Hit Parade

Harvard Medical School researchers are generating a list of compounds that have the potential to become new classes of antibiotics, antivirulents, and even blood thinners.

The work, led by Jonathan Beckwith, the HMS American Cancer Society Professor of Microbiology and Immunobiology Emeritus, builds on basic research discoveries made in his lab in the 1990s.

In 2009, Beckwith’s lab won a Harvard Catalyst pilot grant to help turn this fundamental research into new medicines. Since then, the team has fulfilled the Harvard Catalyst mission and then some. Not only did the grant help them turn their basic discoveries into a high-throughput drug screening machine, it also helped them secure the funding and connections needed to use it to search for new drug leads.

A screen to find drug candidates for one disease is an accomplishment, but the team developed a screening system that works like a multi-tiered filter that teases out leads for three distinct medical applications, each in need of new therapies: gram-negative bacterial infections, tuberculosis infections and blood clotting disorders.

“It’s a virtually perfect screen,” says Dana Boyd, an HMS lecturer on microbiology and immunobiology and a research scientist in the Beckwith lab. “It tells us instantly and precisely if we have a hit against one of our three targets.”

DsbB Basics

The story began in 1990 when Beckwith and members of his lab discovered two genes for enzymes that regulate a biochemical process in bacteria called disulfide bond formation. Disulfide bonds help stabilize bacterial proteins, particularly the toxins that bacterial cells send out to infect other cells. Neither enzyme is essential for E. coli to grow, so for a while the project took a backseat to others.

Later, Beckwith’s group wondered if these enzymes, DsbA and DsbB, might be common across many types of bacteria, as disulfide bonds are. By then, about 400 bacterial genomes had been sequenced, so Boyd and Rachel Dutton, then a graduate student in the lab, developed a bioinformatics tool to search for similar enzymes in other bacterial species. Dutton is now principal investigator in the Division of Biological Sciences at the University of California, San Diego.

They found that the system was widespread, but in some types of bacteria, such as Mycobacterium tuberculosis, they found only DsbA. Instead of DsbB, they found a different, unknown gene. The only clue they had was a note another scientist had left in the genome: “VKOR.”

They learned that the Mycobacterium VKOR protein is, genetically speaking, closely related to the human VKOR protein, which is involved in blood coagulation. Human VKOR is the target of the anticoagulant drug warfarin (Coumadin), a drug used to thin the blood of patients at risk of clots.

Though warfarin is prescribed widely, it requires close monitoring. Foods such as leafy greens can throw off the dosing, putting patients at risk of clotting if the dose is too low, or bleeding if the dose is too high.

“We thought that if we could find new compounds that act against VKOR, we could potentially find inhibitors that work in a different way, without the same issues with a patient’s diet,” says Beckwith.

The team had also learned that interrupting Mycobacterium VKOR inhibits its growth, making it a plausible target for new antibiotics. In some parts of the world, tuberculosis has become so resistant to antibiotics that it is almost untreatable.

Schemes for Screens

To find leads for new antibiotics against tuberculosis and anticoagulants, Beckwith’s team set out to build a screening system that would detect compounds that interfere with VKOR.

To support this effort, they were awarded the Harvard Catalyst pilot grant in 2009, which was later followed by a Blavatnik Biomedical Accelerator grant from Harvard University.

Even though bacterial VKOR and DsbB don’t share common genetic roots, Beckwith’s team found that VKOR can replace DsbB in E. coli and do its job of promoting disulfide bond formation. The team built a screening system that uses different strains of E. coli, each with different disulfide bond-forming enzymes, to detect inhibitors of the different enzymes.

The original screen detected DsbB and bacterial VKOR inhibitors. More recently the team expanded the screen to include rat VKOR, which closely resembles the human enzyme.

“We’re interested in compounds that inhibit specific enzymes,” said Cristina Landeta, a postdoc in the Beckwith lab and first author on a February 2015 Nature Chemical Biology paper describing the screening system.

“You don’t want to develop an antibiotic that is also a blood thinner,” she said.

In May 2014, the team connected with Hoffman-La Roche, a pharmaceutical company interested in using their screening tool to find ways to block DsbB to treat Pseudomonas infections.

Pseudomonas infections are common among hospitalized patients and are becoming increasingly resistant to existing antibiotics.

In Pseudomonas and other gram-negative bacteria, blocking DsbB is not lethal. Instead of looking for compounds that kill the bacteria—antibiotics—the team is looking for compounds that undermine its infectious capacity—antivirulents.

In this case, an antivirulent would block the disulfide bonds that make bacterial toxins and other virulence factors stable.

“The disabled colony of bacteria could then be overcome by the immune system, or possibly by another drug,” said Boyd.

The team has screened a library of 80,000 potential drug compounds from Roche and is screening a library of 650,000 compounds from the Institute of Chemistry and Chemical Biology at Harvard. So far, they’ve found two compounds that may be viable Pseudomonas antivirulents. Roche has ruled out one, but the team is testing both against Pseudomonas infections in mice.

They have also identified 30 compounds that act against VKOR and could become antibiotics for tuberculosis or anticoagulants. As screening continues, the hits keep coming. “We have to start giving them names,” said Landeta.