Scientist Stories: Luke Lavis, Building Molecular Tools for Imaging

Our second speaker is Luke Lavis. Luke received his bachelor's in chemistry at Oregon State doing synthetic organic chemistry with James White. He then spent 4 years in the biotech industry, first at Molecular Probes in Eugene, Oregon, and later at Molecular Devices in Sunnyvale, California. Luke received his PhD in organic chemistry with Ronald Raines at the University of Wisconsin-Madison, where he designed and built novel fluorescent dyes to image molecular trafficking in living cells. Just out of grad school, Luke accepted a postdoc— Luke accepted a group leader position at Janelia in 2008 and became a senior group leader and head of molecular tools and imaging in 2017. Luke's lab works at the interface of chemistry and biology, developing labels for single molecule imaging, strategies for targeted molecular delivery, and probes to measure cellular activity in intact brain tissue. His lab has developed a set of Janilia Fluor dyes, which are being used in labs around the world, and which he will tell us more about in his talk today.

Okay, so let me start again. So, um, I'll define tool building this way, creating instruments, techniques, reagents, and datasets and computational approaches to push the frontiers of biological research. And as I mentioned, we have this core research area at Janelia, molecular tools and imaging. And so we have optical physicists, we have protein engineers and cell and molecular biologists, and we have chemists. And so this is a group of scientists who don't necessarily focus on biology, but they're at a biological institute. So how does this work? There's a couple key tenets that, that we— that guide this tool-building philosophy at Janelia. First of all, independence. We're all independent experts in our own fields. We formulate and test hypotheses in the areas of science that we work in. But of course, we're always looking for some biological application. Another tenet is collaboration. We work with other tool builders and biologists to solve problems, and Eric gave a great example with the Voltron story that involved several different labs ranging from chemistry, protein engineering, microscopy, and systems neuroscience in flies, fish, and mice. We're patient. We like to say we want to generate industry-level tools, bulletproof tools. And Harold and Eric Bedzik and myself and Abraham all spent time in industry before coming to Janelia. And so we recognize the effort that it takes to transform a prototype tool into something that is truly useful. And then finally, dissemination. We want to freely share tools with everyone. So this idea of dissemination is really hardwired into the DNA of Janelia, and that's because of our founding director, Jerry Rubin. This is Jerry in his current form, but Jerry has a decades-long commitment to distributing the tools that he's developed over his career to the broader community. So I want to tell you a story about an earlier version of Jerry. This is Jerry in graduate school. This is actually too early. I just wanted to show this picture because I think it's cool. But the story I want to tell you about is sort of '80s/'90s Jerry, and this was the best picture I could find on the internet. Sorry, Jerry. But Jerry and his collaborator Alan Spralding figured out a way to introduce transgenes into flies. This was reported in this Science paper from 1982. And the first public presentation of this work took place in a Drosophila research conference at the University of Connecticut earlier that year. And Jerry and Alan showed up not just with slides detailing their science, but also 300 kits with bacteria and a detailed protocol so other people could use their methods. So one thing Jerry always says is the ultimate test for a tool builder is where others— whether or not others will use our tools. And so this really set the tone. And of course, we're continuing that with Ron Vale, our new director. But Ron's only been here a couple months, so I'm not quite brave enough to search the internet and show old pictures of Ron. All right, so we have all these tool builders developing a variety of different tools. And of course, the different tools require different dissemination strategies. So for genetically encoded tools, protein-based tools, Eric already shared the way that we disseminate this technology. We created the GENIE project team. They take the prototype tools that are developed in these labs and then using high-throughput screening approaches They optimize them and test them in different model organisms. And then ultimately they distribute them through Addgene. So they deposit plasmids on Addgene and then that organization takes care of the dissemination. For microscopes, it's a little bit different. Although we'd love to send a microscope and an expert physicist to run it, to every single lab in the world. We simply can't do that. So instead, the idea is to send scientists to microscopes. And so for this, we created the Advanced Imaging Center, or AIC. This is a collaboration with the Moore Foundation, and it houses instruments developed here at Janelia by folks like these guys and then makes them freely available to the community. And Liang, the director of the AIC, will give a talk in this seminar series later and tell you about this exciting effort. For small molecule-based tools, unlike proteins, we don't necessarily need high-throughput screening to optimize them. We can do that in our own lab. But we do need help testing these in cells and animals and then actually making the material and physically sending it out. There is no Agene for chemicals, and so it's up to us to distribute these to the broader community. So I'm going to talk a little bit more about some of the small molecule tools that my lab has developed. So I've been working in this field of small molecule fluorescent dyes for 20 years now, and in this chart I made back in graduate school, I show 30 commonly used fluorescent molecules in biology. They're plotted according to their brightness against the absorption maximum of the dye. That's what each one of these points denote. And then the corresponding structures are colored according to the emission maximum of the dye. Now, the point of this slide is not to go through every structure, but it's just to showcase that we have a large number of fluorophores that span the UV, the visible, and the near-infrared. And as chemists, we can take these molecules and tune and tweak them for specific biological applications. And so my lab has focused on these classic fluorophores, fluorescein and the rhodamines, and we like these dyes because they're near the top in terms of brightness and span a good chunk of the visible region of the spectrum. Now, one thing about these dyes is that they're fairly old. Fluorescein was synthesized first in 1871 in the lab of Edouard von Bayer. And the rhodamine dyes pop up in the patent literature 16 years after that, synthesized by the Swiss chemist Marie Serrasol, who was working at BASF. And so the age of these dyes is something of a double-edged sword. On one hand, the fact that we're still using fluorescenes and rhodomines, these classic fluorophores, in modern biological experiments is a real testament to their properties. We've had 100 years to find better fluorophores, and we simply haven't. But from an organic chemistry standpoint, the age of these dyes is, is a problem. It means that the methods used to synthesize these molecules are often quite old. And so when I started my lab at Janelia, one idea we had was to bring modern chemistry to bear on some of these old scaffolds. And the general idea was if we came up with new chemistry, then perhaps we could make new versions of these dyes with improved properties, and that could ultimately enable new biological experiments. So here was the original classic chemistry. You would take aminophenols and phthalocyanhydrides and literally boil them in sulfuric acid for a day to generate rhodamine dyes. So this severely limited the type of functionality that we could install in these fluorophores, and to basic methyl groups and other very simple substituents. So instead, we used a modern chemical reaction, in this case a palladium-catalyzed cross-coupling reaction developed by Steve Buchwald at MIT, and using this we could transform simple fluorescein derivatives into rhodamines. These conditions are much more mild, and it allowed us to test different substituents on these rhodamines to see if we could make better dyes. So, here's one example where we took this classic fluorophore tetramethylrhodamine containing these dimethylamino groups and transformed it into this dye we call Janilia Fluor 549, or JF549, based on its absorption maximum in water. And we installed these little 4-membered azetidine rings. This is just a net addition of 2 carbon atoms But what we did is greatly improve the brightness of the fluorophore. So the quantum yield of the fluorophore went up from 41% to 88%. So this new dye enabled new experiments. Together with my colleagues in Molecular Tools and Imaging, James Liu and Eric Betzig, along with Bob Tijan, we attached this janelia-4549 using the HaloTag that Eric mentioned earlier. To a transcription factor, Sox2, and then could image the stable binding sites of this transcription factor in live cells. And we found it formed these little so-called enhancer clusters. So this was the beginning of a long journey developing a palette of small molecule fluorescent dyes called the Geneelia Fluorodyes. So, we started with this 560, 550 nanometer excited fluorophore JF549. We found this azetidine trick was general. We could apply it to far-red fluorophores, like silicon rhodamines, to give JF646. And then we figured out ways to fine-tune the spectral and chemical properties of the dyes by using substituted azetidines. And so now we have a whole palette, and we recently expanded this palette a little bit further into the near-infrared and also a little bit farther into the blue. Now, this dye JF525 was, to be honest, just kind of put in here to fill out the spectrum. But Eric mentioned this briefly, this ended up being a fantastic dye for in vivo use with the Voltron system. It was bioavailable and the spectrum matched the opsin used in Voltron. So we have all these dyes in different colors, but more importantly, we have efficient chemistry to make them, so we can synthesize them in reasonable quantities. And that raised the possibility maybe we could start giving these things away for free to everyone. And so we started doing that. In the early days, it was just John Grimm, who's a senior scientist in the lab, the co-inventor of these gemmellia floridae, and John would make the compounds and aliquot them, and I would field all the emails and put things in bags, and we would send stuff out. So after a while, this started taking up more and more of our time, so a couple days a week, and we realized we needed help. So John got help from a fantastic series of technicians over the years Natalie is unfortunately leaving us in a few weeks to start graduate school. These two guys are already in graduate school. And we also got a lot of help from the head of this tool translation team, T3 Tim Brown, administrative help from Anastasia, and then IT help from Megan Beamish, creating essentially an e-commerce portal so people can log in, request stuff, and it keeps track of our inventory and basically keeps us quite organized. So, uh, together over the last 5 years, we've given away about 12,000 aliquots of our dyes to well over 1,000 labs. And if you think about the list price of some of these fluorophores from companies, it represents about $50 million worth of stuff that we've given away in the last few years. So we were shut down over the pandemic, but we're back. So John just tweeted this a couple weeks ago. He revamped our freezer containing now thousands of vials of dye, and we're ready to go. So if anyone wants to try things out, you just need to email me, we'll send you a link, and we'll send you some stuff. All right, so I'll just end with a little bit of science. So what's interesting about these dyes, in addition to being extremely bright, is that some of these fluorophores are fluorogenic, meaning when they bind their cognate biomolecular target, they turn on. So here's an example. This is JF635, the HaloTag ligand. This is this little chloroalkane piece that binds the HaloTag protein. And when this binds, it goes from this colorless form, which predominates in water as the free ligand, to a highly colored fluorescent form. We get about a 2 order of magnitude change in absorbance and fluorescence. And so this is useful if you just want to dump some dye in your sample. You'll only see the stuff that's bound. But Eric and I had this idea, perhaps we could exploit this change in optical properties and use it for sensing. So we knew that this change in properties was due to the protein environment, and we also did some careful spectroscopy and found that this form on the protein is only about 70% of the material, 30% is still in this colorless form. And so there's an equilibrium on the surface of the protein that we might be able to exploit. So Eric helped us solve the crystal structure of the HaloTag bound to a ligand, and then that allowed Eric to design rationally proteins with sensor domains flanking the dye, and then changes in conformation of these sensor domains could change the environment around the dye, causing an increase in fluorescence. And so Eric mentioned this GCaMP sensor, so we took cues from that and made something we call HaloCaMP. This is a circularly permuted halo tag with the calmodulin motif that Eric mentioned and a calmodulin-binding peptide. When calcium binds, you get a change in the environment around the fluorophore. It turns on. So this works in vitro. It works in cells. These are neurons being electrically stimulated. And one cool thing about this is we can optimize these sensors not just by high-throughput screening mutagenesis of the protein, but we can also change the dye. And so my lab is busy making new variants of this JF635 with interesting properties, and Eric's lab is busy making different protein variants, and we're screening combinations. And finding indicators with specific properties. All right, so that's the brief story. It's been really a pleasure to work with fantastic scientists in my lab, other labs here at Janelia, and outside. So thank you very much. I'm happy to take any questions.

Thank you, Luke. That was a great talk. Let's get right into some questions. First question is, you know, it's— you're doing a lot of dye production and dissemination right now. And is there any thought of outsourcing the production of the Junilia fluor dyes?

Yeah, I mean, my, you know, my dream is to have an ad gene for chemicals. There are many challenges with that. But I think it would be useful because there are tools that would be beneficial to the biological community, but don't have quite the margins or the volume to make them a viable product for a company. So I think there's a sweet spot there where an ad chem type thing could fill. We do partner with companies, so in some cases, Some of our stuff can be viable, so a few companies do sell some of the Janelia flora. So when it makes sense, we do partner with companies.

Great. So the next question is a little bit more programmatic. Can you tell us, the molecular tools and imaging program at Janelia, is this mainly driven by the biology going on at Janelia, or is it more self-driven, or is it some combination? Can you speak a little bit to that?

Yeah, I think, you know, when I tell people I'm a chemist at a place that does mostly biology, I think the assumption is, oh, you run a shared core for chemistry. And the idea is a little bit different. Now, you know, we don't necessarily have to collaborate with anyone at Janelia. But we're here, it would make no sense to just be here and do our own little chemistry thing. So it's sort of, I guess, a bottom-up thing. We're all here, we are all experts in our own field, as I mentioned. But because we're in this environment, immersed in this environment, we get access to problems that other folks may not realize are biological problems. And so we can direct our program to address these problems while still spending our days and nights, in my case, doing chemistry.

Right. So somewhat related to that, I'm curious to know, you have a lot of backgrounds, people from different backgrounds in molecular tools and imaging, But do you— are you seeking, are you looking for tool builders with expertise in areas that are not already filled in your group with as far as like the scope design, fluorescent sensors? Are you looking for additional expertise?

Yeah, we're always— I think we're always looking for other people with expertise in other areas. For example, We recently hired Abraham, who has a more material science background. It might seem odd for a biology institute, but it works. I think we would love to have a diverse array of scientific disciplines at Janelia. And it takes a particular type of scientist, someone who wants to do their own thing, but has an eye for biological applications. But it could be— beyond our optical physics, protein engineering, chemistry that we have right now.

Let's move on to a question from Jacek Kolonowski. Jacek has a question about the HaloTag. Go ahead, Jacek.

Hello, thanks a lot. So I just wanted to ask you if there is a reason why you prepare your fluorophores as a substrate for HaloTag only? I know you showed the structure of the protein and said that this matches there, but wouldn't you expect also the same fluorogenicity behaviors with other protein tags if you attach, if you attach a fluorophore to them? Yeah, we like the HaloTag for a couple of different reasons. So the HaloTag was developed by scientists at Promega, and they did a really great job with the directed evolution. So we like the HaloTag because the labeling kinetics are quite fast, and unlike some of the other tags, there's a very tight association of the dye with the surface of the protein. And so across the board with these fluorogenic molecules, we often see a higher degree of fluorogenicity with the HaloTag than some of the other things like the SnapTag. And our thought is that with some of the other tagging systems, the dye is sort of flapping out in the wind and it doesn't have as tight association, so that change in environment is not so dramatic. So you don't get these large increases in absorption and fluorescence. So we like it for a couple different reasons. It's extremely fast, which is important in vivo, and the tight association magnifies this fluorogenic effect.

All right, thanks for that question, Yasek. Let's move on to a question from Eric Zhao. Eric, are you there? Would you like to ask your question? Yeah, yeah, go ahead.

Okay, so, so, you know, quantum dots are well known for its bright— for its, you know, pretty impressive brightness and, and the you know, resistance to the photobleaching. So are these organic fluorescence dyes, you know, better than the quantum dots? No, quantum dots are way better. But quantum dots are way bigger. So here we're, you know, the state of the art is fluorescent proteins. It's a little clunkier to have to use a small molecule dye, but as Eric mentioned, we can introduce it either IV or IP, so it doesn't scare biologists too much. Going with quantum dots, I think in vivo is much more challenging. They're large, you have to coat them to make them biocompatible, and then trying to direct them to specific cells is difficult. So we think, This is, again, a sweet spot where we get modestly improvements in brightness and photostability, but still compatible with the in vivo experiments, which are a mainstay at Janelia.

All right. Thanks for that great question. I think I want to close with a question that came up during Eric's talk, but can really be directed to either one of you. You guys both made it look very easy, the development of these sensors and dyes, but we all know that's not the case. Can you just tell us a little bit about some of the challenges and obstacles that you faced during the course of developing these sensors?

Yeah, I think, so, you know, when I showed that tweet from John with the freezer, So one thing we had to do is clean out a freezer to fill it up with a bunch of vials to send out. And I was going through basically a decade's worth of stuff. And one thing that struck me as I was throwing away a lot of aliquots is that about half of the aliquots were from things that we had made that ultimately really didn't work. So, you know, in all of these presentations, you always show the stuff that works, but there's a lot of trial and error. And, you know, especially with chemistry, you have an idea, you have to make the thing to test your idea. And it's sometimes disappointing because you spend all this time doing all this synthesis, maybe the synthesis is clever, and then ultimately, you know, you try this thing out in a cell or an animal, and it simply doesn't work. So there is a lot of pain along the way. There's a lot of failed ideas to get to the ones that work.

Yeah, I can imagine that's the case. I don't know if Eric might have a comment on that also. Eric, did you, you know, are there any sort of obstacles or challenges that you want to share with us, you know, quickly in the last minute or so?

Sure, I would say— can you hear me okay?

Yeah, we can.

Okay, great. I would say similar to what Luke said, most of what we make doesn't work, right? We don't have— actually, we do have freezers full of things that didn't work. They're not in the form of aliquots of chemicals, but they're in the form of glycerol stocks of E. coli expressing different different sensors that weren't useful. As I mentioned, we end up sort of combining strategies of rational design and screening to most efficiently get at things that hopefully are useful. But yeah, the vast majority of tweaks that we make or new types of things we make by combining different protein domains end up breaking the function of one or both of those protein domains and are essentially useless. So yeah, the main challenge is a lot of trial and error, and there's no guarantee that even if you have a solid idea for a thing that would be useful conceptually, that you'll actually be able to make it, or that something exists out there in nature that's close enough to to give you a starting point.

Got it. All right. I think that's all we have time for today. I just want to close by once again thanking our speakers, Eric Schreider and Luke Labus. We really enjoyed your talks today. Those of you who still have questions that you would like to have answered, Eric and Luke have promised me that they will head over to Slack after the seminar so you can ask all the questions you like over there, and I put a link to Slack in the chat box. I also once again want to thank our audience for joining us today, and I also want to call out next week's seminar. We will be hearing from Cold Spring Harbor Laboratory next Wednesday, and the speakers will be Zachary Lippmann and Rebecca Leishan. So please join us next Wednesday, 9 AM Eastern, at the same link, and we look forward to seeing you then. Thanks everyone.