Hello. I'm Steve O'Rahilly. I'm from the University of Cambridge. I'm interested in what makes us fat, and when we get fat, why do we get sick? And in today's talk, I'm gonna talk about why obesity leads to its adverse health outcomes. Obesity is really the effect… the end effect of an excess of energy intake over energy expenditure. This leads to an expanded mass of triglycerides in our fat cells, in our adipocytes, and that's how we define obesity. But really, we're worried about obesity not because of the cosmetic effect of this but because of the health effects.
And I think it's useful to group those into three. Firstly, there's mechanical and gravitational effects, if you like: the weight influencing the fact that we increase the risk of osteoarthritis of the knees, the increased intra-abdominal pressure leading to reflux esophagitis, the narrowed airway leading to sleep apnea. So, there's clearly a direct link there between the expanded mass of triglyceride in fat cells and the adverse effects. But when it comes to cancers and metabolic diseases, it's not quite so clear why having an expanded mass of triglyceride in fat cells leads to those. In fact, it opens the question of, does it? Is the obesity just a marker? And is it something about the chronic energy intake over energy expenditure that's directly leading to these cancers and metabolic diseases? The way we've tried to address this question over the years is to find one key phenomenon very closely associated with obesity and predisposing to metabolic disease and, indeed, cancer.
And that's the biochemical marker of insulin resistance. Now, as individuals get fatter, they tend to need more insulin to control blood glucose than they did when they were slimmer. So, here's a graphical representation of that. And you can see that the amount of insulin actually varies enormously across the population. There are some people who are obese who don't become particularly insulin resistant, but some do become severely insulin resistant and require a lot of insulin to maintain normal… normal blood glucose. Insulin resistance comes with a lot of additional baggage. Gerald Reaven in the United States, some 20 years ago or more, pointed out that individuals who are insulin-resistant have high cardiovascular risk.
They have abnormal circulating blood lipids. They have abnormal clotting factors. They have fatty liver. They're more prone to a range of disorders, not just simply an increased risk of insulin resistance and type 2 diabetes. So, a key question is, how do we get from obesity to insulin resistance? Because insulin, in the whole body, affects predominantly the muscle and the liver. The adipose tissue itself doesn't really concern itself with taking up too much glucose compared to those major glucose-controlling organs, liver and muscle. So, how do we get from sustained positive energy balance to insulin resistance in liver and muscle? If you were to ask a hundred diabetes researchers these days, or a hundred practicing diabetologists, they'd probably say, well, we think we know that now.
We think it's because the fat cell becomes large, the adipose tissue depot becomes inflamed, and the fat cell produces a range of circulating or paracrine factors which somehow influence insulin and glucose handling in liver and muscle. And there's abundant evidence from mouse experiments that this is true, but the support of evidence for humans, as I'll come to later, is somewhat less secure. There's another and somewhat older idea about how the… how we might translate this sustained energy balance into insulin resistance. And that's one where… I some… I sometimes say it's where your fat cell is not your enemy; your fat cell is your friend. Your fat cell is by far the safest place to keep positive energy… energy balance and keep calories. And it's only when you start reaching the limits of safe fat cells… fat storage of triglycerides, and those nutrients start to go to non-professional storage tissues such as liver and muscle, then it's the ectopic lipid and the ectopic nutrients that are really causing the mischief and causing insulin resistance and defective glucose handling in these tissues.
One of the reasons we've got to the notion that this is an interesting idea is partly the fact that we've been studying patients with severe insulin resistance for over 25 years. And over those years, we have found… we've looked for people who have extremely high levels of circulating insulin, and we've looked, genetically, to find signaling defects, defects in how insulin works in the cell to control blood glucose. And we found a number of individuals and a number of genetic defects as… as sort of illustrated in this slide here. But I think one of the most important things we've learned is from taking those individuals and then measuring the variety of physiological variables that we find in those people with defective insulin signaling, and then compare them to people with the typical metabolic syndrome. So, in the top two rows, you see that in both our individuals with signaling defects and typical metabolic syndrome, insulin levels are high, diabetes risk is increased.
But below that, we start seeing, really, quite marked divergences. When you have a defect in insulin signaling, you actually have no abnormalities, often, in triglycerides or HDL. You very often have no abnormalities in liver fat. And you often, indeed, have an increased serum adiponectin. These all diverge markedly from individuals with, if you like, common, or garden, metabolic syndrome or insulin resistance. But there was one set of conditions we've studied over the last a couple of decades where we really do see a complete metabolic phenocopy, and that's quite individuals with lipodystrophy. These are individuals who are insulin resistant because they either cannot make fat cells or they cannot keep triglycerides within fat cells, and they develop as pictured here: an absence of body fat, fats elsewhere in the… in the body, as shown under the skin of the feet or in the… in the… in the eye circulation. And with my wonderful colleague, David Savage, and others over the last few years, we've discovered many of the genes, in collaboration with others, and indeed, other groups around the world have discovered other genes.
And so, we now understand a lot more about the genetic architecture of lipodystrophy. Here's one example, I think, which is particularly illustrative. It's human perilipin-1 deficiency, a specific subtype of mutations within this adipocyte triglyceride coat protein. All the members of this family who only carry mutation in… in perilipin-1 all develop partial lipodystrophy, insulin resistance, and the full range of downstream abnormalities: fatty liver disease, cardiovascular disease, etc.
The individuals carry… carry heterozygous mutations, which cause one copy of perilipin-1 to be made normally and then another mutant copy with an aberrant C-terminal tail. So, what's happening in these individuals? The important fact is that perilipin-1 is only expressed in white adipocytes. So, here we have a disorder where you simply have one copy of one gene made only in one tissue — the white adipocyte — and yet every feature of the metabolic syndrome is present in these severely affected individuals. What's happening? Well, what's happening is as follows. Here's where perilipin-1 is. It sits on the surface of the lipid droplet in the white cell with its partner molecule, CGI58.
And when you're making fat… when you've eaten, your lipid droplet is concerned with making fat, and the breakdown enzymes, ATGL and HSL, are kept away from the lipid droplet. Then, when you fast overnight, the hormonal milieu of fasting, and indeed sleeping, induces hormonal changes in the body which induce phosphorylation changes within the adipocyte. The C-terminal phosphorylation site in perilipin-1 knocks off the binding site for CGI58, which then goes zipping around the surface of the… of the white fat droplet and grabs ATGL, the initiating lipase, and that starts to break down the triglyceride droplet. Then, at the same time, the second phosphorylation site, nearer the N-terminus of perilipin, binds hormone-sensitive lipase, the second initiat…
The second lipase. And then the beautiful cascade of lipolysis starts to break down the triglyceride, down to three individual fatty acids. So, that's happening in all of you, in everybody listening to this talk, in between their fed and their fasted state. And that is natural metabolic health. The unfortunate people who have this C-terminal perilipin-1 mutant have, as you see in the bottom here, with the green box… they've lost that phosphorylation site at the C…
At the C-terminal end of perilipin-1. And throughout the day — irrespective of whether they've been fed or fasted — CGI58 is free to roam and… and binds promiscuously to ATGL. And therefore, throughout the day, whether you're fed or not, your fat cell is breaking down triglycerides and releasing molecules of fatty acid. Now, remarkably, this in itself is enough to cause every single feature of the metabolic… the metabolic syndrome: Insulin resistance, fatty liver, atherosclerosis. This is a rare condition, but I think one which is really illustrative of the very important point that the health of your white fat cell — and indeed the health…
The healthy dynamic of your white cell fat droplet — is really crucial to your overall, organismal metabolic health. Is this, again, more relevant… outside these rarities? And is it… is this sort of process relevant to the wider population of people with the metabolic syndrome? Well, to address this, we collaborated with our colleagues in the MRC Epidemiology Unit — whose names are listed here — led by Nick Wareham. And we asked about genetic variation influencing lipodystrophy-like metabolic phenotypes in the general population. So, we looked at people who had high insulin, low HDL/cholesterol, and high triglycerides as a composite index. And in nearly 200,000 people, we found 53 genetic variations in the genome that influenced those traits in the general population. And these are shown in the next slide. Now, on the far left of the slide, in red and blue and red, there are the 53 genetic loci that we found to associate — either red being positive, and blue being negative. And of course, that's how we defined our phenotypes, so of course you see the consistent red, blue, and red. Now, if you go to the further end of the slide, you should be able to see independent populations, where those SNPs are looked at against type 2 diabetes and coronary artery disease.
And of course, it's reassuring, scientifically, to find that those genes are associated, in totally independent populations, with these adverse outcomes, these… coronary artery disease and type 2 diabetes. What we found puzzling and exciting is when we looked in the middle of the slide. There, we took independent populations and said, are these variants, these SNPs, associated with more body fat or less body fat? And rather remarkably, you see a lot of blue. And that lot of blues actually tells you that these individuals SNPs and these indivi… the genes, when put together, actually are associated with lower amounts of body fat, not higher amounts of body fat. And this is taken further into an independent analysis in people who have had compartmental measurements of their body fat using DEXA scanning, so we can actually measure the fat in the arms, trunk, etc. And again, looking at the genetic factors that put you at risk of developing these metabolic syndrome phenotypes, what we found quite remarkable was down at the bottom of the slide, here. We see that individuals carrying more of these risk alleles actually had less fat… less fat in the gynoid distribution, i.e., buttocks and thighs, less fat in the legs, and really, not very much difference in the places we thought would be really important, the visceral fat…
Only a little bit of an increase in the… in the visceral fat. This is… the changes in body fat are driven much more by lower body fat in the legs and thighs. And indeed, when you take these SNPs and using bioinformatic tools ask where these variants are expressing their actions most, you see, quite remarkably, that they do so in adipocytes and adipose tissue. In other words, we chose these variants in a completely hypothesis-free way. We didn't ask the adipose tissue, was it abnormal? The adipose tissue spoke to us spontaneously. And then, when we take some of those variants and manipulate them, we can show that manipulating these genes influences adipose tissue fat accumulation. So, in a further body of work led by Luca Lotta and Nick Wareham, they asked I think what is a very interesting and rather piquant question. And the question really is, is it worse to have a small bottom or a big belly? And they separated out those variants. In red are the variants that…
That cause… that are selectively associated with a waist ratio. And in blue are the variants associated with only hip ratios. So, you can see in the bottom there that fasting insulin, a measure for insulin resistance, is much more strongly associated with the hip SNPs rather than the waist SNPs. And then, when it comes to the bottom-line data, type 2 diabetes risk is greatly influenced by hip, much more than waist, circumference.
So, what's actually determining these risks of diabetes is not the big belly; it's the absence of fat on the buttocks and legs, much more so than the larger belly. They're much more equal when it comes to coronary artery disease, but for diabetes it's the absence of fat in buttocks and thighs that's much bigger… having a much bigger influence than the positive presence of fat in the visceral deposits. So, that really led me to develop a kind of useful model, an explanatory model, or at least an illustrative model, for what's going on. So, you can think of this as a… as a bathroom in a hotel you happen to go into that… somebody's left the plug out, and somebody's left the… left the tap on. But that's fine, because the bathroom is…
You know, it's an old B&B. It's got a [unknown] carpet, and the carpet's perfectly healthy because the flow is fine. You've reached a steady state, and you've got a perfectly healthy, non-soggy carpet. But we often think of metabolic disease as being due to too much energy, i.e., pouring too much in, or a decreased energy out, restraining how much we let out. And of course, then, you overfill your bath and you get a terrible, messy situation. And that's kind of how we tend to think about metabolic disease. I think what our data, both in lipodystrophy and now… and al… now, in the common forms of insulin resistance, is showing that… that really we haven't paid enough attention to the size of the bath. In other words, as well as how much energy we take in or how much energy we expend, we differ between us in our abilities to safely expand our adipose depots, so that some individuals become severely insulin resistant and diabetic at rather low levels of energy imbalance whereas others are capable of tolerating a large amount of obesity without developing any metabolic complications.
So, here's one example of what one might be able to do in individuals who've got a small bath. We look after, in our clinic, patients with severe insulin resistance and lipodystrophy. Here's an example of one such patient and a DEXA scan. This person has very low fat in their… in their legs and limbs compared to their central depots. They have severely uncontrolled diabetes on hundreds of units of insulin a day. And what we've done in this individual is undertake bariatric surgery. But this individual is only very modestly obese — he wouldn't be really defined as severely obese — but has got a really poor ability to handle any excess calories. Now, what bariatric surgery does is it chronically suppresses food intake, largely through suppressing appetite and changing the signaling to the brain. And just within a few months… here's April, just a few months after the surgery in December. That individual has lost a large amount of their central body weight. But effectively, what they've done is turned the tap off on their small bath.
And rapidly, they have… their diabetes has gone away, effectively. They're off… they were on hundreds of units of insulin. They're now on no units of insulin. So, individuals with this very limited capacity to store adipose tissue safely are particularly bet… have great benefit from bariatric surgery, even, indeed, more so than individuals with severe obesity. And now, this has been spun out in much broader clinical studies to see how… how widely applicable this is in partial lipodystrophy. We're now moving, in our next scientific phase, away from, if you like, vague hand-waving about these sorts of genes that influence adipose mass, trying to drill down and find precise molecular regulators of fat cells, particularly so that…
If we can identify those, then perhaps they might be therapeutically manipulable. We might be able to use them as therapeutic targets to increase our safe fat cell storage. This is just an overview slide showing that, in using UK Biobank, we can find rare nonsynonymous variants that are strongly associated with waist hip ratio. Because Biobank is now so large, and the sequencing and SNP data is so extensive, we can really start drilling down into individual genes. Here, for example, are some variants in the serine kinase Alk7, encoded by the gene ACVR1C. The two on the right are actually missense variants. They're truly genes that influence… genes that… variants that cause changes in the amino acid sequence in the… in the genes. And looking at the two of them here, N150H is probably a bit less interesting, because it's not so conserved across species.
But I195T is highly conserved across many, many species. And if you look at what I95T does, it has a profound structural impact on the putative functions of Alk7. It influences the… the so-called GS domain, here in green, and really would prevent it flipping away when the ligand binds to this receptor, and prevent the access of the next downstream signaling molecules, the SMAD… prevent access to the… to the catalytic loop. So, when we went ahead and reconstituted this mutant — a very talented postdoc in the lab, Nuno Rocha, did this — and really wanted to see whether this variant truly does influence signaling. And what Nuno showed, in red here, is that this variant, which is associated, as I mentioned, from protection… from protective phenotypes, a low waist hip ratio and protection from type 2 diabetes, actually is a loss of function allele compared to a constitutively active, in gray, and one of the other variants, in yellow, which doesn't seem to have any effect, but the red is our 195T variant, which is markedly impaired in signaling compared to our wild type.
So, this is surprising, because in mice, at least, inactivation of Alk7 has been reported to enhance lipolysis. So, we're further working, now, in human adipocytes to see if this finding is different in humans. But I think this data provides evidence that large-scale human sequencing efforts are now empowering the study of human coding variations and their links to phenotype. And these rare missense and nonsense variants can now be pretty unequivocally associated with human phenotypes. And these were provide an… play an increasingly important role in drug target validation. So, does insulin resistance always affect all insulin tissues equally? And the answer to that is, no. Because glucose is handled in different ways in different tissues. There are, for example, tissues where… where insulin is not required, particularly, to… to get rid of glucose: the brain, the kidney, hemopoietic tissues. But two of the key target tissues I mentioned to you before are liver and skeletal muscle.
Actually, insulin works in different ways in these two tissues. In liver, insulin reduces hepatic glucose output, reduces glycogenolysis and gluconeogenesis. But it does so through post-translational modification, by inducing transcriptional events, and, indirectly, through substrate delivery from fat and muscle. Insulin action in skeletal muscle happens in a very different way. There, you get very rapid stimulation of glucose uptake. And here, preformed GLUT4-containing vesicles sitting underneath the plasma membrane rapidly flip to the surface… plasma membrane surface in response to insulin. And so, there are some key differences in how insulin works in… in these two tissues. And this is illustrated by a rare family that we discovered a few years ago, in which… in comparison to most people with insulin resistance, reflected in the blue line above, who have high fasting insulin and high insulin after a meal, these two siblings had extraordinarily high insulin postprandially but normal fasting insulin. So, how could that be? Well, that was because they had a process, and a mutation affecting a process, that only occurs in skeletal muscle and, in part, in fat.
And that is the translocation of GLUT4. And there's a key molecule, a so-called Rab GAP, TBC1D4, otherwise known as As160, which is a key regulatory step in regulating how insulin translocates GLUT4 to the plasma membrane. And individuals who are heterozygous loss-of-function for this mutation have a continuous leak of GLUT4 to the cell surface so that, actually, when insulin comes to act, there's inadequate intracellular GLUT4 vesicles, and you get a severely impaired postprandial response. Again, we looked for many years and could never find a second family, so this is an exceptionally rare albeit very interesting rarity. But in the last couple of years, it's become fascinating that if you look elsewhere, for example in the Greenland Inuit, the allelic frequency of a very similar mutation is 17%.
A very large number of Greenland Inuits carry a very similar mutation. And indeed, in those, you can see that those that carry two copies of that have a much higher 2-hour glucose than individuals who carry one or no copies of that… of that. So, a rarity in the UK has become a phenomenon of epidemiological proportions in another population. And that's another lesson I think we're learning in this field: the importance of studying rare population isolates, and how illuminating they can be, and how different from each other the architecture of genetic variation causing disease can be between these populations. So, the carriers here, interestingly, also have marked postprandial hyperinsulinemia. And many people have thought that if you had, for example, impaired insulin signaling in skeletal muscle and had very high insulin, that might put your weight up. These individuals have no difference in body weight and, indeed, no differences in circulating lipids, making us think again about what hyperinsulinemia might actually be doing to things like body weight and lipids. So, I'm gonna end with an overview, really, of how organs work together.
The wonderful thing about endocrinology and metabolism is they're really not conditions of cells; they're conditions of systems. And you have to understand how the system works to really get a true understanding and true knowledge of disease pathophysiology. So, if you were gonna avoid getting type 2 diabetes, how might you go about it? Well, first of all, you'd maintain energy balance. You wouldn't let your energy in get more than your energy out. And that, as we've discovered from our genetic studies, it… not discussed with you today, but in other contexts, is really a job of the brain and the gut-brain interface. So, the brain is really the most important organ determining who is perpetually hungry, who is easily satiated — the variation between individuals in energy balance. The next step is that of the… to store… if you… if you can't manage to stay in energy balance and you do accumulate an energy imbalance, then you're fine so long as you can safely store that in fat. So, adipose tissue has really turned out to be a very important…
And not just a passive store of lipid but a very important determinant of how we… how our metabolic balance or imbalance can be main… can be maintained. If we can't store lipids safely in our adipose tissue and it goes to liver and muscle, then a key factor there is whether that lipid produces insulin insensitivity of glucose metabolism. There, we find, at the moment, fewer genetic variants that determine that factor. That may be a methodological issue. I'm sure these processes are key.
But some individuals become insulin resistant, some becomes severely insulin resistant, and yet some severely insulin resistant individuals don't get type 2 diabetes. And in work, again, I haven't discussed today, the key transition point between those individuals who are insulin resistant and those who then decompensate to diabetes is the genetic architecture of your pancreatic beta cells. Work from colleagues in Oxford, Exeter, and elsewhere around the world… have really done beautiful science showing how important the genetic health of your pancreatic beta cells is to determine whether you transition to type 2 diabetes or not.
I'll return to the immune system and the host defense. Much work in mice has really illum… illustrated a critical role, and suggested a critical role, for the immune system in maintaining metabolic health. And yet, much of our genetic work and, indeed, very much pharmacology really hasn't come up with that much human evidence that this is terribly important. I mean, for example, if you look, as Bernstein's lab did a few years ago, at the… at the genetic SNPs… at single nucleotide polymorphisms associated with a variety of inflammatory and a variety of metabolic phenotypes, there's really very little overlap.
There's almost a complete separation between the inflammatory diseases and the metabolic diseases. And in the same paper, looking in a different way at which tissues might be implicated through… through genomics and transcriptomics, you can find, for example, in the top left-hand corner here, Bernstein's lab is looking at neurological conditions and showing that they cluster around brain tissue — not surprisingly. And then a whole range of immune conditions. Very, very wide… widespread immune conditions. And of course, again, unsurprisingly, the genetic variation influencing them is having its major effect on immune cells — on lymphocytes and macrophages. Again, like with the SNPs and the disease associations, metabolic disorders such as high triglycerides, low HDL, fasting glucose, etc, tend to cluster away from the immune phenotypes. So, I would say that it's still an interesting hypothesis that our host defense and immune system is critically important for metabolic health, but we still need more and more detailed genetic investigation, and more pharmacological proof, before those undoubtedly beautiful murine experiments are seen as deeply relevant to human disease.
So, finally, how can we use this information to prevent insulin resistance. We are… at the moment, we know that we need to reduce calories in. And that's easy to say. We can tell people to lose weight and reduce food. But sometime, now, we get to a stage where drugs will help. And there is an… and there are exciting developments in anorectic drug developments, and I think this is likely to get more effective. We can tell people to exercise more, increase their expended calories. There's much work going on trying to develop drugs to increase brown adipose tissue thermogenesis, for example , or exerc… act as exercise mimetics, but they're some way behind the development of drugs for suppressing food intake. What about making our bath bigger? Can we increase safe storage of calories in adipose tissue? Well, actually, one class of drugs, one of which is still available, has proved the concept.
Thiazolidinedione drugs do indeed improve insulin sensitivity while at the same time increasing your fat mass. Sadly, with some of the drugs, there are other safety issues. But I think we have got proof that if we can cleverly target adipose tissue to make it a safer store for our fat, that is another route towards rendering individuals metabolically healthier. So, I'd like to finish by thanking my colleagues and collaborators, our referring physicians, our funders, but particularly, our patients, participants, and families, without whom we would not have had a research program.