By David N. Leff

Erysipelas is a nasty but non-fatal bacterial skin infection. Many decades ago, an American physician named William B. Coley, on the staff of Memorial Hospital, in New York City, noticed — and reported — that a patient with end-stage cancer suddenly improved when he caught a case of erysipelas.

Coley extracted toxins from the infective microbes and used them to treat cancer, with some positive results. But such anticancer bacterial extracts dropped out of vogue with the advent of antibiotics against infection in the late 1940s.

Melanoma biologist John Pawelek, at Yale University School of Medicine, in New Haven, Conn., explained that "in the early 1950s, children with leukemia often had infections, and they would go into remission. But as soon as they started getting antibiotics, the infection went away, but their leukemia relapsed. So people thought maybe bacteria would be a good thing against cancer."

Pawelek agrees, but for quite different reasons.

He is senior author of a paper in today's Cancer Research, dated Oct. 15, 1997. Its title: "Tumor-targeted Salmonella as a novel anticancer vector."

To develop their genetically engineered, tumor-killing bacterial construct, Pawelek told BioWorld Today, he and his co-authors started with two basic observations:

"Adenoviruses, the current vector of choice in cancer gene therapy, are severely limited in their ability to reach metastases. You have to inject them directly into the tumor in order to be effective. And even then they populate only perhaps five percent of the cancerous cells.

"When we put wild-type Salmonella typhimurium into a mouse with a tumor, it would populate that tumor in great abundance. However, the wild type went all over the body, and killed the animal very quickly."

Defused, Gene-Armed Salmonella Takes On Tumors

Those two facts strongly suggested to the Yale researchers that if they could deprive Salmonella of its lethal virulence to mammals and equip it with a tumor-killing capability, they would have a tumor-targeting anticancer system competitive with viral vectors. So far, they've reached these milestones in mice and foresee Phase I trials within about 15 months.

That estimate comes from one of the paper's co-authors, parasitologist David Bermudes, now associate director of biology at Vion Pharmaceuticals Inc., in New Haven.

"Vion," Pawelek said, "is the underwriter of all this research. It funds my lab, and has licensing rights from Yale to the things that come out of it."

The first thing was to weaken S. typhimurium so it wouldn't kill off mice before they had a chance to test the group's new anticancer approach.

"So we set out to modify this Salmonella," Pawelek recounted, "by making repeated genetic changes and testing a lot of clones till we found some strains that went into tumors but were less and less toxic to the mice.

"Then we genetically crippled the bacteria," he continued, "so they are incapable of growing unless they have certain essential nutrients. Finally, by repeated mutation, and isolation of multiple polyauxotrophic [specific-nutrient-dependent] mutants, we got organisms that were safe for the animals but still had great tumor-targeting capabilities."

Into this basic vector, the Yale/Vion team inserted a diabolical gene construct, designed to kill whatever tumors the bacteria encountered.

That gene expresses an enzyme, herpes simplex virus thymidine kinase (HSV TK), which recognizes the precursor molecule of a potent antiviral drug, gancyclovir. When HSV TK bumps into that prodrug sequence in a tumor cell, it releases gancyclovir, Pawelek observed, "perhaps 10,000 to 100,000 times better than its mammalian cell counterpart. So when we added that prodrug to the HSV TK gene, it converted that precursor at a much greater rate than the mammalian cell was able to do it.

"That prodrug by itself doesn't bother the tumor cell at all, which doesn't know how to convert it," Pawelek pointed out.

"But when we populated the tumors with Salmonella that were carrying this HSV TK gene, then injected the mice with gancyclovir, the thymidine kinase gene in the tumor recognized the gancyclovir prodrug and converted it into its toxic form. This fooled the tumor cell into thinking it could incorporate it happily into its DNA. But it's a very bad toxin when it gets into the DNA, and kills the tumor cell."

Pawelek considers his main results so far to be: "First, multiplication of the Salmonella within the tumor. We've reached extremely high densities, 10⁹ bacteria per gram of tumor cells. And second, targeting multiple tumors in the body. One can do it from afar. We entered the vector into the mouse bloodstream and they could find all the tumors in the mouse."

S. typhimurium can cause enteritis, a sudden, acute intestinal infection marked by nausea, vomiting, diarrhea, slight fever, and eventual recovery.

But when Salmonella invades the bloodstream, it secretes a sinister toxin, lipopolysaccharide (LPS), which causes septic shock, a potentially fatal nosocomial (hospital-borne) infection.

"Which led us to a big question," Pawelek recalled. "'Why do we think we can put Salmonella into the bloodstream of a human being and have it go to the cancer, but not set off septic shock?'"

Mini-Pigs Model Toxin-Free Mutants

The answer: "What we have to do in order to achieve that goal is alter the LPS of our special Salmonella strains. That's what we and Vion are working on now. In fact," Pawelek continued, "we are testing new strains of Salmonella in a herd of miniature swine at Texas A&M University, in College Station, Texas."

Bermudes expects Phase I human trials in melanoma patients "in possibly 15 months or so," and observed, "Phase I is safety, but it has other endpoints as well. One that we hope for is actually to show tumor targeting. Because if you have melanoma, you can have tumors that can be biopsied." *