Well into the 20th century, before the advent of antibiotics, people routinely died from simple bacterial infections.
In 1924, for example, President Calvin Coolidge’s son died within a week because of an infected blister he developed while playing tennis, despite access to the very best medical care available. The rise of antibiotic-resistant bacteria threatens to take us back to an era when that sort of thing was the norm.
This isn’t only true for people; recently the bacteria in farm animals have been showing increased resistance to antibiotics as well, as we keep hammering livestock with overuse of these drugs.
Clearly, we need to be much more judicious about the use of antibiotics, but we also need new approaches to fight bacterial infections.
Besides antibiotics, there is another natural enemy of bacteria, one that hasn’t been commonly exploited in therapy: bacteriophage, or just “phage” for short. These are viruses that infect bacteria, and they are devastatingly good at it.
A bacteriophage starts by injecting DNA into a bacterial cell:
Then, the phage’s DNA commandeers the machinery of the cell, causing it to make about 100 new little phage particles in as little as 20 minutes. Then, the cell explodes!
Ouch.
So if these bacteriophages are so awesome at fighting bacteria, why haven’t we used them like antibiotics?
For one thing, bacteria evolve resistance to phage pretty quickly. In a one-on-one battle with phage, bacteria will generally win, because they are better and more creative at evolving than phage are. In the big picture, that’s OK with the phage, because if they eradicated their hosts entirely, they’d have nowhere left to go. So it’s a balance.
Also, phage are very specific. That’s great news for your cells, because bacteriophage don’t recognize them at all. But phage will generally only go after one species or even subspecies, so for any particular bacterial infection, you’ll need a really good phage match. That leads to sad stories like this:
The researcher couldn’t get Mallory Smith’s story out of her mind. Smith was a 25-year-old cystic fibrosis patient, and she was near death at a Pittsburgh hospital, her lungs overwhelmed by bacteria. All antibiotics had failed. As a last resort, her father suggested an experimental treatment known as phage therapy.
That meant giving her viruses known as bacteriophages — phages for short — which naturally parasitize bacteria. But not any phage would do. Smith needed one perfectly evolved to kill the microbes in her lungs. Urgent messages were beamed around the world, over email and Twitter, from one phage researcher to another. Was there someone, somewhere, who had the right virus tucked away in a fridge at the back of a lab?
That’s how Jessica Sacher heard about Smith. As a grad student experimenting on phages at the University of Georgia, she’d stumbled across a STAT article about the desperate virus hunt. She mentioned it to a friend she’d met through swing dancing, a tech consultant and developer named Jan Zheng. He knew there had to be a more efficient way to find the right virus when a patient’s life depended on it. “Using Twitter for this kind of stuff is ridiculous,” said Zheng. “It won’t scale to more than one or two patients.”
Over breakfast on Nov. 13, the two hatched a plan to create a Phage Directory to accelerate the search for a perfect match. That same day, scientists found a phage that could kill Smith’s bacteria, but she died two days later at the University of Pittsburgh Medical Center. The website was launched just two days after that.
Not to mention that releasing an exotic phage that’s untested in humans could lead to something akin to toxic shock syndrome if the immune system decides it doesn’t like what you’ve done.
Things like the Phage Directory are noble efforts, to be sure, and they will probably help save a few people, but we should be able to do a lot better than this.
Enter Timothy Lu and his team of researchers at MIT. They’ve been hard at work on this problem, and their October 3 paper in the journal Cell is a promising advance. As I usually do, I’m giving all of their names, because they deserve that: Kevin Yehl, Sébastien Lemire, Andrew C. Yang, Hiroki Ando, Mark Mimee, Marcelo Der Torossian Torres, Cesar de la Fuente-Nunez, and Timothy K. Lu.
They’re using synthetic biology to attack this problem, and it’s sharpened up the power of evolution for these phage, enabling them to outfox the bacteria. Synthetic biology is a marriage of engineering and biology that takes advantage of our recent huge improvements in making and testing gigantic libraries of biological components like DNA.
Let’s look at a phage particle to understand what these researchers have done:
We’ll zero in on that “tail fiber”. It’s the landing gear the phage uses to recognize its host. Here’s what the tip of that tail fiber looks like when it’s pointing at you, magnified approximately a zillion times:
The tail fiber is made of three copies of a single protein called gp17. That’s why it looks kind of triangular. The legend shows you the colors of the “loops”, which are short little connectors that hold the structural (blue) parts together. The blue parts don’t change much from phage to phage, but the loops do. The authors figured that these variable loops must be what gives a phage its ability to recognize a particular bacterial host.
The loops correspond to specific little regions in the gp17 gene:
The colored line (which reads like text, from left to right) represents the DNA, and the letters below it are the amino acids that the DNA’s instructions call for. So if you look at the red region (the “HI loop”), you see “DAPP” below it. That means the gp17 protein is going to have an “HI loop” made up of those four amino acids. Let’s go to our amino acids chart to see what they are! Even biologists forget the abbreviations sometimes….
So the “DAPP” in our “HI loop” is going to be aspartic acid-alanine-proline-proline. (Why did they abbreviate “aspartic acid” with a “D” anyway? I don’t know!)
The “HI loop” is really small, so it’s going to take a long time for natural random mutations (accidental changes in the DNA that happen from time to time) to alter that teeny region.
Meanwhile, bacteria have literally hundreds of ways they can change what their surface looks like. Random mutations can occur in lots of places within bacterial DNA that would cause their surfaces to look different and not be recognized by the phage anymore.
So the phage has the key, but the bacteria can keep changing the lock without too much trouble. By the time the phage stumbles onto the new answer, the new-look bacteria have already multiplied like gangbusters. Not going to work too well for curing an infection. You don’t have that much time to piddle around.
This is where synthetic biology comes in!
Instead of slowly making and trying one random key at a time, we can make ALL POSSIBLE KEYS at once, and as Montgomery Burns would say, “Release the hounds!” We can make a collection of phage with every possible “HI loop”, for example. There are 20 amino acids, so that’s (20)(20)(20)(20) = 160,000 different combinations just in that teeny little loop. For longer stretches like the “BC loop”, there are billions of possibilities.
We can now use this huge collection of phage to pick out our Top 10 list. We set the whole thing loose on normal bacteria, but also on bacteria that have already changed their surfaces in different ways in response to previous phage treatments. Any phage particles from the mix that are good at infecting bacteria are going to emerge from our tests very quickly. If a single phage particle makes 100 copies of itself every 20 minutes and keeps going, it can infect a trillion bacterial cells in 2 hours. Yipe. At the end of our test, the most successful phage types will have the highest population, because they were great at feasting on the bacteria and making more of themselves.
Then we can take our Top 10 Phage Mix and use it on a real sample of bacteria, like an infection. We’ve now got a collection of phages that have already anticipated how the bacteria are going to try to weasel their way out of an attack.
So how about some results?
Lu and his team used normal phage but also their amped-up library on some E. coli. I’m pulling a bit of an Adam Schiff “parody” graph here that gives you the factual gist of what they saw, to streamline things a bit:
Without any phage (purple line), the bacteria grow right up as usual. When normal phage is added (green line), most of the bacteria are indeed killed, but a few mutants that were probably already in there still grow, and after about 12 hours it looks like you never added any phage. But the synthetic mix (red line) does something cool. Those few bacterial mutants do start growing a little bit, but there are already other phage present to anticipate their tricks. Those phage go to work, the clever bacterial mutants are killed, and they don’t come back. Infection gone!
They found that the little “HI loop” was really good at this, but the “BC loop”, for example, wasn’t. So in the midst of these experiments they also discovered a bit about what regions are truly the most important for the phage to recognize their hosts. That means you could also start with phage you know nothing about and whittle it down to the most important regions by trial and error, then focus like a laser on those to make your synthetic phage mix really potent.
They also tried their synthetic-phage treatment on actual E. coli skin infections in mice. A mouse infection should actually be a pretty good model for a human infection, because it’s more about the bacteria and less about the mechanisms of the animal. It’s not nearly as complicated as cancer or diabetes. Even when they used a mix of normal and phage-resistant bacteria to infect the skin, they found that their synthetic phage that did best against bacteria in culture also cleared the infection in mice.
They even evolved phage to attack different species of bacteria entirely, from the same basic phage chassis. That means you wouldn’t need to go hunting for exotic phages to treat infections of different bacterial species, but rather you could start with a basic phage system that tests out well for safety and reliability in humans.
This is still preliminary, and we’re not quite in Miracle Cure Land just yet, but the approach that Lu and his team have developed here looks like it’s got the potential to do very well. There will be hiccups along the way, of course, but this makes too much sense not to ultimately be successful.