Pushkin.

There are these amazing cells in tiny human embryos.

The cells are called pluripotent stem cells, and they're amazing because they can become any kind of human cell, a red blood cell, a skin cell, anything, any cell in your body.

But pluripotent stem cells only exist for the first fourteen days of embryonic development.

After that, they're gone forever.

At least we used to think they were gone forever.

And then about twenty years ago, researchers figured out how to take regular cells from adults, blood cells or skin cells or whatever.

Take those cells, bring them into the lab, and then turn them back into pluripotent stem cells.

These cells are called induced pluripotent stem cells, or ipsc's, and the possibilities they present for human health are both kind of obvious and awesome.

If a person has a disease, you could use that person's own cells to grow any kind of new cells that they might need.

You could grow new brain cells for Parkinson's disease, or new heart muscle cells for heart failure, or new bone marrow cells for leukemia patients.

It has taken a long time to put that dream into practice, and in fact, it's not really solved yet, but researchers are getting close.

A bunch of clinical trials are now underway using iPSCs to treat everything from Parkinson's disease to cancer to macular degeneration.

But even if these clinical trials are successful, there will be another problem to solve.

Turning a patient cells into iPSCs and then into whatever kind of cells they need takes months of work by highly trained science.

It costs hundreds of thousands of dollars for each patient, and so even if those clinical trials are successful, the process of making the seals will still be too expensive and too labor intensive to ever benefit you know, millions of patients.

So if the dream of IPSS is going to come true, somebody needs to figure out a faster, cheaper way to make them.

I'm Jacob Goldstein and this is What's Your Problem, the show where I talk to people who are trying to make technological progress.

My guest today is Nabiha si Client.

She's the co founder and CEO of a company called Selino.

Her problem is this, how can you make iPSC therapies quickly and cheaply?

Before she got into the induced pluripotent stem cell business Nibiha was studying to be a physicist and she loved physics.

Was going off to get a PhD in physics at Harvard, but just before she started school, her grandmother died, and she told me that had a really profound effect on how she thought about her research and her career.

My grandma died due to severe diabetes that was not possible to control with insulin and other medications, and that just I felt really helpless.

I felt helpless, and I felt this urge too.

Okay, I don't feel comfortable going down this intellectual curiosity path of becoming a physicist, and like, what can I do to build better tools?

There must be better tools that are necessary if my grandma died pretty odd, you know, due to diabetes, and there was nothing anyone could really do about it.

So she started thinking about how to use physics to improve human health.

For her graduate work, she figured out how to use lasers to make tiny holes in cell walls, and she started talking to all the researchers that she could to try and find useful applications for her research.

Eventually, someone told her about induced pluralpotent stem cells, about iPSCs, and she realized that she might be able to use lasers to help automate the process of cultivating iPSCs.

IPCs are extra complicated and special.

I thought, I like to call them special because you actually have to go and like scrape bad cells away with a pipe pedant.

It's super artisanal.

So you're basically having these brilliant scientists looking under a microscope, holding cells in the dish, and then scraping with a pipe petter.

And they're literally working ten ten hours a day scraping cells, not taking vacations, trying to get to work during the craziest snowstorms.

Because if they don't show up, that run dies.

Uh huh.

And presumably, I mean if you think of actually getting it to the point where you can treat patients, it would be sort of impossibly expensive slash small scale, right, Like you could never do that for one hundred thousand people or something, right.

Heart, we don't have enough scientists in the world.

And of course that like highly skilled labor means like even more expensive than expensive drugs usually are.

Right.

Presumably that's right.

And I think the cost estimates have gotten down over the past couple years because people have figured out how to do better biology but we're still in the hundreds of thousands of dollars.

So you're setting out to solve this problem of how to grow iPSCs at scale, and in particular right this problem of getting rid of the bad cell colonies without disturbing the good ones.

How do you figure out how to do that?

So one of the things we did at the time was we spent quite a bit of time with biologists, so we were trying to understand their workflow.

What do they do, what do they do in the lab, what are they looking for?

How are they doing their hand gestures?

And there's a lot of tilting, there's like tapping.

This is the artisanal sort of separating the sales process exactly.

The entire IPS manufacturing process can be on the order of three to four and there are many different parts to it.

And certain scientists have certain features that they see by eye.

So some will describe smiley faces, some will describe other features that they're seeing.

Meaning like they look through the microscope to distinguish the good from the bad.

That's right, and they sort of know it when they see it.

That's right.

And you have a whole range of how good our scientists are globally.

So the best scientists are sitting at inside cleanrooms looking at these cells and trying to decipher the best cells.

And the stakes are high because you don't really get in the way they're manufacturing their cells.

They don't get to necessarily test the cells until you've done the entire process.

Yeah.

So, just to be clear, like this is a disaster for drug manufacturing, right, like not even for safety, but just like it's never going to work, right, It's never gonna work at scalesh Sure it could work for research for a while, but like you or somebody like you needs to come along to automate this, right.

I'm so glad I never thought about how hard it was going to be and just went for it.

If I know how hard it was going to be, like startups are hard, but I think we had this curiosity.

We're a bunch of physicists, We had energy, we were passionate, and my co founder Marine and I we are very passionate about being in medicine and using our physics knowledge for medicine.

So we were like, Okay, why don't we try why don't we figure out how to do it?

And between twenty twenty and now, so many things have happened on the science and technology front at Seleno that I would label as impossible.

I think there were at least the first few years.

Of that phase, I was like, I don't know if any of this will work, and it was hard for me to come to terms with given that we had raised a big round.

You know, a lot of people were talking about.

This, but I also felt I kept telling my team, it's like, we have to give it our best shot, because if a team like ours isn't brave enough to try to go after this, this might not be resolved for a few decades.

And what that means is we are at risk of falling into this pattern of pushing through complex manufacturing, of selling gene therapy products, even getting them through a phase through approval, but once they hit commercial, they're not meeting the patients.

Meaning even if it works with the kind of technology that existed before you came along, even if it works therapeutically, it'll be so sort of bespoke and expensive that most people in the world aren't going to get it even if they need it.

Is that what that means?

That is what that means, And you know iPSCs, it's still to be determined.

We don't have any first approval, so let's see how that goes.

But it is the most complex manufacturing process that I've seen so far.

Other areas, other examples I can point to are cart therapies, Cancer therapy is curative, incredible.

This is another cell therapy and the domain of like let's take cells and develop them and give them to the patient.

Right therapy is the sort of the signal achievement of self therapy so far.

Right, it was.

Huge, and that was happening that approval was coming up when I was starting Seleno, so it actually was very motivating to see, Wow, now we can actually reach the point of curing previously in kurbal disease.

But it's been interesting to see about the cart space.

Every year, the maximum number of patients being dosed with a CARTI on the year is on average still ten thousand patients annually.

Which is a small number.

Which is a small number, what hundreds of thousands or millions who might benefit from the tree correct.

And now the scope is expanding to solid tumors and autommunity, So like, why are we stuck?

We're stuck in this number.

There's an infrastructure problem.

It's a scale problem.

It's a lack of manufacturing at scale.

That's correct, and so I do feel very strongly that it's important to push forward trials and build for scale from the get go.

But it's hard because it's tempting to take shortcuts and like, oh, we could do this and this will be much faster.

And the question I'm always asking my team is Okay, great, so let's take a step back, but how will this solution?

Could this address a million patients annually?

Yeah, because there are a bunch of sort of cluges where you could make a thing that works and it's more efficient than what people do today, but it's still kind of clue g and artisanal and not great to scale.

Like, that's the trade off.

That is the trade off, and that's an everyday trade off.

And I feel very blessed that we've been at it for quite some time, building the different layers, and we're at this point now where we're building our closed system.

It's called Nebula, Okay, and it's it's it's the next version.

It's the version I never imagined even when we were building the laser based technology, which we call the optical bioprocess.

But the idea is in order to ultimately scale down cost.

You know, it's always great to run high precision manufacturing, so we have that down, but the cleanroom costs are still.

High because again you're dealing with living cells and they can't get contaminated because that would be disastrous.

So you have to have a crazy super sterile environment.

And we're doing many patients, and you want to parallelize as much as possible, and you want to protect all the patient samples from each other.

So what we're building now, which is even harder but exciting and we're making some great progress, is the closed version of this so building cassettes that could ultimately be the clean space that all the manufacturing happens.

Is the size of your iPhone, which is very exciting, huh.

So it's basically, the size of your iPhone is a tiny clean room where you are cultivating induced player potent stem cells.

It all happens inside that little box.

That is what's being designed.

I just want to put lots of little stars next to it, because this is all in collaboration with the FDA.

You know, the type of vision we're portraying is not how cell therapy manufacturing is happening today.

So we're gonna work with the FDA and we're going to present data and we're going to work with our clinical collaborators.

But yes, wouldn't that be awesome because then we can we have these autonomous systems with close cassettes.

We don't need to establish high grade cleaner So you get a ton of flexibility on setting up manufacturing in places where you might not have clean room, you might not have academic centers of excellent, you might not have the workforce.

Yeah, Like the dream is it happens inside a machine essentially, like you buy the machine and all the artisanal knowledge and the clean room and everything is in the machine is embedded.

In the machine, intelligent machines that make your best them cells.

Yep.

So that so we've talked about the past, and now we've talked about the dream for the future.

Let's talk about the present.

I read Is it right that there's a Phase one trial in Parkinson's that you're involved in?

What's with that?

Yes, we had a very exciting year.

We found our first clinical collaborator.

It's the Parkinson's Cell Therapy team at mess General Brigham and it's been great to work with them.

They have a phase one running where they're making the patient's own IPSS and their own neurons, and these neurons are being transplanted into the brain, okay, and we're working together to do the transfer of the manufacturing into our automated platform.

And there's a lot of back and forth right now on getting that collaboration up and running.

And so just to be clear, that trial exists now and they're currently doing it in the artisanal, old fashioned, hard and expensive way, and the hope is that you will come in and do it in your faster, cheaper way.

That's exactly right.

They're incredibly brave scientists and clinicians and they've put in the hard work over the past few decades to get this bioprocess up and running, invent new surgical techniques, design the clinical trial, and yeah, we want to support them to scale through the trial and then how does it get into commercial So it's been really exciting.

And one of the things I love about this collaboration is we're geographically very close because we're in Cambridge and they're across the River Master.

I think it's like a mile or two from us.

So the ability to connect dots has been incredible because the manufacturing is happening inside the hospital.

Interesting, like you're building a machine in the hospital.

We will be deploying the machine in the hospital.

Yes, the machines are being built at our headquarters.

Now I shouldn't say built.

Yeah, so you're going to drive put the machine in a truck.

How big is the machine by the way, you know.

Maybe like one or two refrigerators.

Yeah, so that's how big they are.

And the cassettes are about the size your iPhone.

And initially we'll be deploying the machines to be doing making sales for one patient at a time, and then as we build more evidence, we'd love to make more and more patients on each machine.

And will that be the first time your sales are going into patients.

Yes, I believe.

So.

We do have two other amazing collaborators, one in South Korea.

They're working on a peripheral artery disease.

They also have a phase one trial.

They're going to have a US expansion and then a spinal cord injury company as well.

But yeah, the mass General team is doing great.

We're also leaning into collaborating with the FDA We've had a very positive experience working with them.

Earlier this year we were granted our Advanced Manufacturing Technology designation for iPSC generation and it's I think that maybe that's the biggest surprise of the year at how technology forward the FDA is and they've been paying attention.

They've been asking great questions and all of our meetings.

They have a full stack AI team that is fully tuned into how our models are being trained, like what's what's behind the hood.

So it's really really compelling.

Yeah, because I think the last ten years of selling gene therapies have been transformative in terms of curative medicines, but everybody is missing the impact of.

Scale, like including the regulator.

So they've now taken upon themselves to sort of help the field think about scale and work with technology companies like ourselves.

So it's incredibly it's special and.

Just briefly, like, from say the patient's point of view, how will it work?

Simply?

Yeah, I mean you know, we can take the mass General Brigham.

Yeah, in that case, what will happen?

In that case, they'll be they'll see a physician and they'll get a diagnosis for their disease.

And in this case it's Parkinson's.

Then they'll get a prescription that says you're going to get an atologous cell therapy.

And meaning your own cells.

Yes, yes, exactly, personalize your own match with your DNA.

And then they'll show up to or maybe it's the same day they get the prescription and maybe they're at their doctor's office, but they will have to either provide blood draw or a skin biopsy.

I can imagine really far into the future.

It could be hair cells, it could be saliva.

But for this one, yeah, they take a little bit of your blood or a little bit of your skin and then.

And then they say, okay, we'll schedule your surgery and come back in a couple months and.

We'll be ready to go.

Essentially, you're starting with the patient's cells, their skin cells or their blood cells.

You're ending up with brain cells that match their own brain cells.

What part of that transition happens inside your machine automatically.

Yes, So the getting to really high quality iPSCs is the first product we've established.

We're establishing end to end, Like.

You put the whatever the blood cell or the skin sell in the machine and out the other end comes a high quality ips is it really like that.

It's lots of AI and lots of biology and lots of fluidages, but it is kind of like that.

Yeah, okay, it's true that part is automated.

That's really what I'm asking, Like.

We're working on it.

Yeah, it's and it's going to be automated.

There will be human human experts in loop as always, and there will be end QC.

But the day to day operations, I mean, you know right now, like Exeleno, things here are running automated and they run pretty much twenty four hours a day, so they're like, you know, proming imaging, fluids, laser processing because the cells need different actions at different time points, so a lot of those things we've established to be algorithmic.

Yeah, it is automated.

You know, we have a we have a very small and mighty team of bio engineers, but we definitely do the work.

Of ten fifty x more.

I would say, yeah, through automation, through.

Automation, that's right.

Yeah, we'll be back in just a minute.

One quick note before we get back to the interview.

Near the end of the conversation, you will hear Nabiha mentioned something called allogeneic therapies.

Those are therapies where ipsc's induced player potent stem cells are developed based on generic cells rather than based on a patient's own cells.

Those are easier to cultivate, but in many cases they have drawbacks similar to organ transplants, because patient's own immune systems tend to reject those cells.

So I just wanted to clarify that in advance.

Okay, back to the interview.

How are you using machine learning or AI in your automated process.

The way we use machine learning and AI is we take a lot of photos of all the cells every day.

How are they doing.

We've trained a bunch of algorithms in Google Cloud that tell you, oh, this is a good stem cell, this is not a good stem cell, this is good density, this is not good density.

And then those algorithms feed into other algorithms that make decisions on what to do with the cells.

So that your expert scientists don't have to sit and make all these decisions, and it's hard to make when it's at this massive of a scale.

So they can review what the algorithms are doing, they can intervene it whenever they want to.

I think like now, a very timely a similar field would be self driving cars.

You know, there's just a lot of imaging that's being used, and then the car can make its own decisions.

And at least in a tesla you have you have a driver and they can override any time.

Wimos are running autonomous.

So that's what we do.

We use imaging to help with all the decision making your manufacturing.

So it's largely pattern matching, right, which is essentially what the expert scientists are doing.

As you said, some of them talk about whatever a smiley faces something, it's just because the individuals have seen whatever thousands have done it thousands of times, and it's a it seems like a good thing to use AI for, right, like a classic like, here's lots and lots of good ones and lots of lots of bad ones.

Now here's a new one.

Pickwa are the good ones and one of the bad ones?

It's that you're absolutely right.

So what's great about it.

Usually if humans can see something by eye, we're able to trend an algorithm to do that.

So that's real number one.

And then the second thing that's very exciting about what we do is this time series data.

This is really important because you can really start to draw patterns through time and even figure out predictions.

You can go back in time and predict the future.

And our algorithms have been able to because we've fed in a lot of time series data, and we've fed in also genetic data of how these cells look.

At the end, these algorithms can go.

Back into the early stages of the process say actually, this one is this cell or this colony is going to be bad, so you might want to eliminate that.

So, just to be clear, time series data is like we can think of it like a time lapse video of the life of a cell colony, and so then you learn patterns of like if it is evolving in a certain way or developing, I should say in a certain way, it's going to be good, or if it's developing in this other way, it's going to be bad based on the past.

Is that what you mean?

That's exactly right.

You just made me think of like Netflix.

I don't know why, but you know, like different movies, like I know what I'm going to get when I'm watching a rom com or I want to watch a murder mystery.

So you have the but you know, not every rom CAMM is the same, but I know how the story should flow.

And I'm making my choices.

Yes, so it can start to make those kinds of look in the crystal ball and it just increases the efficiency and we still do the end processing is the same, and the more data we see it in, the better they get.

So it's exciting.

Data is important.

Data is important for everything in AI right now, and it's no different for us.

And I should add the data that we generate a lot of our friends are generating in biotechnology companies.

It's very expensive data.

So we do a lot of hacking to figure out what is like the optimum data said that we can take that it's cost effective and like timegated and like we're not losing any resources.

We don't have access to infinite amounts of data.

So okay, So that's that's where you are as a company.

You talked a little bit before about iPSC therapies more generally, right, but let's return to that now for a second.

So, like, have any iPSC therapies been approved by the FDA?

Not yet?

Okay, The first approvals are going to be conditional approvals are going to be in Japan, okay, And you know, I think Japan is a very passionate about iPSCs, given that they want to know about price, so they've been working hard at it when other nations and countries.

Maybe slowed down.

I started a little bit, you know, just being like, you know, not.

Sure what are those therapies going to be.

Yeah, Parkinson's, there's heart disease.

Those are the two that I think will be up first.

And then there's just incredible trials running all over the world around vision laws.

I mentioned spinal cordanentry diabetes.

Yeah, there's some really interesting programs out of China where it was an autologous patient derived pancreatic cell transplant that they carried through, which was quite incredible.

Yeah.

So I do think the volume picks up and sort of creates even greater urgency to start putting all the pieces together and getting to scale.

Urgency, because once people start figuring out therapies that work, there will be a need to actually make the cells, which is where you come in.

Yes, make the cells, scale them, and give therapy developers options because right now a lot of them are budget limited, resource limited and can only dose one patient every two years.

Wow, just because it's so expensive to essentially make the cells.

And high failure rate, it's not a high yield rate, and they don't They're understaffed, there's been budget cuts, there's just a lot of problems.

So it would be great to have even more trials running.

You know, I think until Phase three trials happen, it's really hard to know how the trial's going to go.

We just don't have enough volume right now to have enough shots on goal.

Well, and to run a phase three trial, you kind of need the automation, right, I mean Phase three trials tend to be quite large, a lot of patients, and if you have to have scientists making sales by hand, it's going to be hard to run a phase three trial, right.

That's exactly right, which is why we haven't seen any ATOLL news programs get that far yet.

But the ALO ones are getting there, which is exciting because it builds evidence for the mechanism of action.

So it's easier to make sort of generic cells at scale as opposed to sort of bespoke sales for each patient.

That's right, That's why the alloy So Okay, if things go well, what for you and the field, I guess you want you need both right, people need to find therapies at work, and you need to be able to sort of implement those therapies at scale.

Like what will the world look like in what is the right amount of time to say ten years?

It's five enough.

Will the world look different in five years?

In five years, I do think the world look will look quite different for Parkinson's pace And that's incredible because it is a pretty horrible disease that leads to lack of independence.

Just it's just sad to see what patients have to go through and have lots of family members on our team who.

Have Parkinson's.

So yeah, you're going to go to your doctrin and say I want this this therapy and that will be an option.

And I think we will see more of those.

In ten years.

I think we'll see at least five more diseases where there is one allogenetic therapy that's available and a lot of the autologous trials are getting into phase three in a scalable way.

And what's what I'm hopeful for in the next five years.

My mom just turned sixty, and in the next five or ten years, I don't want to have to worry about her diabetes all the time, and I would love to have the option of having her own cell replacement to manage her diabetes.

That would be incredible, and.

I think that that is happening, and I think the questions around how scalable will it be and where will we get it and who's.

Going to make it?

I mean, I think people are going to figure this out.

Even this year.

This year was not happier for biotech and biopharma.

The markets were down.

There's a lot of concerns around scale and investor returns.

But I still find it incredible that gene cell therapy regard of medicine, companies and scientists are still chugging away.

We're still getting into trials and brute forcing it because everybody is so passionate.

I think the passion really comes from creating a paradigm shift from treating symptoms or accepting the status quo of a disease trajectory to wow, can we reverse disease?

Can we get to curative medicines?

And that collective will pay is incredible And I think a lot of us have experiences around aging and loss in our families, so it's driving this movement.

And I think in the next couple of decades, we'll have lots of humans on Earth.

We're going to be above sixty five and getting into eighty.

So this does become an economic concern as well.

So how do we keep everybody healthier for longer and using everybody's own regeneration, their own cell tissue and even organ replacements.

I'm very up.

I just tend to always be optimistic.

That's what gets me up every day to work on these hard problems.

So I'm quite excited, and I think what I'm encouraging myself, my team, and my friends, like, let's keep the optimism high.

Let's problem solve together.

We don't have to do this alone.

They'll be back in a minute with the lightning round.

I want to finish now with a lightning round.

Okay, were you aware when you chose the name of your company that there is a personal injury lawyer who advertises a lot whose name is Seleno?

Yes?

Ish, not really so Yeah.

I think when I named the company, I wasn't clear that anybody was going to care about us, honestly, Like, we didn't have a website, we just incorporated.

We had a terrible logo that I made on PowerPoint, so it became more obvious a lot later.

But I knew it was an Italian last name when we put that way, So I did know then, and I think Seleno and Barnes that person is an Italian last thing too, Cellino.

Did anybody ever call your company?

I did, like happen like maybe two percent of the time.

But the way I got to the name was cell and Innovation Selena and it is also a star in the Pleiades star cluster and says it's.

A nod to my astronomy.

Love.

You called your mother's approval of your curry a proud moment, and so I'm curious, like, what is the secret to making a curry?

Your mother approves.

Of being very focused on the taste and the smell.

Uh, it's always experience.

That's not a secret.

Of course, it should taste good and smell good.

It's not a secret, you know.

It's interesting.

Yeah, my mom is an excellent shift.

She's excellent and many things, and her curry just taste a certain way.

So I'm always trying to get as close as I can, and when it matches that she makes the best Bengali food in the hut, So like that's my measure.

I know in my wife's family, there's a tradition where if someone asks you for the secret to your recipe, you mislead them, you don't actually tell them the secret.

And I feel like that's what's happening here.

I feel like I haven't got any information out of you on making a great curry.

Okay, so let me maybe just general notes about Bangladeshi curries.

Like it's a lot of turmeric, garlic and ginger paste.

We like whole spices like cardamom and bay leaves and things, but it's.

Turmeric, cuman chili.

It's not very complicated, but the ratio is different than other regions of South Asia.

So turn it up on the turmeric and then you'll get to Bangladesh.

Good.

Are lasers overrated or underrated?

Underrated?

What is one underrated thing about lasers?

They are so incredibly precise and thanks to optics light based manufacturing, we have all like all semiconductors.

That's why you and I get to have this conversation online through through a laptop and the internet.

Yeah.

So the precision and the scale that they've delivered, it's incredible.

Light is incredible.

I mean I just I just find it remarkable that can be a wave and a particle at the same time.

Actually, my dog is named Photon.

Oh great, I like that is your dog full of energy?

Incredible?

And she is a particle and a wave at once.

We all are I guess if I understand correctly, she's.

Very high energy.

Yeah, we kept My husband and I are both physicists, so we went with Photon because we both use a lot of lasers in grad school.

And then our daughter is Kiara and that also means light in Italian.

Oh.

I love that.

Nabihas the client is the co founder and CEO of Seleno.

Please email us at problem at pushkin dot fm.

We are always looking for new guests for the show.

Today's show was produced by Trinamnino and Gabriel Hunter Chang.

It was edited by Alexander Garriton and engineered by Sarah Bruguer.

I'm Jacob Goldstein and we'll be back next week with another episode of What's Your Problem.