The 3rd webinar in our education series will focus on how Artificial Intelligence can expedite media formulation and speed the time into the clinic.
This is truly a breakthrough in enabling scientists to optimize and control their media. In Webinar 1 we explored how custom media, and even one component change, can improve your cell potency. In Webinar 2, we showed how custom media is actually more cost-effective than using Proprietary Off-The-Shelf (POTS) media with similar development cycles.
In this webinar, we’ll teach you about NB-AIR™, the first-of-its-kind, AI platform for media formulation based on peer-reviewed articles. We’ll discuss:
- How machine learning can optimize cell performance by a meta-analysis of ~75M scientific publications looking for relevant compounds that are proven to impact critical quality attributes (CQAs).
- Demonstrate how NB-AIR gives you the ability to create a formulation based on your desired CQAs and the latest peer-reviewed articles. Instead of using POTS media with an unknown formulation that is a decade old, you leverage the latest peer-reviewed science.
- How these formulations can be transferred directly to NB-Lux™ our cloud-based portal for real-time quotes, ordering, and tracking of formulations. You can order lot sizes from 2L- 2000L.
- These tools will finally allow therapy providers to own the IP on media, a critical component in their ecosystem. Media is a competitive advantage in your therapy development.
Breathe new life into your cells with NB-AIR™. Did you miss the previous webinars? They’re available!
Webinar Audio Recording:
Listen to the audio:
Rachel Gauvin: All right. Welcome, everybody. We have more folks rolling in, but I think we should get started.
Thanks for being here. It’s the webinar: “Intelligent Media Formulation Using Machine Learning.” We’re really excited to be here and share this information with you — great group of panelists, some great information to share.
For those of you who have been with us for all three webinars, thank you. We’re really excited that we have this trilogy here for you. But, just in case you aren’t aware, we had a great webinar in September called “Better Media, Better Outcomes” and one in November, “Faster Media, Faster Outcomes,” both of which are available on our website under our Education tab. So
highly recommend taking a look at that. We’ll do a quick recap today. But it will be great — it’s great content for everyone to have background for today.
And, today, we’re, like I said, talking about intelligent media formulation using machine learning. Quick question for you all here. We are curious, and you’ll see why later on, but what cell types do you work with? And what are your critical quality attributes or your CQAs? So, if everyone wants to take a moment and just answer a couple questions here…
Great. I see T cells, dendritic, MSCs popping up high. That’s exciting, and the critical quality attributes pretty set across the board with proliferation and viability being up top. I’ll leave this up for another moment or so. But, yeah, pretty good distribution with all the cells and CQAs. That’s interesting. Yeah, proliferation and viability are way up there.
All right. Well, thanks everyone for that information. If you can share real fast, you’ll see HEK cells were just knocked out above, but T cells, MSCs, iPSCs way up there. And then critical quality attributes pretty sat across the weights with proliferation, viability being up top. So thanks for that information. That will help us, and you’ll see why in a moment.
All right, yeah, we have a great group of panelists here with us today. And I’ll let each of them introduce themselves. Roddy, why don’t we start with you?
Roddy O’Connor Ph.D: Okay, thanks, Rachel. Hi, everyone. Great to be on board again for the third in our series. I’m a Research assistant professor for Center for Cellular Immunotherapies, and I really focused on CAR T cell metabolism.
Rachel Gauvin: Alex?
Alex Klarer: Hi, everyone. Glad to be back. Name is Alex Klarer. I’m the Head of Cell Therapy Development at a CDMO based in Newark called Biocentriq.
Rachel Gauvin: Great, David?
David Smith Ph.D: Hey, everyone. Welcome back. David Smith, VP of Technical Operations for Ori Biotech, a pretty small startup company based out of the UK developing technology hidden on all of these manufacturing problems. Prior to that, run an R&D team at a CDMO devoted to cell therapy.
Rachel Gauvin: Great, and Dave?
David Sheehan: Hi, I am Dave Sheehan. I’m the Founder and CEO of Nucleus Biologics, and thank you to everybody for attending and really excited to share what we’re doing on the AI platform.
I think one of the things that we really maybe need to get at the heart of is, What problem are we solving? And, every day, there’s data coming out in peer-reviewed journals that is telling you what components are driving performance of different attributes on your cells, and those attributes affect the final therapy.
And so, on top of that, we’re in this very rapidly growing industry, and time to market is going to be a critical metric for any therapy developer. How fast can they get into the clinic and prove their therapy? And, historically, we’ve been hampered as an industry because a lot of the media has been proprietary off-the-shelf. And it’s convenient. I mean, it made a lot of sense on the first-generation cell therapies. But a lot of these formulas are decades old, and in one of our social media feeds, we talked about how an ex vivo was first cited in 1999. It was really designed at the time probably to proliferate the cells. And so not only is the formulation created for a different purpose, but it’s also not disclosed. And so, as a scientist, it’s really difficult to optimize.
And then you look at that in combination with the fact that there’s just all this advance going on and knowledge around pathways on specific cell types and the number of papers that are out there in the public domain. And it’s overwhelming for a scientist to try and figure out how to optimize their ecosystem. And media is a key component. And, up into this point, there’s really never been a tool that allows you to do that. That’s really based on not anecdotal evidence but proven science.
And so what we’re going to talk about today is NB-AIR. It’s the only cloud-based AI platform for media configuration that’s based on peer-reviewed articles and community research. And that is going to be something that we start with and then build off of.
So, maybe, now that we’ve covered what the problems are, we’ll talk a little bit about and recap some of the prior webinars. Roddy?
Past Webinar Summary
Roddy O’Connor Ph.D: We’re highlighting one of those 400, 000 papers on T cells. That’s our own work. I did mention I’m very interested in CAR T metabolism and, of course, the metabolic properties of T cells are not fixed. We kind of showed that previously. And, in this kind of figure and paper we’re referring back to, it was really important because we showed that with kind of a physiologic adjusted medium, so culturing T cells ex vivo in this medium, could give rise to progeny that really had enhanced anti tumor potency after we transferred or infused them into a xenograft model, a neuroblastoma. So that’s really what we’re highlighting there on that slide, that there is better engraftment and cytolitic activity in T cells expanded in this physiologic adjusted medium. So don’t take for granted the qualities of the media, or really conceptually think about it and optimize those qualities, okay?
Yeah, and this does lead us to really one of the central paradoxes that we’re faced with here at Penn, specifically the immediate challenges: How do we improve CAR T cell advocacy in a solid tumor environment?
So, in many models, we’re able to show that CAR T cells, they can effectively infiltrate solid tumors, and they’re there beside the tumor cells largely present while also largely dysfunctional. So it always leads us back to this complex immunosuppressive aspects of the solid tumor. And you could dive deeper into this, but it’s — as the figure shows, it’s largely complex that promotes this T cell exhaustive features. But it makes us wonder: What are the qualities that we can convey to the T cells while we have them in our hands? So, while we’re expanding them nine to 14 days ex vivo, what kind of features can we convey to them or can confer to them that will lead to a better anti tumor function following infusion into an environment?
And really we have to consider what’s been done historically in terms of the medium formulations, and we have kind of touched on that in the previous two webinars. And I’m going to show you kind of a new, something kind of new, emerging, in the literature about the media formulations and how kind of a correlative feature with a glucose level in the tumor environment.
We’ve touched on this before that within that complex features, that there can be high degree of mitochondrial dysfunction in T cells that do kind of infiltrate and traverse solid tumor environments. And this is really new findings from Greg Delgoffe’s lab. So he’s able to show that considering the glucose levels in media formulations… and he’s shown previously, and we can see on the figure, on the upper finger there, that when T cells are exposed to glucose levels, very low glucose levels that would be found in the interstitial or extracellular spaces of tumors, that there’s an impaired mitochondrial function or mitochondrial stress, as measuring that kind of charged separation, of course, the intermittent mitochondrial membrane.
So I just included that figure in the bottom for those who like to dive deeper into that. So that separation of charge is a key determinant of T cell quality, mitochondrial function. And Greg’s lab was able to show that there’s really shared features of mitochondrial stress if the glucose is very low as what you would see in a tumor or too high as you would see in these kinds of historic medias. And Dave did point out at the start that the historic media formulations were really designed to promote cell viability and propagate large numbers of cells. And I think it doesn’t confer the optimal features to T cells for things like potency, engraftment, less differentiation ex vivo, and enhanced anti-tumor function following infusion. So there’s a lot of work coming out even just looking at glucose levels in this example and how it impacts T cell mitochondrial function, which is going to be permanent for a high-quality T cell in a solid tumor environment. So I’ll leave it there.
And I’ll leave you off with that, David.
Alex Klarer: So we might be…
David Smith Ph.D: — We’ll see if we can get through these slides quickly and move on to the next one. But I think, taking from what Roddy said, there’s obviously a number of key features within your media that is vital to have in there, vital the concentration of it, vital probably the time point during culture that it goes in there. And so I don’t think that’s sort of up for debate, then really your posed with the question of Do you go for a custom media where you’ve got complete control over that, which sounds great but potentially has a long lead time, it can be very costly, and, really, when everyone’s sort of running to get to the clinic, do we have time to stop, pause, really create the optimum media for ourselves?
The other option is this POTS, proprietary off-the-shelf. And so this is everything obviously that you saw on the slide before regarding sort of aimed five ex vivo examples. So you don’t really know what’s in them and pretty hard to change what’s in them. You can obviously do additions, but removing anything isn’t possible. But there is a huge benefit of time, and we know how important that is to everyone. So really up to now what we showed in that first webinar is there is no real perfect solution there.
So, as we moved on to the second webinar, started looking at speed being crucial here, and so something that Nucleus has put together is NB-Lux. So this idea of actually a development time that is equal to that of proprietary off-the-shelf media. So roughly 22 weeks. That’s giving you time to test out variants of media within there. So instead of maybe picking the five top companies out there and testing all their media over a couple of weeks, you can do that by just having those built specific for you. So you know exactly what’s in them. You know the concentrations. You know the variants of everything in there. So sort of really important that they managed to bring the timeline. So now they’re the same, so time isn’t a factor anymore.
And then it’s cost. So, historically, custom’s extremely expensive. Custom of anything in the world is extremely expensive. Then, when you actually break it down, is it really? And so the platform that again Nucleus has managed to build is we can do it faster, often time is equal to money, and so that means that, automatically, is a bit cheaper. But then the idea that they could add every single thing that you need into the formulation. And so now there isn’t a need to go into a clean room to add additional reagents to it at point of use or anything. You’re not buying from all these different vendors. When you look at your base media, maybe that’s cheaper, but then you’ve got to add your protein sources, you’ve got to add cytokines, growth factors, things like that. It’s actually not that much cheap, so what we saw in that webinar, when we broke down some of the cost analysis, actually, proprietary off-the-shelf generally is about 33% more expensive. And you rule that out, that that might not mean a lot during development. But you start going into commercial scale, where you’re using liters upon liters t that actually an average CAR T therapy this is done off of. It could be about 4 million a year. So now the life cycle of your product, 10, 20, 30, 40 years, that really has a significant effect.
And so now what sort of we managed to show in there is that creating a custom media a, allows you to study single components — and that may affect your culture — allows you the IP, the knowledge of everything that’s in your media, so you can take that anywhere you want, you can have it produced by any manufacturer out there, and it can also save you a lot of money towards the end as well.
So the next bit that we kind of left at the end of that webinar was, “Okay, that’s great. I have no idea what I need to put into my formulation.” And you’re asking me and as a lowly engineer that I am, I know that they need some glucose and probably some amino acids and things, but I have no idea what concentration, what’s important based on the CQAs that I’m gonna get out there.
So that’s really [lost signal 00:16:19] and back over to Dave now to generate that knowledge.
Start of Presentation
David Sheehan: So we kind of feel like we’re at this pivot point as an industry, and scientists have been dependent on industry to provide them with proprietary off-the-shelf media and then do testing of multiple proprietary off-the-shelf medias and figure out which one works best. And it’s a very much a black-box approach. But what I think you heard from the first two webinars is changing individual components has a huge impact and, when you do it and you build the media from the ground up, you control it. It’s your formulation.
And so we’re proud to introduce NB-AIR. And the name NB-AIR comes from Artificial Intelligence Research. And, basically, what we’re doing is we’re saving you time. We’re putting time that you would use for purposes of researching, how to do iteration, how to optimize your media, and allowing you to use it for other purposes because we’re cutting the formulation time and allowing you to go into what is your cell and your critical quality attributes, figure out exactly which compounds you want in order to drive cell performance.
And it’s based on published papers. So everything’s from PubMed. Recommend those multiple formulations to you based on concentrations that are extracted from data analysis and then enhance your cell performance and ultimately your therapy by allowing rapid ordering and testing of formulations that are going to allow you to identify those key performance contributors to your ecosystem.
And so the idea is you may — by going through this process, it doesn’t replace your wet lab work; it just makes it a lot more efficient. And how are we doing that? There’s three primary components. One, we have a machine learning algorithm that searches PubMed articles for cell types, critical quality attributes, and compounds. We basically go through and parse out the conclusion of the research paper and score the article and the contribution to the critical quality attributes. And then we use machine learning, and the machine learning basically recommends formulas based on what the input was from the user.
And so what I’m going to do now is step you through what it looks like. And I’m going to stop sharing for a second and pull up the actual live site. Can everybody see this? Okay, so what we wanted to do with this was create a very simple user interface. And there’s a lot — don’t — understand that behind this, there’s a lot of work going on. There’s a lot of databasing and a lot of machine learning algorithms that are supporting this. But, right away, you see there’s four steps. You pick a cell type, you pick a critical quality attribute, you click your components, and then you output formulas.
And so, for cell types, we heard multiple versions. You can pick really any cell type. You can pick T cells and K cells, and each one is going to have a different distribution of components that have been studied based on CQAs. And, in this case, for this example, I’m going to pick NK cells. And you can also input in here what media you’ve been using historically. And, for us, that’s just helpful because it guides us on what to do. So, even if you’re using a proprietary media…
And then over here is a reference to the components and what they do to the CQA and an impact score. And I’ll explain a little bit about that. So, once you’ve picked your cell, you move on to the next step, and you can pick your CQAs. And we heard from the poll at the beginning that there were several critical CQAs. It sounded like probably viability was one of the top ones. And, again, you click on viability. You want to maintain or increase that.
And, what you see here, often this side is the side panel, is which are the components that most contribute to this critical quality attribute? And the green increases, and the red decreases, while the yellow maintains. And there’s information in all of that. But then what’s relevant here is that there’s actually a link to the research paper. So, if you want to know exactly what that research paper said, you can click on this link, and it’ll take you through to the research paper, or you can just review the quote that we pulled out.
And so the idea is that you can go in and select multiple components. The other one that we heard that was pretty big was proliferation. Usually people want to increase that. So you pick increase and probably cytotoxicity, given this NK cells, and you want to increase that. You can also rank the importance of each one of these. So you have pulldown menus on rank on each one of it. But you can see that, for cytotoxicity, all the components that have a contribution to it, and then the score, which in this case is set up to be the max of 100, and it’s not an indication of how impactful that component is on that cell as much as it is an indication of the number of times that that components been studied on that CQA and how many papers reference it. So it’s an indication of the strength of the component based on a review of the literature.
And so once you’ve picked all of your critical quality attributes and selected what you want, the next step is to move on to picking your media, your media base, and also all of your components. And so is the chat thing popping up on the screen?
Rachel Gauvin: Yeah, we have some questions coming in I’d like to discuss later.
David Sheehan: I know there’s a lot of chats going on, but… And, so here, you can select media. You can actually select the most cited, the second most cited, or what we recommend, or you can put your own base in here. And so for this circumstance, then you can go in and say, “Okay, what components do I want to search?” Well, clearly, in the prior presentation, IL-15 was big, IL-2 was big, IL-7, PDGF, and guanosine.
And you can see each one of them and where they’re mentioned. And these are all individual references of papers. And the papers pop up over here. So this is like a meta-analysis. This is giving you the ability to weed through thousands of articles and distill it down into something where you can know exactly which components you want to add to your media formulation.
And then once you’ve selected the component, the last step is to go to the formula. And so what we do in this step is we list up on the top what are your critical quality attributes. And then you can go in and select the formulas, but what we’re giving you is a recommended concentration based on the published papers. And so you can kind of see where are we coming out on each one of these. And what’s really nice about this is that you can name these. And so you can save that. And what ends up happening is, once you’ve looked at this, you can make modifications or save all these formulations directly to NB-Lux and then order them.
And so I’ll show you that functionality here. NB-Lux is our online portal. So now that you’ve got the four different formulations, this will take you over. The formula have all been saved. They’re now in NB-Lux, which is our platform for configuring and ordering media. And you can go in and open this one and basically customize exactly what you want. So, if you want to customize it, you can go in and make modifications while that price looks a little bit high. But, basically, you can go in and select exactly what you want.
And so once you’re done with that, you can save it. And that’s in your profile.
So now you can order these or modify the formulas if you want to.
Okay, great. So let me stop share for a second, and I’ll go back to the slides.
So, basically, again, to highlight, you pick your cell, you pick your CQAs, you pick your base and your components, you finalize your formulations, and then you export to NB-Lux. And you can order them in any size from two liters to anything up to 2000 liters.
And then I’m going to turn it over to Alex, and he’s going to talk a little bit about how we go about creating this analysis tool and what kind of learning modules we’re using.
Alex Klarer: Thanks, David. So when Dave and the Nucleus team first showed me this, you see all of the options you have here to select your cell, your current culture media basal medium that you’re using, the different media components that you might be interested in implementing or may already have as part of your media, and your desired CQAs, and the algorithm is going to give you a new formulation that you could try and test.
And the big question from that is how is it generating this recommendation? And what criteria is it using to determine what might be an increase or decrease or maintain on that web chart that you saw in the demonstration? And it turns out this is actually a very intuitive process and it’s — augments what we already do when doing a lit review on any topic for instance.
So, on the next slide here, what this confidence score is it’s a relative value of the potential of that component like Dave described. And, just like if you were to go through and review these papers, it weighs the contribution of that paper by the impact score of the journal it was published in and the number of references. And so you can imagine having yourself or a PhD level scientist spend hours to review hundreds of papers on this subject, and what they’re doing is implicitly creating this impact score. And they’ll use that to decide what components and the media composition is for the screening study that they want to run.
And the machine learning that Nucleus has implemented is providing that entire screening process upfront and saving you those hours of review, and allowing you to get an idea of where your focus should be before even starting that process. And you saw in the demonstration also that it provides you those papers. And so it guides you in the choices that you’re making, following the process that you would already be using. And, that way, you end up with a high confidence in the formulation that you end up with.
And you can pre-screen a weight, low impact papers, a low impact conclusions, without having to spend the additional time to do that. What it doesn’t provide is a quantitative magnitude of the effects. It only provides that relative confidence in the direction. It’s a qualitative look at what the impact of each component might be.
And so when you combine the Nucleus Biologics ecosystem, you see that the NB-AIR helps you with that initial screening process and determine what media you should be using, you should be experimenting with, at the beginning in the NB-Lux platform, then takes that learning and gives you a process to purchase and receive that media in a timely manner.
So I’ve been looking at some of the questions in the chat and “How are you speeding up the developments?” And it does start with the NB-AIR process to remove a large portion of that literature review process, and then the NB-Lux platform then takes all of that learning and gives you your pre-formulated medium in the same time periods that you would expect to get proprietary off-the-shelf media.
And then they also have the Krakatoa system, which is coming soon, that would augment your end users’ ability to store and have at their fingertips these custom formulations.
David Sheehan: So I think it’s probably worthwhile to — thanks, Alex — go through Q&A, and I might loop back and make sure everybody understands. NB-Lux has been in the marketplace for over a year. It is an active site. And you can order any media. And you can order from sizes from two liters to 2000 liters. So that capability exists and the two-liter ordering is typically weeks in order to get small lot on media, and that’s part of one of the key differentiators.
Do we want to maybe take questions now, or…
Rachel Gauvin: Sure thing. Yeah, we have a number of questions. I’ll try and get through them all but be respectful of time.
The first question here is The information gathered from research papers, does it take into account patent content that may have licensing fees around the concepts or component concentration?
David Sheehan: It doesn’t right now, and it’s looking at PubMed. But, historically, most manufacturers have not patented formulas. They keep them as trade secrets.
Rachel Gauvin: The next question is For additional cell types not discussed today, is there a critical mass of literature that must be available to provide useful recommendations? How easily can additional cell types be added, like macrophages, monocytes, and this like?
David Sheehan: Yeah, now that we’ve figured it out, it’s fairly straightforward to do that. And I think what you saw today, the run through on the NK cells, there was clearly a bug in one of the concentrations because of the price that came across. But, I think, for us, adding additional cells is really just listening to our customers on what’s important and what cell types do they want to see.
Rachel Gauvin: Thank you. Next question is Do you provide information on material sources for the raw components that go into our NB-Lux formulations?
David Sheehan: We do. And so one of the — and maybe I’ll stop share for a second and go back to NB-Lux. So you have an option. So I think this was the formula that we priced out, and I’m gonna guess that there’s something off on one of the concentrations here, but you can load this into the Customizer and probably see… All right, so we — 10 billion dollars. I’m gonna guess it’s probably one of the cytokines. And, yeah, sure enough, IL-2 is off. So you never would put that much IL-2 into a… but it’s probably more like 0.0001. Okay, that makes sense.
So now you’re getting the idea. All of these raw materials, you can select USP grade. And then if you want cGMP manufacturing, you can select it here, and it allows you to basically — will give you a full spec package. So you’ll know basically all the ingredients. And so you can not only select the grade, but you can also select the documentation level.
Rachel Gauvin: Great, thanks. Next question: Is there a possibility to search for specific cellular subsets, such as gamma, delta T, or regulatory T cells? And, if not, are there plans to extend the functionality to include the expanded function?
David Sheehan: Yeah, again, if anyone has cells they want to look at, we can always do that functionality. As long as there’s published articles, the machine learning algorithm has the potential to harvest anything that’s in the public domain.
Rachel Gauvin: Great. Back to documentation a little bit: Do you help clients that need a Drug Master File or DMF file?
David Sheehan: We do, yeah, and a great example is we manufacture a protein called Physiologix, and we will frequently create technical documentation packets for customers that are submitting for a BLA or any kind of IND.
Rachel Gauvin: Great. Switching gears and going back to the critical quality attributes, On the CQA metabolism, can you explain a little bit more about that? Obviously, metabolism is a large field, and this is much more involved than, say, viability?
David Sheehan: Hey, can you ask the question again? I’m…
Rachel Gauvin: Yeah. Can you talk a little bit more about the critical quality attribute of metabolism? It’s a larger field and more involved than, say, viability.
David Sheehan: Maybe, Roddy is in-, —
Roddy O’Connor Ph.D: — Yeah, so definitely one of my interests. So you’re exactly right. It’s changing every day, and it’s quite a rabbit hole. You could just dive into and drown basically. So I think there’s a great opportunity when you use this platform to identify kind of the latest and greatest cytokines.
Let’s say that we’re using here at Penn… so the field of adoptive T cell therapy is moving away from IL-2. And, okay, what’s the suggestions to go to? So you could use this platform to maybe get a hint of what people are using and not just that but what concentration will be optimal to use. So that’s always going to be a hot new area, the type of cytokines to condition the media during this ex vivo phase.
We have spoke a lot about the glucose. So it’s simple energy source for cells, but look at the complexity of altered levels. So this is ideal once again. It gives you all the insights and papers into do you want higher levels to boost proliferation, or do you want any trade off with quality? You saw the impact of mitochondrial stress. So there must be a sweet spot. What would you like based on the attributes that you want? So that’s just two prime examples: the cytokine conditioning factors and glucose levels. And there’s a ton more. You name it. The info will be there, and the insight will be there. So it’s just a great platform.
Rachel Gauvin: Thanks, Roddy. Question: Could your platform give recommendations for establishing a serum free formulation, given that most of the study published are based on cells grown in media formulations containing serum?
David Sheehan: Yeah, I think it will. What we’ve seen is that there’s in the — our work with the AI platform, there’s a lot of growth factors that are being indicated, and that’ll start to send you down a path of either reduced serum or serum elimination, once you understand the growth factors that are driving the cell performance. And so in the example that I gave, we picked some cytokines, but we also picked one growth factor, and there were other growth factors that were listed in NK cells. So I think that’s the benefit of the platform is it’s telling you all of the research that’s been done. And if you start to put some of these together in different combinations and test them, the idea is that this becomes a rapid iteration tool.
Rachel Gauvin: Yeah, and this one was addressed in the webinar somewhat, but I think it’d be nice to maybe summarize again. The question here is Classic media development takes normally years, how can Nucleus’s approach help media development? So if we just want to drive that point home?
David Sheehan: Yeah, I think, part of what this does is it shortens — I think the key with media development is figuring out what’s most important. If you want to test everything, you could be in a Do Loop on iterations. And one of the things — and, Roddy and I, we’ve worked together with our science team to develop formulations, and it’s knowing when to put your pencils down and say this is good enough. And I think what we’ve been able to do is figure out how do we use the AI platform to inform which components to focus on. And it’s easy to create the perfect formulation, but it may be 70 or 80 or more components, and it’s not manufacturable. It’s focusing on the things that are absolutely critical and getting rid of those components that don’t really contribute.
Roddy, any thoughts on that?
Roddy O’Connor Ph.D: I mean, I think we have — I totally agree. We’ve done a lot of work on it, and to address another — it hits on another question, maybe on optimal CAR T cell media. We’ve worked maybe a year and a half on this. So that’ll be coming out soon, NB ROC, so… and it’s… we’ve just been going through this exact process, looking at the factors that are present, what levels are they, and what will be optimal to confer these qualities to T cells expanded ex vivo. And like, you say, eventually, you do have to put the pencils down and test it out and really where the rubber meets the road, kind of, right? So…
David Sheehan: Yeah, and I think it’s worth saying we did that development prior to having the AI platform. So there was a lot of trial and error and —
Roddy O’Connor Ph.D: — old school.
Rachel Gauvin: Too funny. Great. Quick question here. Just, Do you sell CAR T cells and media together or separate?
Roddy O’Connor Ph.D: Yeah. We don’t. Well, we don’t sell CAR T cells, but the media that we’ve optimized for CAR T cells will be available from Nucleus Biologics, NB ROC.
David Sheehan: Yeah, and just so everybody knows, we’re going to put that formulation on NB-Lux, so you’ll be able to see the formulation. And this is part of being transparent and making sure everybody’s aware of it. And I did get a question: When will this site go live? It’ll probably be live within the next day or two. We wanted to do this webinar first so that everybody could understand exactly what it is. We’ll put this recording up online, so you can go back and look at it, and then the site will be going live.
Rachel Gauvin: Great, and then can you speak a little bit, Dave, about the pricing model?
David Sheehan: Yeah, the idea that we wanted to do was encourage use. And so it’s a price per year for use. And for academics it’s 950 a year and for industry it’s 5000.
The idea is once you use or sign on to it, you actually will have credit so that if you do want to buy media and transfer those formulas into NB-Lux, you’ll be able to buy the media and will credit you for the fee for access to NB-AIR.
Rachel Gauvin: Great. And then a question, Does NB-AIR cover only information on PubMed or published books as well?
David Sheehan: Yeah, we’re figuring out how to access everything in the public domain, and we’re also incorporating our own research. And, ultimately, what we want to do is build more of a community where the people who are logging into the system can also give feedback. So that’s kind of phase two, is to increase the level of functionality and create more of a knowledge base that people can tap into.
Rachel Gauvin: Great.
David Sheehan: Especially with the pace with which — I mean, Roddy talked about it that Greg Delgoffe’s glucose information just came out in December. And so things are happening so fast. We want to get that information into the hands of the scientist as quickly as possible.
Rachel Gauvin: Yeah. Uh…
David Sheehan: And I would also say we’re doing a lot of wet lab formulation work, and so if you need help with formulations, we do a lot of it. The way we do our formulation, our wet lab formulation work, is if you pay us for it, it’s your formula, you own the IP, we don’t own anything about it. And all the work that’s done on NB-Lux is scrubbing published literature or our own research that we did. So the wet lab formulation work is just a service that we offer, and we’ve built out quite a capability around that.
Rachel Gauvin: Yeah. Dave, if you want to put up the Q&A slide again, there’s a little bit more information on how people can access that, which might be useful. Great. And one other question we have that came in is How accurate/optimal are the exact values of each component in the formulation? Is the idea that the platform is best used to identify these components, and we go through several iterations to find the best concentration of each component?
David Sheehan: Yeah, it’s a great question, and it’s one thing we’ll learn as we get into it, is — Clearly on that NK cell example, the IL-2 concentration was off, and we’ll fix that. But the idea is those are the concentrations that are listed in the article, and then what we’ll recommend and start to build algorithmically is the ability to recommend doses around that level based on a more in-depth view of what the concentrations are that have been used in the different papers.
Roddy O’Connor Ph.D: Yeah, I think that’ll save a lot of time for people, right? If they can get close to a ballpark figure, that they could then maybe optimize themselves, or at least they know what’s — no reviewer would look at their paper and say, “Oh, you’re way off. This is exactly what people are using for IL-2 for T cells. For IL-7 and IL-15, it’s 10 nanograms per mil.” So it’s great to have that, even if you’re just an information addict, to be able to look at the platform and see that rather than diving through so many papers and scratching your head, “Well, is this right. Is that right?” So it’s a good help.
David Sheehan: Right, and I think it’s a great — and coming back to the glucose analogy, Roddy, it’s like, if glucose is truly at really low concentrations and at really high concentrations not good for mitochondrial stress, where is that sweet spot? And we know that a lot of the proprietary off-the-shelf medias have very high glucose levels and not physiologic. So definitely figuring out where is that. Is it five millimolar? Is it 10 millimolar? What is that level that’s really where you want to be?
Rachel Gauvin: Yeah, and then there’s a question kind of on the logistics of getting access to the account, which I can answer. In order to access NB-AIR, you have to have an NB-Lux account. So we encourage everyone to go in even today to sign up for NB-Lux. And then, from there, in a day or two, as Dave said, we’ll be ready to add on the layer on the feature of NB-AIR. So that’s kind of how everything is interconnected.
I think that addresses all of the questions that we’ve been given. If I missed anything, feel free to type that into the Q&A or chat, but like we said, we would love everyone to sign up for NB-Lux so that you can get access to NB-AIR in the next day or two. And we’re really excited about this launch and really appreciate your time in joining us.
David Sheehan: Thank you. Thank you, everybody. And thank you to the panelists.
Roddy O’Connor Ph.D: Thank you, guys. It was great.
David Smith Ph.D: Thanks, everyone.
Rachel Gauvin: Thanks, everyone. Have a great day.
Alex Klarer: Thank you.