E. Otis Kendall Professor of Biology, University of Pennsylvania
 Director, Penn Genomics Institute, 2001 to 2006

David Roos had been studying nucleated parasites such as Toxoplasma and Plasmodium (malaria) for several years when he decided to ask a simple question: How do antibiotics such as clindamycin work in treating both malaria and toxoplasmosis? The answer turned out to be a discovery that simultaneously solved three biological mysteries, rewrote biology textbooks, and helped to launch the field of evolutionary cell biology.

Clindamycin and related drugs kill bacteria by inhibiting the ability of bacterial ribosomes to synthesize proteins, but don’t affect ribosomes in eukaryotic cells, including those of humans. Yet both malaria and toxoplasmosis are caused by eukaryotic unicellular parasites, which clindamycin also treats. In 1996, Roos, a professor of biology at the University of Pennsylvania, and then-graduate student Maria Fichera tested three possible hypotheses:...

I want to create tools that allow other users to ask their own questions, rather than my being the ‘big computer scientist’ who does the analysis and then tells others an answer.

A fourth, “crazier” idea, according to Roos, was that the parasites that cause toxoplasmosis and malaria harbor yet another type of undiscovered ribosome targeted by the antibiotic. The idea, says Roos, was inspired by prior reports of ribosomal genes on small circular DNA that was mistakenly assumed to have come from the parasites’ mitochondria. The experiments to test this wild hypothesis ended up resolving three mysteries in parasite biology: why some antibiotics are effective against protozoan parasites; the source and function of the circular DNA originally observed in the 1970s in Toxoplasma and other eukaryotic parasites; and the identity of an uncharacterized organelle-like structure with multiple membranes.  

In two back-to-back papers in Science and Nature in 1997, Roos, then-postoc Sabine Köhler, and Fichera established that clindamycin and a few other antibiotics target ribosomes encoded by these novel DNAs found in the parasites and that these sequences are associated with a distinctive organelle surrounded by four membranes. They dubbed the newly discovered organelle the apicoplast. The organelle was acquired, evolutionarily speaking, when an ancestral parasite “ate” a eukaryotic alga that already harbored a plastid through the endosymbiosis of a free-living cyanobacterium. Unlike plant and algal plastids, the team showed, the apicoplast had lost the ability to photosynthesize, but the organelle was essential for the parasite, including for protein synthesis.

The acquisition of apicoplasts by certain parasites studied by the team was the first proven example of secondary endosymbiosis—a “nesting dolls” phenomenon in which a eukaryotic cell swallows another eukaryote, which itself had a prokaryotic cell already residing in it.

For Roos, the story highlights the “remarkable cell biological and evolutionary insights that can be gleaned only by studying eukaryotic diversity.”

An early computer programming start

Although Roos made a name for himself when the discovery of the apicoplast made it into textbooks, biology was not his initial career ambition. He enjoyed the natural world growing up—hiking, skiing, and exploring the mountains and forests around his hometown—but his interest in computers also had an early start.

Born in 1956 in Boston, Massachusetts, Roos was raised in Hanover, New Hampshire, his father a biology professor at Dartmouth College. Roos’s elementary school participated in an innovative pilot program initiated by Dartmouth’s John Kemeny, a computer scientist who co-invented BASIC, one of the most widely used computer languages worldwide. Kemeny had a vision that computers should be accessible tools for everyone, and Roos, along with his second-grade classmates, was taught how to program.

“This was visionary in the 1960s when computers were seen as tools for computer scientists that would benefit the public indirectly,” says Roos. “Most people didn’t think that computers would be something everyone would use in their daily lives.”

Roos continued to program through high school but also had many other interests, including art and math. During that time, he took college math classes at Dartmouth, and when he applied for college his senior year in 1973, Roos was “shocked to discover the backwards computer science technology at both MIT and Stanford” compared to what he was used to at Dartmouth. “I visited MIT, and they told us with great pride how they could have twenty users time-sharing their computers, when at Dartmouth, we had ten times that many computer users at the same time.”

Dismayed, Roos decided to enter the workforce instead, becoming a computer programmer for the computer company Honeywell. After a year and half on the job, he applied to colleges, entering Harvard University in 1975.

First art, then science

Initially, Roos was an art major, taking both studio art classes and art history. “It’s a pet peeve of mine that some find a contradiction between interest in art and science. I find that science is a very aesthetically pleasing and artistically driven discipline in many respects,” he says.

Eventually, though, biology won out. Roos was singled out by a biology teaching assistant who suggested he approach Morris Karnovsky, a cell biologist and morphologist, about working in his lab. Roos accepted the nudge, and in Karnovsky’s lab he researched how cell membranes change during polyethylene glycol (PEG)–induced cell fusion. He also characterized the membranes of mammalian cell lines he had isolated that could resist PEG-induced fusion in Richard Davidson’s lab, then at Children’s Hospital Medical Center in Boston. These hands-on experiences made Roos switch his major to biology and also apply to graduate school.

Starting from scratch

Initially, Roos applied to MD/PhD programs, but the admission committees gave him a hard time, he recalls, questioning his commitment to the field based on his prior meanderings in computer science and art. Even though he thought he was only interested in MD/PhD programs, on a whim Roos accepted an invitation from Rockefeller University in New York City to interview for the bioscience PhD program and began his graduate studies in Purnell Choppin’s lab in 1979.

There, Roos found that the fusion-resistant mouse cell lines he had generated at Harvard had membranes with a distinct mix of lipids that directly correlated with their reduced ability to fuse with other cells, suggesting that the types of lipids present in the cell membrane control whether cells could merge with each other. He also found that cell fusion was not guided by changes in the fluidity of the cell membrane, which was previously hypothesized to be the way membrane fusion occurs.

It was a really terrible idea to do something you’ve never done before when starting your own lab. You can’t train people in something you have never done before!

In 1985, Roos joined Robert Schimke’s lab at Stanford University as a postdoc, to get the molecular biology training he was lacking. He studied the action of antifolate drugs—a class of chemotherapy agents that block the activity of folic acid—and developed an assay to test sensitivity and resistance to these compounds in cultured human cells.

Roos applied for a Markey Trust fellowship, which provides funding for newly minted professors to start their own laboratories, but initially had no idea what research questions he wanted to address. He settled on parasitology for the “not very good reason that the research was relevant to parts of the world where I wanted to travel,” says Roos. After perusing the literature, Roos concluded that what was missing from malaria and other parasitology investigations were molecular genetic tools.

He came across the work of Dartmouth’s Elmer Pfefferkorn, who had developed a plaque assay in which Toxoplasma would infect and lyse mammalian cells, in turn infecting their neighbors and generating plaques—pockets of lysed cells. The assay, Pfefferkorn had shown, demonstrated that generating Toxoplasma mutants was possible and could be used to do genetic crosses and identify biochemical pathways. “I proposed to develop an in vitro Toxoplasma culturing and transfection system to be able to add genetic markers in order to do molecular genetic experiments,” says Roos.

Roos’s funding proposal failed to convince the reviewers, but his interest in parasitology nevertheless became invaluable to his career, establishing the research direction he wanted to pursue in his own laboratory.

A bold move leads to biomarker success

Researchers had previously tried to create genetic tools to manipulate Toxoplasma and Plasmodium by using the same type of anti-biotic-resistance biomarkers that were successful in yeast. But these attempts failed, and molecular biology experiments had stalled in the field of parasitology. So Roos took a different genetic marker tack, taking advantage of what he knew about antifolates as a tack to clone the organisms’ genes for dihydrofolate reductase, an enzyme required for the synthesis of the purines adenine and guanine and of some amino acids, and a target for certain chemotherapy drugs and antimicrobials. Roos began the genetic marker project while still in Schimke’s lab, but was only successful after he started his own laboratory at the University of Pennsylvania in 1989.

He exposed parasites to antifolate drugs and selected cells that were resistant to the treatment and therefore must have had a mutation in their dihydrofolate reductase genes. In 1993, he cloned the gene for dihydrofolate reductase–thymidylate synthase (DHFR). With postdoc Robert Donald, Roos created a plasmid with a mutated version of the gene, which, when introduced into the parasite, results in antifolate resistance and can be used as a selectable marker. In 1998, using this genetic tool, Roos and postdoc Mary Reynolds uncovered how mutations in certain amino acids within the gene for DHFR alter interactions between the DHFR protein and antifolates, resulting in drug resistance.

I find that science is a very aesthetically pleasing and artistically driven discipline in many respects.

Looking back, Roos sees his younger self as overly bold. “If I had had any sense, I would have done a second postdoc and learned how to do this work with Pfefferkorn,” says Roos. “It was a really terrible idea to do something you’ve never done before when starting your own lab. You can’t train people in something you have never done before!”

Roos also learned an important lesson from his Penn colleague Lewis Tilney. Roos initially wanted to have his postdoc advisor, Schimke, listed as an author on the dihydrofolate reductase paper. “The first draft listed Schimke because I started the work in his lab, but then Lou read the paper and said, ‘You’re the one that did all of the work. If you don’t call Schimke and tell him that he doesn’t need to be an author on the paper, then I will.’ So I called and of course Bob was very gracious and said that this was always my project. Lou had given me good advice, and I try to maintain that if I am not directly involved in the work, even if it’s done in my lab, then I am not an author on a paper.”

Launching a bioinformatics career

Even with his roots in laboratory biology, since 1998 Roos’s work has become subsumed by data management and integration. These days, with just a few postdocs and graduate students, his lab has “dwindled almost to vanishing,” he says.

Since 2000, much of Roos’s time has been spent leading a team that is responsible for supporting EuPathDB, the Eukaryotic Pathogen Genomics Database, a catalog of parasites and other pathogens. He helps grow and manage the database as new omics data are generated by researchers around the world. A full-time staff of almost 50 people now maintains and updates the collection, funded mostly by the National Institute of Allergy and Infectious Diseases.

The database began as a research question. After discovering the apicoplast in 1997, Roos’s lab wanted to understand the functions of the organelle in the context of other endosymbiotic organelles, and turned to evolutionary biology. Roos’s team identified some of the key apicoplast and nuclear genes necessary for the function of the organelle.

In 1999, as the human genome was being assembled, the full genomic sequences of Plasmodium and Toxoplasma were still incomplete. Roos recognized that the parasite sequences that were emerging needed to be compiled, but that manually comparing sequences was not feasible. He also realized that the small computer programs his lab members had written to analyze their sequences of interest could be modified and used by others to ask their own scientific questions. That was the origin of the initial Plasmodium falciparum Genome Database (PlasmoDB) that Roos, along with then-postdoc Jessica Kissinger, launched in 2001. “Its success has come to dominate my life ever since,” he adds. The database continued to grow and now includes genomic and other omics data on 285 organisms, from fungi to pathogenic parasites.

Roos’s initial interest in computer science has come full circle. “I want to create tools that allow other users to ask their own questions, rather than my being the ‘big computer scientist’ who does the analysis and then tells others an answer,” he says. “That philosophy is in line with what John Kemeny had in mind when he wanted anyone to have access to computers as a resource and to use them the way they want.” 

Greatest Hits

  • Isolated and biochemically characterized the lipid membranes of mutant mammalian cells altered in their membrane fusion properties
  • Created assays to determine drug targets and resistance mechanisms in mammalian cells and protozoan parasites
  • Developed molecular genetic and cell biological tools, including selectable markers for drug resistance, enabling the functional dissection of Toxoplasma and Plasmodium parasites and helping to launch the field of evolutionary cell biology
  • With lab members Sabine Köhler and Maria Fichera, uncovered a novel organelle unique to parasites belonging to the phylum Apicomplexa, including Toxoplasma gondii and malaria-causing Plasmodium  
  • Developed the Eukaryotic Pathogen Genomics Database (EuPathDB.org), a data-mining resource facilitating the interrogation of multi-omic datasets relating to hundreds of microbial species, hosts, and diseases

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?