Some of the classic symptoms associated with illness (inflammation, anorexia, fever, etc.) help the body fight disease and return to its normal state (homeostasis). This session will help you understand the molecular basis of inflammation. In summary, the body has specialized immune cells (sensor cells) that detect homeostasis disturbances and activate a series of responses that help clear the inflection and restore homeostasis. Finally, this session evaluates how some of these defenses, e.g. anorexia, can aid or disrupt recovery depending on the type of infection the body experiences.
00:00:16;00 My name is Ruslan Medzhitov.
00:00:17;00 I'm a professor at Yale University School of Medicine, and an Investigator from
00:00:21;17 Howard Hughes Medical Institute.
00:00:23;10 In this lecture, I will discuss inflammation, and I'll give a brief overview of the
00:00:28;21 field of inflammation and its role in both host defense and homeostasis, as well as in pathologies.
00:00:37;19 So, inflammation is an enormously huge field with a lot of details known about some aspects
00:00:44;00 of inflammation.
00:00:45;00 And to be able to summarize it in this short overview, I will use a couple of perspectives
00:00:51;28 that summarize some key features of inflammation, and specifically I will start by putting inflammation
00:01:00;09 in the context of the better-understood and better-defined phenomenon of homeostasis,
00:01:06;15 and I'll show the similarities, parallels, and how the two types of processes
00:01:13;13 interact with each other, causing both beneficial and detrimental outcomes.
00:01:18;07 So, just to remind you, homeostasis maintains stability of biological systems
00:01:25;13 in the face of perturbations.
00:01:27;00 And perturbations could be either external or internal to the system.
00:01:31;15 And inflammation is induced when these perturbations exceed homeostatic capacity of the system.
00:01:37;09 So schematically, this could be summarized as follows.
00:01:40;15 If we imagine the position of this ball as the state of the system, here, in the center
00:01:46;26 -- you can see, in the normal state -- the homeostasis is maintained by keeping
00:01:54;13 the state of the system in the desired position.
00:01:57;22 And when it deviates from that position, homeostatic mechanisms will bring it back.
00:02:02;20 But if perturbation is large enough and the system goes outside of its normal control zone,
00:02:10;06 outside of its normal homeostatic range, then homeostatic capacity is no longer sufficient
00:02:17;08 to keep the system in a desired state, and that's when inflammation is induced,
00:02:22;14 to force the system back into the homeostatic state.
00:02:25;17 That's one way to think about connections between homeostasis and inflammation.
00:02:30;21 So, inflammation is something that forces the system to go back into the homeostatic state,
00:02:36;00 when perturbations are large enough and when they overwhelm homeostatic capacity.
00:02:43;04 In modern terms, we can describe homeostasis using the idea of a control circuit.
00:02:49;22 And this is a very simple but very fundamental concept.
00:02:54;04 So here, what is summarized on this slide is key components of a homeostatic circuit.
00:03:00;26 And whenever we speak about homeostasis, that means that we talk about maintenance of
00:03:08;02 some variable of the system.
00:03:09;09 It could be blood sugar; it could be temperature; it could be sodium; it could be any of the
00:03:15;03 variables of the system that the system cares about and wants to maintain.
00:03:19;11 So, that's what's denoted here as X.
00:03:21;24 And when we refer to homeostasis of this variable, that means we want to keep it close to
00:03:27;26 some desired value.
00:03:29;14 And that's what's called the setpoint value, X' here.
00:03:33;02 So, that is... it is the value of that variable, or the difference of that variable value from
00:03:39;20 the setpoint value, that is monitored by the sensor.
00:03:43;00 The sensor is the component of a homeostatic circuit that monitors the value of the variable
00:03:47;28 the system cares about.
00:03:51;01 And the second essential part of the system is the effector part, and that's the part
00:03:56;24 that can change that value.
00:03:58;26 So, the sensor monitors the value; the effector can change the value.
00:04:02;06 And they need to communicate with each other through a signal that's denoted as C, here.
00:04:07;16 So for example, in the case of systemic homeostasis of blood glucose, X would be the actual concentration
00:04:15;23 of glucose in the blood, X' would be the setpoint value, which is in humans about 5 millimolar,
00:04:22;23 and the sensors would be pancreatic alpha and beta cells that monitor how much glucose
00:04:27;04 we have in the blood.
00:04:28;08 Are you eat, glucose level goes up.
00:04:30;26 Beta cells in the pancreas will detect that and will start producing insulin, which is
00:04:35;06 the example of the signal, shown here, which will go on to act on its effectors,
00:04:41;06 which include skeletal muscle, fat, and liver.
00:04:44;04 And the effect of insulin on these target tissues will be to lower blood glucose level,
00:04:49;04 for example by inducing uptake into those tissues or conversion into glycogen or lipids.
00:04:54;24 If the glucose level is lower than the setpoint value, then alpha cells of the pancreas
00:05:01;08 will detect that, a low lev... a low level, and start producing a different hormone,
00:05:06;05 which is glucagon, which will act exactly, again, on effector cells... effector tissues and organs,
00:05:11;25 for example liver, and cause them to start producing glucose to raise it...
00:05:17;07 to raise the level to the desired value.
00:05:20;06 So, that's how a homeostatic circuit works at... at an organismal level, a tissue level,
00:05:27;10 and a cellular level.
00:05:30;12 Now, the origin of the concept of inflammation goes back to... it can be credited to many people,
00:05:38;14 but the two that I want to highlight here are Rudolf Virchow Elle Metchnikoff,
00:05:44;01 who were contemporaries and colleagues.
00:05:49;01 And so, Virchow, of course, is credited with the development of the modern science of pathology,
00:05:53;25 of cellular pathology.
00:05:56;15 And he was an extremely influential scientist in Europe at the time.
00:06:01;01 And Elle Metchnikoff, of course, is known for his discovery of phagocytes and its...
00:06:07;10 their role in innate immunity.
00:06:09;12 But in the context of inflammation, these two individuals provided very important
00:06:15;15 conceptual contributions.
00:06:16;15 But there was one important difference between them, in that Virchow primarily viewed inflammation
00:06:21;15 as a pathological process, whereas Metchnikoff recognized early on that, in addition to these
00:06:27;17 pathological outcomes of inflammation, that the... the primary reason for an
00:06:34;03 inflammatory response is to provide protection from infections.
00:06:38;24 And he visualized and conceptualized the inflammatory response as being part of a spectrum,
00:06:47;28 where at the base of the spectrum would be what he called "harmony/disharmony balance",
00:06:53;28 and this is what we currently would call homeostasis, but the term homeostasis wasn't coined yet,
00:06:59;22 until 1929.
00:07:00;27 Then, the next level would be physiological inflammation, when inflammation plays
00:07:07;11 beneficial roles in host defense.
00:07:09;20 And then pathological inflammation, and finally immunity.
00:07:13;08 And that... that concept of physiological inflammation and the spectrum of
00:07:18;28 inflammatory response from homeostasis to immunity is actually a very profound insight which was
00:07:25;10 largely forgotten until very recently.
00:07:27;11 And only now we are starting to rediscover and realize these fundamental connections
00:07:32;27 between physiological processes and inflammation.
00:07:38;18 So, taking that, Metchnikoff's idea, and putting it in... looking from different dimensions,
00:07:46;13 we can summarize it as follows, as the spectrum of degrees of deviation from homeostatic states.
00:07:53;06 So here, on the... on the left side, you see the range of conditions of a system that
00:08:00;02 would be within a homeostatic state.
00:08:02;23 If it deviates far enough from that, that's what... what we would call a stress response,
00:08:07;11 or we could also call it a physiological inflammatory response.
00:08:11;07 And if it deviates much further than that, that's what we would call inflammation proper.
00:08:16;23 And so this... any deviation from a normal state, therefore, can be... can lead to
00:08:22;07 the induction of the inflammatory response.
00:08:24;18 So, the causes of inflammation from that perspective can be summarized as follows.
00:08:32;09 In the center here, in the middle, you can see that loss of homeostasis per se is sufficient
00:08:36;27 to lead to inflammation as... as I just mentioned.
00:08:41;02 But in addition, there could be exogenous perturbations that can lead to loss of homeostasis.
00:08:48;11 And the two major types of such perturbations will be pathogens (during infection)
00:08:53;25 as well as toxins and allergens and virulence factors produced by pathogens.
00:08:59;02 So, both pathogens and... and toxins can cause loss of inflammation... loss of homeostasis
00:09:06;14 that... and that can lead to inflammation.
00:09:09;10 But in addition, the immune system developed two pre-emptive mechanisms to trigger a protective
00:09:15;04 inflammatory response, even before pathogens or allergens can cause damage to the system.
00:09:22;07 And there are two fundamental ways that the immune system detects these inducers of inflammation.
00:09:27;10 At the top here, what I call structural feature recognition is the property of the
00:09:32;17 innate immune system to detect invariant structures associated with microbial cells.
00:09:40;03 This is sometimes called the pattern recognition system, where receptors of the immune system
00:09:46;10 detect conserved structures that have found in most microbes, for example we lipopolysaccharides
00:09:53;10 of the cell wall or peptidoglycans, lipoteichoic acids, and so on.
00:09:58;07 And detection of these structures is sufficient to trigger inflammation.
00:10:02;23 On the other hand, allergens and toxins and virulence factors, they're extremely diverse.
00:10:07;00 There... there are many different types and there is no way to detect them all based on
00:10:11;08 structural features, because they don't share any structural features.
00:10:15;09 And the strategy of recognition here is what I would call a functional feature recognition,
00:10:20;16 because what is detected is not specific structures but rather specific biochemical activities,
00:10:27;02 such as protease activities, lipase activities, lipid binding, membrane perturbations,
00:10:33;25 pore formation, and so on.
00:10:35;16 Those functional features are detected by that system, and that also can lead to inflammation.
00:10:40;13 And that type of strategy is particularly important in allergic inflammation.
00:10:46;07 So, based on these ideas, we now can summarize how the immune system operates based on
00:10:53;20 the simple logic of the control circuits.
00:10:57;05 So, we know that the immune system... one of the major functions of the immune system
00:11:01;18 is to detect pathogens and to provide a protective response against them by, for example,
00:11:09;01 destroying them or expelling them from the organism.
00:11:14;00 To do so, the immune system has to have two essential components.
00:11:17;04 It has to have a pathogen-sensing component, or pathogen-sensing cells, and it has to have
00:11:25;10 antimicrobial effector cells, and sensor and effector cells have to communicate with
00:11:31;06 each other through a signal, and once the effector receives the signal from the sensor, it elicits
00:11:36;22 a response that leads to defense from the pathogen.
00:11:40;02 So, that's a very simplified view of the immune system.
00:11:45;13 And the signals that are involved in communication between sensors and effectors are what contributes
00:11:51;02 to this enormous complexity of inflammation and understanding of the immunity.
00:11:57;10 There are many different types of signals in the context of inflammation.
00:12:01;28 The signals are usually called inflammatory mediators.
00:12:06;13 And two major types of inflammatory mediators are signals called chemokines and cytokines.
00:12:14;14 Chemokines are short polypeptides that are produced upon infection by sensor cells
00:12:24;05 that detect pathogens or tissue damage.
00:12:27;08 And what chemokines do is they recruit effector cells to the site of infection.
00:12:33;15 For example, macrophages that function as sensor cells, when they detect bacterial pathogens,
00:12:40;04 will produce chemokines that will recruit neutrophils to the site of infection,
00:12:43;24 and then neutrophils will take care of the pathogens.
00:12:47;11 The second type of inflammatory mediators are cytokines.
00:12:50;26 And this is a... again, a very diverse group of signals that belong to different structural families,
00:12:56;14 but basically what cytokines do... they... they, again, are produced by sensor cells
00:13:01;14 when they detect infection, and they activate effector cells to elicit various
00:13:09;16 antimicrobial functions.
00:13:11;25 So, with this in mind, we now can summarize much of the inflammation and diversity of inflammation
00:13:22;02 into these simple and universal components of the inflammatory pathway.
00:13:28;00 Any type of inflammation includes these four universal components.
00:13:31;12 There is always some type of an inducer of inflammation, for example, pathogen, toxin,
00:13:36;22 tissue damage, or loss of homeostasis.
00:13:39;13 There are sensors that detect the inducers.
00:13:42;19 These include various types of cells of the innate immune system, such as macrophages
00:13:48;04 and mast cells, but also various types of sensory neurons.
00:13:54;27 And the sensor cells produce inflammatory mediators, which include cytokines, chemokines,
00:14:01;15 as well as bioactive amines like histamine, peptides, like bradykinin, as well as
00:14:07;28 lipid mediators called eicosanoids, which include, for example, prostaglandins.
00:14:13;07 And these mediators then act on various target tissues.
00:14:16;25 And almost any tissue in the body can be a target for different types of inflammatory mediators.
00:14:23;01 So here, I'm showing liver, vasculature, epithelial cells, and neutrophils.
00:14:30;21 When mediators act on these effector cells, they cause appropriate changes in their
00:14:36;16 state and their function, or in their positioning.
00:14:39;02 Again, chemokines can recruit neutrophils to the site of infection.
00:14:44;11 Cytokines acting on hepatocytes sites or vascular endothelium will cause their activation,
00:14:52;21 changing protein secretion or permeability of the epithelium.
00:14:56;02 And in the case of mucosal epithelium, they can change the production of antimicrobial
00:15:02;21 peptides or mucus.
00:15:04;21 So, this is this inflammatory pathway.
00:15:08;00 And as you... as you may notice, there is... this is very much related to... it's the
00:15:17;21 same kind of a control circuit we just discussed for homeostasis, where we have a sensor,
00:15:23;26 a signal that connects sensor to the effector, and the effector.
00:15:28;06 The only difference is that in this case what is monitored is not a homeostatic variable,
00:15:33;15 but rather some inducer of inflammation, such as a pathogen or toxin.
00:15:38;22 So, there are these clear parallels between homeostatic and inflammatory control circuits.
00:15:45;04 The reason for that has to do with the fundamental importance of these type of control circuits.
00:15:50;02 They're everywhere, from engineering systems to biological systems.
00:15:55;04 And again, the differences between them are related to the types of inducers that are...
00:16:03;18 that are detected by sensors, or homeostatic variables detected by homeostatic sensors.
00:16:10;07 But we should also keep in mind that sometimes the differences between homeostatic and inflammatory
00:16:16;10 control circuits can be arbitrary, because inflammatory mediators used by homeostatic...
00:16:21;16 by inflammatory circuits can also have some homeostatic functions, and homeostatic signals
00:16:29;11 used by homeostatic circuits can participate in regulation of the inflammatory response.
00:16:37;10 There are actually two different designs... versions of the control circuits.
00:16:45;24 Here on the top is the control circuit I just mentioned.
00:16:48;19 That's the simplest one, where you just have sensor and an effector, and the signal
00:16:52;24 that connects them.
00:16:54;21 There is another type of a circuit which has an additional component in between.
00:17:00;12 And that's what's called a controller or integrating unit.
00:17:04;04 So, here we have a sensor that monitors an inflammatory inducer or homeostatic variable.
00:17:11;21 It produces a signal that then acts on the controller, and then the controller does
00:17:15;23 some type of a computation, and then sends a second signal to the effector.
00:17:19;28 This type of a design is particularly prevalent in both immune and nervous systems.
00:17:28;02 In the case of the immune system, the control... the role of a controller is typically played
00:17:31;17 by a lymphocyte.
00:17:33;03 And in the case of the nervous system, it's played by various types of interneurons.
00:17:39;04 So, sensor cells, again, after detecting the inducer, produce one signal, and then
00:17:46;24 the controller produces a second signal.
00:17:48;20 And these two types of signals are distinct in the immune system, as we will discuss.
00:17:55;10 So from that perspective, we can summarize the entire operation of the immune system
00:18:02;00 as... as connections between pathogen sensors and effectors.
00:18:09;11 And there are three types of disconnections.
00:18:11;17 The simplest one is shown at the top, where the sensor and the effector is the same entity,
00:18:17;10 the same cell.
00:18:19;16 The sensor would be, for example, a receptor, and the effector would be, for example,
00:18:24;05 an antimicrobial enzyme.
00:18:27;11 The second type is when the sensor produces a signal that acts on the effector,
00:18:32;07 as we just discussed.
00:18:33;16 And the third type when there is a lymphocyte in between.
00:18:36;22 And the first two types belong entirely to the domain of the innate immune system,
00:18:41;22 and the second... the... the third one, it can be either innate or adaptive immune system,
00:18:48;07 depending on the type of lymphocyte involved.
00:18:50;26 So, we will go through the different versions of these circuits to illustrate how they operate
00:18:58;09 in the context of infection.
00:18:59;25 So, the simplest one is when a cell like a macrophage encounters a pathogen, like a bacterium, phagocytoses then kills it.
00:19:08;13 So in this case, the sensor would be receptors that detect the microbe, and the effectors
00:19:13;17 would be phagocytic machinery and lysosomal enzymes that will kill the microbe.
00:19:17;24 So, that's the simplest one.
00:19:19;16 And more... more commonly, when macrophages detect pathogens, they will produce a signal
00:19:25;27 that will connect them to the effector, such as a neutrophil, and it will either recruit...
00:19:31;22 recruit or activate neutrophils.
00:19:33;05 And neutrophils are specialized in killing bacteria fungi, and they will proceed to do so.
00:19:39;16 And then the system operates in this manner to provide protection from infection.
00:19:47;14 And finally, the third system... the third design would be when cells... sensor cells
00:19:52;25 like macrophages again detect pathogens, then they produce cytokines that act on, now...
00:19:58;12 on lymphocytes first.
00:20:00;23 And then lymphocytes -- that could involve T cells or different types of innate versions
00:20:08;02 of T cells that I'll describe in a second -- which then produce the second-order cytokines.
00:20:13;21 In this case, a first-order cytokine would be IL-12 produced by macrophages, which acts
00:20:18;11 on lymphocytes, and causes lymphocytes to produce second-order cytokine
00:20:22;08 such as interferon-gamma,
00:20:24;08 which will then act on effector cells -- that will be macrophages -- and cause them to
00:20:28;23 become activated to kill bacteria.
00:20:30;23 So, this design is actually... it captures most of the operation of the immune system.
00:20:37;28 And most of the complexity comes from the generation of lymphocytes and their
00:20:42;24 functional heterogeneities.
00:20:44;12 So, we will now... we'll go quickly through different components of these systems,
00:20:49;15 starting with sensors.
00:20:51;04 There are several cell types that can function as sensors in the inflammatory and immune
00:20:57;15 These include macrophages, mast cells, epithelial cells, dendritic cells, and plasmacytoid dendritic cells.
00:21:05;10 So, these are different sensor cells that have different types of specializations.
00:21:11;06 Macrophages, mast cells, and epithelial cells are kind of general-purpose sensors.
00:21:16;13 They detect a large variety of pathogens and other types of inflammatory inducers.
00:21:24;23 Dendritic cells are specialized on activating T cells.
00:21:28;07 And plasmacytoid dendritic cells are specialized on antiviral responses.
00:21:33;21 The lymphocyte part is... that's where a lot of complexity comes in.
00:21:40;11 They can be... there are two versions of circuits, depending on what kind of lymphocyte is used.
00:21:46;20 And broadly speaking, there are innate lymphocytes that participate in the innate immune system,
00:21:52;15 and lymphocytes involved in the adaptive immune system, which are T and B cells.
00:21:57;22 The innate lymphocytes, again, come in two versions.
00:22:00;04 There are so-called innate lymphoid cells, which are... have been relatively
00:22:04;16 recently discovered.
00:22:05;16 They don't have T cell receptor.
00:22:07;26 They reside in tissues and they respond to cytokines produced by sensor cells,
00:22:13;15 and in turn produce cytokines that affect effectors.
00:22:16;07 Then there are inmate-like lymphocytes that have T cell receptor, but it's not a random receptor;
00:22:23;05 it's invariant, so it's designed to detect very specific subsets of antigens.
00:22:29;10 And finally, the adaptive immune system of course has antigen receptors, T cell receptor
00:22:35;00 and immunoglobulin receptor for B cells, and these are the most complicated cells of the
00:22:41;21 immune system because of the way that they develop and because of the way that
00:22:46;02 their receptors are assembled, and all the additional steps that are involved to make the cells functional,
00:22:51;18 because their receptors are generated at random.
00:22:55;15 Again, when lymphocytes detect cytokines, they respond by producing cytokines.
00:23:02;03 And what's summarized here are some of the types of cytokines that... on the left side,
00:23:07;21 that act on lymphocytes and the different types of lymphocytes
00:23:12;24 and the second-order cytokines produced by lymphocytes.
00:23:16;02 And then these things... cytokines produced by lymphocytes and, again, act on the effector cells,
00:23:21;07 which are... examples are shown here: macrophages, neutrophils, basophils, eosinophils,
00:23:27;06 mast cells, and epithelial cells.
00:23:30;06 Depending on the type of cytokine produced, there would be different type of change
00:23:33;18 in these cells, effector cell types.
00:23:37;06 And in addition to these specialized effectors of the innate immune system, practically
00:23:43;06 any cell in the body can be an effector, because most cells express receptors for at least
00:23:48;15 some of the cytokines produced by lymphocytes.
00:23:52;10 So, now we will quickly go over... with these concepts in mind, we will go over some of
00:24:01;00 the key features of the inflammatory response.
00:24:04;08 And we have to start with one of the oldest notions in the field of inflammation,
00:24:11;05 which is the cardinal signs of inflammation.
00:24:13;10 These were first defined by a Roman physician, Cornelius Celsus, in the first century AD.
00:24:22;12 He defined them as redness and swelling with heat and pain.
00:24:27;25 That was his description of how to diagnose inflammation.
00:24:33;10 And much later, Rudolph Virchow added a fifth cardinal sign of inflammation, which is disturbance
00:24:39;14 of function or loss of function of tissues.
00:24:42;22 The four cardinal signs described by Celsus are a consequence of the changes that
00:24:49;09 occur during acute inflammation.
00:24:52;07 And these are local changes due to alterations in the local vasculature, as we will discuss next.
00:25:01;15 So, this is what typically happens during the most common types of inflammatory responses,
00:25:06;27 when you have a mild infection or papercut or some other splinter or some other injury
00:25:15;03 to the epithelial surfaces.
00:25:18;04 So, microbes or damage to the tissue are detected by sensor cells such as macrophages, dendritic cell,
00:25:28;27 and mast cells, as I just mentioned.
00:25:31;24 And once they detect microbes or tissue damage, these cells start producing inflammatory mediators
00:25:37;00 such as cytokines and chemokines.
00:25:40;00 And one of the effects of these inflammatory mediators, locally, within the tissue,
00:25:44;07 is to act on the local microvasculature.
00:25:47;26 And specifically, they... by acting on postcapillary venules, they cause several characteristic
00:25:54;26 changes of the endothelium of the venules.
00:25:59;04 They cause vasodilation, so there is increased blood flow.
00:26:03;13 And increased blood flow causes heat and redness.
00:26:07;20 And it causes increased vascular permeability, so that plasma starts going from the
00:26:15;02 inside of blood vessels into extravascular spaces within tissues.
00:26:19;12 And that causes swelling, or edema.
00:26:23;12 And together the edema, and effects of inflammatory mediators, also can cause pain.
00:26:29;13 So, redness, swelling, heat, and pain are all [results] of these vascular changes
00:26:35;09 that occur locally.
00:26:37;04 Another important change that happens is that endothelium within postcapillary venules becomes
00:26:42;19 activated, in the sense that it now becomes... acquires adhesive properties such that neutrophils
00:26:51;09 and monocytes and other cell types that go through blood vessels... normally, they would
00:26:56;13 pass through.
00:26:57;16 But when... if there is a local inflammation, the local endothelium becomes sticky, so that
00:27:03;02 these cells now adhere or attach to endothelium, and ultimately they crawl through the endothelial wall
00:27:11;06 into the tissue.
00:27:13;00 And that's the process called extravasation.
00:27:16;01 And the point of that process is to deliver the circulating effector cells to
00:27:21;05 the site of infection.
00:27:22;05 And actually, Elle Metchnikoff was the first to recognize that that's the point of vascular
00:27:27;08 changes during inflammation.
00:27:30;00 So, once neutrophils and other effector cells get to the site of the inflammation,
00:27:35;00 where inflammation is induced, they will then seek out pathogens and will destroy them or
00:27:41;21 repair the damaged tissue.
00:27:44;27 Another important point that was realized probably in the last decade or so is that
00:27:55;11 once inflammation accomplished its goal, which is elimination of pathogens, for example,
00:28:02;23 that is not enough to get back to the normal state.
00:28:06;25 If you just eliminated the cause of inflammation, it doesn't mean that the system automatically
00:28:11;08 goes back by default into homeostatic state.
00:28:14;22 There is another phase between inflammation and homeostatic state -- that's called resolution --
00:28:19;20 that needs to be actively engaged.
00:28:22;24 This is analogous to a situation if you have, for example, a broken pipe and there's flooding
00:28:27;15 in the system.
00:28:28;15 The cause of the mess would be the broken pipe, so let's say you fix the pipe.
00:28:32;13 That doesn't mean that the system is now back into its original state.
00:28:36;11 Now you have all the water on the floor and you need to get rid of it to return actively
00:28:40;20 back to homeostatic state.
00:28:42;04 So, that's what resolution does.
00:28:44;09 After inflammation accomplishes its goal, there's a lot of mess within the tissue --
00:28:50;11 there are many dead cells, there is destroyed extracellular matrix, and all of that has to be cleaned up
00:28:57;26 and changed back to the original state.
00:29:01;04 And of course this is something that requires a highly orchestrated and regulated process.
00:29:08;06 And that's what resolution does.
00:29:10;06 And resolution of inflammation is a very important but still not fully understood process,
00:29:16;21 but it's... it's well recognized now that it's an active process -- it's not just
00:29:22;00 passive cessation of inflammation -- and that it's needed to restore the homeostatic state.
00:29:27;23 An additional important point to understand about inflammation is that there are
00:29:35;13 not just different types of inflammation based on the causes, but there are also different modalities
00:29:40;04 of inflammation.
00:29:41;04 And they are historically defined as acute and chronic modalities of the inflammatory response.
00:29:49;12 So they, as the names imply, acute and chronic inflammation obviously differ in duration.
00:29:54;16 Acute inflammation can last from hours to days, and chronic inflammation typically
00:30:01;02 can last from weeks to months to years.
00:30:06;22 But more importantly, it's not just the kinetics of the response, but more importantly
00:30:11;21 acute and chronic inflammation are qualitatively distinct.
00:30:17;20 And the common causes of chronic inflammation include failure to eliminate the inflammatory inducer,
00:30:23;05 for example, if there is a persistent infection.
00:30:27;04 It's a failure of resolution of inflammation.
00:30:31;17 And in some cases it could be a positive feedback, such that the consequence of the inflammatory response,
00:30:37;22 for example, collateral tissue damage, may also be a cause for a secondary inflammatory response.
00:30:44;03 And potentially that can sustain the inflammatory state.
00:30:49;07 The qualitative differences between acute and chronic inflammation have to do with
00:30:53;16 the types of cells involved.
00:30:54;27 It's mostly neutrophils and eosinophils in acute inflammation, but mostly lymphocytes
00:31:01;19 in chronic inflammation, as well as macrophages.
00:31:05;13 And there are many other differences related to the type of the mechanism used to...
00:31:12;19 to deal with a persistent inflammatory inducer that's used during chronic inflammation.
00:31:18;01 Like other defenses, inflammation always operates at a cost.
00:31:23;01 And these costs can be divided into distinct categories.
00:31:27;06 The first class of causes of... the first type of costs of inflammation has to do with
00:31:35;14 intentional suppression of physiological functions that are lower priority than the inflammatory response
00:31:43;00 and that are somehow incompatible with the inflammatory response.
00:31:47;24 For example, if you're sitting on the couch and watching TV and there is a fire,
00:31:52;24 then watching TV, as a function, would be incompatible with dealing with the fire.
00:31:57;28 And it also would be obviously lower priority than dealing with the fire.
00:32:02;08 So, you will intentionally stop watching TV, so that would be a cost, but it's a low cost
00:32:07;17 compared to the benefit of putting away the fire.
00:32:12;05 And then the second type of course is unintentional cost.
00:32:16;10 That is, it's not something you want, but it's something you can't avoid.
00:32:21;03 It's unintentional and unavoidable costs, such as collateral tissue damage.
00:32:24;20 So, when you're putting out the fire and putting water on it, you will cause perhaps some
00:32:30;26 collateral damage to the rest of the room.
00:32:33;21 So, these are two different types of costs.
00:32:36;17 And the sum of these two costs has to be lower than the benefit provided by inflammation
00:32:40;15 for... for... for the system to be evolutionarily stable.
00:32:45;20 So, inflammation can be pathological, therefore, for several reasons.
00:32:51;17 And what's important to understand is that even an appropriately controlled inflammatory response
00:32:57;06 operates at the expense of other functions.
00:33:00;04 So, it's often said that inflammation is beneficial but when dysregulated can be pathological.
00:33:05;01 We should appreciate that even perfectly controlled inflammatory responses operate at a cost,
00:33:09;21 and sometimes these costs can manifest as symptoms that we may refer to as a disease.
00:33:17;08 The second reason for pathology of inflammation is when the response is excessive and...
00:33:24;26 either in magnitude or in duration.
00:33:27;01 And the third cost would be when the response is induced when it shouldn't be induced,
00:33:30;26 for example when it's mistargeted against something that is not harmful.
00:33:37;22 And this is summarized in this schematic.
00:33:41;24 When inflammation causes swelling, pain, fever, mucus overproduction, coughing, sneezing, diarrhea,
00:33:47;11 these are all defenses.
00:33:49;26 These are all manifestations of defenses.
00:33:51;25 They are protective from different types of noxious challenges, but obviously all of them
00:33:57;17 are processes that come at a cost.
00:34:01;15 We do feel ill when we experience those reactions, even though they are protective.
00:34:07;06 And what makes it worse is when they're protective but excessive.
00:34:11;08 Then they would be clearly just pathological.
00:34:15;13 And so these are two different outcomes that need to be distinguished: when pathology is
00:34:19;23 due to excessive response versus when pathology is simply the cost we have to pay for a normal response.
00:34:27;03 And then the third type of pathological outcome is more obvious.
00:34:31;06 It's when there is a... just collateral tissue damage or a mistargeted response.
00:34:36;09 So, when we put it this way, it's clear that the three types of pathological outcomes
00:34:42;01 are very different.
00:34:43;05 And... for example, you don't want to interfere with the first one, you want to dial down
00:34:47;16 the second one, and you want to stop the third one.
00:34:50;24 And the challenge is to be able to distinguish which one they belong to so that we know
00:34:55;19 what to do with them.
00:34:57;08 So, the take-home messages in this brief overview is that inflammation is normally
00:35:03;20 a protective response to infection and injury and other... and loss of tissue homeostasis,
00:35:10;06 that it's induced when homeostatic capacity is overwhelmed, and that all of the diversity and complexity
00:35:18;08 of inflam... of inflammation can be summarized in terms of the inflammatory pathway that
00:35:22;22 consists of inducers, sensors that detect them, mediators they produce, and the effectors
00:35:28;10 that eliminate the inducers.
00:35:32;02 And inflammation is normally followed by a resolution phase, which returns the system
00:35:36;12 to homeostasis.
00:35:39;16 And an inflammatory response always operates at a cost to incompatible lower-priority functions.
00:35:48;16 And inflammation can cause pathology when it's excessive, inappropriately induced,
00:35:53;05 or due to collateral damage.
00:35:55;22 And that completes this overview.
00:35:59;02 I will discuss in the next talk some specific examples of inflammation in the context of
00:36:08;08 inflammatory diseases.
00:36:09;27 And thank you for your attention.
00:00:16;00 My name is Ruslan Medzhitov.
00:00:17;00 I'm a professor at Yale University School of Medicine and an Investigator of the
00:00:22;00 Howard Hughes Medical Institute.
00:00:23;13 And in this lecture, I will discuss our recent study on the effect of inflammation in acute illness
00:00:30;11 and the role of disease tolerance in surviving acute illness.
00:00:39;11 As I discussed in the introductory lecture, the costs of inflammation can be broken down
00:00:44;01 into two categories.
00:00:45;10 The first is the intentional suppression of lower priority functions that are incompatible
00:00:51;00 with the goals of the response.
00:00:53;17 And the second type of costs are unintentional but unavoidable loss of function, for example,
00:01:00;05 due to collateral damage.
00:01:02;22 And the sum of these two costs has to be lower than the benefit provided by the inflammation
00:01:07;18 in order to... for the system to evolve the way it is.
00:01:12;28 And this relation between the cost and benefit is what's really essential for understanding
00:01:17;14 many biological functions and their deviations into pathological states.
00:01:22;11 So, every biological trait can be characterized by some benefit it provides and some cost
00:01:28;14 at which it operates.
00:01:30;10 And this can be schematically shown as follows.
00:01:32;26 So, if we plot benefit versus cost for a system, for a biological trait to evolve the benefit
00:01:41;06 has to be higher than the cost.
00:01:43;03 So, any trait that would be in the green triangle part of the plot, where the benefit is higher
00:01:48;27 than the cost, would be evolutionary acceptable.
00:01:54;01 And anything in the red triangle, where the cost is higher than the benefit, would be
00:01:57;25 eliminated by natural selection.
00:02:00;28 And as you can see, the higher the benefit of the trait the higher is the acceptable cost.
00:02:07;27 And this view is from the evolutionary perspective.
00:02:11;14 So, all evolution cares about is that the benefit is higher than the cost.
00:02:17;22 And the same plot from the patient or physician perspective would look like this.
00:02:22;21 Here, again, anything that is in the upper-left triangle, where the benefit is higher than
00:02:29;16 the cost, would be a fair game from evolutionary perspective.
00:02:32;27 But as you move to the right side, when the cost becomes higher and higher, because benefit
00:02:38;09 is high, then this would be associated with conditions we would often refer to as
00:02:44;19 pathological or disease conditions.
00:02:46;09 So, if you're a patient and you're having an acute infection, there are ongoing
00:02:52;07 immune and inflammatory responses, but at the same time you feel very ill as a consequence of
00:02:56;21 these responses.
00:02:59;04 This would be a situation in the upper-right corner, where the benefit is still higher
00:03:04;24 than the cost, but the cost is so high that it makes us feel sick.
00:03:10;03 And it is this position on the plot, on the upper-right corner, where trades that provide
00:03:19;00 very high benefit will have very high acceptable costs.
00:03:22;22 And some diseases can be due to these types of trades, with very high benefits coming
00:03:28;28 with the high costs.
00:03:30;08 And that's the situation we've been interested in investigating.
00:03:34;13 What kind of mechanisms operate when the cost of the response is so high that it makes you
00:03:39;10 feel ill, and you're in fact close to the... to the condition where it could be life-threatening?
00:03:48;01 And one famous set of conditions like this is associated with acute illness.
00:03:53;23 And what's been known for... for a long time is that, during acute illness,
00:03:58;19 humans and other animals experience what's known as sickness behaviors.
00:04:04;04 And these are stereotypical responses that include loss of appetite, social withdrawal,
00:04:11;06 fatigue, altered sleep patterns, cessation of grooming, and suppression of libido.
00:04:17;10 And why they all happen and why there's this particular combination of behaviors is
00:04:23;12 not very clear, but it's clear that they occur in all animals studied.
00:04:28;24 They've even been observed in insects.
00:04:31;23 And it's been concluded, decades ago, that these are not just debilitations of
00:04:40;18 normal behaviors, but rather these are motivated behaviors.
00:04:43;21 In other words, they occur with a purpose.
00:04:46;14 There is some intentional induction of these responses.
00:04:50;26 But what the purpose of these responses is... is... has been less clear.
00:04:54;28 And that's what we investigated in this study I will describe today.
00:04:59;14 So, we were particularly interested in understanding the phenomenon of disease-induced anorexia.
00:05:09;23 We are all familiar with this phenomenon.
00:05:11;19 When you have acute infection, like flu infection or a severe cold, your appetite goes away.
00:05:18;01 You don't want to eat.
00:05:19;02 You want to sleep a lot.
00:05:20;17 And we asked why is it that we don't eat when we are very sick.
00:05:25;18 And to model that, we studied an infection with a bacterial pathogen called
00:05:32;18 Listeria monocytogenes, which is a very common bacterium that causes food poisoning.
00:05:39;11 And what we did... here, we infected mice with a sublethal dose of Listeria and monitored
00:05:47;18 their food consumption.
00:05:49;09 And as you can see, the red line is mice that are infected with Listeria, and as you
00:05:55;18 can see there is a very profound suppression of food consumption.
00:05:59;28 And go on... into this anorexic state until they start recovering from infection,
00:06:06;26 at which point they will regain food consumption.
00:06:11;08 And we asked, why is it that they don't eat?
00:06:13;21 What would happen if they are forced to eat?
00:06:17;27 And to address that, we fed the mice with the same amount of the same type of food
00:06:24;15 that they normally consume.
00:06:26;16 And we only provided them about 20 percent of normal daily caloric intake.
00:06:32;26 So, it's just a small fraction of what they would normally eat.
00:06:36;28 And we used, in this case, a dose of Listeria that kills 50 percent of mice -- so it's called
00:06:43;02 lethal dose 50 or LD50 -- which is shown in the black line.
00:06:47;25 These are mice that are control mice.
00:06:51;15 And then the experimental mice were given food.
00:06:54;03 And as you can see, all of them died within 10 days, indicating that eating during bacterial infection
00:07:01;17 can be lethal.
00:07:03;13 And that result actually is not new.
00:07:05;02 It was first reported in 1979 with a similar model, with Listeria infection, that force-feeding
00:07:11;08 during infection can be... it can lead... can increased... it can increase lethality.
00:07:16;28 So, then we asked, what is it in the food that causes this effect?
00:07:21;14 And we tested, separately, proteins, carbohydrates, and fats.
00:07:26;19 And found that the effect of the... this effect of feeding was due to carbohydrates,
00:07:32;22 specifically due to glucose, because if we would just give mice glucose at the time of the... of the infection,
00:07:39;25 then 100% of them would succumb to infection.
00:07:45;07 And that was very interesting because it indicated that just glucose -- that simple metabolite,
00:07:50;07 an essential metabolite -- is sufficient to cause such a dramatic effect on survival.
00:07:56;24 And then we asked, what would happen if we do the opposite manipulation, if we
00:08:02;20 prevent glucose utilization?
00:08:04;21 And to do so, we used a metabolite derivative called 2-deoxyglucose, or 2DG, which is
00:08:12;02 a glucose variant that can be taken into the cells but cannot be metabolized, so it
00:08:16;18 prevents glucose utilization even if glucose is present in the system.
00:08:22;00 And when we gave mice 2DG twice a day, by injecting it either intraperitoneally or
00:08:29;01 giving it orally or giving it intravenously... it didn't matter which route we used.
00:08:35;16 And as you can see in the blue line here, 100% of mice now could survive this infection
00:08:41;12 that otherwise would kill 50% of mice.
00:08:44;04 So, that was very exciting because it indicated that blocking glucose utilization can
00:08:50;03 protect mice from infection, and giving them glucose can promote mortality.
00:08:57;22 Then we asked whether this is something unique to Listeria or can be generalized to
00:09:02;20 other types of bacterial infections.
00:09:05;28 And to address this, we used a common model of bacterial sepsis that is caused by...
00:09:12;21 not by live bacteria but by specific a bacterial component called lipopolysaccharide, or LPS,
00:09:19;17 which is present in all gram-negative bacteria and which is well-known to induce a very dramatic
00:09:25;20 inflammatory response.
00:09:27;07 So, inflammation caused by gram-negative infections is in large part due to LPS.
00:09:32;24 So, if we just use LPS instead of live pathogens, then we simplify the system and
00:09:38;04 eliminate all the variations due to pathogenicity of different bacteria.
00:09:44;21 And as you can see here, we... when we give an LD50 dose of LPS, which is the line
00:09:50;22 in the middle, we have about 50% of mice that would succumb to sepsis.
00:09:56;11 Then, if we give them either a control -- PBS, phosphate... phosphate-buffered saline,
00:10:03;22 a physiological solution -- or give them food, you can see that there's a dramatic difference
00:10:08;01 in survival.
00:10:09;01 So, mice that received food, most of them would die.
00:10:13;09 And then we asked if this effect, again, is due to glucose.
00:10:16;19 And we performed, again, a similar experiment, giving either glucose or 2DG.
00:10:23;24 And as you can see, now 100% of mice that received glucose would die from LPS sepsis,
00:10:31;00 and 100% would survive if they are given 2-deoxyglucose.
00:10:34;17 So, this was very exciting because this is a very simple manipulation.
00:10:38;22 We're just using either glucose or anti-glucose.
00:10:43;01 And we have this profound, 100% effect on survival in a condition that is
00:10:51;05 otherwise intractable.
00:10:52;16 It's uhh... sepsis is a very complex disease that has a very high mortality rate,
00:11:00;11 and there are still very few treatment options for sepsis.
00:11:02;28 So, we were very excited to see that such a simple manipulation can have such a dramatic
00:11:07;18 effect on survival.
00:11:09;28 Interestingly, these effects of glucose and 2-deoxyglucose were not due to changes in
00:11:17;04 the magnitude of the inflammatory response.
00:11:20;14 So, if we measure the major inflammatory cytokines, including TNF, IL-6, or acute-phase proteins
00:11:28;13 such as serum amyloid protein, you can see that, regardless of whether it's
00:11:33;19 a control mouse or a mouse given glucose or 2DG, the level of inflammatory response was the same.
00:11:39;14 So the fact that mice given glucose died and mice given 2DG survived is not due to changes
00:11:46;02 in the inflammatory response.
00:11:48;04 So, then we asked, what is it due to?
00:11:50;28 And of course, when mice don't eat they undergo a fasting metabolic state.
00:11:57;21 And we then asked whether this fasting metabolism is the one that matters, rather than the inflammation
00:12:05;06 And just to remind you, what happens during fasting is that glucose level goes down and
00:12:12;07 therefore insulin level goes down.
00:12:14;19 And most organs switch from using glucose to switching... to using fatty acids produced...
00:12:23;15 released from adipose tissue.
00:12:25;13 And only brain continues to use glucose, initially.
00:12:28;12 So, during initial stages of fasting, free fatty acids, or FFA, would be released from
00:12:34;27 adipose tissue by a process called lipolysis, or release of fatty acids.
00:12:40;06 And then fatty acids will become main fuel for most organs.
00:12:44;02 And... while brain will continue to use glucose.
00:12:47;24 And then, if fasting is prolonged, then some fatty acids will go into liver and will
00:12:53;15 be converted into a different metabolite called... called ketones, such as beta-hydroxybutyrate,
00:12:59;04 or BHOB here.
00:13:02;06 And this fasting metabolic switch into ketogenesis -- production of ketones -- is controlled
00:13:08;18 by a nuclear receptor called PPAR-alpha, shown here.
00:13:13;06 So, what PPAR-alpha does during this prolonged fasting... it detects fatty acids that are
00:13:19;14 delivered from the adipose tissue and induces enzymes that generate ketones from fatty acids.
00:13:28;06 And the point of that is that ketones now can be used by the brain.
00:13:33;00 And the second thing that PPAR-alpha does, it controls expression of a fasting hormone
00:13:38;02 called FGF21.
00:13:39;15 So, we thought that it's... these PPAR-alpha-regulated processes that are activated during fasting
00:13:47;06 might be involved in controlling survival because when we eat, or consume food or glucose,
00:13:54;12 that would induce insulin production, and insulin will suppress all these processes.
00:13:58;20 It will suppress lipolysis and it will suppress ketogenesis.
00:14:03;00 And we thought that maybe that's why glucose kills and 2-deoxyglucose leads to...
00:14:09;09 promotes survival.
00:14:10;26 So, to test that, we first examined whether, indeed, glucose will prevent these PPAR-alpha-regulated
00:14:20;19 outcomes, such as hydroxybutyrate production and FGF21 expression.
00:14:26;06 And indeed, as you can see on the left side, this is measurement of non-esterified fatty acids,
00:14:32;28 or free fatty acids.
00:14:34;04 So, they are released during fasting.
00:14:35;19 You can see in the black line the fatty acid level in the... in the plasma go up.
00:14:42;25 And later, beta-hydroxybutyrate starts going up and the fasting hormone FGF21 is also
00:14:49;27 strongly induced.
00:14:52;04 But if we give mice glucose, then all these responses are shut down, and none of them happen.
00:14:58;21 And then we asked whether they... this is what contributes to differential survival
00:15:03;13 from feeding.
00:15:05;01 So to address that, we used mice that are deficient PPAR-alpha, where neither ketogenesis
00:15:11;23 nor FGF21 expression can be induced.
00:15:14;19 And we used, also, FGF21 knockout mice.
00:15:18;09 And as you can see in the left panel, both of these mice now succumb to sub-lethal doses
00:15:25;12 of LPS that are survived 100% by control mice.
00:15:30;13 So, that suggested both FGF21 and PPAR-alpha are necessary for survival.
00:15:36;00 And in the absence of PPAR-alpha, on the right panel you can see that there is no production
00:15:40;26 of beta-hydroxybutyrate.
00:15:45;13 But the level of inflammation in all three conditions was the same.
00:15:48;19 So, as you can see in the lower panel, the level of TNF in... in the serum in all three
00:15:55;23 mouse strains was the same.
00:15:59;26 So, what we found also is that glucose supplementation increased, and 2-deoxyglucose decreased oxidative
00:16:09;14 stress in the midbrain area during LPS sepsis.
00:16:13;28 And by performing PET scans to follow where glucose goes during sepsis, we found that,
00:16:20;12 following LPS challenge, the... this area of midbrain was the... showed the
00:16:27;22 greatest difference in glucose consumption.
00:16:29;12 So, there was increased glucose uptake into the... this particular brain area that...
00:16:34;16 where we also saw increased oxidative stress.
00:16:39;08 And so what we concluded from that is that fasting metabolism and ketogenic programs
00:16:45;14 are required for survival of LPS sepsis.
00:16:50;12 And what we noticed also is that the death from sepsis was preceded by seizures or convulsions.
00:16:57;01 And it's also well-known that ketogenic diet is used to treat epilepsy.
00:17:02;01 And all these pieces of the puzzle together led us to ask whether anti-epileptic drugs
00:17:08;04 could be protective from sepsis, which was a very far-fetched idea.
00:17:12;17 But we tested it and, to our surprise and delight, we found that, indeed, the anti-epileptic drug
00:17:20;01 valproic acid could rescue mice from lethal sepsis.
00:17:25;24 And interestingly, the second anti-epileptic drug, called Keppra, did not have such an effect.
00:17:31;06 And that is very informative for us because these two drugs have very different
00:17:34;08 mechanisms of action.
00:17:37;02 So, we then tested whether valproic acid can protect upstream or downstream of the effect
00:17:45;22 of glucose.
00:17:47;10 And what we found was that valproic acid protected against LPS sepsis even in PPAR-alpha knockout mice,
00:17:57;12 which as you remember cannot produce ketone bodies.
00:18:02;12 So... as if valproic acid substitutes to the protective effect of ketones.
00:18:08;01 But 2-deoxyglucose cannot protect PPAR-alpha knockout mice from LPS sepsis because it acts
00:18:13;04 upstream of PPAR-alpha.
00:18:15;17 So, that indicated that valproic acid has its effect very downstream in the fasting pathway,
00:18:22;22 at the same level, perhaps, as ketone bodies.
00:18:26;19 And the conclusion to this part is that during sepsis -- endotoxin sepsis, LPS sepsis --
00:18:37;05 LPS induces inflammatory cytokines, and which of these is most important here is...
00:18:41;16 is not clear, and probably several of them can lead to similar effects of increasing
00:18:47;20 generation of reactive oxygen species in the midbrain area.
00:18:51;25 And glucose promotes this effect and 2DG inhibits it.
00:18:57;08 And ketones also inhibit that effect.
00:18:59;20 And consumption of food or glucose prevents ketogenesis and therefore interferes with
00:19:07;02 the protective effect from... protection from this damage by ROS.
00:19:12;22 And during fasting, a PPAR-alpha-dependent mechanism generates ketones, which lead to
00:19:19;00 reduced ROS production and adaptation to the stress of inflammation, and survival, ultimately.
00:19:26;10 And we think one of the effects, common targets, here, could be histone deacetylase... deacetylases,
00:19:33;11 because both ketones and valproic acid are known to inhibit histone deacetylases.
00:19:38;27 And this is something we are currently testing.
00:19:41;05 So, that's the part of the study that had to do with bacterial infection and bacterial sepsis,
00:19:49;01 where we found that eating during bacterial infection or sepsis interferes with
00:19:55;02 this normal protective effect of fasting metabolism, and therefore the anorexia that we feel
00:20:01;08 when we have infections has to do with promoting these types of protective mechanisms
00:20:09;22 associated with fasting metabolism.
00:20:13;12 And then we... what we found here, therefore, is that this increased glucose level, if it
00:20:21;06 goes above some upper threshold level, can be deadly in the context of bacterial sepsis.
00:20:30;05 And another recent study examined the role of glucose in a very different model,
00:20:35;16 where they looked at the lower threshold level, where they used mice which are unable to
00:20:41;20 produce glucose from the liver.
00:20:44;19 And this was... this... and that also leads to mortality.
00:20:47;20 This was a study by Miguel Soares from Instituto Gulbenkian in Lisbon, where they found that
00:20:53;23 there is also a lower boundary for the glucose level.
00:20:58;10 So, both upper boundary and lower boundary, if they're exceeded in the glucose level,
00:21:03;06 can lead to mortality.
00:21:04;19 So, it's important to keep that in mind, that it's not an excess or depletion; it's a maintenance
00:21:12;00 of the... the right amount of glucose that is required for survival.
00:21:17;00 And this is not surprising, of course, because glucose is still essential for many cells,
00:21:21;25 especially neuronal cells, for survival.
00:21:29;01 And... well, we found this very dramatic effect of glucose and 2-deoxyglucose on
00:21:35;27 bacterial infection in sepsis.
00:21:37;06 We then asked whether this is more a general phenomenon and whether it applies to all infections.
00:21:44;16 And to address that, we used a mouse model of influenza infection, where mice are infected
00:21:52;02 with the flu virus.
00:21:54;04 And in this case, we are giving a sub-lethal dose of flu.
00:21:57;10 And then we follow, again, the food consumption.
00:21:59;26 And just like with bacterial infection, you can see that mice, when they...
00:22:03;19 at the peak of infection, they stop eating.
00:22:06;16 And then as they start recovering from infection, they... they resume food consumption.
00:22:13;02 And we asked again, what would happen if we feed them at the time when they are anorexic?
00:22:20;07 And what we found was very surprising, and opposite to what we expected and opposite
00:22:24;19 to what we found with bacterial infection.
00:22:27;14 As you can see here, if we feed the mice, they actually survive better compared to mice
00:22:32;21 that received control PBS solution.
00:22:36;26 And if we just give them glucose, they also do better.
00:22:40;01 And glucose partially protects from mortality.
00:22:42;14 And we think the... the rest of the protection is provided by sodium.
00:22:50;06 And when we ask the question, the converse question, what 2DG will do, we found that
00:22:56;05 2DG actually was lethal in the context of viral infection.
00:23:00;16 As shown here in the blue line, when mice are given 2DG in the context of viral infection,
00:23:07;04 they all died 100%.
00:23:09;22 Interestingly, this difference in survival was not due to tissue damage that was normally
00:23:16;27 caused by flu virus.
00:23:18;11 So, this is a lung pathology.
00:23:21;06 On the left side is the control and on the right side is 2DG-treated mice, and they're
00:23:26;02 basically the same.
00:23:27;10 There is no difference in the degree of tissue damage caused by the virus.
00:23:31;12 And also there is the same level of hemorrhage, edema, and inflammatory infiltrates.
00:23:37;08 So, that did not explain 100% differences in survival.
00:23:43;06 And also, there was no difference in the magnitude of the inflammatory response, as shown on
00:23:49;27 the left side by measuring interferon-alpha in the plasma.
00:23:55;02 And interestingly and importantly, there's no difference, also, in viral burden.
00:23:59;05 If we measure the amount of viruses during infection, they are similar between
00:24:04;18 control and 2DG-treated mice.
00:24:06;07 And again, 2DG-treated mice are the ones that succumb 100% to this infection.
00:24:12;22 So, what we then noticed, by performing other measurements on these mice, is that
00:24:22;07 the death from viral inflammation caused by either flu virus or some mimics of viral infection
00:24:29;22 was associated with decline in vital functions, such as heart rate, respiratory rate,
00:24:35;00 and so on.
00:24:36;00 And that suggested that there perhaps is a failure of the autonomic control centers that
00:24:40;16 reside in the brainstem.
00:24:43;08 And moreover, when we performed PET scans on these mice, again we found that
00:24:48;09 glucose was preferentially taken into the brainstem area during viral inflammation.
00:24:54;14 And remember, during bacterial inflammation it was preferentially taken into the midbrain area.
00:24:59;24 So, that was puzzling but suggested a possible scenario where viral infection somehow interfaces
00:25:08;09 with glucose metabolism, and the only type of connection that we could find in the prior literature
00:25:15;02 that would suggest the mechanism had to do with endoplasmic reticulum stress,
00:25:21;02 or ER stress.
00:25:22;02 So, ER stress is normally induced by unfolded protein response in the endoplasmic reticulum,
00:25:29;16 and it leads to adaptation to the unfolded protein accumulation through induction of
00:25:35;24 chaperones and various proteases and so on.
00:25:38;06 And that leads to resolution of the ER stress.
00:25:40;13 However, if ER stress is excessive, then the second branch of the pathway is induced by
00:25:49;04 leading to transcriptional induction of a transcription factor called CHOP that
00:25:56;15 leads to cell death through apoptosis.
00:26:00;15 And because glucose availability can also impact on protein glycosylation in the ER,
00:26:05;10 it can also lead to ER stress.
00:26:07;14 So, when cells are deprived of glucose, that can lead to ER stress.
00:26:14;21 And viral infection can also lead to ER stress.
00:26:17;15 So we thought, perhaps these two conditions may somehow conspire to trigger excessive
00:26:24;13 ER stress, leading to induction of this transcription factor, CHOP, leading to neuronal damage in
00:26:30;24 the brainstem.
00:26:33;11 And we tested whether CHOP is indeed induced under those conditions anywhere in the brain
00:26:38;08 and found that, indeed, when mice have viral inflammation -- in this case induced by
00:26:44;24 a viral mimic called poly(I:C) -- either alone or together with 2DG, and then we monitored
00:26:52;02 CHOP expression by Western blot, and we found that it was only induced in hindbrain area,
00:26:59;04 and only when mice received both poly(I:C) and 2DG.
00:27:04;17 And then we tested whether CHOP is involved in mortality caused by infection and 2DG.
00:27:12;18 And to test that, we used either wild-type or CHOP-deficient mice.
00:27:17;12 The gene name for CHOP is Ddit3, so those are Ddit3 knockout mice in open... open symbols.
00:27:25;08 And as you can see, wild type mice that received poly(I:C) and 2DG die 100%; that's the blue triangles.
00:27:32;23 And CHOP-deficient mice receiving the same combination of poly(I:C) and 2DG survive 100%,
00:27:40;10 indicating that, indeed, this particular transcription factor is a critical mediator of mortality
00:27:46;03 from viral infection combined with 2DG.
00:27:50;24 And again, as shown on the right slide, there was no difference in the inflammatory response,
00:27:55;07 as measured here by interferon-alpha in the serum.
00:27:59;12 So, the summary for this part is, during viral infection, or more generally during
00:28:05;03 viral inflammation, because we could find the same exact phenomenon with just using poly(I:C),
00:28:10;26 there is a production of type-1 interferons -- interferon alpha and beta -- and that leads
00:28:15;28 to activation of the unfolded protein response, or UPR, combined with glucose consumption
00:28:29;00 in the brainstem area.
00:28:31;27 Why it's specifically brainstem that's affected in this manner we don't know.
00:28:35;22 It's a very interesting question which we hope to understand someday.
00:28:41;13 But what happens in the context of this response with metabolites is that glucose ameliorates
00:28:48;07 this response -- it prevents induction of CHOP and neuronal dysfunction and... and death --
00:28:55;19 whereas 2DG exacerbates it and leads to CHOP induction and subsequent loss of function
00:29:03;20 of the brainstem and autonomic control centers, resulting in death.
00:29:09;06 So, these are two very different effects of metabolism on bacterial and viral inflammation.
00:29:16;26 And all of it could be tied down to utilization of glucose or block of glucose utilization.
00:29:25;09 And it's completely independent of pathogenicity.
00:29:29;04 It's independent of pathogen burden.
00:29:32;00 And it's independent on the magnitude of inflammation.
00:29:34;25 So, this is what we refer to as being able to tolerate a given level of inflammation,
00:29:41;09 rather than controlling the level of inflammation.
00:29:44;19 So, this study was done by three very talented scientists in my group -- shown from right to left,
00:29:52;17 Andrew Wang, Sarah Huen, and Harding Luan
00:29:57;09 -- and two talented technicians -- Cuiling and Shuang -- who helped with the study,
00:30:03;15 and... as well as Carmen Booth and Jean-Dominique Gallezot, who are...
00:30:08;27 helped with pathology and PET scans.
00:30:11;20 And our funding is shown at the bottom of this slide.
00:30:15;27 And thank you for your attention.
Zheng H., et al. (1995) Resistance to fever induction and impaired acute-phase response in interleukin-1 beta-deficient mice. Immunity. 3(1):9-19
Ruslan Medzhitov iBioSeminar: Introduction to Inflammation
Dr. Ruslan Medzhitov is a Sterling Professor of immunology at Yale School of Medicine and a Howard Hughes Medical Institute investigator. His laboratory studies the signals that initiate and control the process of inflammation, allergic reaction, and immune response. His laboratory also studies tissue biology, and the communication circuits that help to establish stable cellular… Continue Reading