Skip to Main Content

“OK, smell this,” says Jeff Wagner, a Harvard postdoc of immunology and infectious disease. He’s pointing to the two flasks of bacteria in front of me. I breathe cautiously from the first vial of yellowish broth. Nothing. “Now this one,” he says, with an identical looking sample. And I’m hit with it: a distinct mint aroma like someone melted down a pack of breath mints.

These minty bacteria are genetically engineered relatives of Mycobacterium tuberculosis, the bacterium that kills 1.5 million people each year. Thankfully this strain — Mycobacterium smegmatis — is harmless. But it’s a close enough cousin that scientists can use it as a proxy for the real thing.


And though a mint-scented bacterium might seem like a silly achievement, it’s part of a serious strategy by a team of Harvard scientists to speed up discovery of a better tuberculosis vaccine. Their goal: to modify the germ so that it can be safely given to people to test a vaccine – and if the vaccine doesn’t work, that the participants can be cured.

Challenging studies

The only existing tuberculosis vaccine — known as Bacille Calmette-Guérin (BCG) — contains a weakened bacterium that is a different cousin of Mycobacterium tuberculosis. And while the vaccine is 60 to 80 percent effective in children, it works poorly in adults. Partly for that reason, while BCG is widely administered in many parts of the world, it’s not used in the United States.

Multiple other vaccine candidates are currently moving through clinical trials, with some encouraging early results. A better vaccine could save lives on a stunning scale. Annually, tuberculosis kills more people than HIV or malaria, though the number of deaths has fallen over time.


But clinical trials on vaccines can be tricky. The way such studies are typically done, a large group of people is randomized into two groups, vaccine and placebo, and then they’re observed over time. But because only a small subset of people will even come in contact with the microbe, such studies must monitor large numbers of people over long periods of time, which makes trials slow and expensive.

There’s an alternative: infect small groups of people on purpose. That’s what so-called “challenge studies” do, and they’ve been used for a range of diseases, including flu, malaria, and cholera.

But such studies can only be done for diseases that can be detected quickly and treated easily. For example, a doctor can diagnose malaria with a blood sample and a microscope. Thus, if a malaria vaccine or drug doesn’t work, and the volunteer gets the illness, it can be spotted and treated.

That’s not the case for tuberculosis. It can take months for symptoms to develop, after which time treatment is grueling. And because tuberculosis infects the lungs, it’s not easy to tell if infection has taken hold before symptoms develop.

To do a human challenge study with tuberculosis, you’d need a sure-fire way to kill the bacteria if a person’s immune response failed. And you’d need to know the bacteria were gone.

So Wagner has for the last three years been leading a project to genetically modify tuberculosis to make it useable for challenge studies. Genetically modified germs have been used in other challenge studies, but this would be the first case where researchers used the genetic modifications to control the microbe’s survival.

The idea isn’t for challenge studies to replace large studies of natural infection, but rather to precede them, said Wagner. “If you give the person the vaccine and the vaccine can’t even work on this well-controlled challenge strain, then it doesn’t make sense to do the much larger-scale study,” Wagner said.

Any such bacteria would be at least three years away from ever being tested in people — and that’s in a best-case scenario. But the potential is enough that the Gates Foundation has come aboard as one of Wagner’s funders.

Kill switches and minty scents

Wagner’s efforts fall into two main areas: adding “kill switches” to control the microbe’s survival, and adding scents (like mint) to help detect when the bacterium is present.

(Most of this work happens in Mycobacterium smegmatis — or, as microbiologists call it, smeg. Smeg is closely related to Mycobacterium tuberculosis, but grows faster and isn’t infectious — which makes for quick and safe experiments.)

On the first effort, Wagner is taking aim at one of the bacterium’s essential proteins: an enzyme that helps build its cell wall. The microbe can’t live without the enzyme. And Wagner, in turn, is making sure that the enzyme can’t form without his help. He’s tweaked the enzyme to require an artificial amino acid not found in nature but synthesized in the lab. As soon as the amino acid is withheld, the bacteria start dying.

These bacteria, called Mycobacterium smegmatis, have been genetically modified with a kill switch — an artificial amino acid they require to proliferate. Left: switch off. Right: switch on. (Videos shot over 48-hour period.) Jeff Wagner/Harvard University

It’s a strategy that Harvard geneticist George Church and others have already used to control the growth of E. coli. Wagner is now taking this system — which works well in smeg — and putting it into Mycobacterium tuberculosis. In theory, a human challenge strain designed this way would die even if a person’s immune response flops, since the amino acid wouldn’t be found in the body.

But microbes have a pesky way of mutating to avoid things that kill them, so Wagner’s goal is to combine multiple kill switches. Researchers at Weill Cornell Medicine in New York, for instance, have engineered Mycobacterium tuberculosis to burst open on demand. When Wagner combined these two tweaks in one strain of smeg bacteria, only 1 in every 1014 microbes escaped both kill switches.

And that leads to the second of Wagner’s efforts. These switches are important, but researchers need a way to know if they work in an actual human challenge. So Wagner’s plan is to check people’s breath.

Scientists at Dartmouth have shown that compounds made by bacteria in the lungs show up in the breath of patients. Wagner is working with these researchers to detect his tuberculosis strain, but with a twist — he’s making bacteria produce molecules usually found in mint, banana, and cinnamon.

Mint is the front-runner so far, as it’s been difficult to get the bacteria to reliably make the other compounds. The idea would be to scan a person’s breath using chemical tests and identify the amount of mint compound it contains — which would reveal the amount of bacteria present. Wagner and colleagues are setting up equipment to do this in mice.

Long road ahead  

Wagner has spent all three years of his postdoc on this project, but there’s still much work to be done.

Experiments that work in smeg will need to be adapted to Mycobacterium tuberculosis, and may require some tweaking to work. Because tuberculosis is spread through the air, this work will happen in an extra-secure facility — where researchers don full-body suits, two pairs of gloves, and carry their own supply of fresh air. Working in these rooms — where biologists look like astronauts on a space walk — takes twice as long as usual.

The tuberculosis challenge strain will then need to be tested in mice and monkeys. For this step, Wagner will need to find collaborators, as his lab doesn’t work with monkeys. And a single monkey costs over $4,000, making these experiments expensive.

But, if all goes well in that step, it will then be time to test the challenge strain in people.

And Wagner said he’d be one of the first to sign up.

“I would never advocate for it unless I believed it was completely safe … [and] the easiest way to put my money where my mouth is would be to volunteer,” he said.