Sunday, August 9, 2009

A GUT Feeling about Insulin

Ask ten people how to lose weight (fat), and you'll likely get ten different answers. In fact, if you ask ten "experts" the same question, you'll probably also get ten answers (usually attached to some product or service requiring you to part with some money). Why all of the confusion? After all, it seems a fairly simple question at its base: how do you burn more fat than you store?

I believe there's two key failures in critical thinking underlying the confusion. The first is that obesity itself is a "disease", which needs to be "cured". Many other diseases (heart disease, cancer, etc.) are associated with obesity, and the prevailing thought is that curing obesity reduces risk for these other diseases. However, this ignores the mountain of evidence that an organism's metabolism is self-regulating. In this view, obesity is a symptom of of some underlying disease process which causes systemic failure of metabolic regulation. It is this underlying disease which needs to be fixed; further, it is possible that you can have this disease and not be obese (there are plenty of skinny Type II diabetics). Modern medicine is very skilled at treating symptoms and ignoring the root cause; indeed, this effect is rampant for obesity treatments. How many people do you know that have lost large amounts of fat, only to have it come back worse?

The second failure comes from "black box" thinking. When hearing various prescriptions for curing obesity, I'm reminded of a famous Sidney Harris cartoon. For instance, a friend was recently telling me about a lemon juice diet. You drink lots of lemon juice, and the fat miraculously flows out of the fat cells. This supposedly had something to do with changing the acidity of your blood, but of course when prompted this person couldn't supply any actual physiological mechanism to explain this effect.

To understand the problems with black-box thinking, we can use the example of, uh, a black box. It has a hole where you can put stuff in, and lots of different colored lights that blink in response to whatever you provide as input. Your job is figure out the rules of how the input relates to the blinking lights. As we try different things we find many patterns of colored lights, with no obvious patterns. For instance, we supply two different cube-shaped objects, but each elicits a different light pattern. So "cubiness" is not apparently relevant to the lights.

The behavior of our black box may appear complex, but we don't really know if it's inherently complex, or if we just lack enough information to tease out the rules. We might crack the box open and examine how it actually works, and find that there really is a simple rule at the core, i.e. specific lights turn on depending on the molecular composition. The rule turns out to be simple, but it's the variety of different inputs that result in apparently complex behavior. Once you know how the box works inside, it becomes relatively easy to predict its response to a given input.

If you notice, most studies on diet and health take the black box approach: they diddle some inputs, and observe how those inputs are associated the outputs (e.g. fat loss). But if you don't have some understanding of what's going on inside the box, you just wind up with a mass of confusing observations and associations. So the lack of consensus and mercurial nature of dietary recommendations should come as no surprise.

Unification and Symmetry
Science often faces such situations. The core difficulty is a lack of symmetry. Symmetry means "sameness in the face of change". A perfectly smooth cue ball will look the same no matter how you turn it. Paint some dots on the ball, and you break the symmetry.

We often encounter cases where our observations seem to reflect a lack of symmetry, but if we look hard enough we find a deeper symmetry, one that unifies our observations under a common model. Such was the case in particle physics in the 20th century. Physicists had observed a vast zoo of different particles, first in cosmic rays (high-energy particles from space), then in "atom smashers". There were also four apparently disparate "forces" of nature: electromagnetic, weak nuclear, strong nuclear, and gravitation. The drive (which continues today) was to unify these different things by identifying the underlying symmetry. A "grand unified theory" (or GUT) would explain the all subatomic phenomena with a single model. Some progress has been made, e.g. many of the different particles were found to be composed from a much smaller family of more fundamental particles called quarks. The electromagnetic and weak nuclear forces (the latter causes radioactive decay) we discovered to actually be one in the same, the apparent difference occuring because the universe is relatively cold.

A Unified Theory of Fat Storage
Can we find a corresponding unifying principle for how fat loss and gain are related to diet? I think the answer is a qualified "yes". We likely need to restrict the domain of our model to one where the observed effect (obesity) has a common cause. Metabolic regulation is complex, and excess fat storage can have multiple root causes. We'll focus here on one possible cause, because it appears to be common and becoming more so: too much insulin, and/or not enough sensitivity to that insulin. Insulin is arguably the boss hormone for metabolic regulation: it effects many systems, and itself is affected by many factors. By examining the effect of insulin both on the behavior of individual cells and at the level of global metabolic regulation, we can in effect "open the box": see how inputs affect insulin and insulin response, then follow the effects of insulin in the body, particularly on fat storage.

I am going to make the bold claim that insulin is the unifying factor, tying together many different observations about fat gain/loss. I intentionally said "many" instead of all, because there are other metabolic pathways influencing fat storage (e.g. increased adrenaline promotes release of fatty acids from fat cells). I'll make the further claim that just about any successful reducing strategy (one that results in fat loss) can be explained by its effects on insulin, whether that strategy involves diet, physical activity, drugs/supplements, or a combination. We should also be able to explain both the relative efficacy of different strategies both in terms of rate of fat loss and final equilibrium fat mass (e.g. many diets result in fat loss, but all seem to "stall" at some point; we should be able to explain this stall via our model). Some examples are given below.

Our Grand Unified Theory theory then provides a more solid foundation for discussing the relative merits of different reducing strategies, and more importantly for making decisions about which lifestyle modifications are most appropriate. Instead of sifting through piles of observational evidence and "expert" testimony, you simply ask two questions:
  1. Is my obesity insulin related? (The answer is probably "Yes" for most, but not all. Those whose obesity has some other cause, like a genetic leptin disorder, will need to seek other avenues of treatment).
  2. How does X affect my insulin? From here you should be able to make a more informed decision about whether or not to pursue X for fat loss.
Perhaps more importantly, by moving the focus from a symptom (obesity) to an underlying cause, we can begin to recognize that controlling insulin should have wide-ranging implications for health (insulin does many things beyond controlling blood sugar and fat storage).

A Brief Primer on Insulin
The effect of insulin on fat storage has been covered elsewhere in detail, most notably in Gary Taubes' book Good Calories, Bad Calories. But it is probably worthwhile to hit the high points again. Insulin also does not act in isolation, but plays an intricate dance with other hormones and the nervous system. Some of these relationships are covered here.

Insulin is a protein (you can see a computer-generated representation here). Like all proteins, there is a gene that encodes the particular sequence of amino acids for manufacturing insulin. One of the interesting facts about insulin is that it's structure is remarkably consistent across time and species. Thus, species which appear genetically divergent, like humans and hagfish, do make different forms of insulin and the insulin receptor, but they're more simillar than different: human insulin has a large degree of cross-reactivity with hagfish insulin receptors, and vice-versa. So insulin has been around a long time, and the relative lack of cross-species mutation is an indication of it's key role in the survival of an organism.

The effects of insulin are initiated when an insulin molecule binds to an insulin receptor at the surface of a cell membrane. This binding triggers a series of chemical reactions, generally culminating at the cell nucleus, where genes are either up-regulated (meaning they make more of some protein) or down-regulated. Most people are familiar with the role of insulin in controlling blood sugar. One major effect of insulin binding is the manufacture of glucose transport (GLUT) proteins, which move glucose out of the blood, across the cell membrance, and into the cell. But insulin has many other effects. It is mitogenic, which means that it promotes cell division (i.e. insulin is a growth hormone). Insulin plays a key role in the manufacture of cholesterol from glucose, both by up-regulating transport of glucose into the cell and controlling manufacture of HMG-CoA reductase, and enzyme required to transform HMG-CoA into cholesterol (side note: statins block manufacture of HMG-CoA reductase). And there's a pile of other functions as well.

When insulin binds to an insulin receptor, it not only causes a chemical signal to be sent. The entire insulin/receptor complex is also absorbed by the cell (endocytosis), removing the insulin from circulation. A condition in which there is too much insulin in the blood (hyperinsulinemia) could thus result either from too much insulin being produced in the pancreas, or from a relative lack of insulin receptors. Correspondingly, insulin resistance (the failure of cells to respond to the insulin signal) could result from a lack of insulin receptors, a failure in the chemical signal chain, or from some other molecule (like a lectin) physically blocking the insulin receptor.

We should also realize that insulin does it's thing via it's effect on genes. Genetic differences can thus imply diferent responses to insulin. Genes carry the code to manufacture proteins, and a rather small difference in gene activation by insulin can result in large visible differences between individuals. This is particularly true for fat storage. We'll see below how insulin triggers manufacture of lipoprotein lipase (LPL) which is necessary for fat storage. A small difference in the amount of LPL made in response to insulin results in a small difference in net amount of fat storage. But whether that small difference results in net negative or positive storage could determine whether or not an individual will become obese.

On to the point. Insulin controls fat storage primarily through three pathways:
  1. Up-regulation of lipoprotein lipase (LPL)
  2. Down-regulation of hormone sensitive lipase (HSL)
  3. Up-regulation of glucose transporters.
The basic unit of fat is a fatty acid. Fatty acids are not water soluble, as anyone who has tried to mix oil and water knows. Blood is mostly water, and having fat droplets wandering around your blood vessels is not good. So fats need some other water soluble molecule to transport them around in the blood. Individual fatty acids can be transported bound to a molecule of albumin, but this mostly occurs for fatty acids released from fat cells. Dietary fats and those made in the liver are carried mostly as triglycerides in large molecules called lipoproteins. Triglycerides are also the storage form of fat in fat cells. A triglyceride is composed of three fatty acids stuck to a glycerol backbone.

Triglycerides are too large to pass across the cell membrane. In order for fatty acids to get in/out of a fat cell, they must be freed from the triglycerides. Enzymes which perform this task are called lipases. Lipoprotein lipase (LPL) acts on lipoproteins in the blood to free fatty acids for transport into the fat cells. Hormone sensitive lipase (HSL) acts on triglycerides inside the fat cell, freeing fatty acids for transport out of the fat cell. The precise mechanism by which the fats actually make it across the cell membrane isn't entirely clear. Cell membranes are largely made of fatty acids themselves (in the form of phospholipids), so it's like that free fatty acids passively diffuse across the cell membrane (whereas water soluble substances, like glucose, generally require the help of a transport molecule). There is also evidence of fat transporter molecules, though these may be more important in cells like muscle that may need energy faster than can be supplied by passive diffusion.

The fatty acids inside the fat cell, regardless of their origin, are candidates for esterification, which just means they can be incorporated into triglycerides. This in turn requires a supply of glucose to manufacture the glycerol backbone (actually a molecule named glycerol-3-phosphate, or alpha glycerol phosphate; we'll use G3P). Insulin is necessary to effect transport of glucose from the blood inside of the fat cell, and also up-regulates a key enzyme (G3P dehydrogenase) required to form G3P from glucose.

Insulin increases LPL and decreases HSL. The relative concentration of fatty acids inside and outside of the fat cell are thus governed by insulin, as well as the availability of lipoproteins in the blood. Fatty acids tend to move from high concentration to low. If insulin is low, HSL activity is increased, fatty acids tend to build up in the cell and diffuse out to the blood. If insulin is high, LPL activity is increased, fatty acids build up outside the cell and tend to move in. Once inside the cell, insulin governs the relative rate at which fat is stored, not only through HSL, but also by effecting glucose transport and regulating G3P dehydrogenase.

There are other metabolic pathways which affect this process. Some, like de novo lipogenesis, are also regulated by insulin. Others, like acylation stimulation protein (ASP), appear to be independent of insulin. There are ongoing arguments as to the relative importance of the various pathways, but I think the evidence is pretty clear that insulin is king of the hill when it comes to fat storage. For instance, Type I diabetics, who make little or no insulin, basically lack the ability store fat. If ASP were important in humans, Type I diabetics should be able to store plenty of fat (since one of the symptoms of Type I diabetes is ravenous hunger, I think we would have observed this). Any Type I diabetic who injects insulin, however, is familiar with the "fat pad" that forms at the injection site, due to (ta da) the high concentration of insulin in that area.

So, lots of concepts and big words in the above. The takeaway is simple: more insulin means fat cells store fat; less insulin means fat cells release fat. The equilibrium point (at which you're neither storing nor releasing) is thus largely determined by average insulin levels. We should then be able to predict the effect of various lifestyle changes from their effect on insulin. Let's see how that works out for some commonly recommended reducing strategies.

Low Carbohydrate Diet

This ought to be a no-brainer. Of all macronutrients, carbohydrates have the largest direct effect on insulin levels. Protein also stimulates a little insulin release, but nothing like a quantities of readily available carbohydrate (dietary protein also stimulates release of the hormone glucagon, which tends to counteract insulin's effect of driving glucose from the blood into fat cells, thus reducing fat storage). By itself, fat does not stimulate insulin release (in fact it seems to decrease it mildly). But fat does cause release of hormones like CCK, which amongst other things cause the pancreas to release more insulin for a given stimulus of glucose or amino acids (this is called the "incretin effect"). So eating fat and refined carbohydrates together (which is most food in the Western diet) ought to really crank your insulin. High average insulin means more fat storage - look around any public place if you want to see this in action.

Conversely, removing carbohydrates from the diet should drastically reduce average insulin levels (unless you have some non-dietary problem, like an insulin-producing tumor, in which case you've got bigger problems that being fat). The decrease in insulin should move the body away from fat storage to fat release. Since this fat is now available for energy, appetite should decrease and/or activity should increase spontaneously. All of these effects have been observed repeatedly in both animal and human studies.

Low Calorie Diet (Starvation)
Suppose we just cut calories across the board. Say your nominal caloric intake was 2400 kcal/day, including an average of 300g of carbohydrates. Leaving fructose out of the equation (fructose does not directly stimulate insulin release, but does cause the liver to become temporarily insulin resistant, the net effect of which may be to increase average insulin levels), that's equivalent to about a cup and a half of sugar each day (the gut rapidly breaks down "complex carbohydrates" into glucose for absorption into the blood). Since the total amount of glucose in a normal person's blood is about 1 tsp, this 1.5 cups should have a drastic effect on average insulin levels, as the body works very hard to keep blood glucose in a narrow range (too much or too little glucose in the blood will kill you in a hurry).

Now, let's not change what we eat, just how much. We'll go from 2400 kcal/day down to 1600 kcal/day. That implies we're now eating 200g of carbohydrate per day, implying that average insulin levels should drop significantly. Again, this should result fat loss, since we've decreased insulin from the level that promoted our previous equilibrium. And that's precisely what's seen: starvation diets result in fat loss. However, that 200g of carbohydrate still promotes a fair amount of insulin secretion. We would thus expect initially rapid fat loss, tapering off over time, and finally stalling at the new equilibrium point. And once the fat stops coming out of the fat cells, your body is literally starving, and will likely make you fall off the wagon, so to speak. As your body has become used to lower levels of insulin (i.e. your insulin sensitivity has increased), resuming previous levels of carbohydrate and fat consumption should result in rapid weight gain, overshooting your previous equilibrium point. Which, again, is exactly what is seen.

Low Fat Diet

The low-fat diet is an interesting case, and what is called "low-fat" often involves both calorie restriction and the trading out of refined carbohydrates for more whole food sources, which tend to have less effect on blood sugar and thus insulin. Both latter effects of course will drop your average insulin, and result in some fat loss. The interesting thing here is that reduction in dietary fat should also reduce secretion of incretin hormones like CCK, and thus further reduce insulin. So low-fat diets "work", as is often observed. In fact, I would predict it works better than just generically cutting calories. I don't know if this has been observed. The confusion most people have is the idea that eating fat makes you fat, and thus erroneously conclude reducing fat makes you thin. But all of this action is ultimately effected by insulin.

And that's the rub, because it means it is difficult (and probably unhealthy) to eat low-fat forever. If you don't eat much fat, then you need carbohydrates for energy (using too much protein for energy results in nitrogen poisoning). If you get those carbohydrates from the usual sources, like bread, rice, or pasta, your insulin will go up, and you'll get fat again, whether you eat fat or not (note that excess dietary carbohydrate is converted to fat by the liver). Successful maintenance of a low-fat diet means getting carbohydrates from sources which are slowly digested, and/or maintaining a high enough level of physical activity to burn off excess glucose and enhance insulin sensitivity (more on this below).

Physical Activity

We've all heard the old chestnut that to effect fat loss you just need to "eat less and exercise more". We've seen above how calorie reduction can affect insulin levels. But does exercise do the same thing?

Interestingly, the answer is a qualified "Yes". Let's start with an extreme case (which, as it turns out, forms the basis for the very successful "slow burn" type exercise regimens). Muscle stores glycogen, a form of starch, for use as quick energy. The glucose to make that glycogen gets into the muscle cells via the action of insulin. In the case of muscle cells, insulin stimulates the cell to move a preformed store of GLUT4 molecules to the surface, so glucose can be rapidly absorbed from the blood. Now suppose you completely exhaust the muscle of its glycogen stores. What do you suppose its response will be?

Not surprisingly, the cell cranks out more insulin receptors in an effort to rebuild it's energy. After all, you might need that quick energy to escape the next hungry lion that crosses your path. So exercise increases insulin sensitivity of muscle, and we learned above that when insulin binds to an insulin receptor the cell absorbs the whole complex. So, independent of diet effects, we expect exercise to reduce average insulin levels; further, in doing so, the muscles also clear out some glucose. Both of these effects should lead to some degree of fat loss. Any increase in net physical activity should result in this effect to some degree. Your muscle cells will only make insulin receptors if they need to. If you start as a total couch potato, and then start walking a mile a day, your muscles need to adapt to even this small increase in activity (walking a mile burns about an extra 100 kcal).

And of course, that's what people see. How many friends have you known that started a new exercise regime and rapidly lost some weight? This is often accompanied by pronouncements like "I can eat anything I want, as long as I exercise enough". That's true, at least to the point where the new fat storage/release equilibrium is reached, at which point fat loss stops. Since the individual is no longer getting positive feedback of fat loss for their physical exertion, they usually cut back or quit, but continue eating "anything I want", and of course just get fat again.

And all of this ignores the elephant in the living room, which is overall metabolic regulation. If you use more calories than are totally available to you from food and storage (remember that high insulin makes stored fat unavailable), you should get hungry. Further, the body knows what it wants, and will try very hard to make you eat it. If you burn up the muscles' store of carbohydrate, the resultant temporary increase in insulin sensitivity will drop your blood sugar. Your brain senses that drop, and tells you to go eat some carbohydrates. People often "reward" themselves with a food treat after a workout, or maybe have a sugary energy drink or similar. Of course, this tends to defeat whatever gain in insulin sensitivity your exercise created.

The Challenge

The examples above, I believe, illustrate explanatory power of the insulin hypothesis, bringing many approaches which seemed disparate or opposed (like low fat vs. low carb) under a single explanation. My challenge to you, O Gentle Reader, is to provide counter-examples. Are there fat-loss strategies that cannot be explained by the insulin model? Give it your best shot in the comments.