The future of simple rules

Transcript

Russ Altman (00:04): This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. If you enjoy The Future of Everything podcast, please follow or subscribe on whatever app you listen to it with. You'll hear about new episodes and you'll also help us grow.

(00:17): Today, Professor Polly Fordyce, from Stanford University, will tell us about how her lab studies the structure and function of proteins. DNA gets a lot of press. DNA, the genome. It's in my DNA. DNA, DNA, DNA. But proteins actually do a lot of the work in cells, and in fact, they're the ones who read off whatever the DNA is telling us and turn it into action. It's the future of proteins.

(00:43): Before we jump into the podcast, I'd like to give you a reminder to rate and review the podcast and also tell your friends about it. It'll help us get better and it'll help us grow.

(00:53): Many of us have heard about DNA. DNA stores the instructions for how cells should be built and how they should function. But proteins are another important biological molecule. In fact, much of the job of the DNA is simply to encode where, when, and how to make the proteins that do a lot of the work. Proteins catalyze chemical reactions, they interact with DNA to figure out what needs to happen and when, and they serve structural functions of support so that cells are not just floppy bags, but have the right structure and rigidity that they need for the processes of life.

(01:32): Polly Fordyce is a professor of bioengineering and genetics at Stanford University, and her lab studies proteins with high throughput measurements. Rarely will Polly's group measure one thing. They measure 1,500 things at a time, and that leads to a quantitative and qualitative change in their understanding of protein structure and function. She'll start out by telling us about one of her favorite proteins.

(01:56): Polly, your group works on protein structure and function. When most people think about proteins, what they know is they need to eat the right amount of protein in their diet and that there's a lot of protein in their muscles. You make a study of this for your life. I think the best way to start out is, what is a protein and why do you find them so fascinating?

Polly Fordyce (02:15): It's a good question, Russ. I would say it's hard to pick a favorite protein, but what got me into this whole business in the first place was when I was a brand new PhD student, here at Stanford actually, I got here and Steve Block gave a talk about his research. What he was doing was he was studying, you have some cells in your body that extend all the way from the base of your spine to the tip of your toes, and they have a nucleus at one end, but they need to get all of the stuff that they make in the nucleus to the rest of the cell. If they just relied on diffusion to do it, it would take decades. Your cell has this entire transport system of these motor that actually move. They move parts of themselves like walking on a road. They walk on these filaments in your cell to tow cargo everywhere.

(03:10): 20 years ago, that idea, that there were these proteins that actually could tow cargo, they turned chemical energy into motion, they were the most efficient motors we've ever seen, far more efficient than anything we've ever built. That idea that there were molecular machines at the nano scale was just incredibly cool to me. That was the first protein I fell in love with. I actually have a little sculpture of it in my office.

Russ Altman (03:35): That is awesome.

Polly Fordyce (03:41): But now, if I had to pick, there are a whole host that we're working on in our lab, but that was sort of my first love, I'd say.

Russ Altman (03:47): Great. Okay. We should think of these as, obviously, they're molecules, they're made out of atoms, hundreds or thousands of atoms, I presume. When you're talking about walking, this is really happening at the atomic level. These are atoms moving past other atoms.

Polly Fordyce (04:05): This first protein I studied, it's called kinesin, it actually takes eight nanometer steps. A nanometer is a millionth of a millimeter, and it would take eight steps at a time as it was walking along these subunits in the road that it was walking on. The other thing that was really fascinating to me at the time was we built a lot of really complicated machines with microscopes and lasers that allowed us to actually visualize single proteins as they walked, as they took chemical energy from molecules and turned it into this motion and walked in the cell.

(04:44): Those two ideas, I guess, the idea that proteins are these molecular machines that do work in your cell, they do a whole host of really incredible things in your cell. And then, the idea that we could actually watch them in some way as they were doing their tasks, if we could just build the right instruments, all of that to me was really fascinating.

Russ Altman (05:04): Great. Great. Okay. When somebody googles your name, which I recently did in the last 24 hours, what we are struck by is it's all about protein structure and protein function. Tell me about protein structure and function, and if it's like many other things in the world, those two things are intimately related.

Polly Fordyce (05:25): They are. What I think is incredibly exciting right now is that proteins are made of these linear chains of amino acids, and about half of them actually have to fold into a three-dimensional confirmation in order to do their job. They assume a shape, and we always talk about that as their structure. Once they're in that structure, they do some kind of function. Maybe they bind your DNA, maybe they bind other proteins, maybe they receive signals from the environment to tell your cells what to do. They coordinate everything that's happening in the cells.

(06:02): I think in the last year, well, maybe five years, I would say, there's been a really incredible progress in the ability to take all of these models, like models that we train on human language, and get them to read the sequences of the linear building blocks that make up a letter and predict the structure that it's going to have.

Russ Altman (06:26): We've heard about AlphaFold, for example. That's what you mean, I think.

Polly Fordyce (06:30): Exactly. Exactly. AlphaFold is really taken my little world by storm, I would say, where it's really solved a 50-year grand challenge in biology. But I think that what we really want to do next is we want to get to the next step, where we could actually make machines that would do things that we wanted. Maybe we could use them to catalyze green chemistry and we would get rid of waste. We could use them to clean up plastics. We could use them as new drugs or therapies. But in order to get there, we have to get to the next level of the problem, which is going from how sequence encodes the shape to how it encodes the actual function. That's what my lab is sort of all about, is trying to figure that out.

Russ Altman (07:12): Great. I think we have now a good idea of these proteins. They're small, but they're made out of thousands of atoms and maybe millions, but they are still very small. They fold into some shape. There's been progress on that. And then, Polly Fordyce's lab now gets very interested in understanding the function. I know that there's a lot of technology development here, and it's very exciting because, I think, traditionally, people did experiments in test tubes and they measured one thing. I know that your lab has changed that paradigm. Can you take us through what were the ideas that you had, and then what did they turn into in terms of accelerating our understanding of function?

Polly Fordyce (07:54): I would say I'm really lucky to have landed at Stanford where my lab has developed a lot of technologies, where the goal of our technologies, much like integrated circuits, took computers and shrunk them down so we could do electronic calculations in a really small footprint. My lab really specializes in trying to make these fluidic technologies that shrink down the volumes that we need to do. Now, instead of doing one experiment at a time using a big test tube, we can do a thousand experiments at a time using tiny little one nanoliter chambers.

(08:30): That's my specialty, but I was lucky to come here to Stanford where there are people, like Dan Herschlag, who's been spending his career decades trying to understand how enzymes, which they can take a chemical reaction that would take longer than the lifetime of the universe, and they can speed it up so that it happens the second it hits that protein. They're really these incredible catalysts. People like Dan were all of the traditional techniques to try and understand how that works. They had all of these questions about how does this linear sequence of amino acids encode this incredible chemical function. But it was really hard, it would take years to be able to make and characterize a reasonable set of mutants.

(09:14): With Dan, we were able to team up, I was able to learn some of the entomology from him, and then we were able to make a tool where now we can make a thousand, really 1,500, different enzyme variants. We can change one thing at a time or we can change a bunch of things at a time in that sequence. We can make that molecular machine, and then we can ask how that molecular machine, how it catalyzes different chemistries or how it functions under different conditions to really get a broad picture of what it does.

Russ Altman (09:46): Right. Okay. You said the word mutant and you said that this is when you change one of those amino acids in that long string. That means that each one of those proteins that you create and put into these 1,500 little tiny teensy tiny beakers, every one of those is slightly different. Therefore, I'm presuming they may have a slightly different structure and a slightly or very different function. Just to make it very real, what are the actual enzymes that you look at and what are the things that they do?

Polly Fordyce (10:20): We started with, I think this often happens in science and technology development. We started with a protein that I love now. It's an alkaline phosphatase superfamily member called PafA. I have now studied it for six or seven years. There was nothing particularly interesting about this protein when we started, except for that it was going to be really easy to study in the platform. It was super efficient. It takes a reaction that takes longer than the lifetime of the universe and it catalyzes it the second that a molecule-

Russ Altman (10:53): Okay. Tell me the reaction. Even if you think it's boring, I want to hear it.

Polly Fordyce (10:57): It takes a phosphate monoester, and it cleaves the phosphate off.

Russ Altman (11:03): So it cuts off a phosphate.

Polly Fordyce (11:05): It cuts off a phosphate. That's what it does.

Russ Altman (11:06): Okay. Fair enough. You have found that, over the last seven years, to be a fascinating question.

Polly Fordyce (11:12): Yes. Yes.

Russ Altman (11:13): Please go on. Please go on.

Polly Fordyce (11:16): I found it to be a really fascinating question because these enzymes at their... Where they do the chemistry is called the active site. They have a special part, and that's where they make conditions really favorable for this particular chemical reaction to happen. That's the business end of the protein, and we know that's really important. But one of the questions that we had was this protein is made up of 526 different linear building blocks, 526 amino acids, how many of them make a difference? If you change them, is it that only the active site, or do most of them matter? We didn't know that because we couldn't do that experiment before.

(11:59): What I thought was really fascinating was about 2/3 of them, if you make a change, they affect its function. This big enzyme where we've been really focused on the active site, it turns out that in order to get where we want to go to actually design a functional molecule, we're not going to be able to just take that active site and stick it in another protein and call it good.

Russ Altman (12:22): Nature kind of made the protein exactly how it needed to and it didn't have a lot of fluff.

Polly Fordyce (12:29): Exactly. Yeah, it didn't have a lot of fluff. There's even, as is actually, I've joked about this with Dan a lot. At the very beginning, if you told me that I was going to care about some of these things, there's these chemicals called transition state analogs that chemists use to look at very specific properties of the enzyme while it's doing the chemical reaction. There are two different ones, vanadate and tungstate, that differ by less than a 10th of an angstrom.

(12:56): If you had told me four years ago that I was going to care about this, I would not have cared about it at all. But it turns out to be really cool, and that if you make some of these mutations really far away from the active site, all the way on the surface, those mutations change whether it binds one or the other of these transition state analogs, even though they're almost identical. When you're talking about that there's not a lot of fluff, it really kind of blew my mind how little fluff there was.

Russ Altman (13:28): These mutations that you make, they definitely sounds like functional consequences. Are they also changing the three-dimensional structure of what the protein looks like?

Polly Fordyce (13:37): Yes. There's a few different ways that this can happen. That's one of the things that we're developing in our lab is trying to figure out exactly how much a mutation changes the confirmation versus the activity, we call it, confirmation and activity. But in this case, what was kind of interesting, this protein that we started with, which was an unusual protein protein, it's secreted. Bacterial cells secrete it into the environment, and so they want it to be really, really stable. We chose it, because we thought a single mutation would be unlikely to unfold it, but it turns out that because it's so stable, it's really easy for it to get stuck in the wrong confirmation. If you make a mutation, often, some of it will still fold up nicely but it folds up into the wrong shape.

Russ Altman (14:26): Okay. Six or seven years on a protein that you've grown to love, are we done with that protein? Is it a totally understood system? If not, what are the questions that still are driving you and your collaborator, Dan?

Polly Fordyce (14:39): Yeah. I would say one really interesting thing that we found was not only do mutations have effects even really far away from the active site, but there are big, in 3D space, there's big clusters of these residues that all have a really similar impact. So the next question is, were we lucky and PafA was special in some way, and it's the only protein that has things like that? Or are all really stable proteins have it? Is it really big ones that have it and really small ones that don't? If you look at ones that are like PafA siblings, its cousins, do they all have the same sort of, we call it a functional architecture?

Russ Altman (15:23): Right, right. You've learned all these lessons, and you can't tell yet if these are general principles applicable to all proteins or if you're going to have to do this in your little tiny nanoliter wells for a bunch of more proteins before you know for sure that these are the principles.

Polly Fordyce (15:40): Exactly. I think one reason why we care, beyond it just being cool, is, and we think about a lot of the proteins in our cells that are really important for... When we think about cancer or other diseases, a lot of them involve these phosphatase proteins, these proteins that do the same type of chemistry. But in our bodies, it's been really hard because their active site, their business end looks the same. If you have one that's doing the wrong thing, it's really hard to specifically target that one with a drug without messing all the other ones up.

(16:15): I'm particularly excited about the idea that is this a way where we might be able to figure out a far away surface on these proteins that we could target therapeutically? Could we scan using our new tools to try and find those far away surfaces that then drug companies could use to try and develop new therapies?

Russ Altman (16:35): I presume that those surfaces might be specific to the one you're trying to get at and not its siblings that look similar, and therefore you have a much more specific drug, for example.

Polly Fordyce (16:46): That's the idea.

Russ Altman (16:47): Great. This is The Future of Everything with Russ Altman. More with Polly Fordyce next.

(16:53): Welcome back to The Future of Everything. I'm Russ Altman. I'm speaking with Professor Polly Fordyce from Stanford. In the last segment, Polly told us about how she makes measurements of proteins, multiple measurements all at once in little tiny beakers, and this allows her group to understand details about how proteins function. In this segment, she will tell us how proteins actually read DNA and decide what other proteins should be made. This is often in response to SOS signals from the cell saying that something needs to change.

(17:27): Polly, I know you also work on DNA and in fact how DNA interacts with protein. Give us a little primer on how we should think about DNA at these scales that you think about, and then why are proteins important even when we think about DNA.

Polly Fordyce (17:42): One thing that I think is really fascinating is that all of the cells in your body have exactly the same genome. But somehow your eye knows that it's supposed to be an eye, your liver knows that it's supposed to be a liver, and the way they do it is with the same genome, it's like they have a recipe book for different proteins, and that's your genome, is the recipe book. And then, they're deciding which recipes they're going to make in any kind of cell. That is something that's controlled by proteins.

(18:14): There are proteins whose job it is to find specific parts in the genome that are right next to a recipe that you want to make, and then they bring a bunch of the rest of the protein machinery to come, and central of dogma of biology says DNA makes RNA makes protein. They turn on expression of those proteins in order to kind of dictate what the cell should be doing. So, it's a puzzle, how do they figure out where they're supposed to go.

Russ Altman (18:40): Great. Liver cell is going to turn on all the liver recipes, but there are proteins making those decisions about which part of the genome to read.

Polly Fordyce (18:46): Exactly. For me, another question I think is really fascinating is you're a protein, you have this 25 megabase genome, and somehow you have to figure out where you're supposed to go in order to bind the DNA and call the rest of your protein friends in order to start a transcription and making other protein so that the cell is doing what it's supposed to do.

Russ Altman (19:11): That really rings true because at home I do some baking and all my recipes are just in a big pile with no organization whatsoever. I have to figure out that exact same problem every Sunday morning when I figure out where the heck is my recipe for whatever. Sorry. Please continue.

Polly Fordyce (19:26): Yeah. I guess you think, how does a protein know, how do we encode the instructions for when should you turn genes on and off? It's been a long time since we decoded the part of the genome that codes for proteins. We know the triplet code that says how DNA encodes for a particular protein, but we don't understand the code yet for when it should be turning that particular recipe on and off.

Russ Altman (19:54): Okay. What are you going to do?

Polly Fordyce (19:56): We're going to do kind of what we always do, which is we're going to make a whole bunch of nanoliter chambers, and then we're going to systematically vary either the sequence of the DNA or we're going to vary the sequence of the protein, and we're going to try and make thousands or millions of physical measurements of the relative energies of binding and the kinetics of binding.

Russ Altman (20:20): The energy of binding is how tightly they bind, and the kinetics is how long it takes for them to find each other?

Polly Fordyce (20:21): Come on or come off.

Russ Altman (20:26): Come on, or come off.

Polly Fordyce (20:28): One thing that we haven't talked about is I do think it's really an advantage of making the measurements using those terms, like energies or rates, or it doesn't matter whether I make that measurement in my laboratory or you make it in yours or somebody else makes it in theirs around the world. It means if we have those numbers that are on an absolute physical scale, now we can combine the insights from all of our different labs, and then that becomes a data set that now our computational colleagues can use to make the same strides that we previously made in predicting structure.

Russ Altman (21:06): Yes, yes. What you're implying is there is a currency and the units about which these are measured is not up for discussion. The scientific community has agreed, and therefore, if I can get way better data uniformity because of this agreement that you've all made. Okay.

Polly Fordyce (21:06): Yeah, exactly.

Russ Altman (21:29): Okay. You put in all these wells... Oh, it sounds to me like you're going to have to introduce me to another protein that is doing this binding to the DNA.

Polly Fordyce (21:36): Yes.

Russ Altman (21:37): Tell me about the protein.

Polly Fordyce (21:40): We have two, I guess, they're transcription factor proteins is what they're called. We have two of them that we've studied a lot, some favorites. One is called [inaudible 00:21:51]. It is a protein in yeast whose job is to sense when the yeast don't have enough phosphate, actually. It's phosphate again. It's important currency in biology. And then, when they see that the cell doesn't have enough phosphate, they go to the nucleus, they find the instructions that say turn on these genes that are going to scavenge more phosphate, get us more phosphate. That's what they do.

Russ Altman (22:15): Perfect sense. That makes perfect sense. Their task is now clear. Find those proteins that increase my phosphate. And then, tell me about the DNA side. Is it a special piece of DNA in any way?

Polly Fordyce (22:27): It is. We have a special name for it. It's called an E-box, which just means that it's CACGTG.

Russ Altman (22:36): Those are actually DNA bases.

Polly Fordyce (22:38): DNA bases, yes.

Russ Altman (22:39): Say that slowly, because many of us don't speak DNA quite as much as you do.

Polly Fordyce (22:46): CACGTG is kind of... Each of these transcription factors has its own favorite, we call it their consensus site, their own favorite site that they're looking for in the genome. When you talk to people, like [inaudible 00:23:01], who tries to figure out what these consensus sequences are for lots of transcription factors, they can really read DNA. They see a sequence, they can tell you exactly who they think is likely to be there, which is so cool.

(23:15): But what I think what's most interesting to me is it's not simple on your computer where you can control F and look for a CACGTG and say this protein's going to go here. Instead, there's a lot of places where their favorite sequences, and you don't find them in a cell. There's a lot of places where they appear to be bound, but they don't have their favorite sequence. That is a big focus in my lab is how do they know where they're supposed to go in a cell, and a lot of the work that we've been doing is it turns out that the DNA letters that are around a favorite sequence, those ones can have a really big impact on whether or not a transcription factor goes there.

Russ Altman (23:56): It likes CACGTG but it also likes the environment around that in order to set the mood and make it be very happy binding that particular. Do these proteins, they actually in some way can read the DNA, they can tell the difference, obviously, between an A, C, T, and G, but based on the different atomic properties of those letters of the DNA?

Polly Fordyce (24:23): Yeah. DNA, it actually has... You don't see this a lot, but if you look at a sculpture of DNA, which they have in a lot of places.

Russ Altman (24:35): Yes, of course. The famous double helix. The famous double helix.

Polly Fordyce (24:37): The famous double helix. If they did it right, the double helix is right-handed, not left-handed. Sometimes, they mess it up. And then, it actually has two grooves, a minor groove and a major groove. Those two grooves are different widths and different transcription factors interact with different grooves in order to figure out where they're supposed to go. The ones, like these ones that I'm talking about, they're major groove readers and they stick part of their protein into that big wide groove in the DNA and they actually read out the nucleotide bases with physical bonds. Other ones tend to hover around that. They take positively charged residues and stick them in the minor groove, and that gives them information about where they are.

Russ Altman (25:20): All right. That's very cool. Okay. You really painted a great picture that it's a very intimate interaction between the protein and the DNA, basically could hardly get closer physically. If it's a match, if the protein says I found my favorite and the environment and I like what I'm seeing everywhere else as well, what happens in the cell? What is this trigger that would get us excited?

Polly Fordyce (25:46): This is really funny in that we've spent, as scientists, this was an easier problem to figure out which DNA a particular transcription factor likes to bind. They usually have kind of a folded part and then they have a whole part that's not folded at all. It's just like spaghetti in space. For decades, we've kind of cut off the spaghetti in space because it's hard to work with, and we've just asked to how do you find the DNA. But you're right, that their job isn't done. Then they got to turn on transcription. Somehow, that spaghetti part that we-

Russ Altman (26:19): It's funny because... I'm sorry to interrupt, but we had this exact discussion about your favorite protein when you were saying it wasn't clear if the business end was the only thing that mattered or if the whole rest of it mattered too. You learned there that the whole rest did matter. I'm getting a strong feeling you're about to tell me that the whole rest also matters in this case.

Polly Fordyce (26:39): I know. It's funny to have decades of your work reduced to one sentence. Turns out the whole thing matters. But yeah, I mean, it does. The whole thing matters.

Russ Altman (26:51): Please tell me. I'm sorry I interrupted, but it was just a striking parallel.

Polly Fordyce (26:58): We're lucky in the bioengineering department, with Lacra Bintu, and she does these really cool things in cells where she basically tries to ask their job is to turn transcription on. She has these cell-based assays where she makes millions of different parts of the spaghetti and asks which one turns something on. Now, in my lab, we're trying to ask, we know that when they're bound there, they're supposed to recruit particular coactivators. Then those coactivators recruit another protein, RNA polymerase, an enzyme, whose job is to actually copy the DNA and make RNA.

(27:37): We're trying to do a bunch of experiments now to figure out of those spaghettis that talk to these other spaghettis, who talks to who, how strongly do they talk to each other, how fast do they come on and off, and then can we use that information to predict not just where a transcription factor binds in a cell, but what it's going to do? Can we decode the instructions for when to make a recipe?

Russ Altman (28:00): Yeah, no, your description is really amazing because just to review, we had a bunch of atoms interacting to decide that they liked each other, but then in a few sentences, you got us to the decision by the cell to make an entirely new and different protein. I think that gives people a really nice understanding of how these atomic level events that seem inconsequential actually lead to all the processes of life.

(28:25): Well, listen, in the last minute, I want to just ask you, what does the future hold? What are your priorities for the future, and making sure that not only your progress continues, but that the world's progress in understanding these phenomenon continues?

Polly Fordyce (28:40): I think these measurements are super important, and a lot of people want to make them. They're technically hard. My lab has people from computer science, mechanical engineering, the chemistry, all different kinds of people come together to be able to do these really hard problems. I would like to first build a bunch of instruments here where other people could come and do experiments under the guidance of a professional, so that more of these measurements could get made. And then long-term, we're spending a lot of time thinking about how we can make them easier for other labs to make. One way that we do this-

Russ Altman (29:21): Go ahead. I want to hear about that. That sounds good.

Polly Fordyce (29:25): We can make these tiny beads. These beads are little hydrogels, almost like that there was that drink in the '90s called Orbitz that had these floating particles. It's like that.

Russ Altman (29:40): Is it like boba tea? I think that's a more present-

Polly Fordyce (29:41): Yes. Thanks for keeping me current, Russ. Yeah, like tiny boba tea, much smaller than the bobas. We can make them up to a thousand different colors by putting these crazy materials called lanthanide nanophosphorus in them. Now, we can attach a thousand different things where the color of the tiny boba encodes what was attached, and now we can probe binding between a thousand proteins and one other protein or we can start to scale that up more and more.

Russ Altman (30:13): You can send these boba teas all around the world.

Polly Fordyce (30:15): Yes, we can send these boba teas are more portable for other people to use.

Russ Altman (30:20): Right. Great. Well, thanks to Polly Fordyce. That was the future of proteins. You've been listening through The Future of Everything with Russ Altman. If you enjoy the podcast, please consider subscribing or following. It'll help us and it'll keep you abreast of all the new releases. Maybe tell your friends about it too. Definitely, rate and review it so we can get better. We also have more than 200 interviews in our back catalog. There's a lot of good stuff there and I recommend you check it out. You can connect with me on Twitter @RBaltman or with Stanford Engineering @StanfordEng.