Mapping Biochemistry’s Dark Matter: Lessons Learned With Viswa Colluru

Let your experience inform your decisions
 

My journey in natural product drug discovery began by accident. Like any good middle class south Indian boy, I knew what I wanted in life: to study computer science, move to Silicon Valley, and work for Google. But during high school, my mother was diagnosed with chronic myeloid leukemia.

Back then, in the late nineties, the only existing treatment was palliative care. Then, about two years after my mother’s diagnosis, Novartis launched imatinib – the first oral chemotherapeutic against a protein called BCR–ABL kinase. Accessing imatinib proved quite a financial challenge for my family, but once obtained the treatment did send my mother into complete remission. Thus medicines came to signify hope to me.

Ultimately, when my mom won many battles but lost the war, I decided that if I could help at least one family through the course of my work, then I would have lived a good life.

Following that plan took me from sunny south India to sunny midwestern America, where I found the one professor on the University of Wisconsin-Madison campus who was studying a little known field called “immunotherapy.” After completing my PhD there, I decided that the fastest way to deliver hope to patients would be to work at a small, rapidly growing company where I might accelerate both my learning and my impact. So I moved to Salt Lake City to work at Recursion Pharmaceuticals, a company that was integrating technology into drug discovery in entirely new ways.

At Recursion, I worked on diseases where treatments were lacking, such as liver inflammation. During that time, I remembered an incident from my childhood when I went through an episode of jaundice. Unfortunately, there wasn’t a “modern medicine” available to help me, and I endured agony for weeks. But after treatment with medicinal plants, I made a swift recovery. In India, the use of more natural remedies isn’t unusual. In fact, they are often used alongside conventional medicines.

I asked myself if this treatment philosophy was worth returning to. It seemed to me that the pharmaceutical industry’s fundamental problem was an inability to find sufficient efficacious medicines – too many discoveries don’t translate to success in patients, despite costing millions – sometimes billions – of dollars and eating up years in the lab.

While I was ruminating, some pertinent news was hitting the industry headlines. GW Pharma’s cannabis derivative, Epidiolex (a cannabidiol oral solution), had been approved by the FDA – a first in US history. This product of nature helped treat two rare debilitating forms of seizures that affected children. 

Though these were huge milestones in the treatment of both diseases, the therapeutic use of cannabis-based products for seizures is anything but new. Cannabis extract has been used to treat seizures for 5000 years, and a study of the plant’s anti-seizure properties was first reported in a paper published in 1848.

Here was a modern example of the thesis that was already running through my head. I couldn’t shake the idea. I asked myself, “What about the hundreds and thousands of other plants that have a similar historical pedigree for treating human symptoms?” That one question became the basis for my company, Enveda.

There were, however, two main problems to be solved. First, historical human use of these plants was reported in the form of symptoms, whereas modern drug discovery deals in the language of molecular changes. Second, most of the chemistry of these plants was unknown, rendering the search for the relevant drug-like compounds akin to hunting for a needle in a needle-stack.

Respect ancient intelligence; use artificial intelligence 
 

I believe it was the ancient Sumerians who first described the use of willow bark for fevers and headaches. Today, we know that it can serve as an anti-inflammatory that works on specific proteins. But how does one bridge that symptomatic knowledge with modern molecular and cellular information? By building an integrated data source.

Enveda has digitized the history of humanity’s use of plant medicine and integrated it with the sum total of our knowledge of pathways, molecules, and targets. The result? A knowledge graph that we call the “BIOEDGE” which has information on 33,000 plants and more than 16 million relationships to proteins and pathways.

AI helps us predict what kinds of connections might exist between plants and diseases. For example, if we were to develop a treatment for a neurological disorder, we would use AI to identify the best plants to begin looking for the perfect therapeutic molecule.

In addition, we are constantly feeding our massive, curated collection of public information with internal data from our “chemical search engine” – a crucial aspect of our work because many plants contain mixtures of unknown compounds. And if there are two things that scientists don’t like, it’s mixtures and unknown compounds. The industry’s fear of uncertainty is one of the main reasons why many companies have stopped working on natural sources – despite the fact they are roughly ten times more likely to cash out as an approved drug. With the right technology, we can begin to re-engage with these mixtures and unknown compounds.

Nineteen twentieths of the territory lies in darkness
 

Some say the “low-hanging fruit” of small molecule discovery has all been snatched. But what constitutes a low branch is a matter of perspective and the technology in one’s hand.

If you are stuck with the same compounds that everyone else on Earth works with, you may struggle to grab the higher-hanging fruit. But at Enveda, we have about four billion years of evolutionary advantage driving our drug discovery library. In other words, we’re working with an entire forest.

Believe it or not, only 5 percent of this chemical “forest” is mapped, with 95 percent of all chemistry unknown. And we’re not just talking about the chemistry of plants, but also that of the soil and the microbiome. Even our own internal chemistry is largely unknown. We only know about 150,000 compounds, which have given us a third of all FDA-approved small molecules.

The 5/95 figure is really worth considering. Approximately 90 percent of drugs fail in the clinic; 95 percent of the universe is non-observable dark matter; 95 percent of our chemical space is unknown; and of nearly 10,000 diseases, we only have treatments for 500. There is so much yet to be done, and so many fundamental problems to be solved. We’re at the precipice. Now that companies like my own are building the tools and technology that can cast a light into the dark forest – the terra incognita – we couldn’t ask to be alive at a more exciting time.

In fact, my vision for Enveda isn’t just about solving the 5/95 problem – it’s about going far beyond it. With the technology we’re building, we aim to scale the discovery and understanding of the structure and function of natural compounds beyond just medicinal plants to human systems, animals, and food. Such knowledge would change the way we understand the chemical basis for life, and even change how we view society.

Back when the first search engines were launched, they tried to index the whole internet by searching and attempting to retrieve every single thing published online. Endeva’s chemical map of the world operates on the same principle, but by searching through mass spectrometry signatures.

In the process of creating this “map,” we found that the interplay and overlap of chemistry and life is striking. You can take a sample from the sea, from the land, and from our human body, and you can then pose the question: How many of these fundamental chemical fingerprints are similar? The answer is that, often, there is a massive 70–80 percent of overlap. Perhaps you’ve heard that more than 50 percent of human DNA matches banana DNA. Well, now we are learning that the same holds true for our chemical fingerprint.

Self-belief is as vital as hubris is dangerous
 

Every biotech company – especially those that have a particular thesis or platform technology – has to face the reality that their compounds may fail. Even if you take the “1 in 11” compounds ratio that the industry runs on and boost it using your technology to “4 in 10”, you still have to accept a “6 in 10” chance of failure.

For Enveda, that’s no different. Throughout the company’s journey, I have often asked myself about the challenge of building a strategy that harnesses the full potential of our platform without placing our bets on our first compound and our first trial. In recent years, some commentators and philosophers of science have argued that biology will be the avant garde science of the 21st century – that biology is already snatching the baton from pure chemistry and engineering. I am actually a huge believer in the coming “era of biology,” though not for a single overarching reason.

The pandemic is our latest alert to the fact that we are all connected, independent of our differences. We now know it doesn’t matter how cool your technology is – if you don’t have robust healthcare and the associated tools to deal with oncoming challenges, you may die. We are all intimately susceptible and perishable at the hands of a microscopic biological threat. Looking beyond the pandemic and the damage it has dealt us, we need to see the damage we have dealt to our own biosphere.

Increasingly, societies across the world are facing the challenges and dangers of climate change. We can all see the irregular weather patterns. It’s happening, and solving it will involve a fundamentally different approach to material and manufacturing.

On this point, it’s important to remember that the best manufacturers on Earth are living things. That’s why we need to consider the environment and how it responds to us, as well as focusing on changing ourselves and how we respond to it. To survive and prosper in the remaining 78 percent of the 21st century, we must understand the full continuum.

What is Enveda doing in terms of environmental protection?
 

We believe that uncovering the chemical treasures of our planet will be instrumental in generating the narrative and economic incentive to protect our planet’s biodiversity and fragile ecosystems. We are also deeply committed to giving back to the people, places, and territories from which we might draw inspiration from our products. We are working with leading experts in the word to draft a benefit-sharing framework that we are excited to announce later this year.

Did you and your team have to overcome any challenges along the way to building your platform?
 

On the technical side, much of what we are building has never been attempted before. As just one (boring but critical) example, the latest mass spectrometry technology we employ generates a specific kind of data output for which no software code has been written that can be scaled to thousands of samples. In other words, we had to build the table upon which to lay the chemical map of the world. On the business side, entrepreneurship is a phenomenal exercise in self-belief. By definition, outsized success comes from improbable places – which means that most people won’t believe in your idea. The key is to realize that that is absolutely okay. I’m incredibly proud of the kind of people and partners that have signed on to our mission.