Using a library of 100,000 carefully designed molecules, scientists at Compugen’s subsidiary Keddem Bioscience are trying to create a “universal probe” that would act as a key for any enzyme’s active site, Martin Gerstel, the chairman of the board for Compugen, said on Monday during his presentation at the UBS Life Sciences conference in New York.
If such a probe is successful, it would solve the problem of finding inhibitors for proteins in order to study what the functions of specific proteins are, said Dror Ofer, the CEO of Keddem.
The idea of finding a small molecule that just sticks to the active site of a protein sounds very elegant,” said Ofer. “The problem is where do you find that molecule? We will have a way of doing that systematically.”
The idea behind the universal probe, also known as a universal ligand, is that every enzyme has “sticky points,” or chemical moieties that form weak bonds with the chemical moieties in the ligand, Gerstel explained. There is a minimum distance chemical moieties must have between each other in order for them to interact without forming covalent bonds, and there is also a maximum distance that the moieties can have between them before the bond becomes so weak that it is no longer considered an interaction.
By calculating the minimum and maximum distances between the chemical moieties, and dividing that range of distances up into bins, or intervals, Keddem Bioscience scientists have estimated that a library of 100,000 carefully chosen, non-redundant molecules could be used to generate ligands to bind with any active site.
“We started with an understanding of a three-dimensional ligand that fits into an active site, and we developed a theory to which we say ‘I don’t see why this shouldn’t work,’” said Gerstel, speaking during a question-and-answer session at the conference.
But Robert Fletterick, a professor in the department of biochemistry at the University of California San Francisco who studies the functions of biological systems using X-ray crystallography, cautioned that even though a ligand may fit into an active site, it does not mean that “the key will turn the lock.”
“I’m completely suspicious,” said Fletterick. “Binding has conformational changes. The protein changes when it binds to a substrate, and how do you calculate that binding energy change? They can’t solve the problem computationally.”
Gerstel said the scientists’ approach to solving this problem is an example of the difference between having a company run by life scientists and a company run by physicists.
“The physics inclination is to say ‘How do you solve it?’ not ‘How do you make it faster, or better,’” said Gerstel.
Ofer pointed out that one of the biggest problems in the pharmaceutical industry is that there is no guarantee of being able to find drug candidates against targets.
“If you can say, ‘Look, if I put 100 chemists to work on different projects, I’m guaranteed to come up with five drug candidates per year’ — if you could at least make this statistical claim, then you could turn this into a deterministic process,” said Ofer.
In terms of proteomics, having access to inhibitors of proteins would allow researchers to more easily validate whether certain proteins contribute to a disease. Validation of biomarkers currently poses a major bottleneck in going from high-throughput screening to clinically useful therapeutic and diagnostic tools.
“If you’re looking at the human proteome, it will take a lot of work. There’s some basic infrastructure that needs to be created, and it will cost a lot of money and time,” said Ofer. “Basically, we’re trying to get this to work. If it can really succeed, I think it will have an impact that is difficult to imagine.”
Gerstel said that the $15 million that Compugen has in cash flow should be sufficient to fund the Keddem program for the next year. If the program continues to develop successfully, then the budget will have to be scaled up eventually, he said.
Aside from developing a universal probe library, Compugen, a 130-employee company based in Tel Aviv, Israel, is also working on developing therapeutic markers through a predictable computation-based discovery process that company officials say should have a higher chance of success than markers that are found randomly.
“The way research is done in the industry, they do hundreds and thousands of experiments in parallel. The results are all observational, and more than 99 percent of it leads nowhere,” said Gerstel. “Compugen’s mission is to increase the probability of success by using ideas from math, computational science, physics and life science.”
In March, Compugen began experimentally validating CD40, a potential therapeutic protein that it predicted computationally (see ProteoMonitor’s sister publication BioInform, 3-8-04). Gerstel said the company expects to add at least two therapeutic markers to the market each year.