8 Gluconeogenesis and Glycogen

  • Understand how glucose is generated during fasting.
  • Appreciate the tight control of blood glucose levels
  • Understand the reversible storage of carbohydrates as glycogen
  • Glucose homeostasis
  • Gluconeogenesis and its precursors
  • Principles of signal transduction pathways
  • Reciprocal regulation of opposing metabolic pathways

Our cells need energy continuously, yet we eat only every couple of hours or can sleep for 8h without running out of ATP. How is that possible?

Fig. 1 shows the course of blood glucose levels during the day. We learned in chapter 5 how carbohydrates are digested and absorbed. Thus we are not surprised that there is a peak shortly after each meal. But what removes glucose from blood to bring it back to fasting levels within 2 hours? Why does blood glucose levels never fall below 4 mM, even overnight? Insulin levels closely follow blood glucose levels. Insulin initiates the use and storage of nutrients after a meal. The drop of the peak shortly after food intake is caused by insulin. There is another hormone that does the opposite, namely glucagon. It drops after a meal. In chapter 16 we will learn more about these hormones and how they regulate blood glucose.
In this chapter we want to understand what happens biochemically to keep glucose concentrations at around 4.5-5 mM. There are principally five mechanisms that can change blood glucose concentration (Fig. 2), namely food intake, glycogen synthesis, glycogen breakdown, glycolysis and gluconeogenesis.
Fig. 2 Elements of glucose homeostasis in our body.

Food intake can only increase glucose levels. Glycogen reversibly stores glucose and thus can decrease or increase glucose levels. Glycolysis degrades glucose, while gluconeogenesis generates glucose. Acetyl-CoA generated from glucose can be fully oxidised or converted into fat. Of the fat molecule (triacylglycerol), only the glycerol backbone can be converted back to glucose. The fatty acids are broken down to yield acetyl-CoA, which through the TCA cycle can only generate CO2. We will first have a look at gluconeogenesis and see how it maintains glucose levels overnight. The question arises: can we synthesise glucose simply by reversal of glycolysis? Inspect the table below, which lists all steps of glycolysis and their free energy, to find the answer.

Reaction step deltaG0' (standard, kJ/mol) deltaG (cellular, kJ/mol)
Glc+ATP –> Glc-6P+ADP
-16.7
-33.4
Glc-6P –> Fru-6P
1.7
0 kJ/mol
Fru-6P+ATP–>Fru1,6BP+ADP
-14.2
-22.2
Fru1,6BP–>DHAP+GA3P
+23.8
0 kJ/mol
DHAP–>GA3P
7.5
0 kJ/mol
GA3P+P+NAD–>1,3BPG+NADH
+6.3
0 kJ/mol
1,3BPG+ADP–>3PG+ATP
-18.8
0 kJ/mol
3PG–>2PG
+4.4
0 kJ/mol
PEP+ADP–>Pyruvate+ATP
-31.4
-16.7

In glycolysis all but three steps are readily reversible. To synthesize glucose we need to find alternatives for the reactions that have a large negative free energy. This is straightforward for the hexokinase and phosphofructokinase reaction, which in the forward reaction (glycolysis) use ATP: Instead of synthesising ATP in the reverse reaction (gluconeogenesis), our cells just hydrolyse the phosphate, which is energetically favorable (Fig. 3). This alternative is not available for the pyruvate kinase reaction, because its reversal would consume ATP. Thus, we need a detour (Fig. 3). 

Fig. 3 Glycolytic and glucogenic pathways. Red, glycolysis; Magenta, gluconeogenesis from lactate; brown, gluconeogenesis from alanine; green gluconeogenesis from glycerol; blue all sources of gluconeogenesis.
There are four main sources for gluconeogenesis (glycerol, lactate, alanine and glutamine) and their specific metabolic pathways are different (magenta, brown and green in Fig. 3) due to the availability of cofactors. Remember that we normally use NADPH for biosynthetic purposes, but gluconeogenesis uses the same glyceraldehyde-3-phosphate dehydrogenase reaction that occurs in glycolysis and thus uses NADH when running in reverse. As a result, we need to find some NADH in the cytosol when gluconeogenesis occurs and its production must balance its use. This is another example of metabolic coupling. Gluconeogenesis from lactate (follow magenta and blue arrows) is straightforward as it generates NADH in the cytosol when lactate is oxidised to pyruvate by lactate dehydrogenase. Pyruvate is then converted to oxaloacetate by pyruvate carboxylase inside mitochondria. Oxaloacetate is further converted to phosphoenolpyruvate (PEP) by PEP carboxykinase. This enzyme uses GTP and thus will only allow gluconeogenesis to occur if the TCA cycle produces sufficient energy (see chapter 9 for GTP generation in the TCA cycle). Please note that we are not withdrawing intermediates from the TCA cycle, because we use the anaplerotic enzyme pyruvate carboxylase to generate oxaloacetate, not the TCA cycle. Phosphoenolpyruvate leaves the mitochondria and can join the gluconeogenesis pathway.
When alanine is used to generate glucose (during the night while sleeping, follow brown and blue arrows), pyruvate is mainly generated by transamination of alanine, which does not generate NADH. In this case (brown pathway) pyruvate is converted to malate. Although this reaction is part of the TCA cycle, its equilibrium favours generation of malate. Malate can then leave the mitochondria and is used to generate NADH by conversion back to oxaloacetate. Cytosolic PEP carboxykinase finally generates phosphoenolpyruvate (PEP) to merge with the common gluconeogenesis reactions (blue). Glutamine can be converted to malate through the TCA cycle, it is mainly used by the kidney to generate glucose. Glycerol, derived from fat, can be used directly for gluconeogenesis because no NADH is required. First it is phosphorylated to glycerol-3-phosphate, then glycerol-3-phosphate is oxidised by NAD to the gluconeogenis/glycolysis intermediate dihydroxyacetone phosphate (green pathway) generating NADH. Only a few organs can generate glucose for systemic glucose homeostasis, namely the liver and kidney. Gluconeogenesis by the intestine is disputed but it expresses the key enzyme glucose-6-phosphatase. Muscle can carry out some gluconeogenesis (without releasing glucose) to reform glycogen from lactate during recovery, but most of the lactate is removed by the circulation and converted into glucose by the liver. In terms of contribution of different pathways to glucose production (measured in mice) during fasting the order is glycerol > glycogenolysis = lactate > glutamine = alanine.
In organs that can carry out gluconeogenesis, we need to avoid that glycolysis and gluconeogenesis run at the same time. This is called a futile cycle, because glycolysis breaks down glucose and these products can be used as substrates for gluconeogenesis reforming glucose. Such a cycle does not generate useful metabolic intermediates but hydrolyses ATP. You can see this by looking at the first three reactions  in Fig. 3 running down the red arrows and returning via the blue arrows. So what prevents the two pathways from running at the same time? 
Fig. 4 Reciprocal regulation of glycolysis und gluconeogenesis. Regulated proteins are shown as green (active) or red (inactive).
We have encountered two major principles of enzyme regulation in previous chapters, namely allosteric regulation and protein phosphorylation. Both are used to regulate critical enzymes in metabolic pathways. The key allosteric regulator of glycolysis and gluconeogenesis (in liver) is the molecule fructose-2,6-bisphophate (not to be confused with the pathway intermediate fructose-1,6-bisphosphate). It acts as an allosteric activator of phosphofructokinase-1 and as an allosteric inhibitor of fructose-1,6-bisphosphatase. We will see in a moment how fructose-2,6-bisphosphate is generated and when. Another allosteric regulator is acetyl-CoA, which during fasting (when we need gluconeogenesis) is generated from fatty acids. Thus, we have an alternate source for acetyl-CoA and can use other metabolites for gluconeogenesis. Acetyl-CoA acts as an allosteric regulator of pyruvate carboxylase, thereby enhancing gluconeogenesis. Glycolysis in addition is down-regulated by phosphorylation of pyruvate kinase by protein kinase A, which is also activated during fasting by the hormone glucagon. This prevents reconversion of phosphoenolpyruvate generated during gluconeogenesis back into pyruvate.
Despite all this regulation, heat is generated after food intake from futile cycles such as glycolysis/gluconeogenesis, glycogen synthesis/breakdown, triacylglycerol synthesis/breakdown. Each of these processes consumes ATP. In total 5-15% of available energy is lost due to diet induced thermogenesis.
We now want to understand the processes that regulate metabolism in a bit more detail.

Regulation of biochemical processes is generally referred to as signal transduction, because they are typically initiated by signals, such as hormones, neurotransmitters, mechanical stimuli etc. Before we look into details we need to understand an important principle of biological regulation. Biochemical pathways rarely stop and go, they rather slow down or accelerate. This is even true for opposite pathways such as glycolysis/gluconeogensis, fat synthesis/breakdown, protein synthesis/breakdown. To make an analogy, think of biological regulation as pushing the brake and accelerator pedal at the same time.

Increasing the speed is done by releasing the brake slightly and pushing down the accelerator slightly more. Decreasing the speed is done in the opposite way. It is important to incorporate this concept of steady state into biochemical thinking. For example, glycolysis would indeed stop when no ATP is needed (after death), however, there is always demand for ATP, sometimes more sometimes less. Steady state theory is tightly linked to metabolic coupling, which we have come across many times.

Fig. 6 The typical elements of a signal transduction pathway.
As we will see in chapter 16, glucagon is the hormone that increases the activity of pathways that generate glucose and provide energy (gluconeogenesis, glycogenolysis, lipolysis etc.). What happens when a hormone binds to a cell (Fig. 6)? Cells that respond to a particular hormone have receptors at the cell surface. Upon binding of the hormone (the first messenger) a conformational change occurs, which allows the receptor to bind to other proteins or to recruit accessory proteins. This process can then activate enzymes that produce a second messenger, such as cAMP, cGMP, phosphoinositol phospholipids (PIP). The second messenger than in turn activates a protein kinase, which phosphorylates a target protein, rendering it active or inactive (as in the case of pyruvate kinase mentioned in Fig. 4 and shown in Fig. 8). 
Let us look at an actual example, namely the production of glycolysis regulator Fructose-2,6-bisphosphate by glucagon during fasting.
Fig. 7 Regulation of glycolysis/gluconeogenesis by glucagon

Fig. 7 shows the sequence of events. (1) Glucagon (the first messenger) binds to its receptor at the cell surface. This particular class of receptor is called a G-protein coupled receptor (GPCR). The term G-protein refers to a group of proteins that can bind GDP or GTP and are regulated by these nucleotides. In their inactive form GDP is found in the binding site. When GTP is in the binding site they are active, but GTP is slowly hydrolysed to GDP. After binding of glucagon a conformational change occurs in the receptor, opening a binding site for a G-protein on the cytosolic side. When the G-protein binds to the receptor, GDP is released and exchanged for a GTP molecule (2). This causes the heterotrimeric G-protein complex to fall apart (3). The complex will reform once GTP in the alpha unit has been hydrolysed to GDP (4), thereby resetting the whole system. The hydrolysis is slow to allow the signal to last for a while. The alpha unit, which binds the GTP is now activated and can interact with other proteins, most notably adenylate cyclase (5). Adenylate cyclase is activated by this process and catalyses the conversion of ATP to cyclic-AMP (cAMP, the second messenger); cAMP in turn activates protein kinase A (6). The activation is caused by cAMP binding to the regulatory subunits, which then dissociate from the catalytic subunits (7). The liberated catalytic subunits can now phosphorylate the final target, the bifunctional enzyme fructose-2,6-bisphosphatase/phosphofructokinase 2 (8). Before we look at this enzyme watch the video below.   

The colour coding of the subunits is the same as in Fig. 7. Can you identify steps 1-6 (7,8 not in the video)?

Fructose-2,6-bisphosphatase/phosphofructokinase 2 has two catalytic activities. It can convert the glycolytic intermediate fructose-6-phosphate into the allosteric regulator fructose-2,6-bisphosphate or breakdown fructose-2,6-bisphosphate to reform fructose-6-phosphate. Having both parts active at the same time would make no sense, thus one activity dominates while the other is blocked.  To understand what the bifunctional enzyme fructose-2,6-bisphosphatase/phosphofructokinase 2 does, we need to look again at the regulation of glycolysis/gluconeogensis (Fig. 8).

Fig. 8 Hormonal regulation of glycolysis and gluconeogenesis, regulated proteins are show as green (active) or red (inactive).
As outlined above, the key allosteric regulator of glycolysis/gluconeogenesis is the molecule fructose-2,6-bisphosphate (not to be confused with the pathway intermediate fructose-1,6-bisphosphate). It acts as an allosteric activator of phosphofructokinase 1 and thus speeds up glycolysis. In addition, it is an allosteric inhibitor of fructose-1,6-bisphosphatase and thus slows down gluconeogenesis (Fig. 8). When protein kinase A phosphorylates fructose-2,6-bisphosphatase/phosphofructokinase 2, the fructose-2,6-bisphosphatase becomes more active (green in Fig. 8) and phosphofructokinase 2 becomes less active (red). This results in the degradation of fructose-2,6-bisphosphate, which no longer activates glycolysis, and in turn stops inhibiting gluconeogenesis. As a result, the balance tips from glycolysis to gluconeogenesis.
We now understand how glucose is produced while fasting. However, in the hours after a meal we feel energised and blood glucose levels, after peaking for 1-2h, return to normal levels.  It is instructive to calculate blood glucose levels if there was no storage of glucose (Question below).

Question about blood glucose

Assume you had a slice of pizza for lunch. What would be your blood glucose concentration if your body would digest the carbohydrates in the pizza to glucose and absorb it completely within 2h (this is a good approximation of what happens), but could not store or metabolise any of the glucose. A slice of pizza contains 30g of carbohydrates and the molecular weight of a glucose unit in starch is 162g/mol. Assume a blood volume of 5L.  Give your answer in mMol/L without the unit and round to the next full number. Compare this to Fig. 1.

Blood glucose

37 mmol/L

The answer to the question suggests that a large fraction of glucose is stored directly after a meal and released later on. This storage buffer of glucose is glycogen which is synthesized after a meal and broken down between meals. We will discuss this now in more detail.

Glycogen

Glycogen is a branched polymer of glucose attached to the protein glycogenin (Fig. 9). Glycogen particles can be seen in electron micrographs of the cytosol. Figure 9 also shows the loading/use of glycogen in mice during a light/dark cycle. In the dark (black line) mice consume food loading glycogen, while during light (open line) they are active, but do not feed, depleting glycogen stores within 6h.  

Fig. 9 Glycogen. Left panel principal structure of glycogen showing its branched chains of glucose units. Top right, electron micrograph of glycogen particles (G). Bottom right, liver glycogen in mice during feeding and resting periods in mice.

Also in humans glycogen is the main source of glucose during the early hours of fasting (see Table). Overnight gluconeogenesis becomes more dominant.

Length of fast (h) Liver glycogen content (umol/g liver) Rate of glycogenolysis (umol/kg * min)
0h
300
0 umol/kg*min
2
260
4.3
4
216
4.3
24
42
1.7
64
16
0.3

How does glycogen build-up and degradation work? There are two key enzymes, which regulate the process, namely glycogen synthase and glycogen phosphorylase (Fig. 10). 

Fig. 10 Glycogen metabolism (simplified)

Glycogen synthesis cannot proceed spontaneously, because water has to be eliminated in the reaction, which is energetically unfavorable. Therefore, a chemically activated form of glucose, namely UDP-glucose is used. Similar to coenzyme A, which is used to chemically activate fatty acids, nucleotides are used to activate sugars, typically UTP or CTP. In a reaction where pyrophosphate (PPi) is released from UTP a nucleotide-monophosphate is linked to glucose-1-phosphate forming UDP-glucose. This reaction is driven by the subsequent hydrolysis of pyrophosphate into 2 molecules of phosphate. While glycogen synthesis with glucose-1-phosphate is not energetically favorable, it is when UDP-glucose is used instead. There is also a branching and debranching enzyme involved to build up and break down glycogen (Fig. 11), but we do not discuss these reactions here. Final glycogen particles grow until 12 tiers have been formed, each with 12-14 glucose residues and two branches. As a result, there are about 55,000 glucose molecules in a glycogen particle. The branched nature of glycogen is important though to ensure rapid parallel breakdown at many sites.

Fig. 11 Different glycosidic bonds in glycogen allow branching
The breakdown of glycogen is catalysed by glycogen phosphorylase, which uses phosphate instead of water to attack the 1,4-glycosidic bond. Thereby the free energy of hydrolysis is preserved and used instead to attach a phosphate group to carbon-1, which otherwise would require ATP. Moreover, glycogenolysis increases in muscle due to the production of phosphate during exercise.  Phosphoglucose mutase is used to relocate the phosphate group from position 1 to position 6. The resulting glucose-6-phosphate can be used in a variety of pathways, such as glycolysis, pentose-phosphate pathway etc. 
The role of glycogen in muscle and liver is different (Fig. 12). When liver glycogen is mobilised the resulting glucose-6-phosphate is further hydrolysed in the endoplasmic reticulum to generate glucose, which is subsequently released from the ER and the hepatocyte to keep-up blood glucose levels. Why is this step located in the ER? Most likely to avoid intensive futile cycling with the glukokinase reaction in the cytosol.
Fig. 12 The glucose-6-phosphatase system in liver
Muscle does not express glucose-6-phosphatase and therefore glucose-6-phosphate remains inside muscle to generate energy. Alternatively, lactate produced in muscle during exercise can be recycled to form glycogen via gluconeogenesis.
We now want to look at the regulation of glycogen synthesis by insulin (Fig. 13) and its breakdown by glucagon/adrenaline (Fig. 14).
Fig. 13 Regulation of glycogen synthesis by insulin. For explanation see text.

The regulation of glycogen synthase by insulin is fairly complex but follows the same principles of signal transduction as discussed above. Insulin is the first messenger and binds to its receptor (Receptor tyrosine kinase, RTK). This brings the two halves of the receptor together and they cross-phosphorylate (called autophosphorylation) each other on tyrosine residues. The phosphorylated receptor is recognised by the insulin receptor substrate 1 (IRS-1), which in turn becomes phosphorylated itself. The phosphorylated IRS-1 docks onto phosphatidylinositol-3-kinase (PI3K) and activates it. Instead of a water soluble second messenger, such as cAMP, a specific membrane-bound lipid is generated as second messenger, namely phosphatidylinositol-triphosphate (PIP3). It is made by phosphorylation of phosphatidylinositol-diphosphate (PIP2), one of the minor components of the membrane bilayer. PIP3 in turn activates phospholipid dependent kinase (PDK1), which phosphorylates and activates protein kinase B (PKB). Active PKB-Pi, phosphorylates its target glycogen synthase kinase 3 (GSK3), thereby inactivating it. Active GSK3 would normally phosphorylate glycogen synthase (GYS), inactivating it, as well. Thus inactivating GSK3, allows protein phosphatase 1 (PP1), to dephosphorylate glycogen synthase and thereby activating it. Active PKB-Pi has other targets as well, particularly phosphodiesterase (PDE), which becomes active when phosphorylated and hydrolyses cAMP to AMP. This is a safety mechanism to avoid another futile cycle, namely synthesis and breakdown of glycogen at the same time. This is a perfect segue way to look at the regulation of glycogen breakdown (Fig. 14). 

Fig. 14 Regulation of glycogen metabolism by glucagon and adrenaline. PKA Protein kinase A, PPK Phosphorylase kinase, PYG Glycogen phosphorylase, GYS Glycogen synthase, GPCR G-protein coupled receptor
This pathway is more familiar to you as we already looked at its start when discussing gluconeogenesis. The start of the pathway is the same as shown in Fig. 7. Protein kinase A has many targets, here it is phosphorylase kinase (PPK). PPK then phosphorylates and activates glycogen phosphorylase (PYG), causing phosphorolysis of glycogen to liberate glucose-1-phosphate units.  Glycogen phosphorylase is an allosteric enzyme that is regulated by phosphorylation and allosteric effectors. The phosphorylated a-form is active, the dephosphorylated b-form is inactive, but can be activated by AMP. As we saw in Fig. 13, the insulin pathway negatively regulates the protein kinase A pathway and vice versa, the protein kinase A pathway regulates the insulin pathway. How does this work? Protein kinase A can phosphorylate glycogen synthase, thereby inactivating it and preventing insulin action. Vice versa, insulin-activated PKB activates phosphodiesterease, which hydrolyses cAMP, preventing protein kinase A activation. Thus, onset of glycogen synthesis, quickly turns off glycogen breakdown and vice versa. A nice illustration of the accelerator/brake regulation. Glycogen in muscle and liver have different roles. Glucagon is released during fasting and acts on the liver. It initiates a metabolic program to maintain blood glucose levels. Adrenaline is generated during exercise and acts on muscle; it initiates a metabolic program to provide energy for movement. 
Despite emphasizing that futile cycles are avoided, they do occur at a significant level particularly after food intake. This is called diet-induced thermogenesis because the hydrolysis of ATP is not used for useful work and is converted into heat. About 10% of calory intake is converted into heat.
Glycogen and gluconeogenesis are the sources of glucose during fasting and fasting + exercise but in what ratio are the different sources are used? Fig. 15 shows the sources of glucose in mice during fasting and when fasting is combined with exercise.

Sources of glucose

Fasting and Fasting + 20 min Exercise

During fasting glycogen, and gluconeoegensis from glycerol (from fat) and lactate are the main sources of glucose. Amino acids make a smaller contribution. There is a notable jump in the use of glycogen and lactate during 20 min of exercise. 

Another question is which source maintains glucose overnight. See Fig. 16.

Sources of glucose

In the first 4h after a meal glucose is provided by the food that was consumed. The digestion process takes its time, and a large flow of glucose occurs, which is higher than the actual needs. Once the food is digested and glucose deposited (4h) glycogen is used to maintain blood glucose. After 8h of fasting gluconeogenesis starts and becomes the dominant source beyond 16h of fasting. Glycogen runs out after 24h.

Gluconeogenesis and Glycogen
  • Glucose levels are maintained by the body at around 4-5 mM over long periods mainly to fuel the brain
  • In the absence of carbohydrates this is ensured by gluconeogenesis
  • Glycerol, amino acids and lactate are the main precursors for gluconeogenesis
  • Short term maintenance of blood glucose is controlled by synthesis and breakdown of glycogen
  • Glycogen synthesis makes use of UDP-conjugated glucose breakdown is mediated by phosphorolysis rather than hydrolysis
  • The liver is the main site for glucose homeostasis
Signal transduction
  • Signal transduction makes use of (first) messengers (hormones, neurotransmitters) that are released from cells after receiving a stimulus.
  • The first messenger binds to receptors that can be located on the surface of cells or inside the cell.
  • Binding to a receptor will initiate a signal transduction cascade.
  • Protein-phosphorylation and conformational changes are most commonly used to propagate signals.
  • Phosphorylation can activate or inactivate proteins.
  • Signal transduction inside cells involves second messengers. These can be phospholipids that become more phosphorylated and attract proteins to the membrane or soluble second messengers such as cyclic-AMP or cyclic-GMP. 
  • Second messengers can activate protein kinases, which phosphorylate other proteins. 
  • Target proteins are often enzymes or transcription factors
  1. Fig. 1 Mcstrother [CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0)], from Wikimedia Commons
  2. Fig. 2 By the author using Powerpoint
  3. Fig. 3 By the author using ChemDraw
  4. Fig. 4 By the author using ChemDraw
  5. Fig. 5 Michael Sheehan [CC BY 2.0 (https://creativecommons.org/licenses/by/2.0)], via Wikimedia Commons
  6. Fig. 6 By the author using Powerpoint
  7. Fig. 7 By the author using ChemDraw
  8. Fig. 8 By the author using ChemDraw
  9. Fig. 9 A, B, C; A: By Mikael Häggström, used with permission. [Public domain], via Wikimedia Commons; B: Jordi Miquel, Daniel Vilavella, Zdzisław Świderski, Vladimir V. Shimalov and Jordi Torres [CC BY 4.0 (https://creativecommons.org/licenses/by/4.0)], via Wikimedia Commons ; C: By the author: Drawn after data from: Weinert D, Freyberg S, Touitou Y, Djeridane Y, Waterhouse JM. The phasing of circadian rhythms in mice kept under normal or short photoperiods. Physiol Behav. 2005 Apr 13;84(5):791-8. Epub 2005 Apr 12. PubMed PMID: 15885257. 
  10. Fig. 10 By the author using ChemDraw
  11. Fig. 11 GKFXtalk 12:08, 5 September 2017 (UTC) [Public domain], via Wikimedia Commons
  12. Fig. 12 Yikrazuul [CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0)], from Wikimedia Commons
  13. Fig. 13 Yikrazuul [CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0)], from Wikimedia Commons
  14. Fig. 14 Yikrazuul [CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0)], via Wikimedia Commons
  15. By the author using data from Jessie et al. Cell Metabolism, Vol. 36, 2560-2579
  16. By the author after Cahill, Ann. Rev. Nutr. 26:1-22, 2006