 In 1964, Steampart was a conceptual scientist and also an entrepreneur. He's the president of Aluminium-Nay, made by Timothy Mobile available by irradiating milk. And he used the practice from that to start the Wisconsin Alumni Research Foundation, which has continued his lectureship in his name, at the Wisconsin Research Alumni Foundation, Alumni Research Foundation World. He has turned out to be one of the best sectors of an institution in this country and across the world, if we could come to that. And in that way, we are really excited to have Peter Schultz here, because he too is an exceptional scientist and an exceptional entrepreneur. Some of the biggest companies have been started by this, after Max, after Matrix is one example. And he's also contributed to science in multiple levels. Every field that he has touched, he has ended up making several contributions. What I'd like to say, in addition to this, is that he actually was a person who, as a graduate student, undergrad and a graduate student, started the light-up investigation that actually fuels our home, that is polyamorous, a sequence of binding molecules, that can be engineered to target many sequences, and tethered to that, he actually had a DNA cleavage reagent, which is that strategy is now becoming terribly important in human genomes and engineering gene targeting events in mammalian cells. This is almost 25 years ago. Today, he's going to tell us about expanding the genetic backbone of the early things that he started and how, in words of some, has gone past nature into 80 different types of side chains with the natural system, natural organism, from bacteria to man, or human cell types, I should say. Tomorrow he's going to talk about stem cells and how he's doing chemical approaches to engineer and reverse engineer stem cells and tissues. The last thing I want to tell you is there will be a reception. Don't ask me too many questions. He's taking a readout. It's been up all night, and he's here to give you the stock. It will be at the reception, hopefully still a week, but make sure he doesn't get in the line before he finishes. Without every usefulness and tools. Great. It's a pleasure to be back in Wisconsin. There's a terrific science going on here. I think this is the third time I've been here now. I'm in the third department. Anyway, really exciting and terrific to hear all the things going on. What I thought I'd do is tell you about some of the research we've been doing in Scripps. This is actually my title slide. It's a slide down using some high school clocks where they want a top end chemistry. And so the first time I did that, I said, sure, I'll do it. And as I showed up, and I thought it would be a class of chemistry majors, but they opened up the bleachers in the gym. It was a nine through 12th grade, right before lunch. And I was talking about chemistry, talking about a lot of things. Anyway, this is my title slide. Chemistry is a simple but not wildly popular science. And what I use this to point out is what chemists do better than any other field in science is they control the structure of matter in the chemical level. And matter all the way from small molecule drugs to large vinyl molecules and polymers and so on and so forth. So that's what chemists do better than anybody. And in fact, chemists over the last 100 or so years have become remarkably doubtful in making molecular structures from complex natural products like vancomycin. So whole proteins like insulin one can make most recently to an entire genome. Roughly 10 to the eighth adult. So there's always nothing a chemist can't make, but a real challenging chemistry for the next 100 years is what we can't do and we're not very good at is a synthesis of molecular function. So although we can make almost any structure, we can't sit down and say, oh, my catalyst will convert that thing into that long. Or we want a priori to make a selected inhibitor of the rat's oncogene protein or to fabricate a high temperature superconductor with a desired TC or malleability. We don't know how to do that. We really do have a challenge when it comes to making molecules with defined physical properties, biological and chemical properties. So when we got into this area, we said, well, what do we do? And so we looked around and I picked up a copy of Stryer and read it. And you realize the fact that the most, the best, you know, molecules, functional molecules come from nature. I mean nature is incredibly adapted to making functional molecules of your energy. Interested in energy conversion if that was some fact. Satellar molecular recognition, the immune system, the antibody molecule, if you're interested in catalysts from methane conversion and methane monoxygenation. So Mother Nature has figured out how to make incredibly interesting molecules with remarkable functions. So we said, well, maybe that's where we should take the lessons from nature and use nature's synthetic strategies and even the molecules and systems and molecules from nature themselves. Combine those with the chemist's ability to make defined three-dimensional structures and their understanding of chemical reactivity. Use those together so it's really the synergistic use of chemistry and biology together that create new molecules with interesting functions that you can't do by any other route. And so that's really been the theme of our science synthesis at the interface of chemistry and biology from our early work in antibody catalysis where we exploited the utility of the diversity, how much more diversity in the immune system makes selective catalysts all the way through proteins and clay masses, large systems of molecules and materials. What I thought I'd do today is focus on one problem we've been involved in using chemical and biology and biological synthesis together and ask the synthesized large collections of molecules with interesting properties, collections of nucleic acid and proteins and the form of living organisms. And so can we make these large sophisticated systems with no interesting properties and if you're going to change the properties of a living organism, what property would you change? Well, the property that's conserved across all of evolution, the exception of a handful of proteins that use saline system and pure lysine is a genetic code. Every form of lysine uses the same 20 building lys for as long as we've known. And the question is why? And if you look at these 20, I mean, these are just greased. If you only had 20 to play with and you made this molecule and this molecule, this molecule, why would you just move a methyl group over here and it doesn't make sense. The chemists would say, hey, the most important group in chemistry is a ketone group. It's not in the genetic code. You have to bring it in and throw it back. So that's the question. You can even ask why 20? If you discover life on another planet, is it going to be a 20 amino acid genetic code? 21, 25, 30, and it's 30 is a better, more usually better. Okay? And so that's the question. The question really is what would we look like if God had worked on this stuff with that? So that's the goal. It's to figure that out. We're experimentalists, not theorists. So the question is, can we actually begin to make these chemists chemically defined organisms where you add the 21st amino acid of a predetermined structure? And so how do we do that? Well, you clearly have to add new components to the predicate biosynthetic machinery. And I'd like to make the argument this is now, today, is a well-defined chemical problem. You know, the structure is of almost every player in protein biosynthesis that, you know, two and a half angstrom resolution are better. And so as we know, DNA is transcribed to RNA, which is translated out into ribosome. The triplets, which make up the genetic code, three bases code for any amino acid, are translated by a set of adaphramolic fields these transfer tRNAs. It's simply read out the genetic code based on lots of crit based tRNA interactions. At the end of the tRNA is the amino acid which is an ester linkage and we just simply get a trans-isolation reaction to make the protein. The genetic code is enforced by a set of enzymes called the amino acetyl tRNA synthesis. They recognize the tRNA and say that's a phenyl halogen tRNA and load it into a phenyl halogen. Okay? So, what do you have to add? What do you have to do to add new building blocks to the code? You need to do a number of things. First of all, you need a blank codon to code for your 21st, 22nd, 25th amino acid. How can you get these? Well, one way you can get them is to realize there's 64 possible three days codon, 61 code for amino acids, three are stops. You only stop printing biosynthesis once, therefore you have two blanks. We're going to use one of those stop codon blanks. The number of stop codon TAG is our first blank to genetically encode amino acids. You could go to the other stop codon, you could go to a four-base genetic code, you could go to rare codons, and you can engineer a deleted E. coli genome where you take back some codons. The next step is to build a tRNA that recognizes your blank codon and functions on the ribosome that is not, or diagonal to all the other tRNAs. What do I mean by that? That tRNA cannot be associated for any of the amino acid tRNA synthetases in that cell. Otherwise, it will be loaded with one of the codon blanks. You then have to build an amino acid tRNA synthetase that recognizes your new tRNA and loads it with your new amino acid. And again, this synthetase has to be orthogonal. It can only recognize your new tRNA and none of the other 86 tRNAs in, for instance, E. coli. Otherwise, it will load the wrong tRNA with your amino acid. And again, you won't get germ growth. You then have to evolve the synthetase to uniquely charge your tRNA with your 21st amino acid of interest and no other endogenous amino acid in that host cell. Again, a real problem in selectivity. Finally, you have to biosynthesize your transport tRNA amino acid. So that's the challenge. The real challenge is not an engineering synthetase that will guide your tRNA, but in not recognizing the other 86 or the other amino acids. That was really the challenge in this whole process. So first of all, let's worry where we start. Is it the amber codon? Is it the blank codon? Can we get amino acids in? And the first step, we're just going to feed the amino acids to the cell to the media. It turns out E. coli has 56 amino and amino acid transporters as many of which are relatively not selected. And in fact, we actually assay transport. And almost all the amino acids we add get taken up into the cytoplasm. A few don't. And the trick is here, if you've got a phosphonyl or whatever, just make a light x-cytoplasm. And it's taken up. Another degraded, and you simply believe the metabolic enzymes. So in general, getting the amino acids of interest into the cell is not a big deal. The next step is to build your orthogonal tRNA synthetase pair. And this is the challenge. Well, what we did, Dave Lew actually, who's now at Harvard, took a known tRNA synthetase parenene coli and tried to modify the tRNA and mutate the tRNA so that it still functioned on the ribosome but it wasn't as substrate for its pyranthesynthetic than for a loopy or thigem. And he was successful at that. He then tried to engineer that pyranthesynthetase to recognize the mutated tRNA and the mutated tRNA. He was successful at the first part but could never engineer out the selectivity for the original mutated tRNA substrate. And we spent a lot of time and energy on this with other pairs and so forth and so on. It just doesn't project because they're stained red. From all the blood that Dave spills. So the next thing we did is another way to do this. Let's import tRNA synthetase pairs from other organisms and use a heterologous tRNA synthetase pair. And what suggests to that is it's known in the literature that bacterial tRNAs eukaryotic tRNAs and arguably bacterial tRNAs have different and distinct identity elements that have to do with this acceptor stem and this variable arm. So if we import a tRNA synthetase pair from yeast and E. coli or from RV vector and E. coli, we should have a pair that's functional but doesn't cross-reactive in the MD. You guys can see coli tRNAs in this case. So we tried that. We went to yeast to begin with. We imported a yeast tRNA made to corresponding on-sensor presser. It functioned in bacteria but wasn't recognized by the E. coli synthetases and we then imported the corresponding glen synthetase rs and realized it will still love this tRNA in yeast and but it doesn't love e. coli tRNAs so that looked really great but as much as we tried we could never evolve that synthetase to take a known e. coli acid and it's probably because it had really low intrinsic activity in bacteria. And then we got to other pairs and failed again so we were, I think, six years into the project by this time and people were getting just a little discouraged but they had to take it. So finally we removed archae and archae, again, has distinct recognition elements compared to e. coli and bacteria. The nice thing here is there's a lot of information known about archae tRNAs. It was known that the phlegm of the e. coli is expressed well in e. coli. In fact, it was argued in the literature that a tyrosyl tRNA tRNA synthetase pair when expressed in e. coli work and don't cross-react with the environment's e. coli machinery. So we got really excited about this. It was known that this samicodon was not recognized by the synthetase so we thought we could change this to recognize the blank codon without any problem. There was no proof for any theory, so we built the gene and expressed it in e. coli. And long ago, this tRNA was amino-isolated by e. coli and bacteria synthetase. So we failed once again but Leigh Wong, who's now a salt, was doing this and he was a really bright guy and persistent so he said, forget it, I'm going to do what I'm going to do. I'm just going to make a library of these tRNAs and select for synthetase of the right properties. So if you look at the consensus sequence analysis of the tRNA and decided you could mutate these residues that is made in N, made a library of tRNAs and then the question is how did you find one to work? And so what he did is he took this library of tRNAs and put them in e. coli and he actually also added barnase, the gene-coding barnase which is barnase is lethal to bacteria, it's a nucleus and he put the black codons, two black codons and permissive sites where you can put any in e. coli. And so the reason that if in fact any endogenous e. coli amino-isolated tRNA synthetase recognized as any of these tRNAs loads it with a common e. coli and that's if you make full-length barnase and build the cells and they're gone. So you're done. You're not really done. The simplest solution to this selection is just to make it non-functional tRNA. So what he then did is he took the black codon and put it into a permissive site and beta-laconase, an essential gene product when the cells are grown on an ampicillin. He took all the winters from this selection and then took the corresponding archi tyrosyl tRNA synthetase and put that in. So now the idea is that these tRNAs function in e. coli full-translate on the ribosome that will be loaded with tyrosyl and you make full-length barnase with cells to live and you're done. That pair is orthogonal and it works in e. coli and in fact that's what happened. And we've not only made this as a orthogonal pair but we've made about five others using this simple to build e. coli using this simple two-step selection strategy. So then we wanted to ask how do you evolve that tyrosyl tRNA synthetase to take whatever amino acid you're interested in? And what we wanted was a general solution because if every time we want to add a new amino acid to the genetic code it takes two years or three years it's not going to be very interesting. So we developed what we thought was a general strategy for the evolution of specific synthetases. We went into the structure of the archaebacterial tyrosyl tRNA synthetase and randomized all the residues in close proximity to the amino-associated tyrosine. We made a library about 10 of the nice mutates and so how do we find one that's specific for whatever an amino acid you give to? So what we do is we go back and again it's a two-step synth type selection. We put the blank code in the gene-incloding form-prone polysiltransferrate which is required if you grow bugs on form-prone at all. This is a permissive site. We put in our tyrosyl tRNA that recognizes the blank and then we put in this library the synthetases and then we add the unnatural amino acid to the growth tRNA. Now, any of these mutant synthetases recognize the unnatural amino acid and load them onto the erythritol tRNA so that will go in the catapest permissive site and the cells will live when grown on you're done. Again, you're not done. The simplest solution to the selection is to re-evolve the tyrosine from Fennelhallen. So how do you delete all the synthetases of this amino acid? You don't know. So what we do is we again take all the winners from this selection and put them back in E. coli and now we put three blank genes into pharynx. Again, three blanks get into the permissive sites. You put the tRNA back in and now you grow the bugs. The only thing you do is you don't put it in your unnatural amino acid. So now, if in fact one of these synthetases recognizes an endogenous amino acid loads it onto the tRNA and make pharynx in the cell site. So that's it. And it turned out this actually works quite well. We've actually done this with over 50 amino acids with high fidelity and good yields. There's a company now making a kilogram of human growth hormone with an unnatural amino acid that's going to go on a phase two clinical trial. So there's really less work and so there's still issues of contacts that have to work out. So to test this once they got this to work the first amino acid we looked at was on methyl pyrosine which is pretty boring actually. The reason we chose this is because all you're doing is taking a hydrogen off of the pyrosine and replacing it with an methyl group which is pretty trivial for a patient. And what we were most worried about in developing this approach was the fidelity with which we could incorporate these unnatural amino acids because we went with all this work that created these orthogonal on cross-reactive pairs. And the on-protein phylosynthesis makes a mistake in every 10 of the fourth count. So we said, well, this will be a good test case because if we can put this selectively and on pyrosine or phenylalanine that will go as well. So we went ahead and put a blank coat on at the third position in DHFR. A ball of tRNA synthetase pairs specific for this 21st amino acid. And when you add all that to the cells in fact you see a fair amount of DHFR. They contain the unnatural amino acid. This is wild-type. These were grown in metal and media conditions which wild-type is about 10 things per liter or about 5 things per liter. If you believe the amino acid in the tRNA you see absolutely or the synthetase, you see absolutely nothing. So in this case of extremely high fidelity we actually analyzed the mass fact and the mass fact of the whole protein and all those big fragments. And this confirmed that your neomethyl pyrosine went in only at the position specified by the blank. No other position to maximize. So we were pretty enthused. So they said, what can I do next? So Lane J. Nain actually looked at incorporating heavy atoms for structural studies in particular the periodopanolalene which seemed like a relatively simple direction to go in. The nice thing about this is for a single wavelength anomalous diffraction satin phasing studies with an in-house x-ray source. So they were able to evolve a synthetase in paces. Again, it goes in very efficiently with high fidelity incorporated in an internal site for a phenylalanine in T4 licensed on salt destruction. We're able to satiate the structure. So this actually works well. We're actually now extending this with iron and so forth and so on. But then we said, well, let's go in other directions now. What can we do that's more challenging? So Jason Chen said, well, why don't we genetically go to photoproslinkers because there's a huge interest in cell biology now in backing all the cell circuitry. And a lot of interactions between proteins and proteins are high at PMS, but many of those aren't stable. So the transient interactions are kind of substrate. So what you'd really like to be able to do is take a molecular interaction and when the proteins come together they'll lock it to the covalent bond like the palaces. And so one of the best photoproslinkers is this Ben's of the known. If you cite this to the end of high star excited state, it either CH bond inserts so Jason Chen genetically encoded this as the 21st amino acid. Again, it goes in efficiently with high fidelity. He then took this and inserted it in the interface of the GST homodimer. In fact, so all he did was put in mutations, mutations, psychoactive mutagenesis and then genetically encoded this out of the amino acid. He never touched the protein, radiated the E. coli cells and crosslinking. Well, he actually made the mutations that this whole site's got nothing. So this actually works pretty well. We've distributed to about 300 labs. Again, the drawbacks that we're still working through are there are some contacts of facts that we don't understand and we're giving genetic selections to overcome those. But in general, this works pretty well. We've also genetically encoded the aeroglyzids and the crosslinking reagents that you can use to look for organ receptors or ligands and that and so on. Another really useful tool in cell biology is green fluorescent protein and analogs. You can attach GFD to the NRC thermos of the protein in the demo and it serves as a terrific reporter, optical reporter of protein expression, localization and confidential changes at some point. The problem is you're really limited where you can put it in the protein. The NRC thermos is a big ball, so the spatial resolution is not perfect. So we said, wouldn't it be nice if you could genetically encode a really small fluorophore in order to serve 100 million people in any site in the prairie of New York? So, Jan Young stepped into this and the first fluorophore he looked at was hydroxycomorin which is an environmentally sensitive fluorophore. It's also pH sensitive so it could be used as a local sensor of pH itself. He wasn't able to genetically encode this. The yields are about 30 makes per liter in E. coli. So to show the utility of this what he did is he put it in two sites in my alone. It is the E30-7 in kind of the core in his first helix. And so then he unfolded the protein in the presence of urea and if you monitor the unfolding by CD you basically see global melt in the entire protein. If you compare that with unfolding by monitoring the fluorescence of the comorant in position 30-7 you see it basically melts with the core of the protein. And on the other hand you put this fluorophore in position 4 you see that this helix melts before the rest of the protein so you now have a very local pearl of protein conformational changes. Okay? So I just went to that A.I. Chapman in our Harwiches room I actually inserted this amino acid to the FTSD protein which forms this ring the septum when bacteria divide and you can see that you can manage FTSD and these are live bacteria in real time. So we think this actually is going to be pretty useful and you see we've genetically encoded other fluorophores as well. Another spectroscopic probe that one could use is a nitrofoul group a really simple group but nitrofoul residues quench the intrinsic fluorescence of tryptophanes. So you can use them for this endogenous to the proteinous site. So moment genetically encoded this amino acid we put it into again it goes in quite well so we inserted it into the zipper region of GCN4 so we put a tryptophan in position 22 on one helix and a nitrofee in position 22 on the other these are about 6 actrals apart and you can see that when you set up this construct you actually pretty efficiently quench the tryptophanes fluorescence and now what we did was march the tryptophan away and we march it to position 20 and then position 10 so we went from 6 to 15 to 26 actrals and you can actually see that this is the pendant increasing increasing tryptophanes fluorescence. So this could again be a useful conformational probe in D-drone for studying protein structural changes. Another spectroscopic probe is not you need this fluorescent probe it's an IR probe an infrared probe. The problem most doing IR studies of proteins is there are very few bunch of them that are in an unplugged region of the IR spectrum. So we thought well maybe we can use the area of light draw. This is an environmentally sensitive group the IR spectrum absorption spectrum depends on the local environment as I said it's in the free region and it's also a field effect sensor. So we actually genetically encoded this Roshan and we inserted it into my global instead of the phenylalanine 64 we put it as cyanosubstituted phenylalanine and you can see this nitro group right here when you measure its IR stress as a function of bound ligand you can see it's actually a really sensitive reporter of the nature of the bound ligand to the ion. It's also sensitive to the oxidation state of the ion as well. So this is again a good environmental reporter right now we're collaborating with C. Boxer to actually use it to measure local electric fields in DHFR in DHFR mutants. Now the other thing you'd really like to do is proteins as chemists and selectively modify them with exogenous chemical reagents and that's hard to do because most of the cytokines and proteins have relatively non-specific carboxylase aminobrosin so the only one you can really selectively is cysteine and if they're multiple cysteines and they're involved in folding we asked whether we could begin to genetically encode uniquely reactive chemical groups and proteins so for instance you could put in an acetylene and a zinc and carry out a two plus six with this reaction that reaction you can't do with any of the other kind of planing of cysteine and you can do that. You can put in a chirogroup and react it with alkoxumine you can put in a boronine or react it with dials siowasters and react them with side chains or backbone and means to make cyclic proteins and peptides because this is an alfalfile you can ramp it up and this phenylsilanine you can actually treat with hydrogen peroxide and form a dehydro-alanine so if you put this in site specifically treat with a mild ox under mild oxidizing conditions you get a dehydro-alanine chain file or a sugar derived file or even a fissile 1-2 file mean you can make a whole variety of amino acids side chains including acylated and mentholated analogs of lysine that we're putting into histones to look at histone for translational modifications and this works quite well you can generate these proteins reasonably yielded in high levels of to the point where we actually now are starting to understand whether we can actually replace disulfide cross-links with landfiling cross-links or you can imagine putting this dehydro-alanine in a loop in a type 2 beta turn form a dehydro-alanine and then come back in with a Michael with a Hickley file and ask whether you actually would preference for D or L configuration of that site so you can actually use it as a well boronate-splined adiols ok so if you have a carbohydrate and you want to target it you can use a protein containing a boronate of amino acid boronates also flying to steering proteases are terrific if you can use it to selectively target proteases and so forth so we genetically encoded Mark Bushin and Eric Breskin we actually encoded this boronate and then we put this into a protein Z-alanine protein we're running down a dial containing pollen and sticks like glue with the affinity curation you oxidize you get off the tyrosine containing protein and reduce you off phenylalanine so you can use it as a really simple affinity curation trick the other thing you can do is put this boronate into a protein and then add a fluorophore that has a dial now if you add this fluorophore then it doesn't fluoresce I was there in red if you add it to the boronate and amino acid and you complex these dials you actually get intense fluorescence ok so this could be a really simple way to introduce a huge variety of different fluorophores um into cells and could be probably um simpler than using the flash ligands perhaps less time now I can be getting slightly more of my proteins so erics it up I can take this keto amino acid and there's a proxy name to put in a laxative protein 8 is a fluorophore at one site I can then take a melanin to sanitize the laxative 594 and put it in here and so now I can make pairs fluorophore pairs they define sites in a protein very simply because I have two and then we carry out a single long on unfolding sites in the presence of the nature and we're actually trying to get the map the structure of the unfolded state by various stress pairs as I said um you can actually make three pairs of things this way so if you go in hopefully if you take their cases which is something people inject themselves every day and you look at it on an HDLC it's five peaks it's a mess if you have a city activity those five peaks only one are really active so you know what kind of I mean the FDA should be shocked but yeah because if you did that was lived before people would have a heart attack but the problem is it's the only game in town so big pharmacist says hey I'm going to kind of modify this with a pad and you know this is the best I can do so we're making really terrible protein reagents so the question is do you actually use the gene chemistry can make an incredibly selective modified there are two proteins so we can actually begin to do medicinal chemistry on proteins like we do on small molecules so the folks at Amherst other children Daniel and others actually went in and made a variety of human growth hormone mutants pagolated those and the pagolated proteins are absolutely chemically pure they have wild type activity and two week half-lives and they're going into phase two in human trials they look like the best human growth hormone therapy out there and again as I said they're going actually into a kilogram, make a kilogram of the modified protein now you can also put this oxygen in and put a little oligonucleotide on say a HER2FAB put an oligonucleotide on what ever target on whatever other protein you want to add and you can form homodymes so you can begin to mix and match a targeting protein with a toxin or what have you or HER2 and HER3 and make all kinds of hormone heterozoic which might have interesting therapy activities so that's synthetic modifications there's also a whole family of post-translational modifications and nature puts in the proteins solpation phosphorylation glycosylation and many cases we don't really know what these do the real challenge in studying post-translational modifications from a chemist perspective it is hard to make selectively post-translational amount of proteins we don't have the selectivity of the enzymes and some of these are dynamic processes so for instance many proteins you can add proteins now around our sulfate we don't like the sulfation dots okay on CCR5 the sulfate so Chang Liu actually came in the lab and said why didn't he solve this problem easily why don't we just directly genetically encode these post-translational modifications and just put them at the level of protein biosynthesis forget about them and match the proteins made and so we just specify the position okay so he genetically encoded sulfylpiracy works really well he put it into a naturally sulfated protein this is a leach protein it's sulfated naturally in pyrosine 63 so what he did is he changed this to a blank genetically expressed it and then actually determined the co-crystal structure and then actually determined the co-crystal structure which he actually winds up doing is decreasing the Ki by about a factor of 10 to the 26th bentomol and it does that by sulfating this pyrosine we still have a whole new series of hydrogen bonding network in the protein so you can actually begin to do this at will about phosphorylation the problem with studying phosphorylated proteins is a kind of assumption and the phosphatase takes it off so it's dynamic so even if you can't make a selectively phosphorylated protein then you go bye bye as soon as it sees itself so what we did, what Jane May did is took an analog and then on the analog it was developed by Darnell and others and it provides some mental phenomena this is a stable mimetic of phosphotidrosine and so they showed you that they coded this and it shows utility he went into the jack-sat signaling pathway and when you activated this pathway you ultimately get phosphorylation step one in homo-diagonization which leads to transcription so what he did is he replaced the naturally phosphorylated pyrosine 701 which this provides about the phenylalanine you and what you see is you get a constituently activated and homo-diagonized protein and it can't appear to be a little less it's only one nail more versus one nail more for the phosphotidrosine but I think you can see this is probably a pretty good tool for the phosphotidrosine another useful thing to be able to do in cell biology is to turn on protein activities with light in a spatially and temporally controlled fashion and one can do this by making photocase protein photocase small molecules depending on if I ride the chat others have been really useful so well in the tools because you can only count some of the specific sites in the cells so can you do the same thing with proteins? well some work has been done in this regard to do it inside living cells what we get is photocase searing cysteine and pyrosine residues and so forth with a nitro-venzyl derivative in this case with a searing group we use the cyanoboxy-substituted nitro-venzyl group which you can take off very efficiently with a 405 nanometer laser which is a kind of photo microscope so I edited this and left key so what he did to show his utility is to use a bit of a sentiment and he put it into the yeast transcription factor which is responsible to cellular phosphate levels so in the presence of high phosphate the faux4 in the nucleus the searing 128 gets phosphorylated it's sent a substrate for an export receptor MSN5 and gets exported out of the nucleus into the cytoplasm so if you genetically encode this photocase-searing analog and put this inside the cell it's going to go into the nucleus and it's going to stay there because it can't be phosphorylated and that's what happens so it stays in the nucleus there it is and then you photolive the 405 nanometer laser and you can just watch it get phosphorylated and in real time photoinitiated phosphorylation in an export and you can actually follow the patterns of this in real time and so on so again I think you can do these with kinase to become pathway as you can probably photo-initiate transcription factor binding to DNA and so forth and so on Dockery in Pam, England actually showed you could also look at your names and encode the amino acids and insert the amino acids so we cleave the protein backbone so they use this nitro-palalamin derivative so we made this and inserted it into a peptide-delicited chemistry and when you photolize this the 369 nanometer light you actually cleave the backbone you get the free acid-neuron water and the cytoplasm compound they want to yield a little low but it works reasonably well so then Frank Peters genetically encoded this inserted it in position 82 and T4 lysozyme and then photolized and we get these two fragments and indeed we get these two fragments the overall yield is about 30% and we're not quite sure why it's not low yet whether it's being quenched by those side changes of salt but I think this wind-up being is the way you can turn you basically destroy the protein in a specific site and stuff you can do this reversibly too Dan Roth actually used an azopensin-based amino acid busy genetically encoded to make proteins to reversibly turn on and off with light using something developed by Erlinger a number of years ago and the idea here is what genetically encoded was an azopensin amino acid and this amino acid exists in the trance state in the cis state and you can convert these with light at different wavelengths so you can build up by radiating with 320 nanometer light you can build up a photo-stationary state of 80-90% of the cis-azopensin so that's what Dan did and then to have a light-activated protein cyclic AMP dependent DNA binding protein and put this azopensin residue right outside the cyclic AMP binding site and it turns out in the transform the cyclic AMP can get into the site when you bind the DNA but when you photo-irradiate and convert the trance to cis you actually lead to a loss in binding affinity for cyclic AMP state and a loss in affinity for DNA and it roughly corresponds with the photo-stationary state of the cis form now about a third of all proteins bind metals it's a build now I am binding proteins and stuff because you know usually you have build a coordinates ligand sphere four different ligands you have to position those in the right geometry and you have to position the side chains all around that it's really hard to do so we say forget that let's just genetically encode directly metal binding DNA acids pretty build the metal binding spirit so we get that, we genetically encode bicarral aligning, ferrocein and this hydroxyclinoline aligning derivative they all go in quite well large yields of protein can be made so for instance you can put these into antibodies and treat them with a radio ice belt and do imaging of perhaps radiotherapy to destroy cancers metal-mediated oligomerization controls structure metal-dependent hydrolysis oxidative cleavage and heavy atom physician so basically giving an idea we went in and looked at HIV and making an inhibitory HIV so when HIV finds the host cell the GP-41 undergoes a conformational change and we release this structure that has two trimeric helical regions and so one trimer one terminus and the other is the other these are inserted into the target and the viral membrane respectively and then what happens is when this is unveiled these two trimmers come together to form a diamond trimer that brings the two membranes close together and they fuse and you're infected so what was generated is a drug that is used clinically it's something called T20 or Tuzion our peptides that are actually compete with the binding of these peptides to this oil and soil so you bind these synthetic peptides they block this reaction and you block HIV infection but these actually don't work that well they're not very quote-up against and inject hundreds of milligrams that are short of half-flag and they highly act they prompt that activation so what Roshan said is a very simple solution just take this T20 peptide genetically encode this peptide with a bioperial at the end and throw an iron-3 thing on trimeride it's got highivity it should bind like crazy so we did this and trimeride and then we measured this in HIV entry assay based on baby dial expression we actually got an inhibitory in EC50 of 30 people which is about a thousand times better than T20 so this really does is an incredible code of means by DNA to be blocked so we're actually going to need to think about whether this could be clinically useful you can also take a dial binder and do a kind of bridge or e-bride type experiment where you actually take a DNA binding protein simply mutate two residues that are close to the DNA backbone in this case placing 26 in the homodimer replace it with a bioperial group add copper and a reductin so now you have a genetically encoded metal binding DNA cleaving protein you really don't effect the affinity for DNA at all with the bioperial license of bioperial mutations but you see now if you carried out a mapping study you see that you get cleavage directly to Jason for the path binding so again you can do this in vivo, you don't even have to isolate the protein so another thing we thought might be interesting to do what else could you do with genetically encoded I'm actually honest well a huge problem now a lot of people are interested in infectious disease and vaccine infectious disease as well as cancer the problem is if you want to make a vaccine to cancer you have to make antibodies to sell proteins and we have tolerance so it's very hard to make antibodies to our own proteins if you want to make a antibody that blocks malaria and you immunize with malaria what happens is you get a lot of antibodies the malaria pathogen they just happen to not find an equalizing appetite so the problem is there are a lot of things we like to make antibodies to but we either can't make them very well or we make them to the wrong sites so we said can we use an unnatural amino acid but basically go in and paint a bullseye on a protein on one protein and then decide we want the antibodies to buy so we thought it's simple make sure it's been allowing, we can do this again this is pretty immunogenic okay it's a high acid, antibodies love to buy in microfilm groups so what we did is we put this at one site in TNF alpha in mouse TNF alpha we substituted tyrosine A6 with this nitrofee we thought oh this will be really immunogenic with this one mutation and it's so similar to tyrosine that the antibodies that we make for the mutant protein will cross-react with wildfire so we did that made those mutations it's like two atoms and a very large protein are different okay you immunize with TNF wild-type TNF you don't get a mutant response it's tolerance you immunize with a phenylalanine mutation you don't get a mutant response you're immunized with a nitrofee you've got a massive mutant response that cross-reacts with wild-type protein you don't even need oxygen you take these immunized mice and now challenge them with LPS which is a TNF-dependent model the mice that are immunized with wild-type TNF all die the mice that are vaccinated with a nitrofee mutant actually have incredible survival in this model so we basically broke in tolerance in this case so we think this may be a general strategy so now we put it in the beta-anomaly protein the name of beta-anomaly vaccine prostate-specific antigens and we're working on putting it into malaria and HIV at the topes that will lead to a neutralizing response okay, forget about side teams what about the backbone we love the mega-polyester protein so we try to genetically encode these hydroxy acids so instead of making animals you put pesters in and so Chris Anderson tried this realized that in fact as you put it into the cell it gets oxidized so the keto acid transaminate into the virus so Gentile came into the lab and said oh, how do I stop this we couldn't find it in the hydrogenase first of all but Gentile said oh, this is a lot of transaminase and then we'll just go back and forth and that worked well he evolved the synthetase one round of selection it was synthetase that did this so you can use the acid very selectively genetically encode it into high yields if you treat these proteins with K7.8 over 9 you cleave that as a bond right in two so you can use it as traceless affinity reagents or you can kind of get a ligose when we use them as mechanistic growth so you can go out and ask or now the question is going to make an entire lesser protein and well, a full Gentile is now running a DMEL acid and so the question is going to genetically encode DMEL acids he's done at what we actually had to knock out the tyrosyl deacylase some racemases and other enzymes we used some tricks from a literature said hack and others where we made mutations in the peptide transferase center and you can see now we have a synthetase that genetically encodes DMEL acids but it also takes LNEL acids so we actually now have to clean this up and we're going to go to the LNEL acid selectivity so we're going to natively select LNEL acids and also make DMELs Shane Lu just put these into antibody libraries so phase display is a powerful tool for evolving antibodies and peptides with interesting properties using peptide libraries and CDR antibody libraries because you can amplify and accuse libraries sequence the answer so what Shane did with Vaughn Schneider is he took a human germline library antibody library and then ranilized six positions in CDR 3 of the heavy chain which introduces TAGs of specific rate and then he selected for antibodies that found to GP120 now there have been naturally isolated antibodies that find a GP120 that have sulfated pyrosynthesis that's because the natural receptor is sulfated so the sulfation is a general trait so then he just carried out panning experiments and now he's isolated in the antibody that actually binds the sulfate where you need the sulfation and it binds HID 1.20 with a higher affinity so you can begin to evolve unnatural antibodies and now we're doing the same thing he's put the morning elastin into that CDR library and now this guy should target carbohydrates on proteins so we're now panning these antibody libraries against carbohydrate binding proteins carbohydrate containing proteins because now we have a warhead that'll just hone in and bind it with the morning and then the rest of the antibody will get selectivity hopefully so I think this is kind of neat why does this work we didn't think we would get this many amino acids in so we saw the crystal structures of about 8 of these synthetases now this is a wild type synthetase found in tyrosine here you put in anaphylamine what happens is the site rearranges you lose a hydrogen bond to the tyrosine okay and you reorganize the active site with new hydrogen bond donors and steric interactions in fact in this case you also put a mutation into the lube that actually changes the entire information of the south that you look at the lube the residues that were outside come in and the ones that were inside flipped out so the whole active site reorganized not only the side teams of the backbone but the other side teams of the lube that you usually never see I mean you see them germline antibodies but anyway we think these proteins are remarkably plastic in fact we see this in about six of the structures in other cases here's a benzophenone the benzophenone went in and actually found this huge hole that we generated in this synthetase and we're taking out these amino acids now so you can probably put the fission sink in here and it's going to be a G20 okay so I think again that's promising how do you go from 20 to 21 or 25 will you go with four base genetic code so Tom McAvoy and Chris Anderson set up a selection to look for a fission four base decoding tRNAs and they found for instance AGGA is decoded by UCCU because AGG is a rare codon so what they did is flip the AGG into the gene so now they're specifying the 21st amino acid with a four base codon they built a four base decoding or five multRNA synthetase paramoled it to take homeoblutamine and now they're putting it in homeoblutamine and all that will pyrosine two unnatural amino acids into the same protein ideally to go even further we've actually taken out a 52 kV piece of the E. coli genome and it has 41 testings in it we've deleted four redundant codos just taken them out of the genetic codon we've got it so we're down to 57 and now we're putting this back in and the question is if the E. coli growth is fine I think you could do an entire genome believe the 25 amino acid E. coli can you move this to higher organisms the answer is yes we've developed the same kind of two step synth selection here he uses suppression and stop codons in the DNA binding domain we've got four to drive both natives and monitor the selection so you just play, play, play, play, play and then you have to isolate the DNA and you can add new amino acids to the yeast genetic code both to the DCA and Pika you can actually move these into the malium cells the identity elements between yeast and malium cells are the same so you can just take out of all the synthophase and yeast, move it into the malium cells the tRNA doesn't work so much so we use a trick by Yoko Yoko and the literature where we actually use the spherzomophilus tRNA that expresses better with malium and planking sequences so we've built in two pairs a lusulperide E. coli pair and a tyrosil pair and so Jen Powell and when she did this so what you can see is they made both of these expression patterns so in show cells using this these evolved tyrosil pair in synthophases you can put in a range of amino acids in show cells in two to three D cells the lusulae you can put in even more bizarre things so we put in damsel and progan and so forth and so on and this actually works pretty well surprisingly you can make stable cell lines expressing these things and so that's kind of neat because it doesn't kill the malium cells even though like 25% of the codon is terminating K to G for reasons we don't know it turns out this is an endogenous amino acid and mice and us and so we evolved the tRNA to make a pair that's specific for this amino acid and now we're putting it in making a stable transgenic so we actually think we're going to be able to make a 21 amino acid transgenic mouse that specifies this is the 21st amino acid I'm not quite sure what good it is but it'll be a milestone in organic synthesis I think okay so here's the progan so what we did is this is an environmental for you can see we're going to wander a more useful wavelength so gen糰 puth it's an account modeling like an ensue and now when you buy in calcium you actually get a recorder in calcium binding and we think we can do this in a value of cells so now we're making a recorder protein instead of small amounts of it and so the last thing we did was we said you know I remember these as a feeding experiment so you said hey let's just say you're completely autonomous so what we did is we genetically encoded the phenylalanine and then we built in the biosynthetic pathway for this amino acid by taking three genes biosynthetic genes from shut-in-the-line season which would take the reasoning all the way to this phenyl acid we've sent transgenic with a specific transgenic this works well you may come to this phenyl acid in five genetically encoded brain you can throw this up in the dirt and it's on its own so that's really the next step is now that we actually have these organisms that can evolve is to evolve them we wouldn't do it in the I don't think we'd do it outside we might do it in Boston so that's what we're doing and the question is now I think we're in positions 21 better than 20 and we're better flying down so finally what we have all this is the expanding genetic code I'm kind of this one this nitro just to show you how easy this is my daughter when she was 16 worked in the lab and genetically encoded this she's good she's not that good it's hard to do so we've added a huge number of phenyl acid to think 50 or more and they don't end great there's still some things we have to work out but I think since the social system is selectable that's no longer a problem and I think for about a few billion years life on the planet has been pretty much limited by 20 phenyl acid genetic code given to us by Asia that limitation and so now the real interesting challenge is now that we can pretty much just kind of say what other structures we want or many that we'd like the question is can we have all proteins and maybe the whole organism properties we can't gather and finally I'd like to thank my group at our union actually there are about 300 ex-co-workers but these are the people in my current group and I try to mention the names clearly I didn't talk about the work of all these people I try to mention their names throughout the project I probably have the opportunity to work with the very best people over the last 20 years or so so these people make everything more just the main and it's a terrific group of collaborators it's Scripps, it's UNF and Berkeley and elsewhere and finally I'd like to thank you for your attention there's a reception by the chemistry agent right now free wine, free beer and you're welcome to join us there and tomorrow there will be a talk with you again we have two questions and let's take him over here how many of you have to go over there I ain't a direct face on this so when we got this to work one of my ex-students a lawyer and that business guy suggested that we could make friends with this and I see you're crazy with that usually when I tell somebody they're crazy and they work for me that means they get it to work so he went and started a company called Ambrick's which is kind of Daniel and Hochow and others what they did is scaled this up fidelity and the yields and so they're doing a 4,000 liter fermentation right now and they've regulated growth hormone the clinicians just looked at the phase one results and it looks like all the clinicians in the growth hormone field say this is the best growth hormone out there so far it looks really terrific so we are likely to have that they also have an interferon alpha and interferon gamma are making an FTF 21 that would work right now and they're making some people in the longs with Eli Lilly I don't know what I was supposed to say all that stuff so I just forgot it the fact is it's actually robust now and they're really making therapeutic proteins and putting them into prying some people that looks really pretty good but again we're going to take sites and get variations and efficiencies that aren't totally dependent on context and so there are factors probably chaperone folding and so on and so on there are other things we need to do to improve it so we did a complementation experiment where we randomly introduced pieces of the equal age genome and we actually got increased levels of the tractable system now we can use a lot of E. coli genetics to figure out how to make it better and that's what we're doing now but it's pretty robust surprisingly self and anti-old cyber now if I go over so now so now not only on my group of scripts I run a biomedical institute the Norris Research Foundation funds and they're able to stay and get a large amount of money we have about 600 people and complete freedom to work on whatever I want so I'm now working on all the diseases I think I've been yes we'll talk a little about tomorrow but we're working on Alzheimer's because we're trying to make a protease with the grades because there's this barrier hypothesis and we're just going to chew it out as it comes out of the brain and in the first round passing it looks pretty interesting so we're working on getting to you so in E. coli it's pretty clear because there are a lot of suppressor strength and so there are naturally occurring Alzheimer's suppressors the Alzheimer's suppressors work well and in some theory people claim that E. coli GM is a ball of the way amber codons from essential genes and so if you actually go in and look at where amber codons there's probably less than 10 to 20 genes that we might do as essential things so Chad and Groove is a letter hand so we think that'll lead to almost a getting rid of R out of 1 and the mammalian cells it's crazy because 25% of the stock goes into TAD and mammalian stable cell lines and you can grow these cells and so we don't get it our value back it's not clear the systems evolve away so we don't quite understand why that works so well and I think again some of these experiments now are going to be really neat because we're going to find new biology and things without translation then we can know about it I can sort of understand why the simple things could be evolved and Chad and Groove here that doesn't occur in nature so I'm a little bit surprised that we can get things through the peptide echo channel that on the long side changes can't have so the things get some box contact yeah you know some of the biggest things we're putting in go in the most efficiently okay so you know some of these floors and some of these metal binders and some of these extended molecules don't pretty well work we've done for like C16 so I don't think so um and uh you know a lot of the in vitro work we and others did suggest that that was the case so where we think we may be getting a problem is the EFTU has a binding site and if you look at the structures it may have a problem with binding so I mean it'll ask to have more limited depending so we're now making use of the EFTU to open up that site and we're going back and doing selections again so again we're going through stepwise the EFTU we use peptide transferase and so on and so on and I think that's the nice thing now is we just select for improved efficiency but you know size is surprisingly uh not a factor yet um taxes remarkable is there an evolutionary imperative to have that sort of flexibility isn't the EFTU yeah you know we've done a lot of work in this area especially with like germline antibodies where you can argue that it evolved flexibility to the binding site we think there's a lot of flexibility there because that helix is there's a cavity and a helix is here and then there's the solid and exposed surface so what happens is in a normal back bone change as soon as you change what's on the inside and there's no way you can compensate for all of that with eight mutations so now what happens is we mutate on the inside to accommodate the enamel acid what goes on on the outside and solve the deals with it and so it's just completely reordered so we think as long as you're within one shell of the surface you're going to see a huge amount of back bone flexibility that's the supposition we got it