Endometriosis Foundation America: Medical Conference – 2010
Linda G Griffith, PhD
Thanks very much, for the invitation to speak today. I am excited to attend this because I am actually new to research in endometriosis.
Within about the past year, we have really established our first research projects, bringing in some approaches we have used in a variety of other tissue systems. This in collaboration, and with great inspiration, from Dr. Keith Isaacson who has been very generous in giving his time on nights and weekends to come in and meet with the students and post-docs. What I would like to do today, since we really are just getting ready to submit our first papers in the area of endometriosis research, is describe how scientifically we, being a whole bunch of faculty at MIT that I could corral into this endeavor with Keith, got interested in endometriosis as an incredibly exciting opportunity to develop new tools and systems biology in tissue engineering, and in particular to study chronic inflammation. I will say at the outset, I am also Keith's patient actually.
One thing that struck me as a professor at MIT over the past 20 years is that there has been an increasing sense of unease at the number of women involved in academia. An increasing sense of, well we need to do something about this, and let us think about the way we say things to women, and I am supposed to get upset if somebody mistakes me for the secretary, and so on. In the environment that I am in, you know, if somebody thinks I'm the secretary, I say, “oh no, she is not here right now and she will be back soon”.
What really struck me is that there is this huge unmet need to address the health needs of women, especially women of the age of the students that I teach. I teach a very large course sophomore class. I teach physical chemistry and I see women who end up dropping out of MIT because they have endometriosis, and it is not diagnosed and they are not taken seriously.
A bunch of us got together and felt it was time to have a very visible research effort in the school of engineering, centered in the school of engineering in MIT, to try to bring some new approaches to the study of endometriosis, and also raise awareness within our kind of community about the disease itself. I am going to start out by talking a little bit about what engineers do. Particularly what engineers today are doing in biology by using an example from other areas of my research program, and then I will close with our emerging efforts in endometriosis.
A lot of people, certainly at NIH, when I go do things, I am frequently invited as an engineer because they think I make things. They think I make tools or devices or widgets or whatever, and it is true, I do. In fact, when I was an assistant professor, I got pulled into a project with Joe Upton at Children’s Hospital to create a piece of cartilage in the shape of an ear. If you are about my age or older, you may remember the human ear on the back of the mouse, made simply by combining cartilage cells with a degradable scaffold. We figured out how to make the scaffold in the shape of an ear. This generated enormous excitement that, oh my goodness, now we can just do the same thing and get heart in a box, liver in a box. We will fix all these terrible problems with donor scarcity for liver, et cetera, by just combining cells on scaffolds and like magic growing these organs. Of course, that has not happened and predictions of a lot of people 15 years ago have not panned out and I am not sure we ever will reach the stage where we will have a liver in a box or a heart in a box et cetera. I think we can ask the question, is it really a good goal to do that?
Maybe we should be thinking about better ways to study the diseases that lead to these catastrophic failures of our organ systems, and in order to do that, we need to think a lot harder about the way the biological systems work from a quantitative perspective. If we take the approach of the ear, we sort of went from a phenomena that a little piece of cartilage can kind of grow on its own, to what we perceived as an intervention, let’s make a scaffold to make any particular shape, without really going through the middle where we understood exactly how it was happening. When we tried to apply it to other problems, it did not work.
That motivates us to now go in and think very hard about those intervening steps in the box. Engineers typically with any kind of system, whether it is biological or other, go in and understand phenomena from the standpoint of mathematics. You analyze and you do enough experiments so that you can build a model of how the system is working, and with that model you can predict how things will behave outside places that you can measure. That gives you design principles to build new things. Here we are trying…we actually started a department at MIT where we are focused on trying to build new molecular therapies, and other interventions that will take medicine into the new age.
As an example of this and I think, I hope, you will be able to see how the same kinds of approaches will readily be applied, hopefully to endometriosis, I am going to give a complete story that we published in the area of liver.
This is an area I have worked in for about 25 years and it is an area where we are trying to blend tissue engineering and systems biology into drug discovery. And now we are trying to do the same thing for endometriosis, but let me take you through the complete story, so you can see how we think about these problems.
The specific challenge is not necessarily to grow a liver that is going to go into a patient but can we prevent the kinds of things that lead to liver failure? One of them is idiosyncratic drug toxicity; this is a huge problem in drug development. If you are from a pharmaceutical company, you probably appreciate this very well that you get a drug that made it through preclinical, and phase I, and got out into a larger patient population. All of a sudden these idiosyncratic, not predicted toxicities, appear. Many drugs get pulled through this and it costs billions and billions of dollars. There is hypothesis about this, and how this works.
One very intriguing one, and one for which there is some clinical evidence, is that it is a synergy between the drug and some kind of liver inflammation. That liver inflammation could be a lot of things. Somehow the stress the drug places on the metabolic pathways intersects with signaling pathways set up by cytokines, to cause catastrophic failure of the liver.
To illustrate this graphically on the left, there is a picture of how the liver is stimulated by things that leak in from the intestines during normal operation, during normal homeostasis. There is always a little bit of leakage of bacterial lipopolysaccharide across the gut walls to stimulate the liver. In normal circumstances, drugs will interact just fine, so the hypothesis is shown on the right that in cases where you have increased gut permeability and more inflammatory stimuli now leaking across, you get this interplay of the inflammation processes responding to this LPS with the drug and now you synergize to create a failure of the liver, death of the liver cells. Now this is a very complex process. You can see that there are many cell types involved that are temporarily, probably, controlled process. How can we possibly try to understand this with the kinds of systems we could put in early drug discovery?
This is actually an engineering problem. How do we take complexity and start to tease out the essential features of it, so that we can build rapid high throughput ways to understand it?
Now let's go in and let's focus on the hepatocyte even when many cells are involved in this, we will view the other cells as secreting cytokines. We can think about a cytokine landscape combining with a stress or infection or a hepatotoxic drug. The cells are going to generate intracellular signals in response to these, including kinase signals, and they are going to have an outcome when these two things impinge, the drug and the inflammation, that could be survival or death, maybe proliferation even.
Now, we want to try to understand this in a predictive way, we want to build a math model of this process, okay. That sounds pretty intimidating, but let us step through how you build a math model for any complicated process. I picked one that a lot of people know and if there is just one take home message, it is that by modeling systems, taking some data, and building this kind of model, you can start to predict how the systems will operate where you cannot measure or where it is too expensive to measure, et cetera. You do a limited set of measurements.
Measurements are always hard and expensive, and then predict how the system will behave. An example is to signing a new airplane engine wing. This has been done for over 100 years and you do not put a whole 747 into a wind tunnel to figure out how to improve the wing. You put a model of it in there. Let’s think how the model can let you predict how to build the actual airplane. You go in and you have your wind going across, and the photos here show the actual lines of flow using smoke. To highlight them, you can image the flow over the wing and it is just showing a different tilt. Now, there are a lot of things that would influence lift of that wing. So lift, let’s just say simply, is the variable you care about. There is temperature of the air, there is pressure, it is all kind of issues about the fluid, the velocity, and there are also parameters of the wing, a huge long list. We cannot possibly measure all those things. In fact what we do, is we go in and we model the fluid mechanics, mathematically. Now most of the math here, it is differential equations, but you can see over on the sides there is chaos. We cannot model that precisely. There is different kind of math we use for this. What you do is you take all of this huge long list of variables that is over here on the side, and you group them together. The math tells you how these variables are related. You get simpler variables, dimensionless, that group things together. These things change in tandem and influence the process. Then you get your desired outcome and you can build your plane. Now what lets you do this is that this grouping of variables is the same for your small wing here, and for the plane that you want to fly, that is the way the math will work. You can do this kind of simultaneously thing by going through this mathematical process.
Can we possibly apply this to biology? Okay, how do we? We are just at the beginning of doing this to understand how cells respond to cues in their environment. Here the complexity is arguably a little more complicated. You have cues, things like your temperature, your velocity, et cetera, or your cytokines, for example epidermal growth factor, insulin, tumor necrosis factor. They impinge on the outside of the cell, transmit signals through these signaling networks, and there are different networks highlighting the different colors here, and then there is a response. This is sort of like the wiring diagram for a plane. What are the currents going through the different wires that are controlling the plane. You could think of it that way, but we do not know those currents because we are not setting them. We want to try to put volt meters on and measure them and understand how the cell makes the decision to live or die when it is getting these cues from the outside. What we do is we start out, we do not know anything at this stage, we are going to measure a lot of stuff. We are going to give it a lot of cues or a big landscape, and then we are going to measure a lot of things in this network. We are going to use a toy model system to then try to develop our models. Let me show you how we have done that for this problem for idiosyncratic drug toxicity.
You have simple cell cultures, we start with those. Can we do it with simple cell cultures? Forget initially capturing all the complexity of different cells, combine our cytokines and drugs, make lots of measurements including the signals, and do that. The first thing we ask, is can our in vitro system capture this synergy that we think is happening with inflammation. To do that you have your cells and culture, this is a metric of death of the cells. You add the drug because when you add the drug, you see you are killing cells. Death goes up, that is this line here. If you add those inflammatory cytokines when there is no drug, you actually do kill the cells a little bit. This is the amount you kill them, but if it is only additive, you would just add that amount of death to the initial line and get the red line. In fact, it is synergistic, so we can see that.
Now we went and did this with 90 drugs and you can see a lot of them show this kind of synergy. Now we say, okay, can we go in and start to measure all the signaling networks and get this to be predictive for how the cells respond. All the yellow dots are where we can measure 17 different places, and in fact we go in and find in that wiring diagram, when we measure signaling, this is just one. We again see these synergies drug alone, cytokine alone, and the mix. Now we go in and do it again. This is just a huge data set, you can think of this as measuring over a wide, wide range of wind velocities and so forth in the wind tunnel. We can ask, these are the death signals, how are these things related; it is not obvious. There is not just one thing. We use our math models, apply them, and find in fact we can instead of measuring an enormous numbers of things, there is really these four pathways that are operative. We can go in and now take those four pathways, we do not need to measure everything. We measured those four pathways, and we did an experiment where we took donor cells from two different people, tested them under different conditions, and in fact we can predict these outcomes. We cannot predict all idiosyncratic toxicity. We are now asking questions about do we need to build more complex tissue engineering models, and in fact have a huge effort on building these 3D models to predict toxicity. As you can see here, the complex tissue models actually pick up toxicities we cannot pick up with the simple models. We are now bringing these kinds of approaches, and I will just take two minutes here and be done, bringing these kinds of approaches into endometriosis research. We have assembled a team at MIT that includes biologists, polymer scientists, and others on both the tissue engineering and the biology side, together with a set of clinical and collaborators, including prominently Keith, but a variety of other people, not only in the Boston area, but throughout the country, and even in Singapore. The concept of the mission is, starting out, of trying to bring these approaches into discovering the etiology, of bringing new tools into study of etiology, particularly of endometriosis, and developing similar kinds of models for both drug identification and drug development.
We started with just a Foundation grant but recently have gotten NIH grants. A couple of example projects is we are bringing new approaches to do analysis, both culture and facts, into studying the cells in peritoneal fluid to really try to get a handle on things that may be influencing lesion initiation and progression. Again, taking this kind of systems biology approach I just described, what causes cells to invade versus being more quiescent. And finally, building new tools to parse these kinds of signaling networks, including developing very sophisticated activity assays to multiplex these kinase activities so you are not just measuring the phosphoproteins, you are actually measuring the activity. This is in collaboration with Barbara Imperiali. You just now can take a 30 cell lysate and really start to get very, very rapid data, very accurate data on how these cells are responding to inflammatory signalling. They combine that with microfluidic devices so that we can integrate this and get high throughput data.
Again, today, I just wanted to give you a sense of the efforts, particularly of the Endometriosis Foundation, which really got us to start thinking that it is not just us thinking that there is a need in endometriosis, but also that there is a huge community out there willing to support these kinds of efforts. Finally, we also have efforts in classical development of new surgical tools to treat the patients now, and we are trying to pull together a lot of engineering faculty, with biology, to bring some new kinds of tools and approaches into endometriosis.
With that, I will close.
