Executive Interview with Joseph Wu, Professor & Director, Stanford Cardiovascular Institute, Co-Founder, Greenstone Biosciences

[0:00:00] Transcript

[0:00:22] Hi everyone. Today, I will be interviewing Dr Joseph Wu, the Director of the Stanford Cardiovascular Institute, a cardiologist, and also the Co-founder of Greenstone Biosciences.

[0:00:44] He has extensive expertise in developing organoids and new alternative models for drug discovery. So welcome Joseph. It's a pleasure to have you here today. And yeah, I guess no problem.

[0:01:06] And I guess the first question from me is, as someone who's an expert in the cardiology field, what pieces have you observed in the field of cardiovascular research over the last few years.

[0:01:28] I think in the past, cardiology has been more physiologic and functional, meaning that we want to know where the arteries block. We want to know what the cardiac function is, and how do we come up with drugs to improve the cardiac function? How do we come up with better stents and pacemakers to help the cardiac to help the heartbeat?

[0:01:50] But right now, with the evolution of genetics, you'll see that more and more medications that we come up with that targeted to a specific pathway. For example, right now, you know, the hottest drugs are GLP drugs, the whole mechanisms, it's all worked out and how it benefits cardiovascular disease that's actively being pursued. Also, the drugs are PCS canine.

[0:02:13] Again, it's all based on genetics, and then understanding the disease pathway allows the big pharma to come up with better drugs that can target our patients. Thank you very much, and you're using IPSCs to generate organoids, and in our previous conversation, we touched on the importance of genetic history.

[0:02:35] So how would you say that you know having knowledge of genetic history has allowed you to uncover, potentially uncover new targets.

[0:02:57] Yeah. So I think traditionally, when we think about how we do biomedical research in the past, we use a standard cell line. It could be HeLa cells, Cho cells, HEK cells, whichever cancer cells you're working at, or use some type of mouse that you tested drugs on.

[0:03:19] But when you think about it, if you do 1000 experiments on a single strain of a mouse. It's essentially N equals one because it's an inbred mouse. And the deviations that you see is not due to the genetic variability that exists among humans. The deviations that you see is actually due to a technician's tailbone injection skill or intraperitoneal injection skill or [unclear].

[0:03:41]  What I really want to know when I develop a drug is, how does that drug respond in a variety of different patients at different ancestry, genetic background, sex and so forth. That's otherwise impossible to recapitulate, capture using a inbred mouse string or using a single transformed cell line. This is the reason why it's important to create a large repertoire, large bio bank, of these iPS cell lines that comprises of different genetic backgrounds.

[0:04:03] On top of that, if you're interested in diseases that are related to, for example, rare orphan diseases. It's also important to use these iPS cells from rare orphan diseases to first test your drugs first, rather than using our patients as guinea pigs at the test at the drugs on and so for a variety of reasons, it's important to understand the genetics.

[0:04:26] It's important to have a test bed to validate your genetics, and the test bed will be your IPS cells, your organoids, and it's important to have better technologies such as AI software to help you accelerate the whole progress.

[0:04:48] Thank you very much. And I guess you've kind of touched on some of the reasons why it's important to move towards organise and raise awareness around these and new alternative models.

[0:05:10] Do you think there's any resistance from some pharma and academic communities, and if so, do you think it will be possible to overcome this?

[0:05:32] Yeah, I think right now this day has been some initial resistance, and mainly because of lack of familiarities to these platforms. Symptoms you know, as you know, Pharma has been used to use amounts of strings in standardised cell line for a long period of time. And I think another reason may be that they're not used to the variability that exists, right?

[0:05:54]  Because when you test a drug, you want to have a standardised and a single cell line or a single mouse or strain, and it's all standardised, and they're not comfortable with the data being, quote, unquote, all over the place, but the reality is that in clinic, in the hospital, the clinical data will be all over the place because of the genetic background differences among all of us.

[0:06:16] If I give a medication, a statin, a beta blocker or ACE inhibitor, to 20 different individuals, I'm going to get 20 different responses because of our differences so but I do think that you know, with these new mandates or recommendations by the NIH and also by the FDA, more pharma folks will be pivoting toward using these organoids and stem cells, and also more academia folks will be using these as stem cells and organoids, because, after all, a lot of the products are initially developed in academia that then get carried over adopted by pharma.

[0:06:39] Great. Thank you very much. And I asked then also in our previous conversation, we touched on the use of AI, and you mentioned, it's useful, but not good enough on its own yet, particularly in like target validation. So I guess in your experience, can you expand on where AI has helped you the most and where it has fallen short?

[0:07:01] Yeah, I think in my own opinion, I think AI can help you. Help point roughly the direction, the direction meaning help you point toward North, South, East, West, right. But once it points toward North, finding the right target at the north side really requires you to have good understanding of the biology, about the pathophysiology, about the mechanism of action, the disease that you want to go after.

[0:07:23] Because for any given disease, there are a lot of targets, and there are a lot of big players, medium players, and small players, and all these players are interconnected and trying to sort out what's a big target, what's a medium target to go after that gets harder, and that really requires a human intuition and judgment, rather than just, you know, basing it completely on the AI.

[0:07:45] For example, we recently ran a docking screen, and the AI software predicted 20 compounds for a anti fibrosis target that we're going after. Out of the 20 compounds, 17 of them with that there's just no good three of them that worked. They're not as good as it's work I can get from Sigma.

[0:08:07]  But it doesn't mean that the AI failed. It just means that the training data set that the AI had to derive these 20 different formulations were not as good. And so therefore we just go back to the AI and say that, hey, you predicted 20, 17 were really bad, and for these 17, make sure you remember these designs. And you know, train yourself that these designs don't fit for this particular molecule that says pocket, this particular structure and these three so, so make sure you kind of put it into algorithm that they said three or so and so you have to train yourself.

[0:08:30] And that's why, I think for somebody to say that I can just use AI to come up with a drug, and this drug will work, in my opinion. I think it just makes no sense.

[0:08:52] Yeah, perfect. Thank you very much for your insights. And you're also the co-founder of Greenstone biosciences, you recently raised series A funding, and I believe you have drugs advancing into Phase One trials. I guess. What would you say sets your company's approach apart from others in this particular space?

[0:09:14] I think, you know, many companies have their own strengths and weaknesses, and we certainly also have our own strength and weaknesses.

[0:09:36]  I think the thing I want to, I do want to mention, is that even though we are a platform company, I think because of. Of our company's background and the founders and the scientists that we have, we kind of have a broader understanding of what it takes to get to a drug, meaning that really, when it develop its work, it's going from A to Z, and a is your platform, and then you then use your platform to come up with better target, better drugs and so forth.

[0:09:58] But I would challenge it by saying that it's actually Z to A meaning that you got to understand what is the need. Because you're not going to go after the disease in which there are already 30 different drugs, and the market is already very crowded. You're going after disease in which there's a met need, not a lot of drugs out there. You feel like you can make a difference, and then go backward and say, Okay, what is the disease? And I'll just merge a and z together somewhere in the middle, meaning, what is the disease? What is the target? How can we use AI to come up with the drug designs?

[0:10:20]  How can we, once we synthesize the drugs, how can we test these drugs in our organoid models? How can we refine these drugs, and at the same time, while you're thinking about that, you're already testing these drugs in terms of its safety, its PK/PD, its absorption, distribution, metabolism, CMC, a lot of times all these drugs that you have, they all drop out because it just poor absorption, poor penetration, poor solubility, whole bunch of these factors that's not predicted just because you have a quote, unquote, AI software upfront.

[0:10:43]  And so I think we kind of take everything into perspective and then that's how we kind of decide we're going to go after this drug, that drug.

[0:11:05] Yeah. Great. Thank you very much. And I guess for if any early-stage biotech founders that decide to watch this interview, What lesson would you say you've learned about attracting good investors early on and  building credibility around your company?

[0:11:27] I would say you got to have a clear message to the investors in terms of what you want to do, you got to be able to kind of distinguish your startups from the others. But at the same time, to me, I think it's very, very important, at least to me, it's very, very important that we don't over promise and under deliver. And because I think a lot of times, I mean, I'm coming from academia world, right?

[0:11:49] We will tend we tend to be more conservative, we tend to under promise and over deliver. And so for us, I think if we're going to go after something, it's after a lot of careful thoughts, considerations, and we're going to say this is our target.

[0:12:11] We're going to go come up with small molecule, come up with some kind of therapeutic going after this target for this particular disease. So it's a lot of communications with the investors and trying to tell them what your vision is and they have to believe in the vision. And luckily, we have investors who believe in our vision are very patient to us and are very supportive to us.

[0:12:33] That's great. Thank you very much. And I believe, Joseph, you have had some experience in the regulatory field. So, on that note, how do you think regulatory agencies can better support or validate IPSC based and organoid platforms in drug discovery and development?

[0:12:56] Yeah, so I've served on the FDA cell and teaching the IP advisory panel, I think since 2017, so we've seen a lot of these phase three trials that came up to our panel for going yes or no in terms of recommendation.

[0:13:18] And I do think that the FDA recommendation for new, alternative methodologies. I think it's a good one. But I'm also a believer that we should take all technologies into our playbook and then figure out what's good for what. I don't think stem cells and organoids can completely, replace the animals.

[0:13:40] I think it's just part of a pipeline in which, after you do certain validations, and let's say, for example, you got five lead compounds that pass your toxicity safety in animals and cell lines, and before you. Of the clinical trial, you don't know which one really, really works.

[0:14:02] So in that, in that context, you could take all five and add a fraction of the cost, of cost, put it on 1000 iPS cell patient organoid from, you know, 1000 different patients, and then show that, for example, drug number one doesn't really work.

[0:14:24] It works in two to three out of 1000 drug number two works in five out of 1000 but drug number three works reasonably well in 200 300 out of 1000 that's the one you probably want to go after, because you can kind of consistently see it in this large pool of IPs, cell organoids from different patients.

[0:14:46] And this is the whole concept of clinical trial in a dish, a chip in a dish that we've been advocating to the FDA and also to the to our academic colleagues for a long period of time.

[0:15:09] Yeah. Thank you very much, Joseph. And final question from me is, we're obviously very excited to welcome you at Oxford Global's Discovery and Development us event. So what is the one key message you would like the audience to take away from your presentation?

[0:15:31] I think they are exciting new tools that are coming up. And I think during my presentation, I will talk about how we integrate genetics. You always got to understand the genetics of the disease, stem cells and organoids, because it gives you a test bed to validate your genetics, and then AI, because it gives you large, higher throughput screen. And I will talk about how we integrate all three to come up with some of the pipelines that we have at Stanford and also Greenstone,

[0:15:53] That's great. Thank you very much. Great takeaway message, well, thank you, Joseph, for sharing your insights with us, and it's very clear that you have a lot of expertise in organoids, AI and drug discovery, and we very much look forward to welcome you, welcoming you at the event taking place at the start of October.

[0:16:15] Yeah, thank you, Lucia, pleasure to be here.

[0:16:37] Thank you very much.