Building T Cell Receptor Therapies with Generative AI

Immunotherapies such as checkpoint inhibitors and chimeric antigen receptor (CAR) T-cell therapies have revolutionized the field of oncology over the last few years, but they are just the beginning. A new wave of immunotherapies is now hitting the clinic.

In 2022, Immunocore’s tebentafusp (Kimmtrak) was the first TCR anti-cancer therapy to receive FDA approval. While most companies developing TCR therapies are using a method called phage display to design new treatments, ex-Immunocore scientist Michelle Teng and colleagues at Etcembly have developed a method of using generative AI to discover and design a range of different TCRs to fight cancer and other conditions. They have already developed a candidate bispecific T cell engager therapy, to target a cancer antigen known as PRAME, in less than a year.

Teng, CEO and co-founder of Etcembly, is an entrepreneur as well as an experienced scientist with a background in T cell immunotherapy research. A few months after founding Etcembly in 2020 she also co-founded rare disease biotech SynaptixBio, where she retains a position as CSO, to try and find a cure for her daughter Sofia who has a rare neurodegenerative condition called TUBB4a Leukodystrophy.

Inside Precision Medicine senior editor Helen Albert spoke to Teng about her work, her inspirations and what she hopes to achieve at Etcembly and SynaptixBio.

Q: What inspired you to become a scientist?

Michelle Teng: I’ve always been driven by curiosity about how things work, where they’ve come from and how they come into being. That innate sense of curiosity, and my relentless drive to find the answers to questions moved me into science. I was at an intersection at one point in my academic life where I was considering being a lawyer. My parents thought I had the aptitude for it, because I was also very analytical and logical in my thinking. It could have gone both ways to be honest.

I did my first degree at Imperial College London in biochemistry and that consolidated and crystallized my desire to become a scientist. Having that great nurturing educational environment really helped develop that part of me.

Q: What made you decide to work in industry?

MT: I think a lot of it had to do in my postdoc supervisor at the Wellcome Sanger Institute. He’s no stranger to the field of biotech. His name is John McCafferty and he’s got this great entrepreneurial spirit that you didn’t see in many scientists back in the day. He certainly inspired me to translate whatever I was doing into a business, or at least encouraged me to think that way. He said, “Well, if you really want to be a biotech entrepreneur one day, try industry for a couple of years. If it doesn’t work, you can always return to academia.” So, I took his advice, and that’s why I ended up working at MedImmune, which is now a subsidiary of AstraZeneca.

Q: What did you learn from working as an industry scientist?

MT: I did enjoy it, but I also decided what ‘good’ looked like for me. I didn’t want to be in a large organization where you were just a tiny cog in the wheel and you don’t really understand the bigger picture. I wanted to be part of something small to give me the exposure to what biotech is about.

When I transitioned to Immunocore, that helped cement the direction of my career. I never thought going into industry meant I wanted to spin out a company one day. I was actually quite content to be tinkering in the lab for the rest of my life. At least, that was my romantic view of myself! At some point I realized the work I had done at Immunocore could be done better. If I could find the right resources and funding then I would at least have a decent shot at it. So that’s what I decided to do and why I ended up being a founder.

Q: What was the inspiration to start two biotech companies?

MT: I founded SynaptixBio and Etcembly around a couple of months apart. My daughter was diagnosed with a rare neurological disease [TUBB4a Leukodystrophy] that is life limiting. I initially started a foundation to raise awareness and raise money for research. But I realized that the money that I raised is never going to be enough to take something into the clinic. That’s why the company SynaptixBio was formed, we decided that we had to set up a commercial entity and fund it properly.

So far, everything has actually gone very well. We have a lead candidate; we are manufacturing the drugs. Everything that we have within our control is actually moving according to plan. The regulatory side has been met with lots of obstacles. But overall, from the inception of the company we have met all our milestones with very little delay.

Etcembly, in a way, is a form of escapism. If you found a company to treat your daughter, you need something else to distract your mind, to solve a different problem in a different space. I knew I could apply myself in different ways and that’s what I wanted to do. It is a lot of work, but it’s a lot of fun as well. I don’t regret a single minute of it.

Q: What was the transition like from being a senior scientist at Immunocore to a founder of two biotech companies?

MT: I was a group leader at Immunocore and I had some very elementary training in management while there. I’ve had quite a lot of prior industry experience, which put me in a better place to work out what ‘good’ looks like, it helps me with fundraising strategies, with who we speak to, and how we pitch it, and what we think industry is looking for.

At Etcembly, myself and my co-founder have a pretty clear idea what that looks like. We’re also very fortunate to be guided by the founders of our predecessor company Immunocore who are our mentors, sit on our board and are our investors as well. It’s a privileged position to be in, but it’s also the ultimate endorsement if your former boss says he’ll invest in your company.

As a mother of three you have to be ruthless with your time. It has taught me to be extremely compartmentalized with my tasks. I had to teach myself how to multitask but also compartmentalize each task such that they don’t interfere with my personal life. Between Etcembly and Synaptix there’s a healthy firewall. It’s not easy to do. But if you put your mind to it, it can be done.

Q: What are you trying to achieve at Etcembly?

MT: With Etcembly, when we first started, we went straight into lockdown. That’s when we really did look at the entire TCR landscape. Other than Immunocore, no one can make picomolar TCR. They use a technology called phage display. Phage display has been around for decades and was pioneered by my old boss, John McCafferty. It has evolved to a certain extent, but there are limitations.

Immunocore has stayed with phage display, which works, but we want to look at what ‘better’ looks like. The current process is slow and laborious. It traditionally takes two years from discovery to getting your final product. Using machine learning, we’ve invented a new technology where we can discover and engineer a TCR in 11 months. We are building the ship as we are sailing it, so in that respect we have every reason to believe that when we do it the second and third time it will be even faster than that.

It’s about reinventing how we do things rather than just being contented with how something has always worked in the past. You have to evolve the technology and you need to find novel ways of solving a problem.

Q: Are TCR-based therapeutics going to be the new CAR T-cell therapies?

MT: I think I think that CAR T-cell therapy is a tried and tested technology. They have been very successful in treating liquid tumors, but they haven’t really seen a success with solid tumors, which is something that needs to be addressed. I think TCR therapy is the way forward for treating solid tumors that have high unmet need.

If you look at Immunocore’s first TCR product on the market, Kimmtrak, against uveal melanoma, it was already seen as a rare cancer indication, but their sales figures have actually exceeded expectations. It would suggest that the patient numbers are actually much higher than one would expect. On the back of Kimmtrak, Immunocore has another program targeting PRAME, which is the TCR we’ve engineered against.

PRAME is a $10 billion market, it is huge. Just in the US alone, there are at least 100,000 patients that would qualify for this treatment. Immunocore is obviously first to market. They are the front runner here, because they have product being tested in the clinics right now, that is doing very well. But we are a fast follower. First to the market is good, but there’s nothing to say that fast followers don’t do just as well, if not even better. If you take the analogy of Apple, they are not the first people to invent the iPad, or music streaming services, for example, but they’re doing extremely well, they are one of the market leaders.

Q: How are you using AI to help discover and develop these new therapies?

MT: One of the key methods that we use is large language models. Essentially, it’s training a machine using tokenized words. So, if you say, the fox jumped over the fence, if you blank out the last word, it could be fence, it could be moon, it can be anything. It’s analogous in the case of amino acids in the TCR sequence. You mark every single position of the amino acids, then the machine can learn what amino acid should occupy that position. It will come up with its own TCR sequence based on what’s available in our database.

It’s a bit like ChatGPT when you tell the machine to find something for you, and it gives you answers, but the answers can be slightly different every single time because there is a creative element to it. That’s the beauty of it. You get multiple options and you have a panel of potential candidates that we can then validate in the lab. We don’t take the predictions as ground truth; we always test them empirically to back up what is predicted by the machine.

We are one of the early adopters using machine learning to understand TCR and using it to engineer TCRs for immunotherapy. I don’t think anyone’s really applied it in that way before. I think it’s opened up a whole new field, it’s very interdisciplinary you need biologists and machine learning engineers, bioinformaticians and protein engineers to all talk to each other and find a common solution.

Q: What is next for Etcembly?

MT: It’s a really exciting time for the company. Right now, we are wrapping up fundraising, which has gone really well. But we also are looking to develop the pipeline. We’ll be taking our machine learning engineered TCR to the clinic and will be carrying out the preclinical development stage of that program. We will probably be expanding the pipeline to include other cancer targets and we’ll also be looking at the autoimmune space in the coming year or two. It feels like we’ve been a baby for a while and we’re suddenly growing up because we’ve been successful at what we have been trying to do.

This is just the tip of the iceberg. What we’ve just uncovered, what we can do. I think this will open up a whole new field and whole new era of emerging targets. We are also looking in the blood of long-term cancer survivors for new therapeutic modalities. I think lots of companies have not even thought about that. People who have been successfully treated with immunotherapy clearly have TCRs in the blood that can keep the cancer at bay This is also an area that we want to launch into.

Q: What advice would you give someone who wanted to launch a biotech company?

MT: I think you need to do your homework. Don’t just take the first offer that lands on your table. Assess the landscape and the market conditions. Who you bring on as an investor can make or break your company. We’ve been very fortunate that our investor base is extremely supportive, but we’ve also heard stories where companies have not been that fortunate. Whether it’s an institutional or angel investor you must always understand where they’re coming from, and what they want out of the investment. It’s great to have a wonderful idea that people will fund, but you must be very clear what your exit strategy is with yourself and with your investors because they need to be on the same page as you about what a good investment looks like.

Q: There’s been a lot of talk about needing more women in biotech and pharma, particularly at senior levels. What’s your position on the topic?

MT: I think the female biotech CEO is a rare breed. I don’t see many of them. But, in recent times, I’m starting to see more and more women coming forward in leadership roles, I think we’re still so far away from where we should be in terms of equality and having a 50:50 split in the boardroom. But, I’ve always championed and supported women in leadership positions. It’s my desire to see more women moving up the ranks in leadership roles and having a say in the boardroom. We’ve come very far, but there’s still a lot of work to do. I think the tides are slowly turning, but we have to be patient. I would also say that women absolutely have to support each other, otherwise, there is no hope that we will ever achieve equal rights.

Q: What advice would you give yourself if you could go back in time to when you started in your career?

MT: I would say everything happens in its time and just be patient. It took me a long time to figure out what I really wanted to do. I would tell myself not to be afraid of who I really am and that growing up and growing older is about becoming who you are meant to be in the first place.

Helen Albert is senior editor at Inside Precision Medicine and a freelance science journalist. Prior to going freelance, she was editor-in-chief at Labiotech, an English-language, digital publication based in Berlin focusing on the European biotech industry. Before moving to Germany, she worked at a range of different science and health-focused publications in London. She was editor of The Biochemist magazine and blog, but also worked as a senior reporter at Springer Nature’s medwireNews for a number of years, as well as freelancing for various international publications. She has written for New Scientist, Chemistry World, Biodesigned, The BMJ, Forbes, Science Business, Cosmos magazine, and GEN. Helen has academic degrees in genetics and anthropology, and also spent some time early in her career working at the Sanger Institute in Cambridge before deciding to move into journalism.