In the hunt for novel antibiotics, will new technology overtake underwater exploration?

LAKE SUPERIOR — Choppy, windswept waves slap at the hull as our boat nears the last known location of the Lucerne, a schooner that sank to the bottom of Lake Superior in 1886. The wreck, just off a narrow sand peninsula jutting from the northern tip of Wisconsin, doubles as a suspected habitat for an elusive freshwater sponge called Eunapius fragilis. Finding these tiny aquatic organisms is the reason a trio of young scientists have set off with a local divemaster. The hunch is these sponges could be a source for new chemical molecules, which, in turn, could be the basis for new antibiotics. Looking north under overcast skies in late August, the lake stretches all the way to the gray-blue horizon line, which appears to seesaw back and forth.

“Time for a little search-and-find,” says Brian Murphy. A chemist at the University of Illinois Chicago, he has zipped himself inside a sleek, rubberized dry suit. He stands spread-legged, packing peppered beef jerky into his cheek. (Something about suppressing his gag reflex, he explains.) Then, Murphy straps a dive knife to his calf. “We should drop in on the bow,” he tells Chase Clark, who, at the time, was a doctoral researcher in his lab and his seasoned dive buddy.

Murphy plans to remove sponges off the wreck. Clark’s job will be to catch the organisms into clear, cylindrical 50-milliliter tubes. Antonio Hernandez, who joined the lab more recently, is logging GPS coordinates and has agreed to stay on the boat with Terry Bauer, a white-bearded, retired biology teacher who runs a local dive shop.

The odds of finding a dime-sized organism at the bottom of Lake Superior, North America’s largest lake, seems like a fool’s errand. Murphy’s objective in studying Eunapius fragilis is straightforward: Sponges function as a home for microbes, and these organisms compete for space and nutrients using antibiotics and other chemical metabolites that could potentially be put to use as medicine. These particular sponges have not been well-studied, so their geographic range and the odds of actually finding them on the lake bottom is unknown.

Murphy and Clark pull on masks, plug regulators into their mouths, and slip over the port-side rails. Some 20 minutes go by. The queasy swaying of the boat makes it feel even longer. Eventually, Murphy and Clark surface, lurching about as they find their feet again back on deck. They found nothing: Not the Lucerne. No sponges, either. Murphy, who’s ordinarily chatty and upbeat, has a subdued look. He asks Bauer about going somewhere “softer.” Then, he kneels next to the outboard motor and vomits into the lake.

The third day of the five-day expedition begins with a sample bag filled with nothing. In the field of antibiotic research, that’s a defining feature of our time. From the 1930s to the 1960s, scientists isolated dozens of antibiotic drugs in rapid succession, often by chance, but the field has since plunged into the so-called discovery void — a gap that dates back to 1987, when the last truly novel antibiotic made its way to market.

The rise of superbugs resistant to existing antibiotics adds an urgency to the search for new drugs, and, increasingly, the seekers are not venturing into the field. They’re sorting through massive libraries containing DNA sequencing data, taking advantage of recent advances in computational software and machine-learning algorithms to identify — and then try to predict and synthesize — molecules with potent antimicrobial properties. This strategy is not only more efficient and more likely to be funded than expeditions like Murphy’s, its practitioners say, but it raises a deeper question: Is it even necessary to physically explore new places on earth to find new drugs?

Murphy isn’t ready to abandon ship. He is among those dedicated to pairing old-fashioned exploration with 21st-century technology, believing that biology will continue to drive next-generation drug discovery, that the real world is the ideal crucible for forging potent new compounds. Still, he acknowledges, there’s no way to overcome all the uncertainties. “We have no idea to what degree we’re going to find sponges up here. We have no idea. No clue,” Murphy says. “That’s why it’s total exploration here. You hit the waters and start searching.”

Murphy began systematically searching for new chemistry in organisms when he was a doctoral student in David Kingston’s lab at Virginia Tech. Kingston is best known for his work on paclitaxel, a chemotherapy derived from Pacific yew trees. Murphy did not see himself tromping around in jungles full of plants and “spiders the size of your head” (he is a serious arachnophobe). Eventually, though, he came to see that aquatic environments offered some distinct advantages.

The field traditionally had a terrestrial bias or what’s sometimes referred to as a surface chauvinism, which overlooks the 70% of the globe covered by water. “If you look at the history of where antibiotics came from, 95% are coming from microbes from the soil,” said Bill Fenical, Murphy’s former adviser at the Scripps Institution of Oceanography in La Jolla, Calif. Scientists studied on terra firma.

The search underwater is predicated on two guiding assumptions: Studying a broader spectrum of biodiversity will likely uncover a greater percentage of the world’s untapped chemical diversity; and the vast majority of aquatic organisms have never been characterized and remain virtually unknown. Beyond venturing into previously underexplored ecosystems, testing the waters made sense in other ways. The HIV drug AZT and several cancer drugs were developed from marine organisms.

As for sponges, these animals harbor microbes, and the microorganisms act sort of like hypercompetitive condo dwellers jockeying for prime real estate — except that, unlike terrestrial organisms (say, a fungus attacking a tree), chemistry is dissolved into the aquatic environment. Underwater, the chemical back-and-forth is constant. This unrelenting arms race results in small molecules, which microbes use to fend off each other and could potentially be repurposed into fighting cancer and infectious disease.

Phil Baran, a chemist who co-founded Sirenas Marine Discovery in La Jolla, said there’s another reason to suspect the chemistry produced by marine microorganisms may lend itself to clinical applications. The biological pathways that underpin human development are similar in nature, and these organisms produce molecules that share qualities researchers look for in a drug that dissolves in the human bloodstream. “The things that you need for a good drug — self-permeability, solubility, stability — are often pretty congruent with the types of properties that you see from marine metabolites,” he said.

Finding new organisms is only the first of many difficulties scientists in Murphy’s line of work face. His lab, located on the third floor of a brick tower in Chicago, houses rows of benches, glass beakers, and thousands of Petri dishes containing microbes he’s collected all over the world. Like many other researchers studying microorganisms, Murphy found the old-school method of culturing bacteria and fungi to be frustrating. The process involves too much guesswork. Most microbes refuse to grow in the lab. Organisms that did grow were repeats. In 2018, working in collaboration with Laura Sanchez (then at the University of Illinois Chicago and now at the University of California, Santa Cruz) and several other colleagues, Murphy’s team devised what he refers to as a “pretty kickass method,” and published the details in the Proceedings of the National Academy of Science.

Instead of plating microbes into Petri dishes and seeing what randomly grows, his lab could now prepare a flat steel plate with hundreds of samples, all at once, and feed it into a mass spectrometer. The instrument then spit out a revealing chemical fingerprint. Then, the team analyzed the reams of data with software able to pluck out just those microbes that could be grown, and the ones most likely to produce novel molecules with the potential to be useful antibiotics.

It not only sped up the process, but avoided redundancies. Some of the only published reports of antibiotic compounds found in freshwater sponges are from Murphy’s lab.

The dive trip, of course, was only the first step in generating leads, which represents the starting point for an arduous, yearslong process of developing a drug for everyday use. Time was ticking, and it wasn’t just the emergence of deadly drug-resistant infections. Over the last half-century, Lake Superior is warming at a faster rate than ambient air temperatures in the region.

One morning during the trip, a perky volunteer with a clipboard warned Murphy’s crew about cyanobacteria, an unexpected algae bloom that has been clouding the lake. Sponges are sessile filter feeders. They don’t move around all that much, and respond quickly to any changes in their environment. These animals are not like polar bears or other charismatic icons of cataclysmic change. And yet, the world’s oceans and lakes bear the brunt of human activity: These ecosystems are changing so dramatically that the biological and chemical diversity could be disappearing before we know what’s out there.

Coastal development, nutrient runoff, coral bleaching, and climate change all mean one thing: It’s now or never.

The operational headquarters for the dive trip is a cabin Murphy rented that resembles a prefab single-wide trailer gussied up with wood paneling and fishing-net curtains. The first night in Bayfield, Wis., the crew unloads boxes, growlers of pure alcohol, beakers and glassware, and a case of beer. Murphy chops firewood, and they roast marshmallows around a smoky campfire.

None of them has actually seen a freshwater sponge. Late that night, they pass around a glowing iPhone, trying to glean some intel from an email sent by an Italian researcher who is one of the world’s only experts in identifying sponges. To say that little is known about these creatures is something of an understatement. To say that Murphy is obsessed is to put it mildly. “They’re out there, and, by God, we’re going to find at least one,” he says. Dramatic pause. “I think.” The next morning, Murphy admits he was on his phone until 3 a.m. scouting for potential dive sites.

Two days later — one day of prep, and one morning trying to find the Lucerne — Bauer motors out to Stockton Island. It’s part of the Apostle Islands National Lakeshore, an archipelago of 22 islands containing 42,000 acres of federally protected wilderness, in addition to areas that are underwater. The wilderness designation also obscures its industrial past. Below us, in emerald-green waters, we see shadows of a submerged dock, where the brownstone that built Chicago was loaded onto boats. The island, once clear-cut, now grows with a dense boreal forest that rises steeply from the water line.

“How deep are we?” Murphy asks.

Bauer looks at his depth finder. “Thirteen feet of water.”

Murphy turns to Clark, and proposes their next dive. “We’re going to drop down and run directly west. Go down to 20 or 30 feet, and then slip up into shallower waters and head back east.”

He packs some beef jerky into his mouth, gears up, and slips overboard. Murphy dips his face down, and then shouts triumphantly back at the boat. “Beautiful visibility! Oh, man, you couldn’t ask for better conditions right now!”

They surface with nearly half a dozen sponges. Up close, the sponges look like bright-green, miniature volcanic craters. Suspended in clear tubes, they like floating balls of mucus.

Globally, the race to find new antifungal, anticancer, and antibiotic drugs involves startups that have attracted hundreds of millions of dollars in investment by touting an alternative approach — one that attempts to forgo the guesswork of fieldwork. The shift has been accelerated, in part, by dramatic drops in DNA sequencing costs, which allows researchers all over the world to “digitize” biological samples and transform physical specimens into genetic sequences that can be uploaded to databases.

And so, rather than going out to collect microbes or other organisms in nature, researchers computationally mine existing datasets with algorithms in an attempt to optimize their search results, if you will. Nathan Magarvey, a researcher at McMaster University in Hamilton, Ontario, and founder of Adapsyn Bioscience (which has a partnership with Pfizer), said machine learning software is now being used to catalog and filter vast libraries of data. “You can do the ‘hunting and gathering’ in a much more sophisticated manner with genomic data in hand,” he said.

Algorithms are also being used to predict chemical structure and function from sequence data. Much like online retailers leverage data to serve up product recommendations or personalize films to watch, these tools suggest designs for chemical molecules that might fare best against bacterial and fungal pathogens.

“It will be the future. There is no question,” Magarvey said. “We will not do synthetic chemistry the way we used to.” If the face of drug discovery was once the pipe-smoking professor, he said, then its future will be driven by an AI-driven workflow, guided by brains of silicon.

In 2020, one team of researchers, led by Massachusetts Institute of Technology bioengineer James Collins, announced that a deep-learning algorithm identified a new antibiotic with an unconventional mechanism of action. Artificial intelligence predicted what structure might work as an antibiotic by sifting through a pile of existing drug compounds. Collins sits on the scientific advisory board of Phare Bio, a nonprofit that, he said, is now advancing the most promising AI-discovered candidates toward clinical trials. But he and many others see the application of computational tools as complementary, and he said his findings do not suggest a “legacy approach” is obsolete.

Algorithms need data, and so far, these data still come from microbes and genetic material from environmental samples. “In contrast to synthesizing human designs, which often fail,” said Devin Scannell, vice president for strategy at Zymergen, an Emeryville, Calif., biotech with a huge dataset of novel microbial genomes, “metagenomics gives us access to molecules and materials that have been optimized and ‘validated’ by nature.”

It is, of course, possible to combine new techniques with the old school, tried-and-true. Either way, the biggest problem lies ahead: The vast majority of new molecules turned up in the discovery phase fail before they ever go into clinical trials. As Terry Roemer, a founder at Prokaryotics, a Merck spinout in New Jersey that is focusing on finding entirely novel antibiotics, put it, “I don’t think there’s enough companies swinging for the fences.” There might be a shared optimism in scientific circles, but that enthusiasm had not carried over to the marketplace.

On the last morning of the trip, as dawn light fills the cabin, the crew loads coolers into the van — sponge samples prepped for the lab. Murphy decides to go back and find the Lucerne. Bauer motors out to the familiar dive site under late summer skies. The lake is dead flat and calm. As he pulls on gear, Murphy packs his cheeks with beef jerky. “We’re finding this wreck,” he says. “That’s the thing about the Irish. Slam us down 500 times, and we’re still going to get up off the mat.”

He’s first in the water. Clark follows, and they head eastward, swimming abreast. The sandy bottom is rippled by the current and the wind. No one sees the giant wooden hull of the Lucerne. Murphy motions to stop. He draws a V in the sandy bottom, a crude map of where they’ve already gone. Clark notches another V, proposing a path back to the boat. On the way back, they see nothing resembling a wreck. Murphy checks his compass and gives the thumbs up, the signal to surface. They pop their heads into the blinding light. The search is for naught.

Back on board the dive boat, Bauer is shaking his head. He cannot believe it. He takes off his shirt, pulls on scuba gear, and jumps overboard on a solo dive — a last-ditch effort to find the Lucerne, which has twice now eluded Murphy’s team. The wreck probably harbors a sponge that looks like snot and smells disagreeable but almost certainly contains a previously untapped reservoir of chemical molecules.

Now, though, the boat has come unanchored and, as it drifts away, the shoreline grows distant. Hernandez picks up binoculars. No sign of Bauer. It’s almost noon. Murphy is not about to give up. In fact, he’s confident he’ll still have time for yet another dive before making the eight-hour drive back to Chicago. To his mind, just because the search hasn’t yielded anything so far does not mean there’s nothing to find.

“I’m sure there’s a metaphor in there somewhere,” Murphy says. “I’m too far in to know what it is.”