Lipid Droplets

Part 1: An Introduction to Lipid Droplets

00:00:15.08 Hello, I'm Bob Farese, Jr.
00:00:17.10 And I'm Toby Walther.
00:00:19.01 And we run a joint laboratory
00:00:21.00 at the Harvard School of Public Health
00:00:22.26 and Harvard Medical School,
00:00:25.04 and the Howard Hughes Medical Institute in Boston, Massachusetts.
00:00:27.09 And today, we'll tell you about a fascinating organelle
00:00:29.23 called the lipid droplet.
00:00:32.03 As you all know, life is defined
00:00:34.29 by an open equilibrium of energy
00:00:37.05 that flows through the biological system
00:00:39.01 and allows the system to maintain its structure
00:00:41.03 and to perform work.
00:00:42.15 But, of course, energy availability
00:00:44.14 and the need to expend energy
00:00:46.16 varies greatly over time.
00:00:48.06 And as a consequence,
00:00:50.02 pretty much all biological systems
00:00:52.03 have evolved mechanisms to temporarily store metabolic energy.
00:00:55.26 And they do so primarily, as you can see on the slide,
00:00:59.00 in the form of lipids,
00:01:01.17 that contain most of the metabolic stores in a human,
00:01:05.04 for instance.
00:01:06.23 The major lipid that most organisms store energy in
00:01:09.28 is in triacylglycerols or triglycerides.
00:01:12.26 And they are the universal currency of energy storage
00:01:16.14 because, one, they're energy dense;
00:01:18.03 two, they are highly reduced and available for oxidation;
00:01:22.24 and three, they don't require water for their storage.
00:01:26.08 In fact, as we all know,
00:01:28.03 oil and water don't mix,
00:01:29.14 so oils are partitioned separately from water,
00:01:31.22 as you can see in the movie on your left.
00:01:36.17 As a consequence of these somewhat unique properties,
00:01:39.09 lipid droplets were about the first organelles
00:01:41.25 to be discovered.
00:01:43.23 Because oil and water diffract light,
00:01:45.18 Richard Altmann was able to observe lipid droplets,
00:01:48.19 in this case, in frog liver cells,
00:01:50.11 where you can see the densely stained lipid droplets.
00:01:53.09 This is a really interesting book to look at,
00:01:56.07 by Richard Altmann.
00:01:57.27 He's also the person who discovered by bioblasts,
00:02:00.13 later termed mitochondria,
00:02:02.05 and termed... and coined the term nucleic acid or nucleinsäure.
00:02:07.19 So, lipid droplets are ubiquitous organelles,
00:02:11.16 but they're also highly unusual organelles.
00:02:13.13 As you can see in the electron micrograph
00:02:16.19 on the left from a hepatoma cell,
00:02:18.08 lipid droplets are bounded by a monolayer
00:02:20.29 instead of a bilayer,
00:02:22.13 which is typically what organelles
00:02:24.21 such as the endoplasmic reticulum mitochondria, or the nucleus,
00:02:29.02 which are also shown in the micrograph, have.
00:02:31.19 On the right is a schematic of lipid droplets
00:02:33.14 that shows their composition.
00:02:35.27 So, these organelles have a neutral lipid core,
00:02:39.20 which typically is sterol esters
00:02:41.27 or triglycerides or sometimes waxes,
00:02:44.00 wax esters.
00:02:45.15 And those neutral lipids are bounded by a phospholipid monolayer.
00:02:50.09 And the phospholipid monolayer is decorated
00:02:53.06 by specific proteins that number in the 10s to 100s
00:02:56.25 and are typically involved in lipid metabolism.
00:03:02.18 Now, the presence of lipid droplets
00:03:05.09 converts cells to emulsions.
00:03:06.25 An emulsion is a system where more than one phase
00:03:09.17 is present at the same time.
00:03:11.06 And in this case, the cell becomes an emulsion
00:03:13.04 because lipid droplets form a dispersed phase
00:03:15.19 in the continuous phase of the cytosol.
00:03:18.27 Because such a system is always metastable,
00:03:21.09 and there's a tendency for both phases
00:03:23.21 to just separate out from each other,
00:03:26.05 these is emulsifier or surfactant required
00:03:28.18 to lower the surface tension
00:03:30.13 and prevent all the droplets from coalescing together.
00:03:34.13 Now, as I mentioned, lipid droplets are ubiquitous.
00:03:37.11 They're found in almost all eukaryotic cells,
00:03:39.24 and they're even found in some prokaryotes.
00:03:42.09 These are some examples of lipid droplets
00:03:44.27 in different cell types,
00:03:46.24 and here we're visualizing lipid droplets
00:03:48.29 with a neutral lipid fluorescent dye,
00:03:50.29 such as BODIPY.
00:03:52.10 And as you can see,
00:03:54.06 there are some bacteria that have lipid droplets.
00:03:56.14 In this case, we're showing Rhodococcus,
00:03:58.12 but also mycobacteria have lipid droplets.
00:04:00.09 And also shown here are yeast cells,
00:04:02.08 with their typical 4-8 numbers of lipid droplets,
00:04:05.04 fly cells with lipid droplets,
00:04:06.25 or mammalian cells with lipid droplets.
00:04:10.04 In multicellular organisms,
00:04:12.12 often lipid storage is separated into a dedicated tissue.
00:04:17.19 For instance, in humans,
00:04:19.15 adipose tissue stores most of the triacylglycerol
00:04:21.24 and neutral lipid for energy stores.
00:04:24.11 Importantly, though,
00:04:26.01 pretty much all other cells in the human body
00:04:28.08 have the capacity to make lipid droplets
00:04:30.17 and store energy temporarily.
00:04:32.11 For instance, after a meal, the intestine stores, temporarily,
00:04:37.17 lipids and lipid droplets,
00:04:38.26 here shown in blue.
00:04:40.15 The liver takes up lipids from circulation
00:04:42.29 and from adipose tissue, actually,
00:04:46.24 and temporarily stores those lipids,
00:04:49.25 now shown in red here.
00:04:52.03 These are some of the most metabolically active organs.
00:04:54.28 It's also important to remember, for instance,
00:04:57.05 that the heart runs predominantly on fatty acids as a fuel,
00:05:00.06 and in some instances lipid droplets
00:05:02.15 become very prevalent, when, for instance,
00:05:04.25 heart function is compromised.
00:05:06.15 There are also a number of really fascinating biology problems
00:05:09.17 and processes that involve lipid droplets,
00:05:12.22 particularly the secretion of lipid droplets
00:05:14.24 in an unconventional mechanism,
00:05:16.18 in the mammary epithelium to make milk,
00:05:18.24 which of course has a lot of fat in it.
00:05:21.27 Now, as Toby mentioned, lipid droplets
00:05:25.05 are part of normal metabolism
00:05:27.04 and, in particular, lipid and triglyceride metabolism.
00:05:29.04 But a proper balance of lipid droplets
00:05:31.18 in cells and tissues is required,
00:05:33.09 because, as we all know,
00:05:35.16 basically obesity constitutes a state
00:05:38.02 where there's an excess abundance of lipid droplets
00:05:40.27 in the organism.
00:05:42.16 And conversely, lipodystrophy, or the lack of fat,
00:05:45.11 is a situation in which lipid droplets are scarce or absent.
00:05:48.29 And importantly, both of these conditions
00:05:51.16 are associated with insulin resistance
00:05:54.03 and diabetes and illness.
00:05:57.22 Now, the molecular mechanisms
00:06:00.03 that link the overaccumulation of lipids to disease
00:06:02.26 are still under intense investigation.
00:06:05.10 Generally, it is thought that excess lipids
00:06:08.03 flow over into tissues
00:06:10.04 where they're normally not very abundant.
00:06:12.06 This includes, for instance, the liver, muscle,
00:06:14.16 the pancreas, or the heart.
00:06:17.02 In those cases, it is thought that the lipids
00:06:19.21 -- the excess lipids --
00:06:21.12 interfere either directly or indirectly
00:06:23.03 with the normal physiological function of the tissue,
00:06:24.28 and this leads, then, to a form of toxicity.
00:06:29.14 This has led to a term known as lipotoxicity,
00:06:32.18 first coined by Roger Unger in the '80s.
00:06:36.02 In some cases, as Bob just mentioned,
00:06:38.14 this then results as an end effect in diabetes,
00:06:41.05 for instance.
00:06:43.07 In other cases, cardiovascular diseases,
00:06:44.28 such as cardiomyopathy and atherosclerosis,
00:06:46.28 are the consequence of overaccumulation of lipids and lipotoxicity.
00:06:52.05 Now, lipid droplets and triglyceride storage in lipid droplets
00:06:57.24 are also important in other processes --
00:06:59.18 for example, they have industrial importance.
00:07:01.01 So, for example, in agriculture,
00:07:02.27 seed oil production is a very important
00:07:07.15 aspect of lipid droplet storage,
00:07:09.03 and one that is...
00:07:11.10 that is under intense investigation for optimizing oil production.
00:07:13.29 Similarly, for renewable energy,
00:07:16.07 biofuel production essentially relies on the enhanced production
00:07:21.01 of triglyceride storage in lipid droplets.
00:07:24.25 As we told you, lipids and water don't mix,
00:07:27.13 and the formation of droplets
00:07:29.21 is a biophysical process.
00:07:31.23 However, cells have evolved
00:07:34.11 very complicated machinery
00:07:36.14 to synthesize lipids,
00:07:38.24 organize the formation of lipid droplets in specific sites,
00:07:41.29 and to regulate that process.
00:07:43.27 This is an intense area of investigation,
00:07:46.18 and if you want to learn more about it,
00:07:48.17 please stay tuned.

Part 2: Mechanisms of Lipid Droplet Formation

00:00:15.10 Hello, I'm Toby Walther.
00:00:17.04 And I'm Bob Farese, Jr.
00:00:19.10 And we run a joint laboratory at the Harvard School of Public Health
00:00:22.12 and Harvard Medical School.
00:00:24.05 And in this lecture,
00:00:26.16 we're gonna discuss some of the mechanisms and physiology
00:00:28.23 of lipid droplet formation.
00:00:31.14 As we mentioned in our introductory lecture,
00:00:33.28 cells have this remarkable ability
00:00:36.15 to organize the formation of lipid droplets
00:00:40.24 within the cell
00:00:42.26 in a dispersed and regular manner.
00:00:45.06 And this raises a lot of questions,
00:00:47.10 like, how do cells do this?,
00:00:48.27 what protein machinery is involved in the formation of lipid droplets?,
00:00:53.23 how do proteins target to the surfaces of these organelles?,
00:00:58.01 and what are the functions of these organelles
00:01:01.11 in cells and in organisms?
00:01:03.25 To begin with, in this lecture
00:01:07.07 we're going to show an animation made by Janet Iwasa
00:01:10.27 that illustrates a picture of how we believe lipid droplets form.
00:01:14.17 Here you can see the endoplasmic reticulum,
00:01:16.16 the site of neutral lipid synthesis,
00:01:18.29 and you see fatty acyl CoA's
00:01:21.01 and diacylglycerol substrates
00:01:23.03 that are being covalently joined
00:01:25.13 through the actions of DGAT enzymes
00:01:27.15 -- DGAT1 and DGAT2 --
00:01:29.10 to form triglycerides.
00:01:30.29 These triglycerides phase separate within the plane of the bilayer,
00:01:36.01 and accumulate via Ostwald ripening,
00:01:39.14 and then bud towards the cytosolic surface of the ER
00:01:43.25 to form a droplet.
00:01:45.17 You can see that happening here.
00:01:47.16 And as the lipid droplet grows,
00:01:49.21 the surface angle between the lipid droplet
00:01:53.14 and the ER changes,
00:01:55.14 and eventually the droplet will bud.
00:01:57.12 And the question is, how does this process happen?
00:01:59.18 And it has been our hypothesis, and others',
00:02:02.12 that proteins must be involved in the formation of these organelles.
00:02:08.02 Now, at least conceptually,
00:02:09.23 you can take apart that complex set of reactions
00:02:12.12 into individual steps.
00:02:14.07 Those don't need to necessarily occur sequentially in time,
00:02:16.25 but they can be separated for us to analyze them
00:02:19.20 one by one.
00:02:21.09 In the first step, as Bob just mentioned,
00:02:23.10 triglyceride synthesis occurs,
00:02:26.00 and the initial lens is formed.
00:02:27.25 In the next step,
00:02:30.04 we will discuss how, initially,
00:02:32.26 the lipid droplet grows and buds specifically into one direction,
00:02:35.29 towards the cytosol.
00:02:37.20 In another step,
00:02:40.21 the proteins are recruited specifically
00:02:42.29 to the surface of lipid droplets.
00:02:45.06 And when they're there,
00:02:46.21 they often catalyze reactions
00:02:49.01 that allow the expansion of the lipid droplet
00:02:50.20 in a last step.
00:02:53.14 So, let's start with step 1 of lipid droplet formation:
00:02:56.20 triglyceride synthesis and lens formation,
00:02:59.13 where triglycerides gather within the plane of the bilayer.
00:03:02.25 To begin with, in this step,
00:03:04.29 we have to go through the pathway of glycerolipid synthesis
00:03:08.10 that was discovered by Eugene Kennedy and co-workers
00:03:11.24 and reported in 1960.
00:03:14.07 And this is the major lipid synthesis pathway
00:03:16.13 that gives rise to both phospholipids
00:03:18.07 as well as the neutral lipid triglyceride.
00:03:20.20 In this pathway,
00:03:22.19 which also occurs in the endoplasmic reticulum,
00:03:24.18 fatty acids, which you can see on the far left of this slide,
00:03:28.08 are covalently linked, in a reaction requiring ATP,
00:03:31.28 by ACSL's, or acyl-CoA synthetases,
00:03:35.06 to form activated fatty acids called fatty acyl-CoA's.
00:03:39.15 These fatty acyl-CoA's, then,
00:03:42.06 are covalently linked to a glycerol backbone
00:03:44.17 in a series of enzymatic steps
00:03:46.21 to give rise to phosphatidic acid or diacylglycerol.
00:03:51.07 And those steps are catalyzed
00:03:53.05 by a series of enzymes such as GPAT enzymes
00:03:55.19 -- glycerol phosphate acyl transferases;
00:03:59.25 AGPAT enzymes -- acyl glycerol phosphate acyl transferases;
00:04:02.16 and the phosphate group is removed from phosphatidic acid
00:04:05.11 to form diacylglycerol by PAP,
00:04:08.13 or phosphatidic acid phosphatase.
00:04:11.11 Now, phosphatidic acid and diacylglycerol
00:04:15.04 can give rise to phospholipids,
00:04:17.17 but in the final step of making triglycerides,
00:04:20.06 diacylglycerol is covalently joined
00:04:22.21 to an acyl-CoA to form triacylglycerol
00:04:25.18 through the actions of the DGAT enzymes.
00:04:28.16 So, as I mentioned, the DGAT activity
00:04:30.27 was described around 1960,
00:04:33.24 but it wasn't for another 40 years
00:04:35.28 that we began to understand the molecular processes
00:04:38.15 of making triglycerides.
00:04:41.10 So, at the end of the 1990's,
00:04:45.11 we and folks at Calgene
00:04:49.21 identified the enzymes that make triglycerides,
00:04:51.17 and those are now known as DGAT1 and DGAT2.
00:04:54.25 So, this is fascinating.
00:04:56.16 We have two enzymes,
00:04:58.04 and these two enzymes have convergent evolution
00:05:02.09 in that they both catalyze the same biochemical step,
00:05:04.27 but they evolved separately
00:05:07.03 and are part of different gene and protein families.
00:05:09.24 So, DGAT1 is part of the MBOAT family.
00:05:12.05 DGAT2 has its own family
00:05:14.25 that constitutes DGAT's and MGAT
00:05:17.00 and wax synthases.
00:05:18.25 Both of these enzymes
00:05:20.23 are found in the endoplasmic reticulum,
00:05:22.17 but as you can see, as cartooned here,
00:05:25.02 they have very different topologies.
00:05:27.05 DGAT1 is a... is a polytopic membrane protein
00:05:30.07 with multiple transmembrane domains
00:05:32.10 and is a very hydrophobic protein.
00:05:35.06 DGAT2, on the other hand,
00:05:37.03 has one embedded transmembrane domain
00:05:39.01 that anchors it to the endoplasmic reticulum,
00:05:42.10 as diagrammed here.
00:05:46.06 Both enzymes have been the subject of
00:05:49.22 pharmaceutical inhibitor development.
00:05:51.09 And in fact, there are highly specific
00:05:53.16 and highly effective inhibitors
00:05:55.12 for both the DGAT1 and DGAT2 enzyme
00:05:57.24 that are useful both in...
00:06:00.05 for laboratory tools
00:06:04.10 and have been taken into clinical trials.
00:06:08.11 Now, currently, we only have an understanding
00:06:10.21 of the molecular mechanisms of triglyceride synthesis
00:06:13.23 catalyzed by DGAT1.
00:06:15.28 Specifically, recently,
00:06:18.18 we were able to solve the structure,
00:06:21.03 the molecular structure of this enzyme.
00:06:23.28 And as you can see here by cryo-electron microscopy,
00:06:25.29 what has become apparent is that DGAT1 generates...
00:06:29.25 is present as a dimer in the endoplasmic reticulum membrane,
00:06:34.18 and... where most of the protein
00:06:37.08 is really within the bilayer of the membrane.
00:06:40.01 It turns out that the enzyme forms, like,
00:06:43.11 a butterfly-shaped dimer
00:06:45.23 that is very intimately crossing-over
00:06:49.19 and linked between the two subunits.
00:06:52.00 When you look into one of the subunits,
00:06:53.26 it is fascinating to see how we now believe
00:06:56.14 the molecular mechanism of the triglyceride synthesis occurs.
00:06:59.25 The enzyme has a channel
00:07:02.27 that reaches from the cytosolic side
00:07:04.20 all the way through the membrane
00:07:06.12 to the luminal side.
00:07:08.27 In addition, there is a lateral cavity,
00:07:11.17 or a gate,
00:07:13.25 towards the plane of the membrane.
00:07:15.22 Deeply within the enzyme,
00:07:17.20 and deeply within the plane of the membrane,
00:07:20.02 there is the putative catalytic site.
00:07:21.24 When we look at that
00:07:25.04 in a structure that also contains substrates
00:07:27.10 and a density that most likely
00:07:29.23 is the acyl acceptor diacylglycerol,
00:07:32.21 what becomes apparent is that the acyl-CoA substrate
00:07:35.28 reaches the active site from the cytosolic side
00:07:39.11 either directly from the cytoplasm
00:07:41.10 or maybe laterally from the membrane,
00:07:43.18 and positions the carbonyl group
00:07:47.12 that is used for the esterification
00:07:50.11 right next to the acyl acceptor,
00:07:52.08 the diacylglycerol-like density,
00:07:54.16 right next to the active site.
00:07:56.21 That second substrate reaches through the lateral gate
00:08:00.20 directly to that hydrophobic core
00:08:03.00 in the middle of the enzyme.
00:08:04.24 We now think that in the next step,
00:08:06.20 the triacylglycerol is released,
00:08:08.28 possibly directly into the memory through that lateral gate,
00:08:11.18 initiating the next step,
00:08:14.05 initial lipid drop... lipid droplet growth
00:08:16.07 and lens formation.
00:08:17.29 So, as we begin to have a molecular understanding
00:08:19.26 of how triglycerides are made
00:08:21.28 and released into the membrane,
00:08:23.15 this brings us to the second step of lipid droplet formation,
00:08:25.20 and that is,
00:08:27.21 how do triglycerides phase separate
00:08:30.06 within the plane of the bilayer
00:08:31.26 and then get converted into an initial budding lipid droplet?
00:08:36.14 Now, this is a tough problem to tackle,
00:08:39.22 and we've had a lot of luck
00:08:41.23 taking systems-type approaches
00:08:43.09 to identify the machinery of lipid droplet formation.
00:08:45.22 What's shown here is a screen
00:08:47.24 that we recently performed
00:08:49.18 in human macrophage cells
00:08:51.06 that were loaded with both cholesterol and triglycerides
00:08:54.04 to form lipid droplets,
00:08:55.12 and then we used RNAi
00:08:58.02 to knock down the genes in the genome
00:09:00.06 and determine which of these genes
00:09:02.12 was required for normal droplet formation.
00:09:04.20 And from a screen like this,
00:09:06.10 we obtained, as you can see,
00:09:08.08 500 hits for... that are involved in lipid droplet biology.
00:09:11.28 Now, from image analysis,
00:09:14.22 we can classify the characteristics
00:09:17.11 of how these different gene inhibitions
00:09:21.07 affect lipid droplet biology,
00:09:23.18 and then bin them out into different phenotypes, as you can see.
00:09:26.21 So, we have some classes
00:09:29.10 where there are very small and dispersed droplets,
00:09:31.23 and other classes where the droplets are quite large and clustered,
00:09:34.05 for example.
00:09:36.06 Some examples of these different classes
00:09:38.06 can be seen in these images,
00:09:40.01 where, again,
00:09:41.26 we have a variety of very strong lipid droplet phenotypes
00:09:44.01 caused by individual gene knockdowns.
00:09:46.20 Some of the genes that cause these different phenotypes
00:09:49.25 are shown on the right, here,
00:09:51.21 and obviously, today we can't discuss
00:09:54.01 many of the hits from the screen.
00:09:55.15 I will say, though,
00:09:57.22 all these hits will be made publicly available
00:09:59.16 on a lipid droplet platform
00:10:01.25 that will be launched very soon.
00:10:04.09 One of the hits that we would like to talk about today
00:10:06.25 is this one called BSCL2 or seipin.
00:10:10.04 BSCL2 is the gene that encodes for the seipin protein,
00:10:12.28 and this hit was one of the strongest hits in our screen,
00:10:16.06 and of course attracted our interest
00:10:18.00 because we and others
00:10:21.22 have identified seipin as a very important protein
00:10:24.17 in lipid droplet formation.
00:10:26.19 So, seipin... what is seipin?
00:10:28.13 seipin is an ER protein,
00:10:30.20 as shown in the cartoon on your left,
00:10:32.22 which has N- and C-termini
00:10:34.14 that are oriented towards the cytosol.
00:10:36.18 It has two transmembrane domains
00:10:39.02 and an evolutionarily conserved loop
00:10:42.15 that is oriented towards the lumen of the ER.
00:10:45.26 And seipin was identified in 2001
00:10:48.18 as the causative gene...
00:10:52.00 one of the causative genes for congenital generalized lipodystrophy,
00:10:56.15 which... an example of that syndrome,
00:10:59.07 where there is a lack of body fat,
00:11:01.19 is shown in the image on your right.
00:11:03.09 Now, over the years,
00:11:05.17 we and many other labs have worked on the seipin protein
00:11:08.05 because many labs have identified it
00:11:10.25 as being central to lipid droplet formation,
00:11:13.03 and some of those labs are listed on the bottom.
00:11:16.15 Some of the major findings
00:11:19.10 that are important for understanding
00:11:22.07 seipin's potential role in lipid droplet formation
00:11:24.13 are listed on this slide.
00:11:25.20 One is that the deletion of seipin
00:11:26.28 -- which was first identified as...
00:11:29.09 done in screens in yeast by the Goodman Lab
00:11:32.08 or Rob Yang's lab --
00:11:33.16 show that seipin deletion
00:11:35.25 results in supersized lipid droplets,
00:11:37.00 very giant lipid droplets found in the cells.
00:11:39.23 Joel Goodman's lab also originally showed
00:11:42.01 that seipin homo-oligomerizes
00:11:44.06 to form a large macromolecular complex.
00:11:47.00 And a number of labs also showed
00:11:50.26 that seipin localizes to ER-lipid droplet junctions in yeast.
00:11:55.26 Now, this slide shows a figure in which
00:11:59.06 we're showing you the cellular phenotype
00:12:01.08 for seipin deletion.
00:12:02.28 On your left, you can see that
00:12:04.29 lipid droplets form in these Drosophila cells
00:12:06.18 in an organized manner,
00:12:08.28 as we showed you in some of the previous videos.
00:12:11.10 We load with oleate, they form triglycerides,
00:12:13.05 and droplets form in a dispersed manner.
00:12:16.06 On your right is what happens
00:12:18.01 in seipin deficient cells.
00:12:20.05 And in this case, what we see is something completely different.
00:12:24.01 There are some pre-existing large lipid droplets,
00:12:25.29 but instead of having normal-sized lipid droplets,
00:12:30.05 what we see is a myriad of very small lipid droplets
00:12:32.27 that are found throughout the endoplasmic reticulum,
00:12:37.01 that look distinctly different than the normal-sized droplets
00:12:41.26 found in the wild-type situation.
00:12:43.29 Now, how does the protein actually do this?
00:12:46.02 An interesting early observation in our laboratory
00:12:49.02 was that when we actually look...
00:12:51.19 how the protein looks like,
00:12:53.19 it forms foci in the endoplasmic reticulum.
00:12:56.18 What you can see in the movie on the left
00:12:58.25 is a Drosophila cell, just like the one Bob just showed,
00:13:01.16 where we've genome-engineered the cells
00:13:04.26 to express a version of seipin
00:13:06.20 that has a GFP right at the end of a protein,
00:13:09.01 at its endogenous locus.
00:13:11.06 And what you can see is that this cell now
00:13:14.10 has these green foci,
00:13:16.02 which move rapidly through the endoplasmic reticulum
00:13:18.22 -- now shown in red here --
00:13:20.12 almost as if they were just gonna scan the ER
00:13:23.04 for initial lipid droplet formation.
00:13:25.27 And it turns out that this is a conserved feature
00:13:28.24 for seipin,
00:13:30.13 because it's not only forming punctae in Drosophila cells,
00:13:33.02 but as you can see on the right, here,
00:13:35.17 also in mammalian cells,
00:13:37.01 and the same is true in many other systems
00:13:38.19 where people have looked.
00:13:40.13 Now, what's the molecular nature of these foci?
00:13:42.26 Our current understanding
00:13:45.02 is driven by observations,
00:13:47.02 again, by cryo-electron microscopy,
00:13:49.04 where we have isolated and purified seipin,
00:13:52.27 in this case, from Drosophila cells.
00:13:54.28 And what you observe is that
00:13:58.04 seipin has sort of three parts to it.
00:14:00.13 There are two short extensions on the N- and the C-terminal part
00:14:04.22 that are both predicted to face the cytosolic side.
00:14:08.03 Each subunit also has two transmembrane domains
00:14:10.09 that span the ER membrane.
00:14:13.13 And then there is a large,
00:14:15.09 quite conserved luminal domain,
00:14:16.26 now shown in green in the cartoon.
00:14:20.18 In our cryo-EM structure, what we observe is
00:14:27.00 particularly that the luminal domain folds
00:14:29.27 into a very, very solid alpha/beta fold,
00:14:32.00 that resembles lipid-binding proteins,
00:14:34.07 that arrange into a ring structure
00:14:39.17 that contains 10, 11, or 12 subunits,
00:14:41.08 depending on the species.
00:14:42.29 You can also see that the transmembrane domains
00:14:45.18 are above that
00:14:47.11 and span the bilayer membrane,
00:14:49.17 but in the structure,
00:14:51.01 those were not resolved,
00:14:52.17 presumably because they were quite flexible.
00:14:55.16 Now, analyzing the structure of the folded domain
00:14:58.08 in the luminal side,
00:14:59.23 what we observed is two particular features.
00:15:03.12 One is that this ring really underlays
00:15:06.18 right under the membrane,
00:15:08.03 resembling what could be called a molecular washer
00:15:09.24 that basically anchors the whole machinery
00:15:14.00 -- you know, a very large protein complex at this point,
00:15:16.18 with, like, 12 subunits --
00:15:18.06 right under the ER.
00:15:20.02 The second feature that seems to be specific for seipin
00:15:22.20 in this class of related molecules
00:15:25.25 is that there is a helix
00:15:29.16 that is present in all seipin molecules
00:15:32.06 that have been described today
00:15:34.10 that is pressed and oriented right onto the membrane.
00:15:38.06 This hydrophobic helix, in isolation,
00:15:40.06 suffices to bind triglyceride,
00:15:42.14 leading us to the model that this complex
00:15:45.15 organizes the original...
00:15:47.04 the initial formation of lipid droplets.
00:15:50.05 Having said that, this model predicts, of course,
00:15:55.06 that one can isolate triglyceride in this complex
00:15:57.01 as it is formed.
00:15:58.21 And we've tried that and never succeeded with it.
00:16:01.03 There's a second puzzle here,
00:16:03.04 and that is that this hydrophobic helix
00:16:05.21 is highly evolutionarily conserved in sequence,
00:16:08.12 a feature that you would not necessarily expect
00:16:10.17 if the only role of this was to be hydrophobic
00:16:13.02 and interact with oil.
00:16:14.29 So, based on this, we had a puzzle,
00:16:18.02 and we hypothesized there may be
00:16:20.26 a missing component to this assembly complex.
00:16:22.20 And here is an experiment
00:16:24.28 that Jeeyun Chung in our lab carried out,
00:16:26.26 in which she did an immunoprecipitation,
00:16:29.06 or a pulldown experiment,
00:16:30.28 using either the wild type seipin,
00:16:33.02 which is shown on your left,
00:16:34.27 or a seipin molecule
00:16:37.18 in which she deleted this highly conserved hydrophobic helix,
00:16:39.19 which is shown on the right.
00:16:41.05 And in pulling down all the interactors,
00:16:43.05 she found one hit
00:16:48.01 that was pulled down specifically with the wild type protein
00:16:49.21 but not with the mutant protein that deleted this hydrophobic helix.
00:16:52.06 And that is very easy to see
00:16:55.11 in the upper right part of the figure, here,
00:16:57.15 and that gen...
00:17:00.06 the protein that was pulled down by the wild type
00:17:01.22 is a protein that was named TMEM159.
00:17:04.25 So, what is TMEM159?
00:17:07.01 Well, so...
00:17:12.03 we have now named TMEM159 lipid droplet assembly factor,
00:17:14.16 and so that name speaks to what we think...
00:17:17.27 its involvement in lipid droplet assembly.
00:17:21.03 So, not much was known about this
00:17:24.01 when we identified it as a seipin interactor.
00:17:25.22 TMEM159, or LDAF1,
00:17:27.17 is an endoplasmic reticulum protein...
00:17:29.27 161 amino acids.
00:17:32.14 Its topology suggests both the N- and C-termini
00:17:36.01 are towards the cytoplasm.
00:17:37.21 And the characteristics of the protein
00:17:40.17 suggest that it forms a double hairpin,
00:17:45.02 as we've modelled in this cartoon.
00:17:47.22 The gene and protein were originally identified
00:17:49.23 in murine fatty liver
00:17:51.10 that had been induced by PPAR-gamma
00:17:53.00 in a knockout model
00:17:55.07 where there was fatty liver.
00:17:57.23 And one of the things that caught our attention
00:18:00.00 when we did this pulldown
00:18:01.04 was it turned out that the knockdown
00:18:04.24 of TMEM159 or LDAF1
00:18:06.28 also was highly correlated with seipin
00:18:09.16 in the genome-wide screen that I showed you previously.
00:18:12.22 So, Jeeyun went on to purify LDAF1,
00:18:17.07 as I'll refer to it now,
00:18:20.08 as shown here, from mammalian cells.
00:18:21.22 And when she pulls down and purifies LDAF1,
00:18:23.22 she also copurifies seipin.
00:18:28.15 And these two proteins
00:18:30.19 now form a stoichiometric complex.
00:18:32.24 Importantly, as Toby mentioned before,
00:18:34.24 when we purified seipin alone before,
00:18:37.18 we were unable to copurify triacylglycerol with seipin.
00:18:41.15 But we made the discovery here
00:18:44.20 that when we purify the complex of both proteins together,
00:18:47.16 as you can see in the fourth from the left lane,
00:18:52.20 the complex pulled down in the Coomassie gel on the bottom
00:18:55.08 also shows the presence of triacylglycerol.
00:18:57.14 If we add oleic acid to the cells
00:18:59.20 and drive lipid droplet formation,
00:19:01.11 the complex purifies even more triacylglycerol,
00:19:04.00 as you can see.
00:19:05.19 And if we used DGAT inhibitors
00:19:07.24 in that situation,
00:19:09.09 we blocked the triacylglycerol.
00:19:10.29 So, this is all evidence to us that this complex,
00:19:13.05 together,
00:19:15.06 is interacting with triacylglycerol
00:19:17.23 during lipid droplet formation.
00:19:20.14 We went on to study LDAF1
00:19:23.27 in the context of cells,
00:19:25.22 and one of those experiments is shown here.
00:19:28.08 In this experiment,
00:19:29.26 what has been done is that
00:19:32.09 LDAF1 and seipin have been tagged,
00:19:34.16 using CRISPR technology,
00:19:36.04 at their endogenous loci with fluorescent proteins.
00:19:40.04 In addition, we tagged PLIN3,
00:19:42.23 which we have identified as one of
00:19:45.25 the earliest detectors of a forming lipid droplet.
00:19:48.00 And so, shown in this slide
00:19:49.28 are three examples of lipid droplets forming,
00:19:52.12 as detected by PLIN3 signal,
00:19:55.17 on the bottom set of...
00:19:58.27 in the bottom row of these panels.
00:20:00.15 And in all instances,
00:20:01.16 when we can first detect the lipid droplet formation,
00:20:04.11 it is occurring in the same spot
00:20:06.27 as where we have LDAF1 and seipin.
00:20:08.17 So, this gives us
00:20:11.11 more evidence that the seipin-LDAF1 complex
00:20:14.00 is the site where lipid droplets
00:20:16.27 are forming in these cells.
00:20:19.15 Now, if this model is correct,
00:20:21.14 there are a couple of simple predictions.
00:20:23.18 One of the strongest predictions is that,
00:20:26.06 if the complex determines where lipid droplets form,
00:20:28.27 we should be able to move the complex
00:20:31.11 to a site where it normally is not localized,
00:20:33.10 or where only a few of them are randomly,
00:20:35.10 and now lipid droplets should form,
00:20:37.27 preferably, at those sites.
00:20:39.29 In this experiment, we used a chemical biology trick
00:20:42.21 in which we've generated
00:20:45.07 an artificial chemically induced heterodimerizing version
00:20:49.12 of LDAF1
00:20:51.20 that can be bound to a plasma membrane anchor.
00:20:54.20 When we add to cells that express those two constructs
00:20:59.06 a chemical heterodimerizer,
00:21:00.20 what occurs is that those two parts of the proteins
00:21:03.20 form a complex,
00:21:05.19 and that drags the whole endoplasmic reticulum
00:21:08.04 in close proximity to the plasma membrane.
00:21:11.19 Normally, there is only very little plasma...
00:21:13.29 endoplasmic reticulum right under the plasma membrane.
00:21:16.20 But when we induce this,
00:21:18.23 what we observe is...
00:21:20.07 by total internal reflection microscopy...
00:21:22.19 is that both LDAF1 is targeted,
00:21:25.13 as you would predict,
00:21:27.01 but also seipin is now brought along
00:21:28.18 because it is in a protein-protein complex with LDAF1,
00:21:31.21 not directly interacting with the anchor.
00:21:35.03 This now generates a scenario
00:21:37.26 in which right underneath the plasma membrane
00:21:40.00 we have a lot, a lot of ER
00:21:42.15 that's tethered to it.
00:21:44.05 And what's important and interesting is
00:21:46.27 when we now look at lipid droplet formation,
00:21:49.01 we see that now lots of, lots of lipid droplets form
00:21:51.24 underneath the plasma membrane,
00:21:53.20 because now you have this complex there,
00:21:55.12 which normally isn't found there.
00:21:57.04 In a wild type cell that doesn't have...
00:21:59.01 or in a cell that hasn't been exposed to the dimerizer,
00:22:01.16 a control cell,
00:22:03.02 we see only very, very few droplets
00:22:05.03 forming right under the plasma membrane.
00:22:07.22 But as you can see in the quantitation on the right, here,
00:22:11.07 or in the stills taken from a movie on the left,
00:22:13.03 as soon as you bring this complex right underneath the plasma membrane,
00:22:16.03 now, lots of, lots of lipid droplets form there,
00:22:18.18 because now the machinery is there
00:22:20.04 to catalyze this reaction.
00:22:23.15 So, these are some of the highlights
00:22:25.16 of studies where we have identified
00:22:28.01 at least a couple components
00:22:30.07 of a lipid droplet assembly complex:
00:22:32.17 seipin, which we and others
00:22:34.24 had identified over the years
00:22:36.25 as being crucial to this complex;
00:22:38.17 and this other protein, LDAF1.
00:22:40.10 And what's shown here is our working model
00:22:42.13 for how this might work.
00:22:44.03 So, as we mentioned,
00:22:46.02 triglycerides are made through the actions of, for example,
00:22:48.22 DGAT enzymes in the endoplasmic reticulum.
00:22:50.14 And they are initially released into the plane of the bilayer.
00:22:55.06 It's our hypothesis, then,
00:22:56.29 at the moment,
00:22:58.23 that the seipin-LDAF1 complex
00:23:01.06 becomes the site
00:23:03.28 where these triglycerides can nucleate and phase separate
00:23:06.00 to begin to form lipid droplets.
00:23:07.25 And what's really intriguing
00:23:09.23 is this complex brings together, for example, in humans,
00:23:12.19 66 membrane-spanning domains in a very small area
00:23:16.18 that might create hydrophobic surfaces
00:23:19.04 for this... for this nucleation...
00:23:21.26 catalytic step to occur.
00:23:23.24 So, shown here, then,
00:23:25.29 are triglyceride molecules
00:23:28.07 that have begun to form a lens
00:23:30.07 in the setting of this complex.
00:23:31.29 And as you can see, LDAF1 begins to
00:23:36.14 provide curvature to the membrane,
00:23:38.02 which may also facilitate bending of the membrane
00:23:40.09 and droplet formation.
00:23:42.27 The droplet then grows.
00:23:44.26 And what we didn't show you,
00:23:46.11 but we have data indicating,
00:23:48.02 is that as the droplet is growing
00:23:50.15 LDAF1 coats the surface of the droplet.
00:23:52.12 Here, we think LDAF1 provides surface-modulating qualities,
00:23:56.18 such as lowering the surface tension
00:23:58.06 and regulating some of the protein composition
00:24:00.03 of the developing droplet.
00:24:02.13 So, that's the current model.
00:24:04.01 And obviously, this model will require further testing.
00:24:06.26 And one other aspect of this model
00:24:09.14 is that, as Toby mentioned before,
00:24:11.16 there are domains of seipin
00:24:13.25 that appear very similar
00:24:16.20 to lipid-binding domains,
00:24:18.04 so it may be that this complex
00:24:20.12 also has some role in directing lipid trafficking
00:24:22.21 -- whether that be neutral lipids or phospholipids --
00:24:25.22 during the formation of lipid droplets.
00:24:28.02 That also remains to be further tested.
00:24:31.11 So, in this part of our lectures,
00:24:32.24 we told you about triglyceride synthesis
00:24:34.20 and the initial stages of lipid droplet formation
00:24:37.02 at the endoplasmic reticulum.
00:24:38.24 But what happens next?
00:24:40.11 How do the lipid droplets acquire their specific proteome?
00:24:43.13 How do they grow in size
00:24:45.15 and detach from the ER?
00:24:46.26 And what are their functions in the cell?
00:24:49.09 So, we'll tell you about that in the next part of our lectures.
00:24:52.11 Stay tuned.

Part 3: Physiology of Lipid Droplet Formation

00:00:15.19 Hello, I'm Toby Walther.
00:00:17.08 And I'm Bob Farese, Jr.
00:00:19.03 And welcome to the next part of our lecture,
00:00:21.27 in which we're going to review
00:00:24.02 the process of lipid droplet formation,
00:00:25.29 and specifically, how proteins target to lipid droplets,
00:00:29.03 how droplets expand...
00:00:31.16 how some droplets expand,
00:00:34.09 and the physiological consequences of lipid droplet formation.
00:00:37.26 Now, in the next step during formation of lipid droplets,
00:00:41.26 proteins are targeted to this surface.
00:00:45.24 So far, we know about two distinct pathways
00:00:48.14 how proteins can reach the surface
00:00:50.23 of a lipid droplet.
00:00:52.09 In one pathway,
00:00:54.00 proteins are initially translated on cytosolic ribosomes
00:00:57.04 to generate intermediates
00:00:59.24 that are possibly soluble.
00:01:01.10 Those often have amphipathic helices,
00:01:04.01 or protein domains
00:01:06.20 that can fold into a helix
00:01:08.15 where one side of the helix is hydrophilic
00:01:10.11 and the other one is hydrophobic.
00:01:12.01 Those helices, then,
00:01:14.05 can bind the surface of the lipid droplet directly.
00:01:17.08 In a second pathway, proteins are initially inserted
00:01:20.06 into the endoplasmic reticulum
00:01:21.24 and form often hydrophobic domains,
00:01:26.23 such as hairpins that reach into the plane of the membrane
00:01:29.09 and come out again on the same side,
00:01:31.16 that then, by some pathway,
00:01:34.13 can reach the surface of lipid droplets
00:01:37.00 from the endoplasmic reticulum.
00:01:39.18 So, we first want to discuss
00:01:42.14 a little bit of what we've learned over the past few years
00:01:44.15 about how cytosolic proteins
00:01:46.12 that contain either amphipathic helices
00:01:47.29 or short hydrophobic motifs
00:01:49.14 reach the lipid droplet surface.
00:01:51.19 And one of the principles
00:01:53.29 that we and others have uncovered
00:01:55.29 in this targeting...
00:01:57.27 you know, how do...
00:01:59.13 how do these proteins find the droplet surface
00:02:01.10 from all the other membranes in the cell?
00:02:04.07 So, one of the principles is shown here,
00:02:06.04 and this came, in our case,
00:02:07.23 from molecular dynamics simulation modeling
00:02:10.06 that we did in collaboration with Greg Voth's lab
00:02:12.16 at the University of Chicago.
00:02:14.29 And what Greg's lab did here is model a bilayer,
00:02:19.18 which is shown in the gray and the blue,
00:02:21.00 with the gray being the phospholipid head groups
00:02:23.02 and blue being the acyl chains...
00:02:24.23 and in this case,
00:02:26.09 they've inserted neutral lipids, shown in green,
00:02:28.23 or triacylglycerols,
00:02:30.02 in the middle of the bilayer
00:02:32.06 to simulate what would occur in a lipid droplet.
00:02:34.18 And what they found when they ran these simulations
00:02:36.20 is something unique to having the neutral lipids in the bilayer.
00:02:40.25 And that is, there's flexibility in the phospholipids,
00:02:44.09 so the neutral lipids can poke up through,
00:02:47.20 between the phospholipid heads,
00:02:50.05 and essentially create packing defects at the surface of lipid droplets.
00:02:52.26 Now, this makes intuitive sense,
00:02:54.29 because if you've just taken oil
00:02:56.23 and shaken it up and put it in water,
00:02:58.13 you see there are droplets,
00:02:59.21 and essentially that's one giant packing defect.
00:03:01.27 Okay?
00:03:03.13 Normally, lipid droplets are coated with phospholipids
00:03:05.17 to minimize these defects,
00:03:07.16 but nevertheless, probably all droplets
00:03:09.13 have some degree of packing defects
00:03:11.02 at their surface.
00:03:13.17 Now, to study this process of targeting experimentally,
00:03:16.18 we need model systems.
00:03:18.29 So, Rachid Thiam,
00:03:20.28 when he was in the laboratory,
00:03:22.27 and, later, Coline Prevost
00:03:25.00 set up a system in which they generated membrane vesicles
00:03:27.14 known as giant unilamellar vesicles
00:03:29.10 as a model system into which you could
00:03:30.24 incorporate triacylglycerol, here shown in gray.
00:03:34.09 And in this system,
00:03:36.14 you can now form a small vesicle
00:03:38.23 that also contains lipid lenses,
00:03:41.11 similar to what occurs during lipid droplet formation
00:03:43.26 at the endoplasmic reticulum.
00:03:46.06 We can then add proteins to that system
00:03:48.27 and study where they go:
00:03:50.27 to the bilayer membrane or to the droplet.
00:03:54.05 Now, in this slide, you see an example experiment
00:03:57.09 in which we take an amphipathic helix protein
00:03:59.06 and incubate it with such a system.
00:04:01.23 And you can see that this protein binds specifically
00:04:04.21 to the part of the GUV
00:04:06.25 in which the triglyceride lens is bulging from.
00:04:10.10 You can also see rhodamine-PE, or Rho-PE, here,
00:04:14.17 as a marker for the lipids.
00:04:15.28 And you can see that there is a big bilayer membrane
00:04:18.24 that splits up into two monolayers around that oil lens.
00:04:23.01 With such a system,
00:04:25.20 we can now test predictions from our simulations
00:04:27.23 and ask, directly,
00:04:29.13 whether or not packing defects are responsible for binding.
00:04:32.00 One way to do this is to take
00:04:34.23 a suction pipette to the vesicle and pull on it.
00:04:37.20 As you can imagine,
00:04:40.02 pulling on that vesicle will increase the packing effects,
00:04:42.20 because you increase the tension on it.
00:04:45.00 And what you can see in this movie, here,
00:04:47.04 that Coline recorded
00:04:49.12 is that as we increase the tension,
00:04:51.21 and thus the surface packing defects,
00:04:53.13 more of the amphipathic helix is binding.
00:04:56.12 And conversely, when we release the pressure,
00:04:58.13 or the tension,
00:05:00.09 that then it falls off again.
00:05:02.26 On a molecular level,
00:05:05.12 what Greg Voth's lab has observed here
00:05:07.15 is that during the simulations,
00:05:09.20 amphipathic helices bind the packing defects,
00:05:13.04 particularly using large hydrophobic residues.
00:05:16.04 In this movie, what you're seeing is that a large hydrophobic residue
00:05:19.15 of an amphipathic helix-type protein
00:05:22.20 in an unfolded state
00:05:25.11 intercalates with the surface packing defect.
00:05:27.21 And what occurs after that is that
00:05:30.21 slowly this protein, once it's bound to it,
00:05:33.01 will fold into an alpha helix.
00:05:35.07 This second reaction is on a time scale
00:05:37.25 that's not easily accessible by molecular dynamics simulations.
00:05:42.28 In this process, then,
00:05:44.13 what we imagine is that targeting of cytosolic proteins
00:05:47.26 to lipid droplets
00:05:49.22 occurs in several steps.
00:05:52.19 Initially, large hydrophobic residues
00:05:54.02 read out packing defects that are more abundant,
00:05:56.06 more persistent,
00:05:58.04 and larger on lipid droplets than on other membranes.
00:06:00.02 This then, in a second step,
00:06:02.06 leads to the formation of an amphipathic helix.
00:06:05.06 That in turn essentially renders
00:06:07.27 this process irreversible,
00:06:09.23 because you're not gonna get spontaneous unfolding
00:06:12.17 of this protein on the surface of lipid droplets.
00:06:15.14 So, one of the questions that arises
00:06:18.20 when we look at how amphipathic helical proteins
00:06:21.17 bind to the surfaces of lipid droplets is,
00:06:23.22 why do they go to lipid droplets?
00:06:25.14 Why don't all amphipathic helical proteins...
00:06:27.16 why don't they all go there?
00:06:30.08 And why do some go to other bilayer membranes?
00:06:33.07 And obviously, there are probably multiple factors
00:06:35.17 that control this
00:06:37.20 -- for example, if there's other regions of the protein
00:06:39.19 that might direct it elsewhere in the cell --
00:06:41.19 but there's a unique feature of lipid droplets
00:06:44.29 that's worth talking about as a principle here,
00:06:46.26 and that is that the lipid droplet surface
00:06:49.09 is very crowded.
00:06:50.22 The surface of lipid droplets is finite,
00:06:54.03 and it's a crowded place,
00:06:55.21 with many proteins that adhere and bind to the surface.
00:06:58.13 And what we and others have shown
00:07:02.05 is that the affinity of amphipathic helices
00:07:06.26 is important in regulating
00:07:09.20 the composition of droplets.
00:07:11.04 So, for example, those that...
00:07:12.28 there's a hierarchy,
00:07:14.16 and those that bind with higher affinity
00:07:15.23 will tend to bind to the surface
00:07:17.15 and kick off the ones that don't bind.
00:07:19.02 This phenomenon of macromolecular crowding
00:07:22.28 becomes very important, for example during lipolysis,
00:07:26.08 when lipid droplets are shrinking.
00:07:27.20 And when they're shrinking,
00:07:29.09 the surface area is getting smaller,
00:07:31.20 and as a consequence, some of the proteins are kicked off from the surface,
00:07:34.10 whereas others remain bound to the surface.
00:07:36.12 And we believe that this is one very important mechanism
00:07:39.02 for governing the bona fide composition
00:07:42.00 of lipid droplet surfaces.
00:07:43.20 It's very well possible
00:07:45.23 that abnormal protein composition
00:07:47.13 happens in certain disease states.
00:07:50.09 Now, how about that second pathway?
00:07:52.05 I mentioned to you in the beginning that
00:07:55.21 proteins can be inserted into the endoplasmic reticulum
00:07:57.28 and then reach the lipid droplet from there.
00:08:00.00 But, of course, proteins don't just hop in and out of membranes,
00:08:02.21 and there must be a pathway that allows them
00:08:05.07 to go from one membrane onto the lipid droplet.
00:08:08.20 You can see an example in a cell, where at...
00:08:12.01 in conditions where there are no lipid droplets,
00:08:14.25 both the protein and the ER signal overlap.
00:08:18.01 The protein, here, is labeled in red,
00:08:19.21 and in this case it's the Drosophila homologue
00:08:22.04 of a metabolic enzyme known as GPAT4.
00:08:24.22 You can see, over time,
00:08:26.10 as lipid droplets form
00:08:28.15 and lipid droplet formation is induced,
00:08:30.02 this protein efficiently accumulates,
00:08:31.18 exclusively on the surface of lipid droplets.
00:08:35.27 Now, in this experiment,
00:08:38.06 we're looking at GPAT4 in a so-called FRAP experiment,
00:08:41.24 or fluorescence recovery after photobleaching.
00:08:44.24 And what you can see is a lipid droplet
00:08:46.28 that has a GPAT4 enzyme
00:08:49.07 that's concentrated at its surface.
00:08:51.19 And if I run the movie again,
00:08:53.07 what you'll see is bleaching of the signal
00:08:55.16 at the droplet surface,
00:08:57.05 and then you'll see, in very rapid order,
00:08:59.15 within a matter of seconds,
00:09:01.04 GPAT4 repopulate...
00:09:03.17 repopulates the lipid droplet surface,
00:09:06.00 indicating that there's a very fast kinetics on/off rate
00:09:09.22 for this protein to be...
00:09:12.11 transfer onto the lipid droplet surfaces.
00:09:15.15 So, here's the bleaching,
00:09:17.14 and then you see, within seconds,
00:09:19.00 GPAT4 is again concentrated on the lipid droplet surface.
00:09:22.18 And that GPAT4, we believe,
00:09:24.11 is coming from the endoplasmic reticulum.
00:09:26.19 So, how does GPAT4 get from the endoplasmic reticulum
00:09:29.15 onto the lipid droplet surface?
00:09:33.18 And so, we think that this happens
00:09:35.25 via the presence of
00:09:38.04 ER-to-lipid droplet membrane bridges.
00:09:41.10 This is fluorescence signal, again,
00:09:43.18 and the arrow is pointing to a potential bridge
00:09:46.25 between the endoplasmic reticulum
00:09:48.19 and the surface of this lipid droplet.
00:09:51.12 When we look with electron tomography
00:09:53.14 at a droplet such as this,
00:09:55.19 we can actually see what appear to be
00:09:58.09 ER-lipid droplet membrane bridges.
00:10:00.03 So, this tomogram shows you
00:10:03.04 the endoplasmic reticulum
00:10:05.03 and the surface of one of these large lipid droplets,
00:10:07.27 where the membranes are coming together,
00:10:10.14 and you can see the endoplasmic reticulum
00:10:12.08 is studded with ribosomes,
00:10:14.13 and then it comes right up to the droplet surface.
00:10:16.19 This is color-coded here
00:10:18.23 so you can better appreciate the ER
00:10:21.11 having multiple connections,
00:10:23.09 where the cytoplasmic leaflet of the ER bilayer
00:10:26.11 is contiguous with the monolayer surface
00:10:28.13 of the lipid droplet.
00:10:31.04 Now, we think that this provides the road
00:10:33.04 for proteins to move from the ER
00:10:35.00 to the lipid droplet.
00:10:36.17 But just because there's a road,
00:10:38.04 it doesn't mean all of you have to go across that bridge.
00:10:40.25 So, one really big puzzle for us has been,
00:10:43.23 for quite a number of years...
00:10:45.28 is, why do all the proteins seem to accumulate
00:10:48.14 on the surface of lipid droplets,
00:10:49.25 rather than equiva...
00:10:52.16 equilibrate between both places,
00:10:54.06 which would lead to most of the protein
00:10:56.07 being stuck in the ER
00:10:57.24 because there is vastly more ER surface
00:10:59.18 than there is lipid droplet surface?
00:11:01.25 Now, it turns out that
00:11:03.28 all the information appears to be encoded
00:11:05.17 in a small segment that presu...
00:11:07.13 that presumably can form a hairpin.
00:11:09.15 If you, for instance,
00:11:11.07 take the sequence of GPAT4
00:11:13.03 that just forms this little hairpin sequence,
00:11:15.23 what you can observe in that sequence
00:11:18.06 is two features.
00:11:19.25 There are two hydrophobic segments
00:11:22.04 that are interrupted in the middle
00:11:24.06 by a proline that introduces a kink,
00:11:26.10 and thereby allows the formation of a hairpin-type structure.
00:11:29.06 The second thing you can see is that
00:11:31.16 this is an unusual transmembrane segment,
00:11:34.19 in as much as it contains
00:11:37.02 positively charged residues,
00:11:38.21 usually excluded from transmembrane segments.
00:11:40.25 In addition, a couple of very large hydrophobic residues,
00:11:43.12 such as tryptophans,
00:11:45.02 which are often more oriented
00:11:49.03 to the surface of a transmembrane domain,
00:11:50.20 but in this case are stuck smack in the middle.
00:11:53.20 The sequence is important,
00:11:55.05 because in itself it is sufficient
00:11:57.15 to mediate targeting to lipid droplets,
00:11:59.08 as you can see here on the light...
00:12:00.28 left side of this graph.
00:12:02.16 After three hours,
00:12:04.21 pretty much all of the protein that just contains this signal
00:12:06.29 is relocalized to the surface of lipid droplets.
00:12:10.09 And it is not just the composition of that sequence,
00:12:13.01 but really the linear sequence,
00:12:15.10 because once we...
00:12:16.26 if we modify the sequence by scrambling it,
00:12:21.18 actually this prevents the efficient targeting,
00:12:23.22 even though the physicochemical properties
00:12:26.01 of that sequence remain the same.
00:12:28.18 So, the question is then,
00:12:31.03 if this hairpin is all that's needed to mediate
00:12:33.06 the targeting from the ER to the lipid droplet
00:12:35.16 -- presumably via these ER-lipid droplet membrane bridges --
00:12:38.05 why do they accumulate there?
00:12:40.24 Again, we turned to molecular dynamics simulation modeling
00:12:44.22 in collaboration with Greg Voth's lab,
00:12:47.05 in particular with Siyoung Kim and Jessica Swanson in this case.
00:12:50.29 And from their...
00:12:53.15 what they were able to do is take
00:12:55.20 this GPAT4 hairpin
00:12:57.23 and perform molecular dynamics simulations
00:12:59.26 of the GPAT4 hairpin -- a model structure --
00:13:03.05 inserted either into a lipid bilayer
00:13:05.19 or into a monolayer,
00:13:07.07 in which the bilayer
00:13:09.25 has between it neutral lipids,
00:13:11.08 as I showed you earlier.
00:13:12.23 And when they run these simulations,
00:13:14.08 what becomes apparent is that
00:13:17.08 the final, most stable state
00:13:19.16 that they observe in the simulations,
00:13:21.08 as shown here, is...
00:13:23.25 GPAT4 in the bilayer has one conformation,
00:13:26.18 and it has a slightly different conformation...
00:13:28.08 conformation in the monolayer.
00:13:30.29 So, in the bilayer, as you can see,
00:13:32.28 the proline that Toby described
00:13:35.12 is towards the bottom of the kink of this hairpin,
00:13:37.17 and the... two of the positive charges
00:13:39.18 are located near that proline residue
00:13:43.08 and in close proximity
00:13:46.00 to the phospholipid head groups of the bilayer.
00:13:48.10 The tryptophans that he mentioned
00:13:50.03 are in the middle of the bilayer,
00:13:52.00 where their movement is likely restricted somewhat.
00:13:54.26 However, when the same molecule
00:13:57.26 is modeled in the presence of the monolayer
00:13:59.29 and neutral lipids,
00:14:01.14 it evolves into a different conformation.
00:14:03.02 And so, what you can see here is that
00:14:05.21 some of these tryptophans...
00:14:07.13 and indeed, one of the positive charges
00:14:09.02 has migrated up towards the surface of the model layer,
00:14:12.15 the hairpin has closed slightly,
00:14:14.08 and one of the positive residues R187,
00:14:18.13 remains in the... towards the neutral lipids,
00:14:21.17 but it's solvated by water.
00:14:25.27 And these... this modeling, then,
00:14:28.28 allowed Jessica and co-workers
00:14:31.22 to calculate the free energy
00:14:34.14 of the different residues within these conformations.
00:14:36.22 And the principle that they identified
00:14:41.00 was that these different amino acids
00:14:42.25 -- and in particular the large hydrophobic amino acids
00:14:44.13 such as the W's or tryptophans --
00:14:47.01 find a lower free energy state,
00:14:49.22 as seen in this figure,
00:14:51.16 when they're in the monolayer with neutral lipids
00:14:56.05 as opposed to the bilayer.
00:14:58.13 And so, this gives rise to a model
00:15:01.10 -- which, again, requires further testing --
00:15:03.18 that part of the driving force for the accumulation
00:15:05.21 of such hairpin proteins from the ER
00:15:08.17 to the lipid droplet surface
00:15:10.14 is that conformational changes
00:15:12.12 allow them to...
00:15:14.01 when they reach the droplet surface,
00:15:15.15 to achieve a lower free energy state,
00:15:18.05 and become stabilized there,
00:15:19.23 where there's an energy barrier
00:15:21.20 for going back to the endoplasmic reticulum.
00:15:25.04 So, in summary, we told you about two pathways --
00:15:28.00 how proteins can get targeted to lipid droplets.
00:15:30.10 But what's important, maybe,
00:15:33.08 to note is that this is quite different from how most other organelles
00:15:35.10 acquire their proteome.
00:15:36.25 Most other organelles
00:15:38.26 either use specific lipid landmarks
00:15:40.12 --such as phosphoinositides on the plasma membrane,
00:15:42.12 the endosome, or the Golgi --
00:15:44.01 that bind directly to lipids that are only found there,
00:15:46.25 or they use protein machinery
00:15:48.22 -- such as the translocon in the mitochondria
00:15:50.23 or the endoplasmic reticulum --
00:15:52.12 that recognize specific sets of proteins via their signals
00:15:55.10 and insert them there.
00:15:57.09 In this case, targeting in both principles
00:15:59.05 is driven by the unique physicochemical properties
00:16:02.14 of the lipid droplet surface.
00:16:04.00 In one case, amphipathic helices
00:16:07.12 directly read out the specific surface structure
00:16:09.26 of the lipid droplets.
00:16:11.17 In another case, targeting from the ER occurs,
00:16:13.23 because proteins have evolved unique features
00:16:17.04 that allow them to adopt a lower energy conformation,
00:16:19.11 presumably only on the lipid droplet.
00:16:25.29 So, at this point,
00:16:27.26 we've discussed how triglycerides are formed,
00:16:29.19 how droplets are initially formed,
00:16:31.21 and how proteins target to the en...
00:16:34.12 target to lipid droplet surfaces.
00:16:36.22 The final step we'd like to discuss briefly
00:16:38.16 is how some lipid droplets expand.
00:16:41.12 So, in this image,
00:16:43.04 what you can see is the DGAT1 enzyme
00:16:45.23 and others of the triglyceride synthesis pathway
00:16:48.18 behave like this.
00:16:50.07 They're confined to the endoplasmic reticulum.
00:16:52.21 So, that's the red signal.
00:16:54.20 The green signal are the lipid droplets.
00:16:57.04 However, the DGAT2 enzyme
00:16:59.23 and indeed, also, the GPAT4 enzyme
00:17:01.09 that Toby just discussed
00:17:02.25 have a different property.
00:17:04.24 They start their life in the endoplasmic reticulum,
00:17:06.18 but it appears that they can migrate
00:17:08.27 to the droplet surface,
00:17:10.11 or at least in ER compartments
00:17:12.08 that are very close to the droplet surface.
00:17:14.12 So, this has the effect where you can
00:17:16.08 compartmentalize triglyceride synthesis.
00:17:18.09 Some of it occurs in the ER,
00:17:19.28 but when droplets are busy forming,
00:17:21.15 some of it goes to the droplets.
00:17:24.01 And this has functional consequences.
00:17:26.00 It turns out that when we look,
00:17:28.05 for example, in fly cells,
00:17:29.23 as shown in this figure,
00:17:31.08 that there are two populations of lipid droplets.
00:17:32.22 They all start off relatively small,
00:17:35.00 at 400-600 nanometer diameters.
00:17:37.15 One population, over time, doesn't grow.
00:17:40.01 But the other population begins to expand.
00:17:42.17 And that population is specifically the population
00:17:44.28 in which triglyceride synthesis enzymes
00:17:47.08 are targeted to its surface.
00:17:49.28 And in fact, you can do cool tricks here.
00:17:52.23 You can drive one population or the other population.
00:17:55.19 So, for example, on your left
00:17:59.07 is a control Drosophila cell
00:18:00.13 that has a mixture of both large and small droplets.
00:18:02.05 But if we overexpress the ER-targeted triglyceride protein
00:18:06.05 DGAT1,
00:18:07.24 we create many of these small droplets.
00:18:09.05 If we overexpress DGAT2,
00:18:11.05 which goes to the lipid droplet surface,
00:18:12.28 we can create a larger proportion
00:18:15.05 of the population
00:18:17.08 of these expanding droplets.
00:18:19.19 Now, we told you in the beginning
00:18:23.04 that pretty much all cells can form lipid droplets.
00:18:24.21 And they do that through
00:18:28.00 apparently two different pathways,
00:18:29.20 catalyzed by two different enzymes,
00:18:31.03 DGAT1 and DGAT2.
00:18:32.20 There has been a lot, a lot of work,
00:18:34.08 by many laboratories over many years,
00:18:35.28 studying the physiology of lipid droplets
00:18:37.27 and the role of the specific pathways
00:18:41.00 in human, mammalian, and other physiology.
00:18:46.05 We want to give you a short vignette
00:18:48.09 of how these two pathways
00:18:50.18 participate in human physiology.
00:18:54.03 The importance of that is highlighted
00:18:57.09 by the identification, now,
00:18:59.16 of many genetic mutations in proteins
00:19:02.23 that are either involved...
00:19:04.12 or, in genes that encode proteins
00:19:06.04 that are either involved in the triglyceride synthesis,
00:19:08.11 lipid droplet formation,
00:19:09.24 or that are lipid droplet proteins in the end.
00:19:13.00 Examples include, as Bob already mentioned to you,
00:19:16.06 the seipin/BSCL2 protein
00:19:19.17 that leads to a congenital lipodystrophy.
00:19:22.07 There are other proteins,
00:19:24.04 such as the perilipin proteins,
00:19:25.15 which are thought to coat the surface of lipid droplets,
00:19:27.07 or CCT,
00:19:29.15 a protein involved in phospholipid synthesis,
00:19:31.23 important for maintaining the emulsification
00:19:33.15 of lipid droplets,
00:19:35.02 that are also linked to congenital lipodystrophies.
00:19:39.24 In addition, there are mutations in the enzymes,
00:19:42.12 specifically, that make triglycerides,
00:19:44.27 and they have been linked to a number
00:19:47.01 of different metabolic diseases.
00:19:48.29 This includes, for instance,
00:19:50.15 mutations in DGAT1.
00:19:52.22 Now, a number of years ago,
00:19:54.12 with Harland Winter,
00:19:56.05 we identified a family that has
00:19:58.05 a congenital diarrheal syndrome
00:20:00.14 due to the loss of DGAT1 function
00:20:04.16 due to a compound heterozygous mutation.
00:20:08.06 And what it turns out is that in this...
00:20:10.27 in this family,
00:20:12.18 the kids that have those two mutations
00:20:14.26 cannot tolerate fat in their diet.
00:20:17.19 When they're exposed to fat in the endoplasmic reticulum,
00:20:20.28 the... as the process of triglyceride synthesis
00:20:23.22 is compromised,
00:20:25.17 apparently that interferes with the function
00:20:28.08 of the enterocyte,
00:20:29.28 leading as a consequence to
00:20:32.01 a protein-losing enteropathy,
00:20:33.15 or in other words,
00:20:35.14 a very severe diarrhea syndrome
00:20:37.09 in which proteins are lost from the bloodstream.
00:20:42.17 So, fortunately, those children can be treated
00:20:44.25 by reducing the fat in their diet,
00:20:47.01 and that was an important consequence
00:20:48.29 of that discovery.
00:20:50.13 If we look, just briefly, then,
00:20:52.06 what is the function of the storage of triglycerides
00:20:54.24 in energy and lipid droplets?
00:20:56.18 The... one obvious example is we do it
00:20:58.29 for energy storage,
00:21:00.18 as we started our lecture describing.
00:21:02.23 To illustrate this, we recently
00:21:05.29 have genetically engineered mice
00:21:07.22 that lack the ability to make fat
00:21:09.12 within their fat tissue.
00:21:10.26 So, we call these mice fatless-fat mice
00:21:13.11 or, in this slide,
00:21:16.09 the adipose tissue DGAT1/DGAT2 knockout mice.
00:21:20.05 And just to show you what happens
00:21:22.04 when you aren't able to store fat in your fat tissue,
00:21:24.09 you can see here...
00:21:26.02 well, first of all, if you look at a very small sample
00:21:28.08 of the fat tissue...
00:21:29.26 on your left are control mice
00:21:31.21 -- that's the so-called flox allele --
00:21:33.21 where you can see the WAT, or the white adipose tissue,
00:21:36.05 and the BAT, or the brown adipose tissue,
00:21:37.25 and you see it floats,
00:21:39.12 and it floats because it has triglycerides or fats in it.
00:21:42.03 In contrast, if you delete the enzymes
00:21:44.04 that can make the triglycerides,
00:21:45.25 to no surprise,
00:21:47.24 they don't have triglycerides,
00:21:49.19 and they...
00:21:51.23 it gives rise to fat tissue that sinks in solution here.
00:21:54.26 So, what are the consequences of that?
00:21:56.16 Well, here's just one example
00:21:58.07 of the consequences
00:22:00.03 that illustrates why we store fat.
00:22:02.10 So, if mice that are normal
00:22:05.00 or lack these enzymes in their fat
00:22:06.29 are exposed to cold for a few hours...
00:22:09.10 this is 4 degrees Centigrade cold.
00:22:11.23 You can see, on the left,
00:22:13.21 as long as you give them food to eat, they're fine.
00:22:15.17 They maintain their body temperature just fine.
00:22:17.16 However, if you don't give them food to eat,
00:22:19.20 the normal mouse does fine --
00:22:22.17 it maintains its body temperature.
00:22:24.13 But those that lack the fat stores,
00:22:26.28 after a couple hours,
00:22:28.12 they start to become hypothermic,
00:22:29.29 and you have to stop the experiment.
00:22:31.25 So, this illustrates that,
00:22:33.28 you know, we store fat, obviously,
00:22:36.03 because we need to have that fat available
00:22:38.23 for energy --
00:22:41.04 in this case, energy for thermogenesis,
00:22:42.24 via the uncoupling protein,
00:22:44.14 to maintain body heat.
00:22:46.13 Now, if you look at that reaction, though,
00:22:48.14 it immediately becomes obvious that
00:22:50.19 besides synthesizing triglyceride,
00:22:52.12 the reactions catalyzed by the DGAT enzyme
00:22:54.14 do something else,
00:22:56.02 and that is they remove fatty acyl-CoA
00:22:58.10 and diacylglycerol from the cells.
00:23:00.19 And that is important
00:23:02.17 because both of those are bioactive molecules
00:23:06.03 that, in addition to being metabolic intermediates,
00:23:08.07 can also perturb membrane structure
00:23:10.09 and they can act as signaling molecules.
00:23:13.22 And so, we now know that
00:23:16.13 the synthesis of triacylglycerol in general
00:23:19.06 appears to protect cells from the accumulation
00:23:21.05 of lipotoxic lipids.
00:23:23.21 Particularly, and consistent with this idea
00:23:26.13 that synthesis removes lipotoxic intermediates
00:23:29.12 and puts them in a chemically inert pool
00:23:32.03 of triacylglycerol,
00:23:33.19 there are a number of studies from our lab and others
00:23:37.01 that have shown that if we increase the capacity of cells
00:23:39.18 to make triglyceride,
00:23:41.10 this actually leads to a fat tissue,
00:23:44.11 but, somewhat paradoxically to your initial expectation, maybe,
00:23:48.05 this is protective.
00:23:51.01 For instance, if we overexpress DGAT in the liver,
00:23:53.18 we get fatty liver -- foie gras --
00:23:56.13 and when we do that,
00:23:58.16 that liver is actually protected from insulin resistance
00:24:01.03 and the consequences thereof.
00:24:03.15 Similarly, increasing the DGAT activity
00:24:06.24 in macrophages actually protects them,
00:24:08.27 or helps protect the organism
00:24:11.25 from insulin resistance,
00:24:13.13 and similar observations have been made in skeletal muscle.
00:24:16.21 Others have also overexpressed triglyceride
00:24:20.16 in the heart,
00:24:21.29 and Ira Goldberg's lab has found that
00:24:24.01 under those conditions,
00:24:25.15 the heart is actually protected from lipotoxicity,
00:24:26.28 even though it now stores more fat.
00:24:29.26 So, in this lecture,
00:24:32.09 we've taken you through a series of steps.
00:24:34.07 We've talked about step 1,
00:24:35.27 how the oils, triglycerides,
00:24:38.04 the fats are made within the endoplasmic reticulum,
00:24:40.24 how they initially are released into the ER bilayer.
00:24:44.00 We've discussed some of the machine which...
00:24:47.15 that catalyzes the formation,
00:24:49.08 or the nucleation and phase separation
00:24:51.15 of triglycerides to form lipid droplets,
00:24:53.08 how they begin to bud towards the cyto...
00:24:55.25 the cytosolic surface.
00:24:57.12 We've talked about how proteins target
00:24:59.14 uniquely to the surfaces of these unusual organelles.
00:25:02.26 And we've talked about how lipid droplets expand,
00:25:05.06 finally giving you some examples in physiology.
00:25:09.19 Now, when you listen to this lecture,
00:25:12.09 and as we prepared for it,
00:25:13.26 it sounds like there's a lot of knowledge.
00:25:16.05 And indeed, we now know some things.
00:25:20.03 However, to us, lipid droplets are a frontier in cell biology,
00:25:22.21 because when we talk every day,
00:25:24.14 what is really striking us is that
00:25:27.03 there is a lot more we don't know
00:25:28.21 than the little things we have now figured out as a field.
00:25:31.02 We still don't understand, for instance,
00:25:32.26 why lipid droplets just go to the cytoplasm
00:25:35.11 and don't go into the ER.
00:25:37.12 And maybe what's more, why in some cells,
00:25:39.05 such as in the liver or the intestinal cell,
00:25:41.26 there is a pool of lipid droplets
00:25:44.02 that can go into the ER and become lipoproteins.
00:25:47.22 We also don't really understand
00:25:49.20 what regulates the size and numbers
00:25:51.22 of lipid droplets in a cell,
00:25:53.14 and the composition that is linked to that.
00:25:57.18 Particularly, in this case,
00:25:59.17 it is interesting to think about,
00:26:01.16 what is the actual capacity,
00:26:03.09 and how is the capacity of a cell determined?
00:26:05.08 Why is it a problem
00:26:07.01 if you reach the capacity,
00:26:08.15 and why do cells stop storing more droplets?
00:26:10.21 We also do not really know
00:26:13.03 what happens with the converse process.
00:26:14.21 The removal of lipids
00:26:17.01 has been well studied by a number of labs
00:26:18.29 that have shown how the triglyceride
00:26:21.29 can be metabolized,
00:26:23.11 but we really don't understand what happens
00:26:25.27 to the organelle,
00:26:27.14 where do the proteins go, where does the droplet go,
00:26:28.26 and is that the same in all different cell types?
00:26:32.03 And in many cases,
00:26:34.00 we still don't really understand,
00:26:35.18 what is the consequence of lipid storage
00:26:37.18 in a particular cell type?
00:26:39.17 For instance, there are a number of cancer cells
00:26:41.25 that are characterized by their...
00:26:43.14 by their abundant lipid droplets,
00:26:45.20 such as renal clear cell carcinomas,
00:26:48.00 where the "clear cell" part indicates,
00:26:49.12 as you look at it, that they look clear
00:26:51.02 because they have so much lipid in them.
00:26:53.17 A number of laboratories have provided
00:26:56.10 fascinating descriptions of intracellular pathogens
00:26:58.18 hijacking parts of lipid droplets.
00:27:00.20 That includes viruses such as hepatitis,
00:27:03.07 but also intracellular bacteria
00:27:05.25 such as mycobacterium,
00:27:08.06 that use lipid droplets either as a fuel source
00:27:10.15 or an assembly platform.
00:27:12.27 We also don't know about, for instance,
00:27:15.14 what lipid droplets' role is in the brain.
00:27:18.14 It turns out that the brain has
00:27:20.29 a somewhat isolated triglyceride metabolism
00:27:22.17 from the rest of the body
00:27:24.18 because of the blood-brain barrier,
00:27:26.02 and particularly, glial cells,
00:27:27.23 under some conditions of neurodegeneration,
00:27:29.09 appear to accumulate lipid droplets to an abnormally high level.
00:27:32.04 Nonetheless, we really know very, very little
00:27:34.21 about this.
00:27:36.05 And of course, you probably have noticed as we went through our talk
00:27:38.10 that although we now have some models,
00:27:40.04 much remains to be discovered
00:27:42.03 in how exactly all these reactions occur
00:27:44.20 and what the mechanisms are.
00:27:47.07 Now, finally, Bob and I work together,
00:27:50.10 and we share a lot of our philosophy in science.
00:27:53.02 And I think we both share
00:27:55.24 that we consider the...
00:27:58.07 sort of the things we're most proud of
00:28:01.03 are the discoveries we make,
00:28:02.22 the friendships we cultivate through the years on this,
00:28:04.13 but also, the people we've been able to train
00:28:07.03 and we've been fortunate to work with.
00:28:09.03 Over the years, we've had many students and postdocs in the laboratory --
00:28:12.25 some of them are listed here,
00:28:14.12 particularly focusing on the ones
00:28:16.12 that have generated the data
00:28:18.00 that we've shared with you today.
00:28:20.01 Science is a human endeavor,
00:28:21.17 and it's not isolated.
00:28:23.03 For us, this is particularly true,
00:28:24.26 as we were fortunate to have
00:28:26.16 many, many collaborators over the years
00:28:28.00 that have provided access to technology,
00:28:30.16 critical insights,
00:28:32.08 a lot of discussion,
00:28:33.28 and that is really what makes our days fun.
00:28:36.03 And we would like to thank all of them here today,
00:28:38.10 in particular the ones that are listed here
00:28:40.12 and that have contributed to the data that we showed with you...
00:28:44.06 shared with you today.
00:28:46.06 And finally, we'd like to thank our funding sources,
00:28:48.22 which have made this research possible.
00:28:50.11 It's been a fantastic journey together,
00:28:53.12 and I look forward to a lot more in the future.
00:28:56.12 So, thank you very much.
00:28:58.21 Thank you.