Episode 8: Advances in high throughput protein workflows and automation challenges (Part 1 with Dr. Edward Kraft, Leash Bio)

Andrew: [00:00:00] Welcome to our podcast series, Imagine More, Create Solutions, where we talk about chemistry, biochemistry, and some engineering. For today, our discussions will be covering protein expression, purification, characterization, automating these processes, and an example of downstream application. So a little background before we jump right in.

Andrew: Hello, my name is Andrew. I’m the co founder and chief scientific officer of IMCS. I have a special guest here with me, Dr. Edward Kraft, from LeashBio, Senior Director, Small Molecule Discovery. Edward has worked with IMCS products before, in particular with IMCS Tips. IMCS manufactures chromatography consumables called IMCS tips, or pipette tips packaged with loose resins.

Andrew: And these work on automated liquid handlers on Hamilton, Dynamic Devices, Tecan, and many more, like Beckman and before we dive into the intricacies and complexities of the protein world, a big welcome to Edward. Please tell us a bit about yourself, [00:01:00] Edward.

Edward: Yeah. Hi, Andrew. It’s great being here. I am Edward Kraft from Leash Biosciences, and I have over 17 years of experience in high throughput protein expression, purification, and design, as well as small molecule discovery and characterization. My background, for those listeners who are interested in learning a little bit more about how I started in this field, started with a PhD degree from University of California, Davis, and then working at companies, including Monsanto, which is now part of Bayer Crop Sciences, Genentech, Recursion, and Leash Biosciences. In each of those places, I have focused on pushing the boundaries of high throughput protein technologies to deliver better starting points for R& D. By leveraging improvements in the technology we use, how automation can and can’t be used, and partnering with data science to streamline the data workflows and analysis.

Edward: And I really enjoy automating stuff.

Andrew: Same here. It’s just so exciting to see a higher throughput in the sciences, especially in the protein [00:02:00] world. Um, you actually touched a very interesting point that I was quite curious for the listeners. Your background, you got your PhD from University of California, St. Davis. So did you grow up in California and , what made you decide to do your PhD?

Edward: Yeah. And so, , I’m a Midwesterner. I am from Northern Minnesota, , right outside of Fargo, North Dakota. I grew up on a small dairy farm. , very atypical background, probably for most people who ended up in corporate sciences. But, , 1 of the things that I always enjoyed, in thinking about, , when I was growing up was how to make processes simpler and easier to actually perform. And so growing up on a farm, you do a lot of manual labor and you see all these wonderful machines that have been built around you to be able to make your life simpler. And that was really kind of the driving force to initially, like, develop those thought processes that I’ve used throughout my career. and so from Minnesota, , the winters are long and hard and California has always had this allure of being this like escape to place, right? For for decades. [00:03:00] It’s been

Andrew: Gotcha, gotcha.

Edward: That’s how I really got, , attracted that and I love the outdoors. Avid hiker, climber, backpacking, um, skier, snowboarder, and absolutely love the outdoors.

Edward: Being able to effectively live in a beautiful environment, such as Northern California, and have the outdoors surrounding you year round for activities, uh, was all it took. Uh, sign me up. I was in.

Andrew: True, true. That, that spot of California, you get all four seasons within maybe an hour and a half to two hour commute.

Edward: You can ski and snowboard and, uh, surf on the same day if you want.

Andrew: So I’ll jump right in and start with the book chapter that you, uh, you wrote in Methods in Molecular Biology titled High Throughput Protein Production and Purification.

Andrew: It’s a pretty comprehensive book, uh, published by Humana Press on methods and protocols. So those who are interested in learning more about this topic, I mean, please do see about this book, but can you give us a little snapshot of the [00:04:00] book or the book chapter? I

Edward: Yeah, thanks for highlighting that. This effort was spearheaded by Renaud Vincentelli , and I was happy to be part of sharing our approaches to building a high throughput protein expression and purification QC process across the three major expression systems, those being E. coli, insect cells, and mammalian. And I was also grateful to work for an organization at the time that really supports sharing of information to support career growth and help the field as a whole grow and move forward. The process we described, as used at the time, represented a highly efficient manner to harmonize workflows for intracellular membrane and secreted protein purification to support R& D for non antibody proteins. Using tip based purification, we have built a process to determine protein titers as a predictive baseline for scale up efforts. We removed unnecessary complexity at specific steps. We used one buffer system through, purification processes, [00:05:00] focused on specific affinity tags, His Flag and Strep used high throughput friendly cell strains and viral generation approaches. And this helped us push through most proteins. This let us have the time to tailor approaches to high value, high complexity proteins that didn’t fit within this process. This ensured the organization had a highly successful process to get proteins of all types into the hands of researchers across the organization in a consistent manner. I was fortunate to work with a team of exceptional scientists, data scientists, and automation engineers who embrace the model of continuous improvement of our technology and really enjoy the work that we did.

Andrew: I mean, you covered sort of the fundamentals of how to actually start working in a high throughput protein production. Um, you need the people and sort of the foundation and the approach, uh, you’ve outlined it and it’s also covered in the book chapter. That’s, that’s great. And it also looks like you [00:06:00] presented before at pep talk, but you’ll be another panelist at pep talk in January 2025.

Edward: Yeah, that’s correct. Um, I’ve really enjoyed presenting at pep talk in the past, and so I’m happy to be working back there again with Marianne and the team that helps set that conference up every year. And so I really enjoy sharing in what we’ve been able to learn about the complexities of protein purification across the entirety of the proteome. Um, the majority of. What gets talked about is usually very antibody focused, and so I’m happy to be able to present on everything that isn’t an antibody, but yet they still fit within really great high throughput approaches to be able to do that. So I’ll be talking about most recent workflows that we set up here at Leash Biosciences. And also there’ll be a panel discussion, um, more broadly, on, , the overall kind of state of the field and, uh, kind of open question kind of format [00:07:00] there for being able to ask people. And of course, if you see me there, I love to talk, ask me questions. Um, you know, one of the things I’ve learned over the years is I know less and less as I go forward about the specifics and challenges. Uh, proteins, um, but I’m happy to admit that, , and also appreciate just the, the challenges of trying to figure out how to approach the specific problems that people are facing,

Andrew: I couldn’t agree with you more on the statement that as I, uh, progressively learn more, I realize I know less and, um, it’s exciting. It’s quite exciting, actually, to learn more continuously, learn more and have that curiosity just continuously driving. Um, Anusha will actually be at PepTalk. So she’ll actually have a poster and she’ll attend your presentation as well. So definitely mingled at the event. IMCS actually attends a lot of the automation focus groups. PEPTALK is really [00:08:00] focused on a lot of the protein designs and protein expressions, protein purifications, and then the automation focus is on lab robotics and automation software, a little less on the protein purification.

Andrew: I guess that’s within a niche of that. But we’ve seen a lot of interest That crosses back and forth between the engineering group at the Society Laboratory Automation and Screening Conference, as well as Laboratory Robotics Interest Groups. Um, even like the conference that’s coming up in San Diego, ACS, uh, BIOT, that’s also very biotechnology focused.

Andrew: That’s been pretty interesting events to attend. And, uh, they also cover a lot of these broad protein focused topics. Have you considered presenting any of your work at other conferences?

Edward: , I’ve been focused on conferences, uh, that were more. Focused on protein specifics of our work and less on the automation side. We’ve had people in the group that are automation engineers at places. I’ve worked in the [00:09:00] past that do go to SLAS and and talk and kind of covered that aspect for me. It was always a matter of identifying more with the R and D science, you know, biology, biochemistry and less with the automation.

Edward: Although I very much appreciate it. And I’m a huge advocate of it. I came down to really focusing my time around high throughput, protein purification cohort. Then that group is most represented at, at PepTalk as well as PEGS. And so that’s largely where I focused, uh, my time. If I had the time to go, uh, expand out into those others, I would Certainly love to learn, , more and also present, at additional, places as well.

Edward: And so that’s something I’ll certainly look to do, uh, as my career moves forward.

Andrew: Yeah, you highlighted 2 of the conferences like PepTalk and PEGS that are really protein focused and those are exciting to attend. I’ve actually attended those quite a few times. All right. So before moving into a more serious topic for kind of more of the general audience, maybe some of the grad students and [00:10:00] undergrads curious about you.

Andrew: So you move from San Francisco, a biotech hub to Salt Lake City. What made you move? I mean, of course, Leash Bio is in Salt Lake City, but was that move a culture shock or any sort of surprise?

Edward: Yeah. , I’ll talk a little bit about, you know, San Francisco, and it certainly has changed several times over the last 20 years. Um, you know, I moved to California for grad school back in 2002. And it’s spent a couple of different times, , in California since and so we’ve moved around a fair amount as a family.

Edward: Um, we have the dual science careers, um, challenge in my family. And so we’ve done the tour of the big biotech cities, not just San Francisco, but also Boston and Seattle. Um, we’ve always prioritized family. And family interests as our children have grown, we have really prioritized our love of the outdoors. And so we were part of the big California exodus during COVID. And like many others during that time, we realized we [00:11:00] wanted to live in a place where we could spend more time doing what we love. And that was skiing, snowboarding, rock climbing, and hiking. And less time in traffic trying to get to it. If anybody spent any time trying to get across the Bay Bridge, Um, on a Friday afternoon leaving San Francisco.

Edward: It’s a very painful crawl. That was the aha moment, uh, for my wife and I were like, yeah, we’re moving on. And so cost of living was also a large part of our decision as well, given the realization that the Bay Area has really become unaffordable. Uh, for most people, especially if you don’t have family tides there. Um, and the premium you pay to live there just wasn’t worth what we were getting in return. And so we love many things about California, and I expect we may end up there again. Um, but we can really spend most of our life truly living here in Salt Lake City and being able to spend a lot more time with our kids than we were before. And I think Salt Lake City is a very misrepresented city, [00:12:00] um, and misunderstood. Salt Lake City is really hidden gem. And I hope that more biotech growth happens in this area here. And so , every city we live in truly has its peculiarities. Um, but Salt Lake City has been our favorite place to live. Because kind of the, the talks that I talked about earlier and so I really hasn’t been much of a shock for us. We really identified with that outdoor culture, no matter where we lived. And so it’s very present here. We’ve had kind of this feeling of kind of deep relaxation. You know, being here kind of like, why did it take us so long to realize that Salt Lake City was the place? It takes us 25 minutes to get from our home to Park City, ski bays in the mornings.

Edward: Um, we have the beautiful Wasatch Mountains at our doorstep. Traffic’s a non issue by comparison to where we’ve lived in the past. And if you’re an outdoors enthusiast, there really isn’t a better large metro area in the U. S. than Salt Lake City. My kids can walk around by [00:13:00] themselves without any worries and people are just really friendly here.

Edward: We’re really happy, um, to have made this choice to live in Salt Lake City.

Andrew: It is the ski kind of central park city is a fantastic a very large ski resort

Edward: Yep, largest in North, uh, or at least the U. S. I don’t

Andrew: in the US. I think. Yeah, I think Whistler is the only one that probably beats it by square footage or by the acreage.

Edward: Yep, it’s Absolutely wonderful. The fact that, uh, 25 minutes and if the snow skied out by then, we’d just come home.

Andrew: Yep.

Edward: Um, greatest snow on earth.

Andrew: And I think I was walking around, in San Diego, every San Diego street has sort of a San Diego Mexican fusion restaurant, and it’s always serving guacamole and chips. So I remember offering. Hey, do you want some guacamole and chips? And he said, No, no avocados. So what’s, what’s the deal with avocados?

Andrew: I met a Californian who did not like avocados, but then officially, I guess you’re not a Californian per se. You’ve [00:14:00]

Edward: Yeah, I spent a good chunk of my life there. And so you would expect I would have picked up on that. But yeah, I, you know, I’m a farm kid by background, a dairy farmer. And so if I’m going to put a bunch of fat on something, it’s going to be butter. Um, and so that’s largely it is. I never really acquired the taste, for avocado guacamole.

Edward: It’s, it’s not that I don’t like it. It’s just more of a, like, you know, why add it if it doesn’t really add anything for me.

Andrew: Yeah, understand. All right, so jumping in, a protein expression. Given my short attention span, I kind of browsed through the book chapter, but the first line really caught my eye. And I’m just going to read it verbatim . The expression analysis of recombinant proteins is a challenging step in any high throughput protein production pipeline.

Andrew: I mean that I live and breathe that line every day. Um, it’s an understatement and the adjective of challenging is definitely not doing any justice to the time and energy. That’s so many specialists have spent [00:15:00] boiling and a lot of failures, but a lot of successes in unpublished techniques. Can you cover a little bit more about that that line and expand on it.

Edward: Yeah, you hit the nail on the head here, Andrew. You know, this has been a broken record experience throughout my career and working with scientists across organizations and speaking to those as well at conferences. Protein expression is often thought of as a very simple process that you can just hit the CRO outsource button and get what you need. And in some cases for simple proteins, it really is. However, most of the time it isn’t. And the specific needs of the organization and what the protein is being used for often requires internal efforts. Tools such as Rosetta and AlphaFold have given kind of a false sense that computation has solved the ability to choose what your protein boundaries are that will be successful straight out of the gate. And this simply isn’t so. And in a [00:16:00] recent Nature paper from Lauren Porter at the NIH calls attention to some of the flaws of AlphaFold for being able to perform tasks, , beyond just memorization. And so if you stay within your training dataset, great, awesome, , but if you start moving outside of that, that’s where you start to see things fall apart.

Edward: And that, that’s a kind of a very common trend in MLAI. And protein folding is a dynamic process. And while these tools are particularly useful for well represented structures, question their utility beyond their training datasets. When you extend this to bringing proteins into an expression system, each with their own limitations, you become limited by the inherent ability to prosecute specific folds and the steps leading up to it with the use of strong promoters, often, in those expression systems, and you often overwhelm those expression strains for being able to produce quality protein. It becomes a compromise, and I’d argue, one well worth taking, To use limited [00:17:00] tagging strategies compatible with downstream assays and scale purification efforts done in parallel across your major expression systems in small scale.. Further simplifying by using automation friendly purification approaches to create a valuable process for every organization with a few people needed to run it. And by automation, I don’t mean push a button and walk away. As this is a common misconception of automation, what it offers. These systems help with efficiency and scaling, but require as much, if not more, human time and effort to interface with them and really help them run. It changes the nature of the work you do. but it doesn’t require less human time. And be careful with certain tools, technologies that are quite limited in utility across all proteins. And I’m looking at mentioning capillary electrophoresis as an example. Finding these compromises frees you up the focus time needed to figure out strategies for high value proteins that are [00:18:00] not amenable to basic approaches. And you should expect constraints, buffer optimization, cell culture conditions, construct design, et cetera, to help make them attainable.

Andrew: You cover that point really well, even with the automation and the high throughput. It’s really the people that come together and design that so that it seems seamless. However, there’s so much that’s occurring in the background. Like you pointed out that phrase, push a button and walk away. It’s sort of a fantasy.

Andrew: It can be achieved in certain levels, but, uh, you constantly need to monitor it and adjust. , you also mentioned within the chapter, I think, three different expression systems that you kind of rely on. The E. coli, the bacteria system, and then the baculovirus, uh, expression vector with insect cells, and then a mammalian transient expression system.

Andrew: Um, so it sounds like you go to the bacteria, the E. coli, because it’s the easiest one to go, and you do that first, or do you [00:19:00] run all three in parallel? How do you approach that? These three expression systems.

Edward: Yeah. And so really the, the time constraints here to get information into the hands of scientists is best by doing all three expression systems in parallel. You know, in my experience, if I were to skip an expression host, it probably would be E. coli nowadays. Um, the optimization and strain use can be time consuming and going straight to insect cells.

Edward: You know, if cost isn’t too much of a consideration is the better use of time. For intracellular proteins, the system of choice is insect cells using baculovirus expression. For membrane proteins and secreted proteins, the developments of 293 expression systems make it the go to. And you’ll be better off for glycosylation that is more closely, not always, uh, matched to the native protein. So 293 does a pretty good job at proper folding of most, but not all, complex disulfide structures, common in secretive proteins. I’m also a big fan of BacMam, uh, which is effectively baculovirus that can be [00:20:00] used in mammalian expression cell lines for doing mammalian expression as it avoids the cost of transfection grade DNA and the associated processes of transfection.

Andrew: I’m not as familiar with BacMam. These are larger vectors or how large, I guess the baculovirus itself, the vector is pretty large.

Edward: Yeah, you’re making an active virus there. And so your transfer vector, you can combine that as well as engineering the actual viral DNA itself.

Edward: . And so it’s 120, 130 KB, depending on which version you’re using for the vectorized DNA. And then your transfer vector, um, vectorized can accommodate really large sizes. unlike some of the mammalian type vector systems. And so it gives you a lot of flexibility there to be able to add in and do multi protein complexes. And we’re seeing a lot of development there for additional stuff that can be used for that purpose. And yeah, BacMam is effectively, you [00:21:00] know, a mammalian promoter, using a virus that sometimes is pseudotyped to help with, better with viral fusion. Um, to mammalian cells than GP 64, which is the viral fusion protein for insect cells. And so, yeah, it’s it’s a nice, simple way to be able to do a non replicating virus, within the context of human cells to be able to do gene delivery. Um, and get protein expression, in a transient way that can be a little bit more cost effective, especially if you’re already doing insect cells at a company.

Andrew: that’s very interesting., So you mentioned actually within the E. coli system, that would be probably one of the hosts that you would skip because, is it because there’s so many subspecies and intricacies within the prokaryote that alters your expression solubilities and your yields, I mean, the growth conditions, inductions, chaperones, heat shock gain, osmotic stressors, um, we’ve actually investigated even these, uh, [00:22:00] untranslated regions and prior to the promoter that seems to impact the expression levels and all of those.

Andrew: There’s very limited publications around it, there’s some, but,, it seems like every subtle change to the E. coli could impact your expression,

Edward: yeah, and so to kind of the points I started discussing earlier, it’s largely why I’ve moved away from E. coli, especially if you have limited people to be able to do your process. It’s very, can be a very niche process and the time to discover a way to make it work and then implementing it takes a lot of time and effort. I’ve always focused on standardized approaches in E. coli and then prioritized eukaryotic expression systems if the simple approaches in E. coli do not work. This then compounds in ability to be able to transfer this knowledge to scale up groups or CROs. And so often there’s a disconnect within like a high throughput group and an organization that then needs to then transfer that knowledge to a CRO or scale up group. And so if you have a group that does insect and [00:23:00] mammalian expression, well, then start there. The added cost in those systems is offset by the time you save. They certainly may, can take a very good team and skill set to do well, especially insect cells I found. It’s essential to have downstream scale up options aligned with your upstream expression analysis groups, kind of like what I mentioned, uh, relative to being able to outsource stuff. And so, delaying protein availability for research is many fold more expensive than enabling a good eukaryotic expression effort.

Andrew: So that downstream scale up option, can you touch upon a little bit more on what the high throughput labs should consider? Because, like you said, they don’t work in the scale up side, they work more in the high throughput, smaller scales. Maybe two, three mils of cultures, five mils at most, but then now your scale up is 20, 30, and then potentially thousands of liters that needs to expand.

Edward: Yeah. You know, that’s something that was always a [00:24:00] consideration with developing these high throughput platforms. You weren’t. necessarily, depending on your goals, interested in maximizing your yields out of two to three culture. What you wanted to do is best replicate the scale up options that were going to happen downstream.

Edward: And I think that that’s where you can get yourself in a little bit of trouble is optimizing and finding ways to do things that are not scalable, uh, at two to three notes, and then go to try to scale it up. And then you have a disconnect. And I think some organizations You know, say that they have trouble, , doing small scale and then translating that into large scale. And that’s because you have to stay connected with who’s outsourcing it, as much information as they’re willing to give you about how they’re going about outsourcing it, as well as your in sourcing scale options to make sure that you’re staying connected there. And building a process that best replicates what that scaled version looks like, um, so that researchers are getting in the realm of what they [00:25:00] expect to get. And so staying connected all throughout that process is really essential to having a small scale process that’s predictive for large scale actually work.

Andrew: Now does that also play into sort of the protein design? So some of the expression of your protein targets to improve solubility, you’ll throw in certain tags. I mean, for affinity purification, you’ll have the affinity tags. You mentioned those HIST tags, STRIP tags, FLAG tags.

Andrew: But then what about any sort of fusion proteins to facilitate the expression or solubility of a protein?

Edward: Yeah, I’d say this is an early career mistake most people make is the fact that I’ll just put, you know, GST or MVP on it and it’ll make it so much better. Well, I think fusion proteins with solubility tags are a last resort for intracellular proteins. As I’ve learned over my career, but may have specific uses for extracellular and membrane proteins.

Edward: Um, the downstream processing to remove tags. and have proteins then lose solubility and incompatibility with assays, [00:26:00] make them generally unattractive as a whole. And so more times than not, the fusion protein is masking a flaw in the construct design or boundaries, and the fused protein is actually a poor quality. These solubility tags could give a false hope And can be misleading for research use. If they force dimerization or incorrect apologies, you know, I’m looking at GST and FC as being two tags that are very well known to give false solubility hopes and also force incorrect geometries. That all said.

Edward: There are use cases for large liability tags where specific downstream uses or purification processing methods have been established and, you know, make them essential. FC tags can be an excellent tool for expression of ECDs when used properly. GPCR methods, you’ll commonly find MPP fusions that have been used quite successfully for boosting yields and stability to enable structural work. And so it is important for scientists to be wary of using these approaches and to align with your [00:27:00] scale up purification teams to know the risks. If the small, simple affinity tags are failing to produce protein, it’s telling you something about the protein you’re attempting to make.

Andrew: So what do you rely on? Um, I guess you can’t go into too much of the depth of giving away the secret sauce to improve solubility. I mean, I’ve seen some free software out there like PROS. I guess there’s now some additional software that’s kind of out there to help assist with the solubility or the design of the protein to be more soluble or to improve expression.

Andrew: Are those some of those kind of processes that you also take?

Edward: Yeah, and certainly with construct design, um, trying to choose like The best batteries. Um, there are computational solutions out there , that are really valuable. And I think that’s where, as we’ve seen, I’ll hold improved and Rosetta and other tools, um, that those can be a very valuable in that way to help better choose your boundaries. At the outset, [00:28:00] and it’s also important to consider, you know, the scaling of your process. You can quickly get niche and cute and small scale, but it must translate once again into that skill process to be able to get access to that protein. The 3 expression systems that I talked about with the broadest utility play well with each other in a single lab. They have the greatest options for scaling and outsourcing. I’m a big fan of pichia and fungal systems, but they require a separate space for use given the contamination risks to your other systems. And so they do not play well together. And so all of them should be screened in unison, um, once again to enable research.

Edward: As we discussed in the book chapter, the Ability to combine these systems on tip based purification platforms using automation gives very little reason not to attempt them in unison up front. Small scale expression testing of one versus three systems really doesn’t change [00:29:00] much from the perspective of cost since the post expression steps can be combined. And so there are approaches on the molecular biology side to use expression vectors that can even work in multiple systems. Without too much of a sacrifice in protein yields there as well. And so, yeah, it’s really about combining those different philosophies to really figure out what the path forward is for a protein as quickly as possible.

Edward: And then, like you mentioned, there are some computational tools that can help you. They haven’t solved the problem. But certainly, we should be using all the tools that are available and at our disposal.

Andrew: Yeah, there’s so many tools out there. I mean, computational tools as well as just. Really taking a sledgehammer and just kind of chugging it through as much as possible, too, and building your own database, too.

Andrew: So, before we transition from protein expression to protein purification, uh, we should at least touch upon cell free expression system. It’s lower yielding at this time, [00:30:00] but it’s coupling the mRNA transcription process to the protein translation. So, it consumes a lot of that energy, the cellular energy, so your yields are going to be obviously lower.

Andrew: But if you can decouple that and then just focus on maybe cell free protein expression, uh, there might be sufficient quantities generated with the cell free systems. Have you tried any of these things, or the cell free expressions

Edward: I mean, over the years, certainly these, these have come up, right. And depending on who new people come into an organization, they want to try these things. In general, I’m glass half empty. Uh, on these systems as a tool to get researchers what they need. I agree that, uncoupling mRNA from translation absolutely has an advantage.

Edward: Even still, you need a process to be able to replenish components of the translation, remove byproducts, and so you really need to set up a continuous feed type scenario to be able to maximize your protein production, and they do exist. Um, that they are very expensive. Uh, for the protein that [00:31:00] they can make. Um, and you’re also limited in some cases by the lack of chaperones and cofactors to actually make the functional forms of proteins that you’re really after. And so they certainly do have a little niche, uh, for being able to produce. things, and I’ve used them for that. As a general tool for an organization to be able to scale protein production, I’ve yet to see a good strategy there that’s particularly cost effective, aside from very, very particular niche, uh, situations.

Andrew: Yeah, so all these points, uh, they’re covered further in detail in that book chapter. So, please do go and, uh, read more books, in particular that book. So let’s kind of jump ahead to protein purification. So, technology has come a long way since, I mean, even within the last 10, 20 years, there’s a polyhist tags, twin strip tag.

Andrew: And then there’s that new Intyne, approach where it self [00:32:00] cleaves with the resin from Cytiva called Protein Select, um, and it leaves a tagless protein. So what sort of techniques do you typically go to for, for your protein purifications, your specific buffers and your pros and cons for each of these resins?

Andrew: Then of course there’s Protein A that’s used to purify any FC fused proteins. But again, if you’re using or if you’re going after, you know, a lot of the in intracellular proteins or sort of non antibody based targets, protein A is pretty limited.

Edward: Yeah, aside from FC tagging, um, extracellular domains, it is. And so if I had one tag to work with, it would be Twin Strep. The single step purity, and the small size are really ideal for high throughput applications, and it works extremely well if you’re dealing with secreted membrane and intracellular proteins. That said, uh, given the diversity of proteins, we need other tag options, you know, you have good old hydrophilic flag tag, , workhorse His purification [00:33:00] using his tags and variants of his, and, , those larger solubility tags on specific occasions. I’ve also become a big fan of HiBiT, uh, as well for simple detection of your fusion proteins, and different downstream applications for, for assays.

Edward: Um, it’s not truly a purification handle just yet. Um, but certainly I expect that there’ll be additional efforts built up kind of around that. Intein and other protease based sequences are a great asset when you’re looking for generating tagless proteins or near tagless. However, if I had to couple an affinity tag with something to remove the affinity tag. TEV, , TEV protease is definitely the best choice there. It’s a go to option, , and you have better uniform performance across a variety of conditions, , and localization if you need to generate, you know, a near typeless protein, I guess in that case. Again, depending on your downstream use, removal of Affinity tags, small affinity tags, really might not be a [00:34:00] consideration.

Edward: And you might be spending a lot of time and effort removing these small affinity tags when your downstream assays, um, are completely compatible with them.

Andrew: Have you considered or used other tags beyond the flag twin strip tag or strip tag? Histidine tags. I mean, you mentioned solubility tags too, and, uh, you also kind of mentioned hipbit or HiBiT. Can go a little bit further into that?

Edward: Yeah, , like you mentioned, if I’m going to choose a soluble tag, it’s probably MVP, not GST, because of the GST dimerization problems. And, I started, alluding to it above, and like you just mentioned again, I’d like to see HiBit really take off, um, and be able to actually, get incorporated into more situations.

Edward: And so, we’ll see, you know, what, uh, the company that kind of controls that, ends up doing. Um, but the multifunctional nature of that peg for downstream luciferase type assays and rapid maturation of the restored protein make it [00:35:00] so versatile. Unlike people that may have had experience with split GFP systems, which were kind of slow, um, to be able to restore.

Edward: That split the surface type systems are extremely fast in reconstitution, , instead of a simple consideration for affinity purification. I’d like to see more holistic approach to small innocuous affinity tags that are multipurpose for companies to make their products really

Edward: think about, you know, end to end solutions, , make sure that they work for people in the lab doing the process. It can’t be laborious. It can’t be super niche. I see that’s disconnected and talking to companies that are working on, uh, tag solutions and that they’ve become too isolated

Edward: from their customer base and figure out how people would actually be able to use their systems.

Edward: And then something gets released and nobody uses it. And so there’s a little bit more like customer interaction there to really develop what would be that next tagging system. But overall, I think that we’re generally [00:36:00] pretty well served, uh, for most small little tags. We’re combining small tags and in different scenarios to be able to do purification.

Andrew: Yeah, I wish there was a sort of a higher capacity resin for flag tag proteins. Uh, right now it’s mostly based on the antibody affinity purification. Those are quite costly. I’m sure there’s some way of finding, protein A derivative that binds flag tag.

Edward: Yeah, and as I’ve talked about before at pep talk, you know, companies have to. Define their own solution there,

Edward: um, to be able to overcome what’s been commercially available. Um, and so it’s simply a matter of company being able to effectively choose to take the time, , to develop better options around Flag tag for, , better capacity on a resin. Um, the technology exists. It can be done. It’s just a matter of somebody taking the time to do it.

Andrew: Yeah, so the further parts of the protein purification process, . taking the lysate, doing an affinity tag [00:37:00] purification, IMCS tips for example. You can roll out part of that protein purification starting from lysate, affinity purification, wash, and dilution.

Andrew: That chromatography stuff can be automated. But there are certain parts that are difficult to automate. Can you cover kind of both of these worlds, the automation side and some of those other sides that are still semi automated or manual and that could be feature automated.

Edward: Yeah, , one of the things that I’ve never really spent a ton of time focusing on on the automation side was really the soul culture side of it. I know that there are organizations who spend extraordinary effort and time. In automating some of their cell culture aspects. Um, but that was certainly. , could have been invested in, you know, if ever had the need and an organization to do that, , for the more niche processes that were not antibody. I think the antibody space people certainly. Done a lot more on the [00:38:00] automating of the purchase there. Um, haven’t seen that work necessarily that well for the non antibody space. The automated focus of some areas of sample prep and largely purification steps, quantation, data registration is where I’ve always focused. And you kind of alluded to that, that there are a lot of great things that people have done in the antibody space or even stuff that, , I’ve done. Over time with teams that do play really well there from an automation perspective sample preparation for intercellular proteins for purification, , requires harvested cells license, you know, in the case of integral membrane proteins, you have a stabilization step. So, sample preparation is a little bit more. Manual in that context for a separate preparation for secreted proteins. Um, once you remove the cells and cell debris by centrifugation, , you’re really often running there for being able to more simply put things into. [00:39:00] Purification platforms, such as, you know, on Hamilton with IMCS tips. And so when I area of would like to call out, you know, for more effort. , from companies for the non antibody people in the world. Are better analytical techniques on capillary electrophoresis platforms, , that truly work for a high diversity protein set. , SDS page still remains really the, the go to there. , and so that’s an area where I’d like to see a little bit more, um, innovation to be able to build systems that can really. Better handle the analytical techniques, uh, for just making sure you have an intact protein. , SCC would be great, uh, if you have truly higher throughput, more rapid SCC, where the analysis wasn’t so painful and laborious, um, to be able to deconvolute , what you’re looking at.

Andrew: Yeah, I think Bio Rad used to have like a microfluidic chip that was very similar [00:40:00] to an SDS PAGE gel

Edward: Hmm?

Andrew: and I don’t know what happened to that.

Edward: Yeah, I’ve

Edward: used a variety of these , over time and, and some of them have been discontinued.

Edward: Um, and so there, there’s an element of time, it’s pretty easy to run updated 96 proteins on an SDS page still now. Um, you know, even with the gel being automated versus some of those systems where like sequential loading type

Andrew: Uh, yeah,

Edward: eight or 12 at a time, they’re actually quite

Andrew: not formatted easy for the automated liquid handler to transfer the samples to those tubes or the micro those chips

Edward: Yeah

Andrew: Yeah, have to actually manually do it.

Edward: Yep.

Andrew: It sounded like, , on the front end of the sample prep. You do some filtration or centrifuge, which one’s easier to automate?

Edward: Yeah, filtering , is absolutely the best thing to do upstream of tip based purification, you know, but you still need to centrifuge. Right. And so I haven’t found a truly integrated device. on a [00:41:00] liquid handler, such as like positive pressure devices or vacuum type things that actually work well enough, , for lysates of different viscosities.

Edward: and so , you’re still stuck having to centrifuge things. However, you can certainly dock centrifuge onto a liquid handler and I’ve done that in the past. It’s a minor detail. Um, to be able to integrate that, but it’s not uncommon to see that. And so, you know, in talks, I’ve discussed our work with high throughput compatible depth filtration, uh, to remove debris. And so this has become very valuable innovation upstream of tip based purification. You know, and it is an expense, but it’s well worth the cost. Given the downstream sample uniformity you get there, and how compatible it has been for tip base purification, so you’re not clogging or causing differential flows within your, your tip environment, And so , before that you had dead pass filters or a single, just, , 0. 5, 0.45, or 0.22s and, and for doing them. [00:42:00] And they’re so prone to clogging. And then you’re left transferring it to another plate to try to be able to recover your sample. And it is just. Not worth the time and effort and depth filtration plates, that’s at the time we had developed in connection with the company, , are just a breeze now and I, uh, use them, you know, right now and in my current role as well with building a high throughput, , protein production process.

Andrew: And those depth filters, they, they’re really compatible with the mammalian systems and the insects, I mean, the entire mammalian systems. They’re not so friendly with the prokaryotic systems. Once you lice it, there’s too much debris, isn’t there?

Edward: You have to be careful for overloading them, right? You’ll get channeling through those resins. And so it’s all about making sure that your volumes are appropriate. And so just like anything, it’s, it’s a matter of what your culture volumes are and what you’re trying to filter. And so we ran [00:43:00] Certainly, any expression system through them, it’s just a matter of don’t overload them, , because you will get debris passing through. And so it’s just, , within your process, you can absolutely make them work. , you can also make them not work. Um, And so, but I highly recommend them versus any of the other things, and you can pre centrifuge things to be able to reduce the debris burden, rather than just taking a lysate and going straight through them.

Edward: And so, yeah, there are things you can do at an organization and there are a few companies that are offering this now, um, and reach out to them and talk to them and they probably have a good solution for you.

Andrew: So now moving to the other side of protein purification is What if you just left it on the bead and you don’t elude off, can you do your downstream analysis while the enzyme or the protein, well for enzymes, if it’s on the bead, you could do an enzyme assay, but then if the protein is still on the bead, could you do binding assays or other assays?

Andrew: It sounds like the hybrid has some sort [00:44:00] of luciferase associated assay to it, so you could leave it on the bead.

Edward: Oh, absolutely. You can detect it. You can quantify it, , right then and there. And so, uh, for certain asset types, I think this would be great for subsequent use. It becomes a question of storage. And so at a larger organization, , most companies doing the purification are separate from the downstream user groups. And so it’s, if there’s a means to be able to store for later use at temperatures amenable to protein stability. . Um, I expect this could be a large gain in efficiency, or if you’re at a more flexible organization where you can couple purification directly into, , assays, , then that would be ideal, , and even being able to replace resin altogether, , and have, have the tips, , do some of the work for you in different ways , would be great as well. And so. Yeah, I look forward to seeing, what can be done there, especially have organizations that are really willing to talk to one [00:45:00] another in the different groups and being able to then link up and try this as a truly linked activity, rather than eluding off and then reattaching it in

Edward: some way and, you know, doing your assay, which is highly inefficient.

Andrew: hmm. I couldn’t agree with you more. We discussed briefly earlier about the protein solubility, but then it also touches upon protein stability.

Andrew: I mean, over time, if the protein is unstable, it might crash out of solution, form aggregates, and you might see that in your solubility asset. But, it’s not just, you know, the stability is not just the original functionality retention, but there’s other environmental factors that all cause issues, to the protein stability.

Andrew: So, are some of these issues, and I agree, you know, glycosylations and other, you know, post translational effects can all impact protein stability, but how do you address these? Concerns or topics when you’re doing the high throughput protein purification.

Edward: [00:46:00] Yeah, this is a challenge in the high throughput space and I’d step away from thinking of doing this across a variety of proteins in a single run to look, you know, then at a single protein and variety of different conditions. And so it’s not that different from typical crystallography screen or membrane protein detergent screens, but with a different question on solution properties as a function of the environment that a protein is finding itself in.

Edward: And so, High throughput setups are ideal, , for looking at this question, , but the downstream readouts are going to be truly different. And so in a more typical context of where you’re purifying something in high throughput in a more harmonized buffer system and a variety of proteins, you’re probably looking at very few proteins, in a variety of different conditions.

Edward: And so there are a variety of different assays that you can use here to get around that. There’s for essence size exclusion. chromatography, and there are different things that will bind to tags to be able to tell you, , where your protein is eluding. Um, [00:47:00] nano,-DSF, MALS, SEC, SPR, or if you’ve got known binding agents to be able to read out, , downstream of treating your protein differently. Um, these all integrate really well with protein expression and purification processes. It’s, it’s question on priorities of an organization on where it wants to put this process, relative to the overall screening for purification and titers and in general apparent solubility and when to use it. You can quickly get derailed here in wanting to go into really deep characterization, , of individual proteins, and it just depends on where you want to put that process, appropriately. You see in the antibody space, they look at aggregates, right, right away, in high throughput I think that that’s, um, well placed and functioning really well as an asset there. It just depends on when you’re dealing with the entire proteome where and what are the important questions for those proteins or protein [00:48:00] classes to answer in a high throughput small scale setup versus waiting until later downstream.

Andrew: Awesome. You rolled out the semi automated protein purification process. How much of that is truly automated? And how much of that is still manually intervened?

Edward: Yeah, so I’m happy to call this process out as truly being semi automated, , given I’ve yet to see a truly autonomous protein expression purification process. When you’re talking about, , the variety of proteins that you’re typically dealing with when you’re, you’re not just solely focused on a specific protein type. Where you have an exceptional group of people that aren’t working alongside the equipment. And so, to run a process like this and have the data be representative takes a highly attentive group who are deeply connected to the work that they’re doing. The area of focus for automation where the protein purification and data science side of the analysis and subsequent, , like we talked about the book chapter, , are [00:49:00] talks I’ve given on how you can go about possibly building out some of this. And so, yeah, I have yet to see a scenario that truly warrants a fully autonomous. Process much like you might have or like antibody purification process, which are becoming more and more hands off with still very talented people working alongside that equipment.

Andrew: Definitely, if something goes wrong, somebody has to actually be there to fix it.

Edward: Yes,

Andrew: Or to recognize what could go wrong and be prepared. , in this semi automated process. I mean, that’s a relative term. I mean, you probably relative to some of the academic labs, you probably more automated and relative to some other maybe startup companies. The places that you’ve actually implemented automation is probably highly, highly systemized. So [00:50:00] kind of curious there. I mean, you use our product, IMCS tips in this process, how much of that has helped you for the automation?

Edward: yeah, you know, Andrew, we’ve been working together on an offer for several years now. And so I really enjoyed working with companies. where we are both looking to learn from each other, rather than that product purchaser relationship that you also get so frequently. You never stop thinking about how your technology can be best used, or even updating the design to improve on a prior iteration. In addition to the tips themselves, IMCS makes it so easy to set up the process on a liquid handler by providing onsite application support and methods set up. And so this is allowed for really rapid, uh, set up without trying to coordinate for internal resources , and budgets to be able to then pay to get those as systems set up and see. You really address the fear [00:51:00] factor of biologists and biochemists who walk up to a liquid handler and think as soon as they, they touch it, they’re going to break it. And so it’s been really great, working with you. You, you structured, , this in a way that makes it very accessible, for anyone.

Edward: If anybody’s considering, , doing, , tip based purification, , reach out, , and, , see what IMCS , can do

Andrew: Thank you for that vote of confidence.

Andrew: We just wrapped up part one of our conversation with Dr. Edward Kraft from LeashBio. In this segment, we delved into protein expression and purification. His background in high throughput protein expression and purification is extensive, to say the least.

Andrew: In the next part, we’ll dive into protein characterization, challenges with automating the high throughput protein workflows. And then wrap it up with future directions, and his thoughts in the future. Until then, connect us on LinkedIn and check out more episodes at IMCStips. com or your favorite podcast platform.