
that fasting hyperinsulinemia may develop before an increase in postprandial blood
glucose, causing the release of insulin from beta-cells (7). The latter model, in which
insulin resistance precedes hyperinsulinemia, is perhaps more widely accepted. Patients
who are insulin resistant because of known mutations in their insulin-signaling pathways
provide some of the strongest evidence for this model. In these patients, the primary
lesion is resistance; observable effects, including hyperinsulinemia, are therefore
concluded to be secondary phenomena (8). In this model, with the effects of insulin
blunted, the beta cells of the pancreas compensate by increasing the production of insulin.
In 1963, Randle and others published a paper connecting obesity to insulin
resistance, arguing that glucose oxidation was impaired by the presence of high levels of
fatty acid (9). Today, overwhelming epidemiological data, along with mouse and human
data, illustrate that caloric imbalance is a key causative factor in the development of
insulin resistance, and thus a risk factor for the development of type 2 diabetes (7).
However the specific mechanistic connections between obesity and insulin resistance
remain somewhat unclear.
Notably, one of the many sequela of the obesity epidemic has been a drastic rise
in the prevalence of NAFLD. Today it is a major cause of liver disorder in the Western
world (10). Fat deposits in the liver are strongly associated with type 2 diabetes – more
3
than 90% of obese patients with type 2 diabetes have NAFLD – and insulin resistance is a
common feature of both NAFLD and obesity (10). Two recent studies in particular (12,
13) have advanced the understanding of the mechanisms that link deposits of fat in the
liver and resistance to insulin. Plasma-free fatty acids were observed at higher levels in
patients who were obese and had type 2 diabetes, and further work established an inverse
relationship between fasting plasma-free fatty acid concentrations and insulin sensitivity,
more proximally linking plasma-free fatty acids with insulin resistance (11, 12).
Subsequent studies, using 1H NMR (proton nuclear magnetic resonance) and muscle
biopsy, have more directly demonstrated the strong link between high concentrations of
intramyocellular triglycerides and insulin resistance (12). This can lead to NAFLD by
causing a shift in the distribution of energy substrates, so that they accumulate in the liver
and are stored as fat.
While these results might seem consistent with the Randle hypothesis –
implicating fatty acids as causal to insulin resistance – more recent work suggested an
important mechanistic difference. Under the Randle hypothesis, which was developed
using cardiac and diaphragm muscle, increased citrate concentrations affect
phosphofructokinase. However, recent data support the idea that fatty acids interfere with
an early step in the signaling cascade that ultimately leads GLUT4 to translocate to the
cell surface (9, 12). Diacylglycerides were later identified as a causal factor in hepatic
insulin resistance. In hepatocytes, a novel protein kinase C isoform, PKCε, was found to
be translocated to the plasma membrane, inhibiting the activity of the intracellular kinase
domain of the insulin receptor (13, 14). In myocytes, a different novel protein kinase C,
PKCθ, has been identified (15). The mechanism for resistance to insulin signaling in
4
hepatocytes caused by diacylglycerides was further explained in 2016. The
phosphorylation site (Thr1160) was identified as a substrate of PKCε in the kinase
activation loop of the insulin receptor. In vitro studies showed that a mutation from
threonine to glutamic acid (T1160E), which can mimic phosphorylation, caused impaired
insulin receptor signaling, but a mutation to alanine (T1160A), which resists
phosphorylation, did not show inhibition. Furthermore, in mice, mutation from threonine
to alanine on Thr1150 (the homologous residue in mice) conferred protection in the
insulin signaling pathway in mice that were fed a high-fat diet to induce hepatic insulin
resistance (16). Further experiments showed crosstalk between PKCε and the kinase
p70S6k (17). This work clearly demonstrated that insulin signaling in the liver is
deranged due to accumulation of lipids.
However, changes in insulin signaling do not fully explain the phenomenon of
insulin resistance. The downstream consequence of insulin signaling is GLUT4
translocation to the plasma membrane, and changes to the abundance and distribution of
GLUT4 have been demonstrated in muscle and adipose tissue. Garvey and others studied
muscle and adipose tissues of humans during fasting, and compared patients with type 2
diabetes to healthy controls. Notably, compared to the controls, GLUT4 targeting was
altered in the type 2 diabetic group even in the fasting state, when insulin signaling is
minimal. In adipose, GLUT4 was depleted in all membrane sub-fractions, and the co-
trafficking protein IRAP was similarly altered in its basal distribution among intracellular
membranes in the fasting state (18). These results suggest that alterations in membrane
trafficking may contribute to insulin resistance, independent of alterations in insulin
signaling pathways.
5
1.2. Macronutrient contributions to obesity
The previous section addressed the mechanistic connections between obesity and
insulin resistance, but the contributing factors to obesity are also important to consider. In
2017, the Endocrine Society published a scientific review, making the case that caloric
imbalance has been a dominant factor in the rise of obesity. The review proposed that
irrespective of the macronutrient balance, the calorie amount is the determining factor in
weight gain; in simple terms, “a calorie is a calorie” (19).
A prominent opposing view, often called the carbohydrate-insulin model, holds
that the macronutrient content of ingested calories is important to determine weight and
obesity. Changes in dietary quality in the last 50 years may have caused hormonal
responses that shift calories towards deposition of fat (20). Under the carbohydrate-
insulin model, if calories are stored, then the energy content of blood is reduced, which
causes hunger and subsequent overeating; in other words, a high-carbohydrate diet causes
postprandial hyperinsulinemia and this promotes deposition of fat (20). As Ludwig and
Ebbeling note, the carbohydrate-insulin model does not violate the First Law of
Thermodynamics (conservation of energy). This model sees overeating as a consequence
of increased fat stores, and not the primary cause (20). The details of the supporting
evidence for the carbohydrate-insulin model, including considerable human, animal, and
cell-culture research, are beyond the scope of this thesis (20), but the most recent
supporting study was a randomized human trial, published in late 2018 (21).
Advocates for the conventional model (“a calorie is a calorie”) argue that there
are key flaws in the carbohydrate-insulin model. Specifically, in the carbohydrate-insulin
model, because fuels are being stored under the action of insulin, the level of circulating
6
fuels in the blood is reduced, leading to hunger and thus increased food intake. But as
Hall and others argue, obese individuals have normal or even elevated levels of
circulating fuels, including free fatty acids and glucose, and the adipose of obese patients
actually releases more total free fatty acids and glycerol than individuals of normal
weight. Furthermore, diets with varying degrees of glycemic index and load do not lead
to a significant difference in hunger, either acutely or longer term (22, 23). Finally, Hall
and others note that patients on strict diets with lower glycemic indices had lower levels
of insulin compared to patients on equal calorie diets with higher glycemic loads, but the
patients on diets with lower glycemic loads did not show increased mobilization of free
fatty acids and oxidation beyond the amount that could be explained by the increased fat
intake in their diets. There were not significant differences in weight loss between
patients with lower and higher glycemic load diets, and furthermore, decreased insulin
levels had no predictive value on weight loss (23, 24).
The macronutrient content required to halt the progression of obesity or lose
weight is not only a fundamental question in diabetes research, but a key problem to
solve in order to lessen the burden of the associated diseases. While no one study is likely
to settle the debate, advocates of the carbohydrate-insulin model recently published the
results of a randomized trial in which participants who had lost about 12% of their pre-
weight loss body mass on a conventional diet were assigned high and low carbohydrate
diets for weight maintenance. The primary outcome was energy expenditure, which was
significantly increased in patients assigned the lower carbohydrate diets. The results were
most dramatic in patients with the highest pre-weight loss levels of insulin. The patients
in the highest third of pre-weight loss insulin levels had an energy expenditure increase of
7
308 kcals on average on the low carbohydrate diet, compared to patients on the higher
carbohydrate diets (25).
1.3. GLUT4 transporters and their regulation by membrane trafficking
The phenomenon of increased movement of glucose down its concentration
gradient from the blood to the intracellular space caused by the action of insulin was well
documented in pioneering work from the 1970s and 1980s. A major breakthrough was
made in 1980, when two groups independently showed that in insulin responsive cells,
glucose mobilized a glucose transporting activity from an internal cellular storage
location to the plasma membrane (26, 27). Then, in 1981, it was demonstrated that in
incubated rat adipocytes, insulin increased glucose transport into the cells, a glucose
transporter moved from an intracellular pool to the plasma membrane, and high
concentrations of anti-insulin antibodies largely ablated these effects (28). The first
biochemical characterization of the factor that facilitates glucose transport was published
in 1985 (29). Now termed GLUT1, it is widely expressed, and its location in the cell is
not significantly changed by insulin. Further studies characterized a glucose transporter
that likely increased in concentration at the plasma membrane, allowing for the diffusion
of glucose. Under the action of insulin, transporters would translocate from organelles
inside the cell and embed themselves in the plasma membrane. Once embedded in the
plasma membrane, glucose would diffuse down its concentration gradient (i.e., ATP-
independent transport). Further characterization of the major insulin-sensitive glucose
transporter, GLUT4, was described later (30). The GLUT isoform that is responsive to
insulin was cloned by five laboratories (31-35). and the gene that encoded the transporter
8
was termed SLC2A4 (32). The location of the intracellular storage location of GLUT4 is
currently an active area of research. Interestingly, in the basal state, GLUT4 is not
localized to one compartment or organelle, but two, divided about equally in cultured
cells. Thus, the late 1980s marked a key transition in diabetes research. The physiological
problem of how insulin increases glucose uptake turned into a cell-biology problem of
how insulin signals, and how it causes translocation of the insulin sensitive GLUT
isoform, GLUT4, to the plasma membrane (36).
After the role of GLUT4 as a key effector of insulin mediated glucose uptake was
recognized, an important challenge remained: describing the signal cascade and
mechanism of action that, in the end, leads to GLUT4 being embedded in the plasma
membrane (37). It was demonstrated that insulin binds to its receptor, causing
dimerization or reorientation of the dimer, and thus activation (38). Once activated, the
tyrosine kinase phosphorylates insulin receptor substrates (IRS proteins), in addition to
activating phosphatidylinositol-3-kinase (PI3K), and other effector proteins (37, 39).
Downstream, Akt2 (there are 3 isoforms) phosphorylates numerous targets,
including AS160 (Akt substrate of 160 kDa) GTPase-activation protein (36, 40). The
discovery of AS160 was especially significant because RAB proteins direct vesicle
trafficking, so AS160 links signaling and trafficking pathways (36). AS160 is present
on GSVs (GLUT4 storage vesicles, discussed in more detail below), and interacts with
LRP1 and IRAP (41, 42).
Some of the mechanisms of GLUT4 retention, recycling, and translocation can be
gleaned from kinetic data and total internal reflection fluorescence (TIRF) microscopy.
The TIRF data was acquired and analyzed long after other methods were used to discover
9
significant portions of the signaling cascade and retention mechanisms (43). In the basal
state, the vast majority of GLUT4 is sequestered intracellularly. One fraction of GLUT4
colocalizes with markers of the recycling endosome (such as the transferrin receptor),
while another fraction colocalizes with the Golgi network or endoplasmic reticulum (44).
The dynamics of the GLUT4 transporter have been the subject of much study and
controversy. In unstimulated adipocytes and myocytes, the balance of movement to the
plasma membrane and to intracellular compartments favors intracellular compartments
(45). When insulin (or in the case of myocytes, a contraction or other activating signal) is
present, then the balance favors GLUT4 to be inserted into the plasma membrane. This
could occur because of a decreased rate of endocytosis or an increased rate of exocytosis.
Although the former may occur to a small extent, most data support a model in which the
main effect of insulin is to increase GLUT4 at the plasma membrane by increasing the
exocytotic arm of the GLUT4 recycling pathway (46).
While it is clear that insulin increases the amount of GLUT4 in the plasma
membrane, the mechanism by which GLUT4 is retained and cycled to the plasma
membrane remains unclear. One model, advocated by McGraw and others, proposes that,
even in the basal state, all GLUT4 in the cell will cycle to the plasma membrane. Another
model, advocated by James and others, proposes that in the basal state, some fraction of
GLUT4 is sequestered intracellularly, and will not cycle to the plasma membrane unless
stimulated by insulin (45). The evidence for each of these models is seemingly quite
strong. One notable study showed that all GLUT4 in the cell eventually cycles to the
plasma membrane, even in the basal state (47). Evidence that may reconcile these two
seemingly contradictory models was published in 2008 (48). In these experiments, re-
10
plating adipocytes after differentiation caused a much higher proportion of GLUT4 to
cycle to the plasma membrane in the basal state. To a large degree, re-plating the cells
appeared to disrupt the static retention of GLUT4 in the basal state and cause more
GLUT4 to cycle to the plasma membrane, even in the absence of insulin. These
observations suggested that the cycling seen in re-plated cells might be an artifact of the
disruption.
To allow for the regulation of glucose uptake into fat and muscle cells, GLUT4
glucose transporters are sequestered intracellularly during the basal (low insulin) state.
TUG proteins, encoded by the ASPSCR1 gene, play an important role in this process;
they serve as anchors that retain GLUT4 in a pool of intracellular “GLUT4 storage
vesicles” (GSVs) (49). In addition to GLUT4, GSVs contain insulin-regulated
aminopeptidase (IRAP), sortilin, vesicle-associated membrane protein (VAMP2), and
low density lipoprotein receptor-related protein 1 (LRP1) (50). The physiologic rationale
for having this set of cargos grouped together, all mobilized downstream of insulin
signaling, is not clear. For example, the role of IRAP, another cargo and a transmembrane
aminopeptidase, involves cleaving several peptide hormones. In a mouse model,
vasopressin was shown be a substrate for IRAP, and IRAP-deficient mice had
significantly increased levels of vasopressin (51).
TUG was first recognized as a downstream effector of insulin from a functional
screen that studied GLUT4 localization (52). Subsequent work aimed to understand the
mechanism of its action; while the functional screen identified the TUG protein as having
some role in GLUT4 localization, its relative importance in GLUT4 mobilization to the
plasma membrane and thus glucose disposal was not known.
11
A major advance in understanding the importance of TUG came in 2007, with the
publication of a study that showed that small interfering RNAs (siRNAs) targeted against
the TUG protein mimicked much of the effect of insulin. In cells expressing the siRNAs,
GLUT4 translocated from intracellular storage to the plasma membrane, even in the basal
(unstimulated) state. This effect could be slightly increased by stimulation with insulin
(49). The role of TUG was further illuminated by the discovery that TUG directly binds
to an intracellular loop of GLUT4 (49). It appeared that TUG, under the action of insulin,
released stored GLUT4 from the Golgi to the plasma membrane. But, how, specifically,
did insulin mediate this effect through TUG? Clarification on this point was made in
2012, when it was discovered that under the action of insulin, the intact TUG protein was
cleaved at a specific location, between residues 164 and 165 of the intact 550 residue
mouse TUG protein. The result of this cleavage is to divide the 60 kDa intact TUG
protein into a 42 kDa C-terminal product and an 18 kDa N-terminal product (53). The
current understanding of the role of TUG is that insulin stimulates the endoproteolytic
cleavage of TUG to mobilize the GSVs for translocation to the cell surface, so that
GLUT4 is inserted into the plasma membrane. Once in the plasma membrane, GLUT4
transports glucose into the cell via facilitated diffusion (53).
To gain further insight into the mechanisms of insulin action, the fusion of
individual GSVs in 3T3-L1 adipocytes was monitored in live cells. These experiments
required the development of a new approach, because biochemical assays can neither
determine the location of GLUT4 at individual trafficking steps nor distinguish between
greater abundance of GLUT4 in a compartment due to increased exocytosis or decreased
endocytosis (43). The new approach for tracking individual vesicles containing GLUT4
12
took advantage of TIRF microscopy and a new GSV reporter, VAMP2-pHluorin. The
key motivating questions for this work were two-fold: (1) when GLUT4 is inserted into
the plasma membrane, did the GLUT4 molecules come from a GSV or an endosome?
And (2), what processes could change the relative amounts of GLUT4 from each of these
two source compartments at the plasma membrane?
Although GLUT4 trafficking dynamics remain an area of active research, the
results of this work strongly suggested that GLUT4 resides in two major compartments,
GSVs and endosomes, in cultured 3T3-L1 adipocytes (54, 55). Although distinguishing
between these two compartments biochemically based on the different cargo proteins is
straightforward, tracking individual fusion events in a live cell is challenging. GSVs and
endosomes contain different proteins and also differ greatly in size. GSVs are about 50-
70 nm in diameter, while endosomes are much larger, around 100-250 nm in diameter
(56). TIRF microscopy enabled the size of individual fusion events to be measured and
the moment of docking and fusion to be determined. The experiments suggested that
there were two distinct sizes of GLUT4-containing vesicles at fusion: in pre-adipocytes,
the size of the vesicles was 153 +/- 42 nm (consistent with the size of endosomes), but in
3T3-L1 adipocytes, the size of the vesicles was just 56 nm +/- 27 nm (consistent with the
size of GSVs, which are present in adipocytes but not pre-adipocytes) (43, 56). Finally, in
cells with a TUG knockdown, the exocytotic rate of GSVs was similar to the exocytotic
rate of control 3T3-L1 adipocytes during insulin stimulation. This last observation
supports a model in which TUG retains GLUT4 in the basal state and is cleaved upon
insulin stimulation. A model of TUGs action is summarized in Fig. 1 (36).
13
While these TIRF studies indicate that TUG is important for trafficking in insulin
responsive cells, the role of TUG in protein trafficking may extend to other cell types as
well. For example, in HeLa cells, TUG was found localized in the endoplasmic
reticulum-to-Golgi intermediate compartment (ERGIC) and the endoplasmic reticulum
exit sites (ERES). The vast majority of TUG was bound to p97/VCP, which is a
hexamaric ATPase that is important in membrane fusion and proteolysis. In these HeLa
cells, TUG caused disassembly of hexamers into monomers, controlling their functional
status. Furthermore, a knockdown of TUG disrupted the Golgi. These results suggest that
TUG may have an important role in the early secretory pathway in multiple cell types,
not only in adipocytes and myocytes (57).
Most recently, the mechanism of GSV movement from the Golgi to the plasma
membrane has been described in detail. This discovery was underpinned by the
observation that an antibody directed against the 18 kDa N-terminal product of TUG
cleavage also detected a protein of approximately 130 kDa. This product was not seen
after knockdown of TUG or without insulin stimulation, suggesting that this is a specific
band (49). Analysis of the 18 kDa N-terminal cleavage product revealed that it contained
two ubiquitin-like domains, and ended in a diglycine motif. These characteristics, along
with evidence of the 18 kDa covalently attaching to another protein to form the 130 kDa
product, led investigators to hypothesize that the N-terminal product was acting in a
similar manner to a ubiquitin-like modifier. As a result, it was named TUGUL (TUG
Ubiquitin-Like).
Ubiquitin is best known for its role in covalently attaching to numerous proteins
and targeting them for destruction in the proteasome. Ubiquitination is similar to
14
phosphorylation and acetylation in that they are post-translational modifications that
change protein function. In the context of a cell’s acute response to increased insulin
concentration and translocation of GLUT4, the short timescale of a post-translational
modification is important, as changes in gene transcription and translation are too slow to
either dispose of glucose in a post-prandial setting, or to allow a cell to increase glucose
transport when it has an acute need for more energy.
Ubiquitin is one of a family of similar modifiers termed ubiquitin-liked modifiers
(UBLs). While the short time course of action is similar to small molecule modifiers,
there are a number of differences (58). Ubiquitin is a 76-residue protein, highly
conserved among eukaryotes but absent from bacteria and archaea (59). Interestingly,
while UBLs do not always share high sequence similarly, they have a similar three-
dimensional structure (59, 60). The structure of UBLs, because of their size and diversity,
exhibit functions that differ from small molecules. For example, UBLs can alter protein
conformation or protein-protein interactions (58). In general, UBLs are synthesized as
non-functional precursors. Next, the protein is modified so that there is a Gly-Gly motif
at the C-terminus, the site of attachment to the target. This step is carried out by a group
of enzymes termed de-ubiquitinating enzymes (DUBs). An ubiquitin–activation enzyme,
termed E1, then adenylates the modified C-terminus. The ubiquitin molecule is passed to
a cysteinyl group on the second ubiquitin-conjugation enzyme, E2. The final step is
accomplished by a ubiquitin-protein ligase, called E3 (58). There numerous types of E2
and E3 enzymes, not just one form. As the list of UBLs grows, it has been challenging to
identify which features are unique, and which are common to all UBLs. E3 appears to not
always be required; among other examples, small ubiquitin-related modifier (SUMO)
15
appears to not always require E3 activity (58). Of possible relevance to GLUT4
trafficking is the observation that membrane-protein-sorting factors are mono-
ubiquinated. Ubiquitin thus appears to serve as an important modification to direct many
types of cell traffic to different compartments, depending on cytosolic conditions (61).
The identification of TUGUL as a novel ubiquitin-like protein modifier implied that a
DUB, as well as E1, E2, and E3-like enzymes, may be involved in insulin action, and that
these proteins may mediate TUGUL generation and covalent modification of a target
protein to promote glucose uptake (62).
1.4. Thyroid hormone agonists
Using thyroid hormone mimetics to agonize receptors has the potential to increase
lipid metabolism, decrease low-density lipoprotein (LDL), increase energy expenditure,
and increase thermogenesis (63). Thyroid hormones exert their effects by action at one of
three nuclear receptors: TR, TR1, and TR2, that are distributed in a tissue specific
fashion (63). Naturally occurring thyroid hormone acts equally on all three of these
receptors. An agonist that has desirable metabolic effects could also agonize thyroid
receptors in all tissues, causing ill effects such as tachyarrhythmias, osteoporosis,
agitation, or psychosis (63). One strategy to exploit the positive metabolic benefits of
thyroid hormone actions while negating the harmful effects is to develop synthetic
analogs that agonize only one receptor. The TR1 predominates in the liver, so this is an
attractive target (63). A recent study that tested two thyroid hormone TR1 agonists
showed that while fat accumulation in liver was markedly decreased in the presence of
TR1, sensitivity to insulin also decreased. One of these TR1 agonists, KB-2115,
16
caused a notable decrease in GLUT4 and TUG, yet sortilin, a marker of GSVs, was not
reduced (63). One explanation for the decreased sensitivity to insulin, as well as
decreased TUG and GLUT4 in the presence of KB-2115, could be that even though
GSVs are formed, they are not sequestered in an insulin-responsive pool as is normal
(63).
1.5. Exercise-induced glucose uptake in muscle and the role of AMPK
In addition to insulin, exercise also increases GLUT4 in the plasma membrane.
During exercise, there is a large increase in the demand for ATP. Early in the exercise
period, stored glycogen is the main source of energy. But, as the length of the exercise
period continues and glycogen stores become exhausted, blood glucose accounts for
around 35% of oxidative metabolism and nearly all of the muscle carbohydrate
metabolism (64). Regular exercise improves glycemic control in patients with type 2
diabetes (65). In some ways, this is intuitive: acute increased demand for glucose in
muscle cells reduces the concentration of glucose in the serum. But type 2 diabetes is
marked by insensitivity to insulin, which among its many functions, is crucial to disposal
of serum glucose in muscle in the post-prandial period. Thus it is actually an important
finding that even in patients who are resistant to insulin’s effect on glucose disposal,
exercise-induced glucose disposal mechanisms are still intact. The results of Martin and
others – suggesting that the main cause of reduced blood glucose after exercise in patients
with type 2 diabetes is increased muscle uptake of glucose instead of reduced glucose
output by the liver – further confirmed a key difference between insulin and exercise-
induced glucose uptake in muscle (64, 66). Together, these results could lead to new
17
therapeutic strategies, as at least some mechanisms for glucose uptake are still intact in
patients with type 2 diabetes.
Ruling out decreased hepatic glucose production as the cause of decreased serum
glucose after exercise does not identify the rate-limiting step at the level of the muscle
tissue. In vivo, the rate-limiting step could be glucose delivery to the skeletal muscle,
glucose transport across the plasma membrane, or flux through intracellular metabolism
(64). At first glance, the most likely candidate might seem to be increased glucose
delivery, as blood flow can increase up to 20-fold (at least partly due to IRAP, a GSV
cargo that degrades vasopressin) during exercise (64, 67). In fact, all three of these
elements probably play some role in glucose disposal, but compelling evidence points
towards transport across the plasma membrane as the key regulatory step (68). Two
aspects of this research have important implications for GLUT4 trafficking to the plasma
membrane: (1) contraction-stimulated uptake is normal in subjects resistant to insulin,
and (2) glucose uptake (and hence GLUT4 trafficking to the plasma membrane) is a
crucial regulatory step in exercise. Two intriguing questions, then, are what steps in the
signaling pathway are common to both exercise and insulin-stimulated glucose uptake,
and where is the derangement in insulin resistance such that insulin does not cause
normal GLUT4 trafficking, but exercise does? These are outstanding challenges for the
field, but we address some commonalities between the signaling pathways of exercise-
induced and insulin-induced GLUT4 translocation in the results section.
Upstream in the signaling pathway, AMPK (5′ adenosine monophosphate-
activated protein kinase) acts as important energy sensor in exercising muscles. It is
activated in low energy states, causing the cell to decrease ATP consumption and
18
increase ATP production. In other words, it acts as a metabolic switch, shifting the body
from an anabolic state to a catabolic state (69). The degree of anabolic or catabolic status
is primarily determined by the ratio of AMP and ADP to ATP. Initially, AMP was
thought to be the primary activator of AMPK, but more recent studies have suggested
that ADP is in fact a more important regulator (70). Along with its major upstream
kinase, liver kinase B1 (LKB1), AMPK is the most widely-studied protein implicated in
skeletal muscle glucose transport in response to exercise (71). It is a heterotrimer of three
subunits: a catalytic susubunit, as well as and subunits. There are several isoforms
of the subunits that are expressed in different cell types. The critical serine-threonine
kinase domain is in the ssubunit and the activating residue is Thr172 (69). Mutations to
the subunit have been implicated in cardiac disease. The gene (PRKAG2), which
encodes the 2 subunit, is linked to familial hypertrophic cardiomyopathy (HCM)
associated with Wolff-Parkinson-White syndrome. While the more common forms of
HCM involve sarcomeres, PRKAG2 syndrome is associated with accumulation of
glycogen in cardiac myocytes and does not involve disarray of the fibers of the cardiac
myocytes (69, 72).
In the heart, numerous factors can increase AMPK activity, including
physiological
stress,
hormones,
or
drugs,
including
metformin,
phenformin,
thiazolidinediones, salicylates, 5-aminoimidazole-4-carboxamide riboside (AICAR), and
A-769662 (69). Interestingly, insulin actually decreases activation of AMPK through the
Akt pathway (69, 73). In addition to the short-term effects of the activation of AMPK, the
2 isoform has a nuclear localization signal. Translocation to the nucleus appears to
require activation by phosphorylation at Thr172 (69, 74).
19
AMPK is required for insulin-independent GLUT4 translocation during states of
energy stress, such as cardiac ischemia or skeletal muscle contraction (75-77). In 1999, it
was found that AMPK mediates GLUT4 translocation independently of the PI3K
pathway, an important insight for cardiac physiology (78). AMPK activation also
enhances the sensitivity of muscle cells to insulin-dependent GLUT4 translocation and
glucose uptake, and may mediate the enhanced insulin sensitivity that occurs after
exercise (40, 79). Increased disposal of glucose in the acute post-exercise period (about
2-3 hours) is insulin-independent. But after this acute effect wears off, enhanced muscle
and whole body insulin sensitivity can persist for up to 48 hours (40). AMPK can be
activated by high doses of metformin (80). Phenformin, a similar compound to
metformin, poisons the mitochondria, interfering with complex I of the electron transport
chain, slowing the production of ATP, and thus decreasing the ratio of ATP to AMP (81).
The first known drug that was shown to activate AMPK in cells was AICAR (82).
AICAR, an adenosine analog, is taken up by cells via adenosine transporters and
phosphorylated by adenosine kinase, which generates an AMP mimic, AICAR
monophosphate (ZMP) (83). In a screen of compounds for activators of AMPK, a drug
named A-592017 was a promising candidate after the initial screen. After optimization,
an even more potent activator, A-769662, was developed (82, 84). One rationale for this
kind of search for a more potent and direct activator of AMPK is to gain the therapeutic
effects seen with the biguanides while reducing or eliminating the side effects which may
stem from the biguanides inhibiting the respiratory chain, rather than from the direct
activation of AMPK (82). A-769662 is thought to work by directly activating AMPK
through two mechanisms: allosteric activation and inhibition of dephosphyorlylation (82).
20
The therapeutic effect of metformin to lower the blood glucose level of patients
with type 2 diabetes is due to the action of metformin on hepatic glucose production,
although metformin has also been proposed to increase skeletal muscle glucose uptake
mediated through its action on AMPK (85). The mechanism for decreased glucose
production by the liver is metformin’s targeting of a mitochondrial enzyme that controls
the cellular redox state, as well as ATP production. Thus, activation of AMPK may not
be necessary for this reduction (86-88). The specific enzyme affected by metformin was
recently discovered, after the observation that animals treated with biguanides had
increased levels of lactate. Metformin non-competitively inhibits the redox shuttle
enzyme mitochondrial glycerophosphate dehydrogenase. This process reduces the
conversion of lactate and glycerol to glucose, and thus decreases hepatic gluconeogenesis
(86).
Recent results showed that after induced cardiac ischemia, well-known to activate
AMPK, with subsequent reperfusion (I/R), TUG is cleaved, at least partly explaining
increased GLUT4 and other GSV cargos at the plasma membrane after reperfusion (89).
In addition to AMPK, calmodulin signaling and calmodulin-dependent protein
kinases (CaMKs) have been implicated as vital components of exercise-stimulated
skeletal muscle glucose uptake (90). Whether this path is independent of AMPK is
controversial. In support of a model in which CaMKs are independent of AMPK, studies
have shown that the incubation of rat skeletal muscle with KN-93, a Ca2+Calmodulin
inhibitor, decreased the uptake of glucose after contraction (71, 91). These studies also
demonstrated significant inhibition of exercise-induced Cam-KII phosphorylation
without AMPK inhibition.
