 Cornell University where he's in the Department of Chemistry and Chemical Biology. He did his PhD study. He actually did his BS at the University of Minnesota so he's a hometown boy and then he did his PhD studies at Colorado and during those PhD studies he also worked at the University of Basel for part of his work. Then he did a postdoctoral fellowship with Francis Arnold at Caltech and he started his independent career at Princeton in 2015 and he recently moved to Cornell. His group is known for the development of methods in photo enzymatic catalysis and my PhD studies a long time ago involved photochemistry and organometallic chemistry so I find his combination of using photochemistry to get enzymes to catalyze a different kind of reaction to be a really creative and innovative approach. He photo excites a cofactor like NADH or FMN in order to create and this excited state will reduce the substrate, generate a radical and then the enzyme active site is going to control what's going to happen to that radical so we can get enzymes to do completely different kinds of chemistry and I'm looking forward to hearing some more details about your work. Great well thank you so much for the introduction and for the opportunity to come back to Minnesota. I haven't been here in a few years and so it's a really great opportunity to come back and revisit the campus and have an opportunity to share some chemistry with you all and so today what I want to tell you about is just sort of one story that my group has been working on over the last four to five years and I think what we found through the the course of developing new reactions and actually doing some protein engineering we ended up learning some I think pretty interesting photochemistry and how proteins are able to facilitate some pretty interesting electron transfer events and so my group is I would say broadly interested in using enzymes as catalysts for asymmetric synthesis and so sort of as you can tell from my background I'm more of a synthetic organic chemist and so when I look at bio catalysis what really excites me in my group is the opportunity to use the protein scaffold and directed evolution to really control the selectivity of reactive intermediates and so if you're thinking about sort of the power of small molecule catalysis that would be catalysis that doesn't involve enzymes I think what's appealing to me is the idea that you can take sort of very simple mechanisms merge them together in different orders to gain access to new types of transformations and just to sort of illustrate my point if you think about transition metal catalysis obviously this is an area that that as a field we've studied for say 50 to 60 years and what we know is that transition metals have access to a few fundamental mechanisms I've illustrated a few here oxidative addition migratory insertion and transmetallation just to name a few and I think one of the defining features of small molecule catalysis is the fact that the catalyst can interact and bind with many different types of substrates in bio catalysis we might define this as substrate promiscuity and so you can take this inherent substrate promiscuity with a few fundamental mechanisms merge them together in a number of different orders and you gain access to hundreds of different chemical transformations now the same holds true for organo catalysis where again in this case I'm just showing a secondary amine it can form an enamine a minium singly oxidized enamine you can exploit the substrate promiscuity to again gain access to hundreds of different transformations now in bio catalysis we tend to take a slightly different approach and that's because we tend to think of enzymes as being very specific for a single transformation and so this is sort of illustrated in an enzyme like keto reductase the name sort of indicates what's going on right it's an enzyme that reduces ketones we understand its mechanism it's a very simple hydride transfer a mechanism and I think what we've seen in the context of synthetic chemistry is that these keto reductases can bind and interact with many different types of substrates and so I might decide describe them as being substrate promiscuous but mechanistically limited and so the question that we've had in my group is can we take existing enzymes that we would traditionally think of as only having access to a single mechanism and by changing the substrate can we gain access to new types of mechanisms that would allow us to use these enzymes in completely new ways to solve challenges in synthetic chemistry now of course we are not the only group to think about this and the way that I sort of break down this area is what type of reactive intermediate are you interested in using and so for instance if you're interested in organometallic intermediates then of course you're going to be drawn to natural or artificial metallo enzymes where the metal within the protein is really responsible for the reactivity and then the protein scaffold is going to provide a chiral environment and influence the reactivity of that metal center and so metallo enzymes or artificial metallo enzymes have been known for non-natural chemistry for you know maybe 50 years and have been demonstrated to do a number of of really I think classic organometallic transformations if you're drawn to closed-shelled intermediates that might resemble more closely what we do in organocatalysis and you'd be drawn to enzymes like hydrolysis to tomoraces and albalases and so when my group started we became really interested in whether we could use free radicals within protein active sites now of course radicals are really common intermediates in nature for catalysis because radicals can mediate a number of different transformations but in the context of natural reactions they tend to be quite controlled these are really well evolved systems so that you can avoid you know protein degradation or protein alkylation and so the question we had was would there be mechanisms for forming non-natural radical intermediates within protein active sites and so what what really sort of drew our attention to this intermediate is the fact that it's high in energy and so this ends up being really valuable if you're trying to facilitate different types of bond forming events because the activation barrier for various bond forming events is going to be quite small and so this is why areas like photoredox catalysis and electro synthesis tools that have really sort of revolutionized organic synthesis in the last 15 years have really taken off it's because of the the power of the radical intermediate now of course the challenge when you use really highly activated really reactive intermediates is that it can be very difficult to render these types of transformations asymmetric and so there are small molecules that can do this but I would say that it's quite limited by comparison to other types of intermediates and so the goal in our group has been to find ways to form radicals within protein active sites expanding the types of reactions available to enzymes but then we're going to use the tools of molecular biology and directed evolution to optimize those proteins to hopefully solve some interesting synthetic challenges and so we've looked at this in the context of a few different types of proteins the one that I want to tell you about today are flavin dependent ene reductases and so these are proteins that reduce activated alkenes with really high levels of selectivity ene reductases are some members of that family that are referred to as old yellow enzymes because these were the first isolated and characterized flavo proteins you can buy ene reductases from your favorite enzyme vendors which is of course attractive if you are a synthetic organic chemist that doesn't have experience working with proteins you can buy a commercial kit and think about recapitulating that reactivity without actually having to learn how to do protein expression now the peeling features of this protein family is that they have a very broad substrate scopes they're thought to be over expressed in response to oxidative stress and so they're sort of general reductases and so you can reduce activated alkenes in a variety of different chemical environments with a variety of different electron activating groups now the native mechanism does not involve radicals it's actually just a simple hydride transfer mechanism we tend to simplify enzyme active sites to their key components and so in this case and in ene reductase we sort of view there to be four key components the flavin cofactor which is in its reduced hydroquinone oxidation state two residues that bind orient and activate the substrate and then a conserved tyrosine and so what happens is the substrate binds it's rendered more electrophilic at which point the hydride on on flavin can be donated to the electrophilic beta position forming an enzyme bound enolate that will be can protonated by conserved tyrosine to provide the reduced product and now flavin in its oxidized oxidation state at which point the product can leave an equivalent of NADPH will bind it can donate its hydride to flavin reducing it back to the hydroquinone and priming the enzyme for another catalyst turnover and so in this system the enzyme moves back and forth between the hydroquinone and quinone oxidation states just using a hydride transfer mechanism now I think one of the defining features of of flavin is that it does have access to a third oxidation state this is referred to as the semiquinone and to a synthetic organic chemist that doesn't work with with proteins you might look at the semiquinone and say this is a sort of fleetingly stable intermediate it's known to readily undergo electron transfer events to get to more stable flavin oxidation states but I think this is really the the power of enzymes right proteins have scaffolds that are able to stabilize the semiquinone either thermodynamically or kinetically and so what this means is that you can actually have a long lived semiquinone within a protein environment this is actually essential to the function of reductase domains for for metalloproteins for instance and so the question we had in our group was could we use the flavin hydroquinone as a single electron reductant to form an organic radical within a protein active site and then we could take advantage of the weak NH bond on the flavin semiquinone this bond right here has a bond strength of about 59 kcals per mol this could function as a hydrogen atom source to a variety of different organic radicals to provide the quinone and so if you're familiar with radical chemistry one of the sort of classic reagents used in in organic synthesis is tributyltin hydride this would essentially allow us to do tributyltin hydride within protein environments and allow us to do a lot of that chemistry asymmetrically and so we started with a hydrodehalogenation I don't really want to talk about that instead what I'd rather talk about is some of the work we did looking at this radical cyclization of an alpha halo ketone and so this is a reaction that had been known for for decades was challenging to render asymmetric using traditional approaches and so we imagine that we might be able to use in in reductase to do this and so you know the the the first thing that I want to draw your attention to because this is a point that's going to come back a few times throughout the talk is the reduction potential of the substrate it requires minus 1.78 volts to reduce the substrate to form the organic radical and so what that meant as we looked at the system initially is that the initial electron transfer to to use flavin to reduce the substrate was going to be uphill by over half a volt and so we recognize that this could be a challenge but we didn't entirely understand how the protein would be able to overcome that thermodynamic barrier and so we screened at this point we maybe only had 12 different wild type in reductases in house we found that that the in reductase NCR was able to do the reaction in about 20 percent yield albeit as a 1.71 ratio of the desired psycholized and undesired reduced product and in 7921 er in synthetic organic chemistry typically we're going to be shooting for higher than 955 er and so we engaged in a protein engineering campaign our approach to protein engineering is going to look fairly similar regardless of of the project and so I can just talk you through it right now this is your in reductase active site there's your flavin cofactor there's the two residues that are thought to bind and and activate the substrate and the canonical mechanism we choose not to mutate those because they're also really important for nicotinamide binding which we need to still be able to bind to reduce the flavin cofactor and so we don't mutate those residues but we mutate everything else within five to six angstroms which accounts for approximately 20 residues and so we'll conduct site saturation mutagenesis identify residues that have an impact on the reaction and repeat the process iteratively until we achieve selectivities and activities that we're looking for and so for this particular reaction we found that four rounds of directed evolution resulted in a variant that could provide product in 955 er with 92 yield for the desired psycholized product with less than one percent of the undesired reduced product and so synthetically we had achieved our goal but I think one of the things that had sort of um sat in the back of our heads that we couldn't really wrap our minds around was the increase in activity across the engineering campaign you know in theory we haven't really changed the the redox properties of the substrate we knew in our characterization that we hadn't rendered flavin more reducing throughout the protein engineering campaign and so it wasn't really clear how we were able to to sort of increase the activity for this particular reaction and so it's this sort of thought that was sitting in the back of our heads as we started to develop new reactions and so one of the other reactions that we were looking at uh at the same time was actually the radical cyclization of these alpha halo amides now these alpha halo amides are about a volt harder to reduce than the alpha halo ketones and so when we did our initial screen with a number of different in reductases what we found is that none of them were able to do this reaction in the dark uh you would just get unreacted starting material out at the end it wasn't consumed in any way and so we at this point we were looking for a way of making flavin more reducing and and we actually took a page from from nature nature has a really beautiful photo enzyme called DNA photoliase this is an enzyme that is flavin dependent and uses light to repair cyclobutane lesions in DNA that are induced by UV light irradiation and so mechanistically how that that protein works is is fairly complex I can sort of simplify it to one key step which is where you you photo excite the flavin cofactor to access an excited state that's more reducing by about two volts and so the question we had was could we adopt that mechanism with an ene reductase to access a more reducing excited state to be able to drive this challenging uh electron transfer event and it turns out you can if you radiate the system with 497 nanometer LEDs you can now get the cyclization to occur in 89 yield with 94 6 er and so we were of course from a synthetic perspective really excited to have this really great result but we were sort of puzzled by the wavelength dependence of of this reaction if you know anything about the absorption spectra of the flavin hydroquinone you'll notice that it does not absorb at 497 nanometers it's actually a better absorber around 360 nanometers and so it was clear to us that our initial hypothesis that we were just going to borrow the mechanism from DNA photo lice within an ene reductase was likely not operative and so we ran some UV viz experiments and one of the things we found is that when you take the reduced enzyme and you add the substrate you see this new absorption feature here in orange with a maximum right around 500 nanometers and so this could be a charge transfer complex alternatively it could be some impurity in our starting material that was changing the flavin oxidation state we wanted to rule that out and so what we did is added a competitive binder in this case sodium benzoate the hypothesis being that if this were a charge transfer complex sodium benzoate would bind in the active site preferentially to our substrate it would kick out the substrate and if it's a ct complex that absorption feature should decrease and that's exactly what we see and so this was really sort of our I think first example with flavin that these systems are able to form charge transfer complexes and so what happens is you have your substrate and your flavin cofactor when you mix them together in the presence of the protein you form this charge transfer complex and so I can give you just sort of a brief aside as to how charge transfer complexes work and this will sort of inform what what I talk about next and so in the context of synthetic organic chemistry when we talk about charge transfer complexes usually we're referring to them as electron donor acceptor complexes and so these are molecular aggregates that'll form in solution between electron donors and electron acceptors essentially what happens is you have an electrostatic attraction between the two molecules that force them to aggregate in solution that aggregate is actually a new molecular species with different homos and lumos and so what happens is that when you photo excite them essentially what you're doing is you're promoting an electron from the donor molecule to the acceptor molecule in the aggregate when you do that you form a radical ion pair and then that ion pair can go off and do interesting chemistry and so in the context of our system we think of the flavin hydroquinone as being the donor our substrate as the acceptor when they bind within the protein active site they form this charge transfer complex and when you photo excite that you're able to promote an electron onto the substrate and form the flavin semiquinone and so I should say that the protein is an absolutely essential component of these charge transfer complexes if you leave the protein out and you just have reduced flavin and the substrate there isn't a strong enough electrostatic attraction between the two to form the charge transfer complex and so the protein is absolutely essential for this electron transfer now what we didn't appreciate at the time was how important it is for the electron transfer now now the other part that I want to dive into and I'll be honest this is a lot of physical chemistry and it's probably well beyond what I actually fully understand but we understand it in sort of qualitative terms which helps us understand how this is working so uh Mulliken had described charge transfer complexes using this very simple equation so the wavelength of the absorption of the charge transfer complex is equal to the ionization potential of the electron donor the electron affinity of the electron acceptor and then this coulombic term which essentially describes the electrostatic attraction between the two molecules and so we can we can actually sort of measure a lot of these terms if you think about the ionization potential of the donor that actually relates pretty closely to the oxidation potential of your electron donor the electron affinity can be related to the reduction potential of your electron acceptor and then the coulombic term which is this electrostatic term can actually be simplified to the distance between the donor molecule and the acceptor molecule and so this is the equation that we have sort of thinking or floating around in the back of our heads as we started to do some more protein engineering on these photo enzymatic systems and so one of the things we wanted to do early on was increase the photon efficiency of this particular cyclization and so we developed a protein engineering platform to be able to evolve a photo enzyme uh up to this point no one had actually evolved a photo enzyme because there aren't that many photo enzymes in nature there's only three and so what we developed was a system where you have these 96 well led arrays you run the reactions in white micro tighter plates with clear bottoms it's really important that they're white if they're black they'll absorb a lot of light and then they become very warm and they actually melt and warp uh so you want to use white ones uh we cover them with clear plastic uh covers and then what you don't see on either side are a a sea of usb fans that blow over the top of it to try and keep the temperature below 40 degrees if you do that you get consistent results across the plate and can do protein engineering and so uh these are our modified conditions the wild type enzyme gives us 19 percent yield 91 90 our quantum efficiency of 2.4 percent uh and a lot of our systems early on we found t-36a was the best variant for this reaction uh I'll be completely honest that was an accidental introduction into the protein uh we got our initial hit with that protein realized it was a mistake removed it and it turns out that it had a beneficial effect on the reaction and so we sort of think of it as a an unofficial first round of of directed evolution and so you'll notice that the t-36a variant gives higher yield higher in anti-oslactivity and higher quantum efficiency and then at this point we did one round of air prone mutagenesis tested about a thousand variants and found this triple mutant that we refer to as g6 that has two mutations on the surface one within the protein active site that gives us product in 90 percent yield 973er and now we're getting close to 11 percent quantum efficiency and so we've continued the protein engineering campaign on the system but in truth we have actually shifted towards exciting with red light in this engineering campaign rather than blue light and I'll get into that in just a second and so having found a more effective variant one of the things we wanted to do was understand other changes that we see in the protein we didn't see large changes in the absorption of the charge transfer complex and so we collaborated with Greg Scholes's group at Princeton to try and understand if there were any differences in the radical lifetime and so the way that we gauge radical lifetime is using transient absorption spectroscopy and essentially the spectral signal that we look for in that system is is the flavin semi-quinone because if we have flavin semi-quinone it means that we've got radical character on the substrate and so what we found is that when you look at the wild type enzyme the radical lifetime is about 700 picoseconds if we look at our evolved gluer g6 variant we see that that radical lifetime decreases to less than five picoseconds and so what this suggested to us is that we're shifting from a mechanism that maybe looks a little bit more stepwise in nature to one that's a bit more concerted in nature and so we ran the requisite control experiments and one of the things we found is that it's really important to have the alkene in the substrate and so the importance of the alkene coupled with the the results from the transient absorption spectroscopy what we believe is going on is within the protein active site the electrons in the pi system of the alkene actually interact with the carbon chlorine sigma star in a hyper conjugative interaction and so the protein is actually forcing that pi system and that antibonding orbital to interact with one another and so you get something that looks like this in the radical anion form and so what that means is that those pi electrons can actually help to break the carbon chlorine bond in doing so though you also get carbon carbon bond formation in that transition state and so we think that what we're evolving for is sort of a concerted asynchronous transition state that looks something like this where we actually don't have a lot of free radical character you have sort of a transient radical that that quickly helps stitch everything up to form the product and so that was exciting and really unexpected and with that sort of hypothesis in mind that we went back and started to think about the impact that that those types of interactions would have on the charge transfer equation and so I've sort of simplified it here we're able to quantify the the redox properties of the Flavin cofactor that would be this ionization potential and we know that that doesn't change our hypothesis and this has now been supported with some computations is that the distance between the donor and the acceptor molecule is decreasing over the course of the protein engineering campaign and so we're forcing the donor and the acceptor to pack more tightly with one another and so we think that we're decreasing that distance and then we're thinking about you know if we have this hyper conjugative interaction what impact does that actually have on the electron affinity of the acceptor and I think the way to think about this is actually just to look at the redox properties of alpha halo carbonyl compounds so when you're looking at at the reduction potential of say an alpha halo carbonyl to form an alpha aceratical and the anion that potential does not track with electronegativity of that x group it actually tracks with the stability of the anion and so alpha bromo amids are much easier to reduce than say alpha fluoroamids because of the stability of that anion and so we started to think about this in the context of of a chemistry where you have a more concerted transition state in the case of say gluer t36a if we don't think there's a strong hyper conjugative interaction then you'd have this sort of radical alpha to the carbonyl it can obviously delocalize into the pi system but there aren't sort of other stabilizing interactions if you were to have this hyper conjugative interaction the radical that you form would actually be delocalized over three carbons and so it's actually a more stable radical than what you would form if you don't have that hyper conjugative interaction and so we think that when you have this more ordered transition state where you're packing everything more tightly together you're changing the distance and that also changes the electron affinity of the acceptor and so it's difficult to sort of deconvolute those two terms in the charge transfer equation okay so we've looked at a number of different cyclizations i don't want to go through all of them in great detail what we've done recently is actually moved to intermolecular systems and and this sort of came about prior to our better understanding of how these systems work and one of the the main questions we had was you know if the alkene isn't an essential component of the charge transfer complex we would expect that you would form this hydrodehalogenated product because you wouldn't have both substrates bound within the protein active site you know nonetheless we set out and developed the reaction found that you could take one equivalent of the alpha-chloramide 2.5 equivalents of of alpha-methyl styrene with an enzyme it's actually the same enzyme as before in cyan light you can get the coupled product in 97 percent yield 98 to er in our initial screen we found that no stock er actually favored the opposite in anti-emer but the enantioselectivities were more modest we ran some control experiments and and found that there was a tyrosine within the protein active site that was actually competing hydrogen atom source and so we mutated that out it was the tyrosine at position 219 and now we're able to access the other enantiomer in high yield and antioselectivity and so we're really happy to have both enantiomers but i think the thing that again jumped out to us is the fact that you see less than 5 percent of the hydrodehalogenated product because what this implies is that there's a control mechanism in the electron transfer event and again you can see it in the UV viz this is the absorption specter of the flavin hydroquinone when you add the alpha-chloramide you see this weakly absorbing CT complex and then when you add the alpha-methyl styrene you now see a much more strongly absorbing charge transfer complex and so what we think this is evidence for is the fact that if you don't have the alkene this is sort of a weakly reactive poorly absorbing CT complex it's only when the alkene is present that you form this more delocalized system that's a better sort of electron acceptor for the electron density in the flavin cofactor this forms a better absorbing CT complex accounting for our observed reactivity we've prepared some docking models this is one that sort of recapitulates what we see in the UV viz yeah i think our revised models now probably think that's a slightly different docking confirmation nonetheless it does seem to work and so we've looked at a number of different substrates maybe i'll just conclude by talking about this ketone so when we ran this reaction one of the things we noticed is that the reaction immediately turned yellow we went back and ran the controls and we found that this reaction occurs with outlight and so the question we had was if you were to change the protein could you actually change the degree of ground state reactivity in the system and it turns out you can gluer t36a only gives about 3% yield if you move to ncr you now get 49% yield if you look at the consumption of the alpha halo carbonyl compound with and without alpha methyl styrene what you find is that the the styrene actually accelerates the consumption of the alkyl halide and so again what we think is going on is the presence of the alkene is actually changing the distance between the donor and the acceptor and then changing the electron affinity and so one of the other sort of effects of having a decreased donor and acceptor distance is the fact that when you look at the ground state charge the ground state wave function which should be a component of the neutral and charge transfer form it looks as though when they pack more tightly you get more ground state charge transfer and so maybe I'll just conclude by saying that I think you know all of the chemistry whether it's ground state electron transfer or excited state electron transfer all involve the same mechanism it's all a charge transfer complex it's really just a question as to how much charge transfer there is in that ground state wave function if there's a lot then you can drive the reaction in the dark if there's not then you have to use light and the protein is absolutely essential for dictating how much charge transfer there is and so I had other examples but I talked too slowly essentially what you have is an opportunity to develop a bunch of chemistry using light to drive the electron transfer and then if you wanted to run these reactions on scale you would evolve the protein cause the CT complex to pack more tightly and then you could drive all those electron transfers in the dark which would be better if you wanted to scale up this type of chemistry and so I think there's one unified mechanism and you're just in different continuums of that mechanism based on the protein that you're using and so with that maybe I will conclude just acknowledgement slides and just say that I think there's there's lots of opportunities to use light with proteins I think that charge transfer complexes and bio catalytic reactions are far more prevalent than we would think and so I think there's lots of opportunities to take advantage of those to do really new and interesting chemistry and so with that I just want to conclude and thank my group for all their incredible efforts thank my funding sources and I'd love to take any questions you might have From more UV to more visible I was just curious do you see that your proteins are more stable and are able to do more turnovers as you're using more redshifted light Yeah so I think what we've what we've found is that as we move to red which is the wavelength that we're currently looking at the reactor isn't quite as warm and so the primary mechanism of decomposition is actually thermal decomposition and so what we now see is background non-photon mediated alkylations of cysteines and lysines in the protein and so we're currently going through an additional protein engineering campaign to figure out which of those happen first knock them out and then I think we're going to be able to really drop our enzyme loadings at that point and so I think we're right on the cusp of having a system that would have the metrics that would be sort of more similar to what an industrial bio-catalytic process would look like and so hopefully we'll be able to talk about that in more detail in maybe six months Yeah yeah great talk I had a question about that T36A mutant it's so far away from the active site based off of your picture do you think it's just controlling sort of like maybe this extended polarity network in the enzyme that just if you shut that one off then you induce more charge transfer in your active site Yeah no it's a really good question we've got crystal structures of both of them if you overlay them they look almost identical and so our thought was oh you know it's there's some change in the dynamics of the protein that's causing it something that was really interesting is we started collaborating with Andy Ellington who has this really interesting machine learning approach for identifying residues at the periphery that should influence reactivity and one of the residues that they found was T36A but they predicted a different mutation we went back and mutated to it and within our group we sort of refer to those as the super gluars because they actually work a lot better and have higher activity and so we don't entirely understand how it's affecting it but it looks like there's a couple different ways of being able to identify the importance of that residue Awesome and then I had one more question and I think you alluded to it with Empica but since you're using this these radicals in your protein have you seen any post-trans like different modifications as a result of that of the protein like an oxidation event that wouldn't occur in the wild type protein or anything like that so on reactions that are initiated reductively we don't really see radical alkylation of residues within the protein there is sort of this ionic substitution based alkylation strategy or mechanism but that doesn't seem to be operative for you know modifying tyrosines or tryptophan which you might expect from a radical mechanism what we have seen is that if the protein doesn't have sort of an organized transition state what looks to happen is that you'll alkylate the flavin cofactor once you alkylate the flavin cofactor within ene reductases then its reactivity stops awesome thank you yep thank you