The man making antibodies smarter

Yanay Ofran credit: Eyal Izhar
Yanay Ofran credit: Eyal Izhar

Using artificial intelligence, Biolojic Design founder Prof Yanay Ofran believes he can produce antibodies that are a match for any disease.

Prof. Yanay Ofran's amazing story about the pursuit of an antibody that will save the world from disease Shlomit Lan and Gali Weinreb Professor Yanay Ofran, founder and CEO of Biolojic Design, a company that develops smart antibodies designed to treat a variety of diseases, is frustrated. "Humanity invests $300 billion each year in drug development, and what do we get? At most, we get a few dozen medications a year, most of which don’t solve the problems, and give an additional three weeks of life on average to patients with pancreatic cancer, or manage to inject a medication that to date was given via infusion. Those are the breakthroughs," he says despairingly.

But Ofran does not think the pharmaceutical companies are the only culprit. "The drug companies are portrayed as a devil who says, 'I won’t cure this because it’s not worth my while.' But these companies do have a legal obligation towards their shareholders, not to develop drugs unless there's an economic incentive. "

The problem, as analyzed by Ofran, is much more complicated and therefore far more difficult to treat. "There are three players sitting around the drug development table: science, regulation and the business world. Everyone knows the whole thing is in a jam, but if you ask one player why that is, they’ll point to the other two and say, ‘Because of them.’

"Scientists say business people only go where the money is, and regulators act as if everyone wants to cheat them, and they’re both right. Regulators and businesspeople claim that scientists come up with solutions that are bad and can’t be implemented in terms of either regulation or business. How can we progress? We need better business models for drug development".

So how can we achieve better results? "Drug development costs must be reduced. After all, most of the money doesn’t go on technology but on huge clinical trials - 3,000-4,000 patients in a Phase 3 trial, at a cost of hundreds of millions of dollars. Why are these trials so large? Because in statistics, the smaller the difference between the experimental group and the control group, the more subjects we need to see in order to be convinced that the difference is real.

"But if we do trials on drugs that make a really big difference, we can test them on smaller patient groups. So, the better your product, the cheaper it is to develop. And we believe that these results can be achieved by using tools like ours, artificial intelligence and algorithmics.

"The problem is not just regulatory. When the cost of developing a new drug is $1 billion, and the drug company has to decide which of two drugs to trial, it will go with the one that worked in animal trials. And testing drugs on animals is a mistake, it doesn’t work. Scientifically it's stupid. There’s an old joke about a conference on cancer where the speaker says he’s got great news for all the mice in the audience."

And then, once again, there's a headline, "Cure for cancer found" but in fact, there’s no cure?

"In the end, we will actually solve cancer, because we know what causes it. But when we solve one problem, we replace it with another problem."

Ofran, naturally, believes he has a better solution. Under his management, Biolojic Design has developed a technology for developing and programming antibodies using computational mathematics, big data and artificial intelligence. Unlike the usual antibodies serving the pharmaceutical industry, the promise is smart antibodies, a sort of nano-robot that, like human antibodies, is able to adapt to changing situations. "In two or three months we’re going to inject the first computer-designed antibody in history into a human," he announces jubilantly.

Sounds great. But let's start with the most burning issues: Can your technology cure Covid-19?

"Certainly," he replies emphatically. "Our technology is capable of producing smart nano-robots that can measure things and react to them. From a purely technological point of view, there’s no problem in applying it to any protein system that’s harming the body, whether a cancerous growth, a bacterium, or a virus. The question is, what’s the right way? If you can hit your enemy with a cap gun, you don't need an F16. There may be simpler ways, like a vaccine."

And what about infectious diseases?

"An interesting thing has happened with infectious diseases. Just before the coronavirus, we were looking at developing infectious disease drugs and we saw that most big pharmaceutical companies had closed those departments because infectious diseases weren’t economically worthwhile. Infectious diseases were considered something that could be treated cheaply, or associated with the Third World, which doesn’t buy expensive drugs.

"An antibody sells for an average of tens of thousands of dollars per treatment [annually - S.L. and G.W.]. Antibiotics, today the leading solution for infectious diseases, cost $10-200 per treatment. In a world where insurers are used to paying theses prices for infectious diseases, the chances of selling a product for $15,000, not to mention $ 100,000, is zero."

And we are back, once again, to the market failure that Ofran first talked about.

Reinventing drugs

But while the market, despite its failures, continues as usual, Biolojic Design has already signed collaboration agreements with pharma giants like Eli Lilly to develop antibodies for diabetes, autoimmune diseases, and cancer, in volumes that that could be worth hundreds of millions of dollars. Eli Lilly’s announcement of a collaboration in March of this year for development of an antibody-based diabetes medication, told of the attempt to find innovative methods for developing new drugs, and termed Biolojic 's antibody platform "a promising approach".

In the meantime, Biolojic is still a relatively small company, with 52 employees working in laboratories and offices in Rehovot. Ofran is CEO and founder, and also the person who developed all of the technology.

What happened to the company during the coronavirus period?

"One thing we decided as company management was not to furlough anyone; everyone is working. If we don’t manage to get through this, we’ll all be in the same boat, and whatever happens will happen to everyone. And one of the amazing things that happened is that all three of my companies based there (Biolojic Design, Ukko, and Aulos Bioscience) raised more than $120 million. We started out when things were very uncertain, but actually a lot of investors expressed confidence in us during this period. "

But antibody-based drugs are nothing new. There are already drugs for cancer like Herceptin, or for autoimmune diseases, like Remicade or Enbrel. Just what have you invented? Let's be precise.

"For more than 40 years we’ve marveled at the fact that antibodies can be taken and used as drugs. But the antibodies that medicine uses as drugs do a thousandth of what human immune system antibodies are capable of doing.

"The antibody produced as a drug is a pretty stupid molecule that just knows how to be a spoke in the wheels or a pain in the butt, proteins activated or changed when a molecule that fits their structure exactly binds to them.

"That’s how all communication within the body works. Antibody drugs are capable of adapting themselves to the structure, but interfere only with the protein we want to neutralize, they can’t do anything more complex. A lot of people have been saved thanks to this spoke in the wheels. But much more can be done.

"An antibody is a protein, and proteins are amazing nanomachines. Proteins are what gives color to hair and eyes, what breaks down food. Evolution has brought them to the point where they are wondrous machines, sophisticated nano-robots. We want to use our ability to produce and design proteins, to produce antibodies that do more than just throw a rock at the protein that causes disease."

What can a smart antibody potentially do? In the case of diabetes, for example, it could control insulin secretion which handles the breakdown of glucose in the bloodstream. It would turn on when sugar levels were high and stop working when they were low, like an on-off switch. It’s known that excessive breakdown of sugar can also lead to life-threatening conditions in diabetics. In other words, the antibody mimics the natural mechanism of a healthy non-diabetic person.

The same goes for the immune system. Like many systems in the body, it should be in a balanced state but can lose equilibrium both because of overactivity, as in autoimmune diseases, or inactivity that makes it susceptible to infections and lowers defenses against cancerous tumors. The ideal is an immune system that is sufficient and moreover, automatically adapts itself to situations, as with healthy people. "For cancer patients," says Ofran, "we’ve programmed an antibody that knows only how to activate the immune system and not turn it off, and for autoimmune diseases we have an antibody that does just the opposite."

In other words, Biolojic’s vision is of more effective drugs with far fewer side effects. Of course, they still have to make the transition from computer algorithms and lab test tubes to the real world.

"Four years ago, we threw everything away"

The company's beginnings, in 2008, looked completely different and were actually far less grandiose, if that is the right way to put it. Biolojic Design’s first effort was to be a kind of smart supplier, a service company, with annual revenue of millions of dollars and of course with potential for royalties in the future, to pharmaceutical companies such as AstraZeneca, Novartis, Eli Lilly and others.

"We told these companies to bring us those projects where they’d tried to make a smart antibody and failed, and that we would program these antibodies to do what they asked," Ofran says. Usually, what they asked for was a drug that, like all the rest, ‘throws a rock’ at the protein they wanted to neutralize, but a smaller, more accurate stone. Meaning, gentler drugs with fewer side effects.

"The resulting products entered into the development processes of the partners," says Ofran, "But at some point, we realized we were limiting ourselves to the creativity of the developers at the drug companies. Three or four years ago we stopped everything, threw away that source of income. After all, we want to enjoy what we do."

Three years ago, Biolojic raised a "relatively small" amount of money, as Ofran puts it, from Marius Nacht's aMoon fund, as there was still money left from the previous operation for the company to start building a future drug pipeline of its own. Meanwhile, under the new model, agreements have already been signed with US listed biopharma company Nektar Therapeutics for the development of antibodies against autoimmune diseases, with Eli Lilly for the development of antibodies that fight diabetes (for up to $121 million), and a collaboration with venture capital fund Apple Tree Partners (ATP), to develop antibodies to fight solid cancerous tumors. In addition, ATP has invested $40 million in an independent subsidiary established by Biolojic for this purpose.

The company, to be called Aulos Bioscience, will conduct clinical trials in Australia. This investment will cover Phase 1 of the trials, for safe use in humans. If this part goes well, additional funding will be raised for the next phase.

How do you turn your antibody from a spoke in the wheel into something more sophisticated and flexible?

"We view proteins like sentences in language," Ofran says, surprisingly. "We trained the algorithms in the same way that Google trains its programs to complete sentences. They didn’t teach it the rules of grammar, they just gave it more and more examples, until it learned to offer possible endings to sentences. We don't understand how it happens, but it does happen.

"By the way, the head of our computational department was previously a senior manager at Google. In the same way that Google, Siri and Alexa are able to complete the sentence you started, without having to find new rules and the grammar of a language, so we think the most effective way is to flood the system with examples of human antibodies and what they can do. That’s why see ourselves as a technological rather than a biotechnological company.

"The biotechnology industry is constantly trying to better understand biology, and discover new biology. We aren’t. We say, let's take the old biology, that we do understand, and just do a better and more efficient manipulation. What we’re trying to do, in the lab, is to mimic - particularly on the computer - immune system activity. If we’re seeking an antibody that can, for example, turn off cytokines [an immune system component - G.W. and S.L.], the system proposes suggestions - a million and more antibodies."

The computer’s suggestions are then commissioned from a company that produces DNA according to specifications. These are then "allowed to compete in the lab on a cluster of cells in vitro. Like an accelerated evolution. The few 'winners' of this competition are best at turning off cytokines." The winners go to the next stage, clinical trials, to determine which will be the basis for a new drug. "To this day, the pharmaceutical industry has tried to understand the rules, and discover unknown rules in biology. We don’t deny that there are rules, but we understand that we don’t know them."

"Grandpa was loved and revered"

If Ofran (49) doesn’t sound like your typical technology company founder, and if it seems there’s something almost philosophical in the way he talks about his industry, that’s probably no coincidence. Ofran is the grandson of one of Israel’s most prominent intellectuals of all time, the outspoken, provocative scientist and thinker Prof. Yeshayahu Leibowitz.

Ofran attended a religious school in Jerusalem, then studied at a religious kibbutz yeshiva, was drafted into an IDF intelligence unit, and at the age of 28 had already completed his doctorate at Columbia University - the third doctorate started and the only one completed.

"My first attempt at a doctorate was in linguistics," he says, explaining, with hindsight, the linguistic contexts he attributes to proteins. "I graduated with a bachelor's degree in physics and biology, and I wandered around campus taking courses and arrived at the conclusion that the most interesting question was language. But fairly quickly, I concluded that this was a subject where the questions were more interesting than the answers, a bit like philosophy.

"So, I switched to neuroscience, and my second doctorate dealt with motor control of handwriting. The question is, where does high level processing takes place, which they call the little green man question. We are able to say that this area of the brain sees dots and this one sees lines and this one sees colors. But where's the integration?

"Then I went to a conference and met the high priest of motor control, an old Moroccan Jew at Columbia University, and he said, 'I've been researching feline paw-pad movement for 70 years, which is a straight-line motion, and to do it, the cat's brain has to calculate like mad. And after 70 post-doctoral students and hundreds of articles, I still don’t understand it." And I realized that, once again, I was in a field where the questions were better than the answers. And then I looked at the world of molecules, where it's possible to do research that provides answers as well, and I worked on the use of algorithms in computer science, in molecular research."

It’s hard to resist asking the man who wants to mimic the human immune system using a computer and artificial intelligence to produce an end-to-end revolution in the pharmaceutical industry, whether he has replaced his grandfather's God with a computer. But Ofran, true to his family heritage, is razor-sharp and doesn’t take the bait. He says it has nothing to do with God but with who makes better decisions, man or machine. "My grandfather, too, once said that intuition is another word for the evil inclination."

Leibowitz is Ofran's maternal grandfather. His mother, Dr. Mira Ofran, is a physicist. One of his cousins is psychiatrist and brain researcher Prof. Yoram Yovell. He has six brothers and sisters: doctor and researcher Prof. Yishai Ofran, psychologist Shlomit Ofran, Torah academy founder and teacher Uri Ofran, left-wing activist Hagit Ofran, learning technologies expert Yochai Ofran, and rabbi and network anchor Ilai Ofran.

"My grandmother would say that the greatest gift you can give your children is siblings. We would all meet, the uncles and cousins, at my grandparent’s house. My children have 40 cousins just from my side, and another 12 from my wife's side."

And do you believe in God?

"That’s a meaningless question. These days, the statement, ‘I believe in God’ or ‘I don’t believe in God’ is related more to how you vote than what you think about the world. If you look at philosophy of religion through the ages - this sentence has been given every possible interpretation, from ‘I believe there’s an old man who watches over me’ to ‘There’s nothing out there.’ Everyone defines God differently."

In case there was any doubt, robots and artificial intelligence, as Ofran sees it, are not sublime, even if they do things that humans do not understand. "In the past, religion took a dualistic approach, for example of body and mind. But dualism was banished from intellectual life in the Western world, and the popular opinion was that everything could be reduced, meaning, expressed in scientific language as a collection of rules. Consciousness, soul, life - were also supposed to be explained in scientific terms.

"But recently there has been a new approach called emergent mental construction which posits that things like mind and consciousness, are created from matter, but only on condition that the system is complex. The mind is connected to molecules, but does not exist in every molecule, but only from a certain level of complexity and above. And the more complex artificial intelligence becomes, the more it replicates this. "

What Ofran is saying is that the more complex an artificial intelligence system is, the more successful it is at reproducing the phenomena unique to complex systems, such as a whole that exceeds the sum of its parts. Artificial intelligence is already doing things we can’t explain - we don’t know how exactly they stem from any unit in the system - but we do know that these phenomena originate in the system, and are not something mystical.

Ofran's proteins also fight allergies

Ukko, a company founded by Prof. Yanay Ofran with Anat Binur, uses his computational biology tools for antibody research in a creative way: it designs food products that are non-allergenic, and the opposite, for use as anti-allergy treatments in the future. In January, the company raised $ 40 million in a round led by the Leaps Foundation, Bayer's health and agriculture impact investment fund.

The company's first products are flour containing a protein almost identical to gluten that does not cause a reaction in celiac patients, and a new treatment method for peanut allergy. Its technology may enable it to treat a range of food allergies and sensitivities.

Investors include the Continental Grain Company, Skyviews Life Science, PeakBridge Ventures, Fall Line Capital, and Blu1877 (Barilla’s venture/innovation arm), as well as existing investors Khosla Ventures, Innovation Endeavors, and TIME Ventures, the investment fund of Marc Benioff, founder of Salesforce.

Published by Globes, Israel business news - - on September 5, 2021

© Copyright of Globes Publisher Itonut (1983) Ltd. 2021

Yanay Ofran credit: Eyal Izhar
Yanay Ofran credit: Eyal Izhar
Twitter Facebook Linkedin RSS Newsletters גלובס Israel Business Conference 2018