Episode#16 Proximity Labeling Technologies (BioID) - Natalie Harris

Dr. Barkha Yadav-Samudrala: 0:06

Hello, everyone to another episode of nerd RX podcast, and I am your host, Barkha. Today we are going to talk about a new technique, a relatively new technique called proximity labeling tech technology. And to talk more about that we have Natalie Harris, welcome, Natalie to the show.

Natalie Harris: 0:27

Hi, Barkha. Thanks for having me here.

Dr. Barkha Yadav-Samudrala: 0:30

It's my pleasure. So before we jump right into the topic, why don't you introduce yourself so that we know more about you?

Natalie Harris: 0:40

Yes. So, as you said, my name is Natalie Harris, and I am currently a PhD student at the University of North Carolina Chapel Hill. And my research mostly focuses on lymphatic biology and molecular regulators of how these vessels function. And I have a background in biochemistry and molecular biology, which is relevant for today's talk considering talking about proximity labeling technologies in the cells.

Dr. Barkha Yadav-Samudrala: 1:10

And thank you so much for that introduction. So talking about proximity labeling, I have not come across it. So this is something I'm going to learn with everyone else. So what is proximity labeling? And why what got you interested in it?

Natalie Harris: 1:30

Yeah, so kind of the overview of proximity labeling is kind of exactly how it sounds is that you're wanting to identify, usually, in this case, proteins that are nearby something usually another protein. So the kind of idea behind proximity labeling is a little bit newer in the sense that we're trying to look for things that are around one another. So in the past, a lot of biochemistry has been direct interaction. So like two proteins that are physically kind of touching each other, or within a super small radius. So this proximity labeling is little bit different in that it enables you to identify things that are in the vicinity. So kind of things that are further away. So not necessarily proteins that are directly touching one another, but two that might be in the same kind of compartment within a cell, maybe traveling together as proteins are made and traveled to the plasma membrane. So it kind of gives you more of like an idea of the whole area. So I like to use this analogy, kind of like going fishing. So in the past, direct labeling technologies are like kind of like fishing with a fishing rod and hook. So you're kind of getting one protein at a time. And you kind of kind of immediately identify that fish. So the proximity labeling technology is kind of more like casting a net. So you're kind of seeing everything that was around that area that you're fishing in, then you kind of have to one by one, see what type of fish are in the net. So that's kind of the most basic overview of how this proximity labeling technology works.

Dr. Barkha Yadav-Samudrala: 2:56

Awesome. That was a very nice analogy. Like I could picture that in my head. So why is proximity labeling important? And what are some of the applications you would use for?

Natalie Harris: 3:11

Yeah, so proximity labeling is really helpful in the sense that when you think of protein function, so proteins that you know, they generally traffic to wherever they're supposed to go in the cell, and exert a certain function. But we don't always know like how they get from point A to point B, or when they're at their like location, who else they're interacting with, and whether that's in like a quick manner, like at a certain time, or even within a certain space. So the proximity labeling technology allows you to see what is around your protein at a certain time and space. So for my research, I'm looking at an adherence junction protein, which is a protein that helps cells kind of adhere together like the name says. And in lymphatics, specifically, the protein that we're interested in, maintains these two different confirmations, one where the cells are more permeable relative to one another, so big things can pass between the cells and, and the different conformation where they're more tightly bound together, so things can't pass between cells really easily. But what's interesting is from what we know is that it's the same proteins that make up this junction. So clearly, something else must be going on around the protein or on the protein to make it happen in these two different confirmations. So that's why we're applying this proximity labeling technology to kind of see what the neighbors are more or less than this protein.

Dr. Barkha Yadav-Samudrala: 4:37

So let's talk about the steps involved. Is this could you just give us like a rundown from how would you go on setting this experiment and the steps involved and how long does an experiment usually last?

Natalie Harris: 4:53

Yeah, so the exact name of the technology we've been using is called Bio ID. So it's biotin dip. pendent proximity identification. So essentially, the way that this works is you have a protein of interest, and you fuse it to this promiscuous biotin ligase that we call bio ID. And there's been a lot of different iterations of the bio ID that have come out that are faster, smaller, better. But more or less, you put this promiscuous biotin ligase attached to the end of your proteins that involves kind of cloning, pretty classically, where you would design a construct that has this biotin ligase fuse to the either front end or back end of your protein. And then once you have this construct, and you transfected or transduce it into cells, you give the cells a certain amount of biotin. And then what happens is this promiscuous biotin ligase, kind of takes this by it and eats it up and tags, proteins that are nearby. So basically anything that was nearby, this protein that had the bio ID on it becomes biotinylated. And what's unique about this technology is that it's not just by attenuating, everything in the cell, it's by titillating everything that's near your protein. So then now that you have this cell that has these certain set of biotinylated proteins, you can actually use an affinity pulldown using strep Avidin, which binds very highly and specifically to buy it and Okay, so essentially, what you're doing is you're pulling down these biotinylated proteins very specifically. And now that you have this set of biotinylated proteins, you can send it off for mass spectrometry to actually identify the proteins. So it's kind of like a wrangling the cattle that you've branded more or less, so you take the your cattle that you know which who they are, then you put them in a pen, pull them down, and then send them off for identification. So essentially, it allows you to identify at the protein level, which proteins were near your bait. Okay, and so kind of like timewise, it takes, however long it takes to do cloning, which for me, unfortunately, took many years just because my particular protein was really bad at cloning. But not to scare anyone off, I made other constructs were a different protein. And they were very quick, it takes about a couple of weeks to get your cloning done. And then the actual experiment itself is just transfecting your transducing yourselves, which takes, you know, a couple of days, and then feeding it by a tin, which is a couple of hours. And then the processing for the pull down is maybe about a day. So then you send it off for mass spec, which can take a while just because of the technology. And usually you have to send it off to a core. So that can take like three weeks. So all in all, you could feasibly do the experiment and have results in your hand and definitely less than six months. Okay, so it's it's something that I think is worth doing, if you're really interested in seeing maybe what is interacting around your protein of interest.

Dr. Barkha Yadav-Samudrala: 7:53

Okay. And this might be a naive question, but is there any cut off to how like the proximity? I mean, how close and how far should the other molecule or the other protein should be to interact?

Natalie Harris: 8:12

Yeah, so in this case, the bio ID enzyme itself, it sends off these really highly reactive intermediates that bind to lysines of proteins that are nearby. And so kind of they diffuse out and bind. So this radius is really limited, because they're so reactive, it ends up being about, I believe, 50 nanometers around your cell. So it is a decent distance, but it's not so much that you're picking up a lot of things that aren't actually nearby your protein.

Dr. Barkha Yadav-Samudrala: 8:42

So it's quite sensitive, like 15 nanometers is quite small.

Natalie Harris: 8:48

Yeah, and it's, it seems small, but some of the other proximity labeling technologies like fret, which is like an energy resonance transfer between two things that are nearby, that's like 10 or five nanometers. And it's you have to know what two pieces you're looking at for interacting. So relatively speaking for proximity, it's a fairly decent amount, but again, it's not so unspecific that it will just tag anything nearby.

Dr. Barkha Yadav-Samudrala: 9:15

Okay. Okay, that's great. And you spoke about bio ID. And just from my basic reading, I read, there are other things as well. So are there any alternative techniques to bio ID that you could use to get same results?

Natalie Harris: 9:36

Yes, so there is, of course, the bio ID they've come out with new generations of the enzyme that are smaller and faster and can increase or decrease that radius of labeling. Then there's a fairly other similar technology. It's called apex. I don't remember what it stands for, but it kind of uses a peroxidase interaction to do something very similar in instead of adding biotin you're out Adding peroxidase to the cells and you're kind of like freezing the interactions in time. And it's again, very like controllable. And you could change the amount of peroxidase to kind of get different labeling as well. And but the both of those techniques are really similar in the sense that you have to design your construct to put it in the cells and treat it in terms of alternatives sometimes, like proximity labeling can be to look at really specific interactions, you know, as well. So really common experiment that's done actually, after you perform the bio ID and get a list of proteins, you want to see if they were direct interactors or not. So there's a technique called duo link PLA, so proximity ligation assays, so that involves using two primary antibodies against two targets, you know, and you put this secondary antibody on top that creates this really highly specific fluorescence signal using almost like PCR on the cells, it has these little transcripts that when they're nearby each other actually create like a product that you can tag fluorescently. So essentially, it ends up looking like little dots on the cell, exactly where your two primary antibodies bound. So that's like a proof of concept that these two things are in proximity to each other. That's a little bit different than bio ID in the sense that you need to know exactly what two proteins are looking for. And you are limited to just looking at two proteins.

Dr. Barkha Yadav-Samudrala: 11:28

Yeah. Okay. That is great. And would you say this technique is user friendly, like a complete noob, like a novice could learn this quickly, or does it require some amount of training.

Natalie Harris: 11:44

So I'll say when I started this project, I had no idea what it was either. So you can do it, I think in terms of the user friendliness, it's pretty simple like in what you're doing, you're creating a construct. And a lot of times these backbones for these constructs are already made. So it's kind of like this circular DNA, it has all the pieces together, and you kind of like cut and paste in your protein. So everything else is ready to go, which on paper should be pretty easy. But in reality, it's usually a lot more finicky than you think it is, you have to keep, keep trying again and again. And again to make sure the perfect construct and have a sequence. So it is user friendly in the sense that the directions and things are very clear, it just kind of depends on your protein, and your familiarity kind of with the process of cloning, how long that takes. And then the other thing is eventually, the the goal is to send your pool of biotinylated proteins to mass spec to be identified. So if you're not a mass spec expert, I'm not. Luckily, we have a core on campus that we pretty much send our samples to and they give us data back. So if you don't have that type of service, or if you're running this mass spec by yourself, it can be really complicated in that sense. But we're really lucky here at UNC that we have a core that pretty much handles the entire mass spec part for us.

Dr. Barkha Yadav-Samudrala: 13:07

Yeah, that is a very helpful thing. Even I think we sent our samples to the core facility. Yeah, I don't think every department can afford a mass spec and let alone run maintain it. It's crazy. So let's talk about the advantages and disadvantages of bio it.

Natalie Harris: 13:32

So in terms of advantages, kind of like I mentioned earlier, you're able to see things that are nearby your protein in a certain space and time. So that's really interesting, when you're thinking of proteins getting made and being traffic to wherever they're supposed to go, you'll actually be able to pick up on things that help traffic your protein, like for example, plasma membrane protein, you'll you will identify things that help traffic get to the plasma membrane, which may be known and may not be known. Or once that proteins at the membrane, you'll be able to see what things are nearby because maybe it traffic's to a really particular space on the plasma membrane. So this will kind of let you see those different things. But the other problem is in terms of like a disadvantage, once you get this list back, you get all of it all together. So you're getting things like ribosomes in terms of, you know, having your protein being translated, and transcribed. So then you also get things that are, you know, like actin, which kind of makes sense because ActOns everywhere, most likely proteins have trafficked along actin somewhere because that's how cargo moves in the cell. But it could be good depending on what protein you're looking at. And then you also will get things like receptors of the membranes, so that can be kind of a disadvantage in the sense that you have this whole giant list all together. So it's hard to say, what's real or not. And then I think one of the biggest disadvantages is that in terms of background with any kind of experiment like this, you'll have some Some sort of background things that are pulled down, right. So pretty much anyone that does like an IP, if you use an antibody to pull down different proteins, there is a certain amount of just nonspecific binding to beads, or things that are maybe nonspecifically biotinylated, just by chance that, you know, we're casting a fishing net around our protein, it could just be at random, some protein happened to be in there, in that net, that doesn't mean anything. So it kind of becomes not necessarily a disadvantage, it becomes a little bit trickier when you're actually starting to validate what was on your list. So it's a disadvantage in the sense that you don't always know what's background or not. But they've kind of come out with certain ways to kind of clean that up. So if you know that you have a membrane protein, and you're not interested in how its traffic there, you can get rid of all sorts of translational material, like I'm not interested in ribosomes, I'm going to manually remove all of them from the list. So it kind of depends on your experiment, depending on what you're looking for your list could be really complicated. Or it can be really simple. If you know what exactly you're trying to parse out. And again, like if you're looking at time, if you're just looking at how things change over time, what's in your list doesn't necessarily matter as much as what individual proteins are changing, like with time. So it's very experiment dependent, but there's definitely ways to get around the disadvantages.

Dr. Barkha Yadav-Samudrala: 16:21

Yeah, I think most of the experiments and especially when you start something new, the toughest question is knowing exactly what you're looking for, in most of the things and eliminating the unnecessary data to make your life simple. So I know you explained all the steps in the world, if I may ask, Which step is the most annoying, or like you took the longest to troubleshoot?

Natalie Harris: 16:55

For me, it was cloning for my construct. But again, that was very specific to my proceeding because it took me ash, I hate to say it for years to get out, cloned. But that didn't happen with other proteins that I cloned. And then kind of the most, I guess, then time consuming, or difficult step is once you finally have that list in your hand, you could have a list of 1000s of proteins, you're like, Okay, what do I do with? Yes, so it can be scary, but also kind of exciting, because you get to at that point, decide, you know, what's my next step? What do I want to focus on? So for some people, that can be really daunting, and it definitely is, but it's also very exciting at the same time.

Dr. Barkha Yadav-Samudrala: 17:35

Okay. And what would you say? Are the costs involved here? Like, if there is a lab, very near to proximity labeling, is it a technique that could easily be set up without breaking the bank?

Natalie Harris: 17:53

Yeah, so I think it also kind of depends on your schools, or your labs ability to access and use a mass spec, whether or not you're operating it yourself or with a core. Generally, cloning is not all that terribly expensive, a lot of these bio ID constructs you can buy off of companies like add gene, or even if you had a little bit more money, you could probably go to a place like vectorbuilder, and say, Hey, build this entire construct, I don't want to do any cloning, you make it and send it to me. So that can be actually pretty cheap, depending on if you're going to do some of the cloning manually. So that's maybe like$100 per construct. And then just some incidental costs for like enzymes, depending on how you're doing your cloning. And then doing the various like pull downs. Could be expensive, because you're talking about strep Avidin beads, which you use for your pull down. So those can be a couple 100 per use. But again, you can run a lot of samples on them. So kind of the bigger cost comes to once you've actually, you know, done the full experiment, and you've pulled down to biotinylated proteins, the mass spec, at least at UNC, our core for about a about 12 samples maybe is like$3,000. So once you get down there can be pretty expensive, because they're, you know, running it on the mass spec, they're also spending their time cleaning up the data. So there's special steps that you have to do for mass spec, regardless of what the experiment was to just clean up the data. So like a big contaminant for all mass spec experiments is keratin because even though we think we're the most sterile in the world, we somehow get skin cells and hair cells into our things. So they do clean up steps like that. And then ultimately, at least at UNC, they actually analyze our data for us. They give us quality scores on how good our different replicates were, how clean they were. And they ultimately also do all the searching and pull out all the proteins that were identified. So in terms of their time, the cost is justified for sure, because they're doing a lot of work on our samples. But it can be really expensive. So if you're trying to look at, like 12 Different countries missions, it's really not going to be cost effective. So that's why we do a lot of optimizing before we get to that point, because you can still perform the whole experiment, do the pull down, and then actually run out your proteins on a gel and use strep Davenant, to see kind of how many proteins relatively speaking, were pulled out. And so like, you can compare two different gels for two different proteins. And they will actually look really different. Like they'll look like a smear, but the smears will look different. So that tells you that what was being biotinylated is actually unique to your protein. And so you can kind of look also for known interactors, like on these blocks as well. So something that you know, should appear in this list, that may be a direct interactor of your protein. So you can see if you're pulling that down, there's a lot of optimizing you can do before going to the mass spec, which is cheap to make it cost effective. So you're only sending the most important sample mass that

Dr. Barkha Yadav-Samudrala: 20:53

yeah, that actually brings down the cost so much, because I was thinking like every experiment you run, you have to send to mass spec to know the results. So that actually is very cost effective.

Natalie Harris: 21:07

Oh, yeah. And that's exactly right. So you can really optimized your experiment before you send it there. And for our work, we've only actually sent things to the mass spec, twice like in the head. Now, six years now, we've actually only sent it all the way to mass spec twice. And another reason that you want to optimize is because if you have all these giant mass spec lists, you can waste so much time processing a list, you don't want to spend time looking at a list that is maybe faulty because something earlier in the experiment is bad. So this is definitely the kind of experiment where it is a lot of trial and error in the beginning. But if you have it set up really nice once you go to Mass Spec, that's probably it. Like if you're pretty happy with the list, like maybe even with one mass spec run, you could be done on the mass spec end. So it does bring the cost down.

Dr. Barkha Yadav-Samudrala: 21:56

Okay, that's, that's awesome. So next question is actually one of my favorite question. Is there any fun fact about proximity labeling in general?

Natalie Harris: 22:11

I think kind of the fun fact about it is because it's so new, people are just starting to use it in all these different areas. So for me, I work in lymphatic biology. And so most of the pipe papers I've read with my version of the bio Id have been plant papers. So it's really interesting to see like people are using this technology, all in all sorts of different fields. So that's been really interesting for me. And then my lab had a semi personal connection to the bio ID technology. This postdoc that I worked with, who kind of helped me start this project and introduced me to the technology, his best friend from undergrad was actually one of the scientists that came up with the bio ID technology. So we could go directly to that he could go directly to the source and be like, Hey, I don't get this explained. So it was really nice for us was really funny, because it's just like a small round. He hadn't talked to him in probably like, I don't know, over 10 years. And then it was really funny, like hearing them on their zoom talk. It was like no time had passed at all. So yeah, as it goes to show you that, you know, you never know, like, especially in science, like all these random people you might meet at like conferences or just know in general can help your project out and you can help each other out. So those are show like science is really a

Dr. Barkha Yadav-Samudrala: 23:31

community. Yes, absolutely. And by doing this podcast, I've realized like, it's such a small world, I am realizing and I have had few guests, I was just talking to them about the issues I was having in my Western Blot regarding antibodies, it's always of antibody. And they suggested me something, I bought a trial size. And it worked. I was working on that antibody for like eight 910 months, and nothing was working. And they just suggested a simple thing from a different company. And it has to be like the best western blot I've ever gotten. So it's really like it's a community. And I think we can only excel by talking and communicating basically.

Natalie Harris: 24:21

Yeah, I agree. And it's also like really nice to like, even look at like other fields too, because I feel like sometimes when you're doing your research, and you're just seeing the same names or papers all the time, the same techniques, it's really nice to be able to see what other people are doing. So I really think your podcast is really great because you just are profiling a lot of different technology. So a lot of times like you might see something, then you're like, oh, that's kind of a cool idea. But I have no idea like how it works or even how expensive it is because a lot of times if you go to your PII and be like I want to do this, the next question is how much? How quick is it? And you're like, Yeah, I don't know. They never tell you in the papers. Yeah,

Dr. Barkha Yadav-Samudrala: 24:57

yeah, exactly. Well, and my last question for you would be, could you suggest any articles or protocols for proximity labeling so that I can link that down in the description?

Natalie Harris: 25:12

Yeah, so there's a lot of different bio ID technology out there these days, but one of my favorite, it's called Meet The Neighbors. So it's, it's a really cute title for bio ID. So it's kind of a review article that explains generally how bio ID developed. So it's a little bit, I don't want to call it old. It's from 2016. But that was about the very first bio ID technology. So pretty much if you understand this paper and the basic concepts, any other things you see, so like, they have bio ID, Apex bio ID too many turbo turbo, they have something else called mini something I don't know. But if you understand that most basic original idea, you can really kind of understand any new technology. And I think this paper Meet The Neighbors, mapping vocal protein interact items by proximity dependent labeling with bio ID is probably one of the best explanations, I think they have really good graphic kind of visually showing you that same idea about the fishing net that I've talked about. They really kind of show it visually like what you're looking at. So I think that's a really great place to start.

Dr. Barkha Yadav-Samudrala: 26:19

When Okay, thank you so much. I will make sure to have that in the description below in the show notes. And with that, I'm going to end today's episode and Thank you Natalie so much for giving us your time and explaining what proximity labeling is. Thank you and listeners I will catch you next week on another episode of the podcast and in meanwhile, if you have any suggestions, or if you would like to join me on the podcast, please email me at Barkha@nerdrdpodcast.com And remember, it's good to be a nerd bye.