The Climate Biotech Podcast

Yeast Feast: Transforming Taste through Protein Design with Anum Glasgow

• Homeworld Collective • Season 1 • Episode 9

Share episode

How can protein science shape the future of food and climate solutions? Meet Anum Glasgow, a leading researcher at Columbia University, whose journey—rooted in her Pakistani heritage and a childhood of curiosity on the Jersey Shore—led her to the cutting edge of protein engineering.

Anum shares how her fascination with protein folding evolved into groundbreaking research on designing multifunctional proteins and therapeutics. We explore the hidden elegance of nature’s self-folding systems and how they inspire innovations in climate biotech.

We also dive into the work of Team Yeast Feast, a group pioneering a sustainable approach to flavor. By engineering proteins that enhance sweetness and umami naturally, they’re rethinking how we experience taste—tackling food sustainability with creativity and science. From blind taste tests to leveraging AlphaFold, their work blends playfulness with real-world impact.

Finally, we look at how hydrogen exchange mass spectrometry is unlocking the secrets of protein structure and its role in taste perception. What makes some sugar substitutes fall flat? How do protein conformations shape flavor? We connect the dots between sequence, structure, and sensory experience.

Tune in for a closer look at the Glasgow Lab’s work and the latest in climate biotech, brought to you by the Homeworld Collective.



(00:00) Introduction to the Climate Biotech Podcast
(00:32) Guest Introduction: Anum Glasgow
(02:03) Anum's Background and Journey into Science
(02:40) The Fascination with Protein Folding
(05:35) From Physical to Biological Folding
(08:05) Computational and Experimental Approaches in Protein Research
(13:12) Exploring Taste Perception and Climate Solutions
(14:43) Engineering Sweet and Umami Yeast
(26:06) Technical Insights: Hydrogen Exchange Mass Spectrometry
(33:30) Rapid Fire Questions and Conclusion

Send us a text

[00:00:00] Daniel Goodwin: Welcome to the climate biotech podcast, where we explore the most important problems at the intersection of climate and biology, and most importantly, how we can solve them. I'm Dan Goodwin, a technologist who spent years transitioning from software and neuroscience to a career in climate biotechnology.

As your host, I will interview our sector's most creative voices from scientists and entrepreneurs to policymakers and investors. 

Hello everybody. We're thrilled to welcome Anum Glasgow for discussion about climate biotech. I want to give a few words briefly about Anum's background before we jump straight into it.

We know Anum from working with her through the Garden Grants program, where she's been working on really cool approach engineering challenges. I speak for the home world team. We've had several interactions with Anum over the past year, and she's just one of those people where as soon as you leave the conversation, you want to go back and have another one.

 I've been personally really excited to have this conversation today. Anum received her PhD at Berkeley by the way, I went to Stanford, but I think we can still be friends where she engineered bacterial secretion systems. And Dr. Danielle, Tolman Arshak's group. She later pivoted to computational biophysics as a postdoc in Tanya Kortem's lab at UCSF, where she achieved the first example of a de novo designed, chemically inducible dimerization system.

She also co led a team to rapidly develop a receptor trap therapeutic for treating COVID 19 during the pandemic. Currently, Anum is an assistant professor at the Department of Biochemistry and Molecular Biophysics at Columbia University.

In a research group, Anum leverages experimental and computational approaches to probe how protein systems evolve, undergo conformational changes, and interact with other biomolecules. Anum's long term research goals include engineering multifunctional proteins and therapeutics that respond to binding molecules and correct localization. There's a lot there. I can't wait to talk about it. Thank And so I think on like there's a lot we're going to explore today, but it's always just so much fun to start off as humans and to jump right into it. So on um, welcome, who are you and where did you grow up?

[00:02:07] Anum Glasgow: Yeah, it's awesome to be here with you. Thanks for inviting me. And who am I? I'm a protein scientist. I'm from the Jersey Shore. My parents are Pakistani immigrants, so I'm lucky to have that culture also. And yeah, I didn't plan to be a scientist while I was growing up, but I really like it. I'm kind of a dreamer.

So doing science professionally for me is a way to contribute to society, but also have fun.

[00:02:34] Daniel Goodwin: Wow. I had no idea we were missing the opportunity to make Jersey shore jokes for the past year. So you've said before that proteins are nature's most beautiful self holding systems. And I want to hover on that you as a kid, first thing, did you ever imagine yourself saying this?

Did you see yourself becoming a scientist? And when did you, when was that? If there was a phase change, when was it? 

[00:02:53] Anum Glasgow: Yeah, like I, I really didn't honestly, I think like people have formative stories. I looked at bugs under a microscope or turned rocks over those types of things. No, I was like way into reading, like often my imagination. And in college I studied engineering because. I thought it would be, useful, and, I actually found my way to proteins and protein science in a really roundabout way.

It was from doing undergraduate research. So, like, I didn't know anything about doing academic research, but as a college freshman, my roommate joined a lab and I, she's cooler than me. I followed her there. It was Dr. David Garcia's lab, nothing to do with biology, actually. At the time, the Garcia's lab had just developed a method to build micro scale 3d boxes from two dimensional shapes made of metal.

These are like lithographically patterned micro polyhedra, like how you make computer chips. And at the time people were really interested in making them smaller and patterning holes, like pores in the boxes for precision drug delivery. These types of applications. So like on my first day in the lab, I just saw in a microscope how these like tiny flat arrangements of six metal squares could just be heated up and then it melts the solder in between the faces and they fold up into boxes.

And that was like magic for me, right? Like it was like watching a dozen origami cranes just like pulled up from paper spontaneously. But it wasn't magic, right? It was surface tension forces. And over my four years in the lab, I was totally drawn to the thermodynamics, right? So we did practical things as well, but it was the folding process that really encapsulated, I guess my imagination.

So for me, the best project was we worked on figuring out how there are 11, two dimensional nets that will fold up into 3d cubes. And we figured out which one does the best job. And then we applied those principles to build like way more complex micro polyhedra that have tens of thousands of nets.

That is like an impossible task to test them all individually. And so, , systematically discovering the rules from this choice system was really key. And so, I know this is a really weird way to get into proteins, but I love thinking about how order arises from disorder, right? Like, against all odds.

[00:05:14] Daniel Goodwin: Absolutely. It helps me understand you so well because I feel as if you always see things from a different vantage point. And what's so fun is I think you're the only person who works on protein folding who started from a physical folding perspective. That is definitely the most unique angle I've

heard yet. 

[00:05:30] Anum Glasgow: it wasn't like medicine or anything like that, no.

[00:05:32] Daniel Goodwin: Folding boxes. Okay, great. And so what did you go from the like a physical folding to a biological folding?

[00:05:41] Anum Glasgow: Yeah, so proteins are just beautiful, right? Like, They're much smaller than the system. I just told you about their essential molecules that drive biological processes and they use the hydrophobic effect to fold up. And just because of their huge heterogeneity and their essential nature. I cannot resist them.

[00:06:04] Daniel Goodwin: You can't resist Proteins. That's the soundbite. But where does this fit in your professional growth? Was this PhD student years when you started thinking about finding the ears of stability of proteins?

[00:06:16] Anum Glasgow: I would say so like as a graduate student I worked on a really specific protein assembly. It was a secretion system that bacteria used for pathogenesis. We were repurposing it to secrete biopolymers that are hard to make and other ways for applications and biotech, you know, like spider silk you've probably heard of is strong material.

But I was getting really frustrated at that time with the evolution approach. It's a really good and popular way to make proteins behave differently than how they were naturally evolved to. But we never know what the effect of the mutations that we evolve in our system like why those come to be, right?

Like, it's not that rational of a process. So that drove me in my postdoc. I had one goal, which was to learn how to do computational protein design. In a more rational way, right? Like de novo protein design. that brought me to where we are here. Yeah.

[00:07:09] Daniel Goodwin: So is it fair to say the PhD years was the brute forcing of a secretion system and then the postdoc was trying to form some rational search space?

[00:07:20] Anum Glasgow: Yeah, pretty much like I think my phd projects gave me a really deep appreciation for biological systems I think working through these naturally evolved, super difficult protein systems as a PhD student gave me like a good basis and I'm now starting to build things from scratch.

[00:07:40] Daniel Goodwin: So this is really interesting because once again I love these conversations because I get to learn more about your vector into science and you've worked on many different aspects of biology, but computation from the papers I've seen has always been this through line. But in conversation it feels kinda like thermodynamics was the through line.

So I'm kinda curious about what's unique about the way you and your lab study protein structure and function through the lens of computation?

[00:08:05] Anum Glasgow: Yeah. We combine computational and experimental approaches. Our approach is really rooted in the central problem in protein structure, function relationships, which is how are the functionally important. Confirmational changes in proteins encoded in their amino acid sequences. In other words, how do proteins do their dance?

It's a really fascinating problem because we talked about this a little bit, but like, in order to fold properly, proteins have to be super stable. In general, we optimize for stability when we design new proteins, but at the same time, to do anything interesting, proteins cannot just sit there in their folded state.

They also have to move. So that means that they have hard coded instabilities. It means that evolution has to balance stability with function. And the proof of this is in front of us, right? Like natural proteins. They're only marginally stable at like 8 K cals per mole on average, but they do all of biology and fascinatingly, they do this within the structural constraints of only a few thousand shapes, 

what we call protein folds. And as a field we have. No idea how to design that. So it's up to us and my lab takes this approach of finding patterns and how evolution repeatedly adapts these ancient topologies for an absolutely dizzying array of biological functions and really different environments from the depths of your liver to the ice and Antarctica.

[00:09:29] Daniel Goodwin: Do you work on like ice nucleation proteins or ice finding sites.

[00:09:33] Anum Glasgow: It's on our list. No, I don't. There's this is like, you know, like aggregation is really hot right now. And there's a whole body of work on like how proteins come together to increase their global stability. But we haven't really touched that stuff.

[00:09:48] Daniel Goodwin: Yeah. I've only, it's one of those like wild papers at 3am and Oh, I want to learn more about that. I can't help it. I have to ask you the annoying questions, which is that somebody might listen to this conversation and they say, Oh, we're talking about how do the individual residues on a protein change things in the folds and they get the lazy or like knee jerk answer a lot of people have is Oh, didn't alpha fold already do this?

I think it is really important just to help people like unpack this, like kind of space of where we are today with proteins. Why there's so much important work to be done.

Obviously I've got my own theses, but I'd really love to just hear your take on, this personal listening, hearing you say. We need to figure out what these residues are doing in space, solving folds. Why are we not done?

[00:10:30] Anum Glasgow: Yeah, great question. I'm so glad you asked this question. There is um, so much more to protein behavior than the low energy conformational state and its structure, right? So we are always, we're like the cult of the conformational ensemble here. We're focused on trying to understand that missing link between the protein.

sequence and structure and function we call the protein conformational ensemble. So when I say this, I'm referring to the distribution of conformational states that a protein adopts at equilibrium. And why is this important? Because the protein function depends on how that conformational ensemble changes in response to all perturbations.

 So, for example, a protein might exist in conformation A at 95 percent of the time and conformation B at 5 percent of the time, but when it binds another molecule, for example, confirmation B might become energetically favorable and then is enriched to say, 30%. And this is what unlocks the new function in the protein.

But the obstacle here is that it's really difficult to measure conformational ensembles. You are by standard structural biology techniques. Very likely to see only conformational a whether or not the binding partner is there in my example, just because it's the lowest energy state in both cases, even though confirmation fee is absolutely functionally critical, right?

And even if you could see both confirmations, they might look exactly the same as each other, despite being energetically distinct. So if we really want to know how the protein works, we want to know not just what confirmations A and B look like or C or D. Or their populations.

But we also want to know how every single amino acid in the protein contributes energetically to each confirmation, because if we can see it, then we can design it and then we can diagnose diseases where like there are perturbations such as mutations, you know, so we are, very actively working on these problems, new ways to measure protein conformational ensembles, combining classic biochemistry methods with our own new mass spec and machine learning methods.

and trying to learn the rules for how proteins encode their important conformational states.

[00:12:42] Daniel Goodwin: Okay, so I think this is going to become more tangible when we talk about specific proteins, and then specific experimental techniques. It is worth kind of making the segue through the fact that Homeworld is thrilled to support your work through our garden grants program.

It's been so much fun for us to see the wild problems, and the work that you've produced and all the wonderful people in your lab. I'm starting there and then we can go into a recent paper you guys just put out. And so when the Homeworld does our garden grants program, we do it by talking about problems and then solutions.

And so I'd love to just, put you on the spot to tell the world about what problem you've been working on specifically in the context of climate. And then we can start exploring the solution, which I think leads to a recent paper you just published. So what problem are you trying to solve in climate?

Silence.

[00:13:28] Anum Glasgow: Yeah, like, have you ever wondered how you can taste the difference between a strawberry and a Skittle? We have hundreds of olfactory receptors that help us smell, but we have only a few taste receptors. And our taste cells encode just one protein complex for sensing sweetness, which is really efficient, but it's totally unclear how this works.

The reason this is important to climate change is because so much of what we eat makes demands on the environment, right? Farming animals for meat contributes to 15 percent of global greenhouse gas emissions. And beyond that, sugar overconsumption is a really big problem nutritionally for Americans.

And then, of course, we have global problems of hunger and food instability. And so we're working on trying to solve both of these problems, the scientific problem and the pragmatic problem. We hypothesize that taste perception works like the following. So if you have different compounds in food, they bind your sweet receptor, your one sweet receptor, and they enrich different conformational states.

And that gives rise to the unique biological signaling. So that signaling in the taste cell controls your downstream synapse with your neurons, and that evokes feelings connected with specific flavors. So to bring it together, we're testing this hypothesis by engineering yeast, whose surfaces are decorated with proteins that we've designed to bind the sweet receptor.

And we're hoping that this is going to bias that receptor to occupy hundreds of unique conformational ensembles. And then. We can figure out which proteins that we've designed that are on the yeast bind where in the receptor to activate sweet perception and the umami receptor is structurally really similar to the sweet receptor.

So we're repeating the same process there.

[00:15:16] Daniel Goodwin: And the umami flavor is one that I know the food and beverage industry has been trying to solve for a while. I like the way you're starting with sugar and thinking about more, subtly complex flavors. The name of your project that you proposed was Team Yeast Feast which is hilarious and I think the Homeworld team still laughs about it.

But I think since it started with this funny name, it's why it's always been fun for us. Wait, there's a lot there. Wait, what? There's hyper sweet proteins? What?

And so, um, Yeah. So what are the proteins that you've been engineering in this context? I've been learning a lot this whole time.

[00:15:47] Anum Glasgow: yeah. So these proteins are like de novo designed little topologically diverse scaffolds that we try to tune the affinity for our model of the umami and sweet receptors. And catch them at different epitopes like different parts of the receptor where it'll lock them in different confirmations.

That's our hypothesis. And we have, you know, Alka Fold and other tools to try and predict whether that's the case, especially because an experimentally solved structure of these sweet receptors and umami receptors is not actually available. Yeah, we're hoping that by catching the right receptor in a specific way we can augment the natural flavor of yeast, which is rather nutty and cheesy.

I don't know if you've ever eaten nutritional yeast, but it 

[00:16:31] Daniel Goodwin: a lot of nutritional yeast in our family. Yeah. 

[00:16:34] Anum Glasgow: And I should point out actually that yeast are not just a convenient vehicle for our design proteins. Nutritional yeast are an FDA approved food that is enjoyed by vegans, like our lab members that inspired this project. And, you know, lactose intolerant cheese lovers like myself beer enthusiasts, the health conscious among us.

One and a half tablespoons of yeast has 60 calories. It has 3 grams of fiber. It has 8 grams of protein. All the essential amino acids. Yeast can be eaten raw, cooked, wet, dehydrated. The only problem with yeast as food is that they only taste like yeast.

[00:17:08] Daniel Goodwin: Wow. I actually am embarrassed that it's taken me a year to see that you could be making sweet beer. I don't even drink, but like actually sounds like pretty tasty and So you do use alpha fold in the flow though, right? Is to say to come up with these. Kind of initial candidates, which is, the whole thing is there's a playfulness to this, which I really love, which is that you're taking the same pipeline that people use to make drugs and you're doing it to make something that tastes good, that yeast can produce, right?

That was the workflow I understood, right? Is that you're checking, you're generating scaffolds, testing the fold with alpha fold, and then putting that on yeast.

[00:17:50] Anum Glasgow: Yeah, we actually use a really similar pipeline for developing alternatives to antibodies for like antiviral therapeutics applications. So it's cool that you see that.

[00:18:03] Daniel Goodwin: Yeah. Okay. The funny question is with bio, it's always about the assay. And so what's the assay for we've made sweet yeast.

[00:18:12] Anum Glasgow: Yeah. So you know, before we had more quantitative assays in place, my team would actually set up a blind taste test for me and my husband for their top strains. With many positive and negative controls and We would just eat them. Sometimes you just want to know, does this project have legs

 You want to fail fast as they say. So it was so important for us to do that because it knocks out the biggest question mark in this project at an early stage, is this protein that we designed going to display densely enough to overcome the background flavor of yeast? If not the project, is just going to be dead in the water, right?

So in this taste test, my husband and I would separately sample the yeast from Eppendorf tubes that had labels that we didn't know and our team would write down like whether we found it sweet or not, or like what other comments we made. These food science papers are new for us.

They pasteurize the yeast, this turns out to taste universally bad. They tried dehydrating it like how you would buy nutritional yeast at the store also just like spinning it down into pellets, which requires a lot of yeast. But, you know, personally, I prefer the texture of wet yeast pellets. 

[00:19:23] Daniel Goodwin: I gotta be honest. That sounds so gross. I thought you were good. I thought you were going to dehydrate it. And then you're just dipping your finger in an Eppendorf tube and just

tasting it. 

[00:19:32] Anum Glasgow: I'm not totally uncivilized. We use like a straw, like a spoon.

[00:19:37] Daniel Goodwin: Amazing. Okay. So then the question has to be, can you taste sweeter yeast?

[00:19:43] Anum Glasgow: Yeah. So as bizarre as it is to eat your experiment, I think that was the best thing that we discovered is that project kind of works. We started off with positive controls, which are. Proteins that, for whatever reason, come from West African plants that are super sweet. We know about seven of these that there's enough structural information for them to do something with.

Manellin, brazine, circulin, you may have heard of some of these proteins, but they are tricky little proteins. It's really difficult to make them recombinantly in bacteria, which is usually what we do. yeast actually have no problem. We found out even though they have like a dozen disulfide bonds, they just happily make these proteins and display them on the surface.

So yeah, the answer to your question is that with these positive controls, we found out that we can totally sense the difference between proteins. or yeast strains that have these proteins displayed on their surface and yeast strains that don't, or yeast strains that have, you know, control proteins displayed.

It was also really interesting from these early stage experiments where we just, went straight to eating the product. I didn't want our students to, Do this weird experiment because, you know, just in case, like it was like maybe a little bit unsafe, but like my husband and I were always taste testing the sample separately, and it was really interesting to see how to untrained taste testers experience mostly the same responses to these yeast. So like there was one time where we separately described the same yeast strain as tasting bright. With, like, neither of us really being able to say much more about it. And then, like, now that we actually do have quantitative assays, it's really validating to see that the strains that we perceive as sweet actually do activate calcium signaling the most compared to all of the other strains.

[00:21:31] Daniel Goodwin: Amazing. Okay. There's three questions. I'm trying to not soup them up too much, but the first one is I want to make sure your positive control is to show that your de novo things can be a positive control of something that, is already going to be hyper sweet. 

And got it. And then the dehydrated ACE actually seems really interesting because like lyophilized proteins for therapeutics, et cetera, seems to be really important.

I don't know how well that's understood about how these things operate in dehydrated states. So am I understanding right that in the current status, the dehydration step destroys the signal and so we're having to move forward? Or does the dehydration also work?

[00:22:08] Anum Glasgow: the dehydration also works. We were able to make the same sweet assessments and dehydrated samples. It's just like sort of a stodgier texture.

So most of our experiments now that we have like more quantitative assays in place, we do a simple dilution. We dilute the yeast to the same O. D. For all of our samples and then add them to hex cells actually that are heterologously expressing the taste receptors and their membranes.

[00:22:37] Daniel Goodwin: Wow. Okay. That was exactly what that, that third question was going to be. So that's your quantitative assay is that you're measuring, are you doing like calcium signaling imaging just to see,

[00:22:46] Anum Glasgow: Yeah, absolutely. We have a few assays in place. I like to have several orthogonal measurements for everything whenever possible.

 That's our best assay actually is a microscopy assay of all things. We are pretty new to microscopy, but it is beautiful to watch the cells light up. when calcium is released from the ER after we add the right E strings.

[00:23:07] Daniel Goodwin: Amazing. So you're using the latest GCaMP and then through, wow. Okay. That's okay. I'm totally tickled. This is great. A lot of my, a lot of my PhD was working with GCaMP and microscopy. So it, I did not know this was a part of your research. So that gets me really excited. Okay.

That's super cool. There's two things here now, which is that I think there's also been some technical methods development underneath this to do the development of these candidates. And I think that's recently came up with a paper. And then, um, before we go into that paper on this idea of the sweet yeast, Do you think that's work that might come out?

Is that is that a publication? Is that kind of a funny demonstration video? Where might people be able to follow this story?

[00:23:47] Anum Glasgow: Yeah, I mean, ideally, we wrap up this project at some point and publish it. I think the people need the sweet and umami yeast, right? So we plan to finish up this project, hopefully in the next year, at least as a demonstration that you can use yeast in this way as food alternatives, 

A panel of yeast that can be used for food, but also as useful tools to understand sensory perception.

[00:24:12] Daniel Goodwin: Yeah, I think that's a great point. One of the technical questions about this is how big are these hyper sweet proteins? Are they tiny little peptides? Are they big bulky things?

[00:24:22] Anum Glasgow: Ours are really tiny. And the naturally evolved ones that we know about are also on the small side. When I say that, I mean, on the order of like 20 killed Altons or less, some of them dimerize, which doubles the size, but you know I think that this is like actually a really deep question, who's to say that we can't build more giant proteins to bind to the sweet receptors and cause them to be confirmationally arrested or trapped.

In a particular conformational state, if we follow through with our hypothesis. And I wonder if it's not about necessarily enriching a particular conformational state or forbidding another one and more about just like changing the, or like sampling different parts of the conformational landscape. It's like a subtle difference, but you know, there's no real requirement for high affinities here.

Like there are in some other projects where you might be creating. protein based binders to targets, right? This is not like a therapeutics development. When you eat something, you're saturating your taste receptors. It's not like they're expressed at high concentrations. So there is something else going on here.

And the fact that you can do it with tiny, tiny proteins that likely have four K offs, right? Like operates. Yeah, I think there's more to the story than just like the structure changes. Yeah.

[00:25:38] Daniel Goodwin: I totally agree. I'm going to throttle myself. Otherwise I'm going to ask a thousand questions just on that kernel right there to think about the kind of like the technical work that was underneath this. And hopefully one day people will get to read the paper or go try their own yeast.

Cause immediately you think this can work in yeast. Can this work in rice? Like I'm thinking of is a lot of. Big if true potentials here. But I think it's worth exploring the technical side too. And so it's another one of those cases where oh, aren't we done?

Don't we just have all the data we need to do this with any proteins? And I know that, in the process of this your lab has been working on some really cool stuff and you just published a paper technical title, site resolved, energetic information from HXMS experiments. And I think that's really interesting cause I do think that it's just generally useful for people to understand where this data comes from when we talk about different conformational states.

or confirmation ensembles, excuse me what is the ground truth for that? And what is the gold standard of the ground truth for that? And I think there's a lot of stuff that this is exactly what you work on. And so we just love to hear a little bit about this paper and this link to the projects and also where it could go.

[00:26:42] Anum Glasgow: So this paper that we just put out, it's a way to get the site resolved ensemble energies from a protein. What I mean by that is if you have a protein in your confirmation A versus your confirmation B, And you wanted to know like the energetic contribution of every residue in the protein to each of those conformational ensembles.

This is a way that you can get that information from doing hydrogen exchange mass spectrometry. It's a technique where you drop your protein into a heavy water buffer. And you just let it undergo normal thermal fluctuations and over time, the protein backbone is going to fill with deuterium atoms, right?

Just from the heavy water buffer. The rate at which this happens depends on its local structural context. To solve an exchange phenomenon. And so typically from this experiment, you get a peptide level information. You down the line break up the protein using proteases and then you do a liquid chromatography mass spectrometry.

You separate out the peptides and you can identify them based on their mass over charge ratio. So at different time points in this experiment, 30 seconds versus three minutes versus three hours, you can resolve a peptide rate. For the incorporation of deuterium. So what our method does is it allows you to get site resolved information.

So like single amino acid level information, it's a powerful method because you can get absolute energies from this unbiased energetic information, and it allows you to compare the behavior of, for example. not just like a wild type protein and a mutated protein or like a apro protein and a liganded protein, but even proteins that might have really similar structures, but completely different sequences like orthologs, or for example, designed proteins, 

so you get an idea of like, for example, if a protein family that is extremely ancient has evolved and diverged and you see one type of behavior and this organism and another type of behavior. in this phylogenetically distinct organism what's been evolutionarily maintained energetically such that it is able to perform the same function in each chemical context, right?

So, sorry, this is maybe too much.

[00:28:58] Daniel Goodwin: Oh, no, this is perfect. The best case scenario is somebody listening to this and they think that's really cool. What the heck did she just say? I need to Google this. 

[00:29:05] Anum Glasgow: Yeah, so like, to make it like a little bit more concrete

 In the context of the project that we were just talking about our homeworlds project we would love to know, for example, if we test 100 proteins, half of them cause some kind of taste. But for example, you know, aspartame does not taste good.

Whereas sugar tastes amazing. Why is it that sugar substitutes don't always taste good, right? Like, this is the type of information that can help us finally resolve an answer to these types of questions. Do we see the participation of one substructure in our protein in a particular way to the global behavior of the protein, the conformational ensemble?

Is there a signature behavior? that we can group with this sequence and not with this other set of sequences, so it allows us to finally make a connection between the sequence and the structure of a protein with its ensemble towards understanding the molecular basis for its function.

[00:30:03] Daniel Goodwin: Got it. So to think about this kind of, once again, like level one, it's that you put your protein. Into an environment where there's going to be an exchange rate between heavy water and just what's naturally in , each amino acid to begin with. And from the output of this after treatment, you get a score for each amino acid in the polypeptide.

And that gives you some score on the surface, like the surface availability of 

that protein 

[00:30:27] Anum Glasgow: Yeah, so Common misconception hydrogen exchange experiments don't exactly measure a solvent accessibility. They really measure the participation of the backbone amide group to structure defining hydrogen bonds. Like, that's how I would describe it. 

let's say you have an alpha helix in your protein. It looks like it's totally solvent exposed. It won't necessarily exchange at a fast rate because alpha helices can sometimes be so incredibly strong. It can take even like a year in some really strong folds for exchange to happen at a particular site.

You know, because like, not only does that hydrogen bond have to open up in order for the exchange to occur, something like 10, 000, water bombardment events have to happen in order for that exchange to happen. So it's quite a large dynamic range in terms of time. For solvent exchange to occur in the first place, which is why it's a good measure of local stability.

So you're right about the score part. Like we get a readout. That's basically an ensemble energy. We call it a Delta G of opening. People have been measuring this using a hydrogen exchange NMR for a long time, but NMR. Um, is difficult and tedious and only works on very tiny proteins for a fraction of the exchangeable amide groups, right?

And so our method is, you know, mass spectrometry is much higher throughput and with our method, you can get a lot more information a lot faster. And now at the same level of resolution and quantification.

[00:32:00] Daniel Goodwin: That's really cool. Not a lot of people say mass spec is easy

but I think it's increasingly. So there is a mass spec and, most great research labs. So I think there's something really exciting there. And so the last technical question I have there is, if you want to really get the full confirmation ensemble, do you have to put the proteins mixed in with the small molecules that might change the confirmations?

Or can you just infer the whole ensemble just from a single shot?

[00:32:26] Anum Glasgow: Yeah, um, good question. So um, the timescale of this experiment is like Maybe on the shortest scale, milliseconds, but typically seconds to minutes to hours, right? So these are slow dynamics. So what happens is if you want to know what's the effect of binding a partner in my protein, you would set up two experiments, one with a partner and one without.

And then sometimes if you're really lucky, you might see what we call bimodal distributions in your mass spectra. This is when you actually have two conformational states present at the peptide resolution. So like one region of your protein is actually occupying two separate conformational states that are not interchanging on the time scale of the experiment.

And when this happens, I'm always like really fascinated by it because

[00:33:13] Daniel Goodwin: . 

[00:33:13] Anum Glasgow: you're seeing evidence of different conformational ensembles coexisting.

[00:33:17] Daniel Goodwin: That's awesome. Okay. I've got a thousand other questions, but this paper's online and it's called the Pigeon Feather Technique, right? And so people can find it and I strongly encourage people to do so if you don't mind we're going to just wrap up with four quick rapid fire questions.

And we're just gonna jump right in. So one what's a single book, paper, art piece or idea that blew your mind and shaped your development as a scientist?

[00:33:39] Anum Glasgow: Have you ever been to the Whitney Museum? I'm actually super inspired by brutalist architecture in New York. I think it's a testament to the beauty of a reductionist approach, the kind that we use in our science. Yeah, I just love how you can, see the most essential form for a function and how these, huge concrete buildings, they take up the same amount of space as a baroque building, right, in an old city.

And just like a lot of great ideas, they look really simple, but they're designed in a really thoughtful way and they're heavy and serious. And I think that's beautiful.

[00:34:13] Daniel Goodwin: If you like brutalist architecture, I am thrilled to know that we have a reason to try to pull you up to Boston. We have, whoever built this city loves brutalism. Question two, what's the best advice line that a mentor gave you?

[00:34:25] Anum Glasgow: I think a lot about Jim Wells Punnett square approach for choosing projects. I don't know if you've heard of it, but he talks about yeah, like try to do things that are both innovative and pragmatic. 

[00:34:36] Daniel Goodwin: Underscoring pragmatic. It's really easy to be ambitious. It's hard to land ambitious with pragmatism. I don't think that's a great advice line. Number three, if you had a magic wand to get more attention or resources into one part of biology, what would it be?

[00:34:51] Anum Glasgow: You know, like with recent advances in machine learning, I think there is now incredible potential for methods development for measuring biomolecules so that we can produce more types of data to accurately describe biological systems. Obviously, I'm biased, but I'm talking about analytical chemistry and integrative structural biology.

It's really hard. These days to find money for hardware. And this is work that is at the same time, not very flashy, but also very expensive. And I believe people need to be financially comfortable to do creative science. So, you know, I think that's why breakthroughs are really slow in this area. To come through, even though ideas abound.

[00:35:31] Daniel Goodwin: Full agreement. All right. Fourth question. Lots of developing biotechnologists listened to this podcast. What was something important, say a trend or foundational skill on biotech that you think a junior researcher might overlook?

[00:35:49] Anum Glasgow: So my postdoc mentor, Tanya Cortema, she always encouraged us to really carefully evaluate the structures that are computational protein design methods produced and in this age of computational metrics, I think sometimes people forget to just look with their eyeballs at the data. So I think metrics are great, but that's the thing that might get overlooked, right?

Your chemical intuition is sometimes really powerful.

[00:36:15] Daniel Goodwin: That's an amazing one. I actually lost three months of research about a decade ago where the answer was just look at why your algorithm's failing. Just look at the data. So fantastic. Anum, this has just been so much fun. I really appreciate both the crash course in Hydrogen Exchange while also just getting a chance to hear about your life and what brings you into doing the awesome work that you do.

If people want to find you or your lab or learn more, where can people go?

[00:36:41] Anum Glasgow: They can go to GlasgowLab. org, which is our lab website, and they can email me. 

[00:36:47] Daniel Goodwin: Awesome. Thank you so much, Anum. And we can't wait to talk to you more soon and we're wishing you the best of luck.

[00:36:52] Anum Glasgow: thank you so much.

[00:36:53] Daniel Goodwin: Thank you so much for tuning into this episode of the climate biotech podcast. We hope this has been educational, inspirational, and fun for you as you navigate your own journey and bring the best of biotech into planetary scale solutions, we'll be back with another one soon.

And in the meantime, stay in touch with homeworld on LinkedIn, Twitter, or blue sky. Links are all in the show notes. Huge thanks to our producer, Dave Clark, and operations lead Paul Himmelstein for making these episodes happen.

Catch you on the next one.