Even after COVID-19 is controlled, climate change, growing resistance to antibiotics and lack of clean water will still be waiting for us. We need research that could change… everything. That’s what Blue Sky is all about.
Evolution provided bacteria with a mechanism to overcome our antibiotics, and each time we introduce a new antibiotic, they get faster at becoming resistant to it. At the same time, it can take 14 years and billions of dollars to bring a new one to market. Angela Violi, professor of mechanical engineering, is working to change the game by developing antibiotics in a new way. Her team wants to use machine learning to design nanoparticles that exploit weaknesses in the target pathogens. Not only could they work better, they could make it to market faster.
About the podcast: The Blue Sky podcast is a limited series from RE: Engineering Radio and the University of Michigan College of Engineering. It delves into the four research projects funded as part of the $6 million Blue Sky initiative. Launched in 2018, the initiative gives research teams the freedom to try daring ideas, show results and build momentum to secure further research investment in their efforts to solve global problems. Season 1 is an introduction to each project.
Jim: Welcome to the Blue Sky podcast: nanotechnology versus the superbugs. I’m your host Jim Lynch.
Nicole: And I’m Nicole Casal Moore.
Jim: Today, we’re going to talk about a new way to fight a growing threat to modern medicine. Nicole, did you know that antibiotic-resistant superbugs will kill someone in the United States every 15 minutes?
Nicole: Technically, I did know that because it came up at the news meeting this morning. I remember something in one of your stories about how long it takes to develop a new antibiotic and how difficult it is.
Jim: For years and years we’ve been able to sit back and just rely on old antibiotics. Situations where one didn’t work, you just moved onto the other one. And then in the worst case scenarios where there were multiple drugs that didn’t do the job, at one point you sort of bring in the antibiotic of last resort, you know, there are the one that are the heavy hitters that are supposed to just come in and do the job. And some of those don’t work now either. So we’ve been able to rely on old drugs for far too long. And now that we need new ones, the economic incentive isn’t there. Actually getting something from the early stages through testing, through trials and then onto market, it takes an awful long time and antibiotic resistance is not sitting around waiting.
Nicole: Do you have any good news or is this another episode of Jim’s podcast of doom?
Jim: All I can say is something about your climate change episode, but I’ll take the high road here ’cause in fact, I do have good news. Engineers are getting into the antibiotics game with a mix of big data and made-to-order nanoparticles that they’re calling nanobiotics.
Nicole: Nanobiotics. It sounds small.
Jim: Yes. Nano is actually the Greek word for “wee”. Actually they’re relatively big. Regular antibiotics tend to be small molecules, just a few dozen atoms stuck together. Nanoparticles, on the other hand, tend to be a few hundred atoms across.
Nicole: Okay. But why would we want nanoparticles as antibiotics?
Jim: I’m gonna get to that but first I want you to hear Angela Violi, a professor of mechanical engineering and the leader of the Blue Sky project on nanobiotics. Let’s hear how she explains where we are with regular antibiotics.
Jim: What’s the situation with antibiotics right now? Why is there a need to develop a new class of treatments?
Angela Violi: Over the last 75 years, we’ve been using those miracle drugs that have cured any type of infection that we have been seeing. But there are already infections across the world for which, out of the hundred antibiotics that we have, maybe one or two work. And in some cases, none of them work. The problem is that bacteria are very smart. And so what they have done is become resistant to our antibiotics at this point.
Nicole: I understand that resistance makes bacteria hard or impossible to kill, but how do they actually become resistant?
Jim: Professor Violi explained that evolution gave bacteria the mechanism they need to become resistant to antibiotics, but we’re the ones who gave them the conditions needed to actually do it. When bacteria are exposed to antibiotics, the ones that are susceptible to the antibiotic die. In our bodies, that gives us a leg up so that our immune systems can kill the ones that stay behind.
Nicole: Okay. So too much exposure to antibiotics means that by the time the bacteria are infecting our bodies, it’s already only the resistant ones …
Nicole: … that are left behind?
Jim: Yup. That’s it. And when you dose a person with antibiotics, it doesn’t cut down the number of bacteria. They’re mostly resistant. It’s all on the immune system to sink or swim just like it was before we had antibiotics.
Nicole: That’s pretty grim. So this is all because we go to the doctor with a cold virus …
Nicole: … and often they prescribe antibiotics so we feel like we got our money’s worth for the co-pay?
Jim: Yeah. That’s part of the story. But the other part is in the way we farm. Here’s Angela again.
Angela Violi: One of the main reasons why we have this crisis at this point is that if we look at, for example, the United States, 80% of the prescriptions in the last years for antibiotics were mainly not for humans but more for poultry, for farm conditions—livestock. So, antibiotics are given, you know, to survive the crummy condition of a farm. Once they get in the meat, then they get in the water also, and we are exposed to that once we eat meat for example. Every time the bacteria get exposed to an antibiotic, it has a chance to break the code for the resistance.
Nicole: Eighty percent of the problem is in agriculture?
Jim: Yeah. But things are getting better on that front. As of 2017, antibiotics that are used to treat human infections can’t be mixed into animal feed at low doses anymore.
Nicole: Wait, they were actually doing that before?
Jim: Yeah, weirdly. It resulted in meat that was lower in fat.
Nicole: But we’re still treating sick livestock with antibiotics that are also used by humans?
Jim: Yeah, we also treat groups of livestock when there’s a risk of infection spreading. Other antibiotics used exclusively for animals can still be mixed into the feed. We use a lot of antibiotics in agriculture even now.
Nicole: And then the human prescriptions are the other 20% of the problem?
Jim: Well, not all of that 20% is a problem but the Centers for Disease Control estimate that about a third of those human prescriptions are actually unnecessary.
Nicole: A full third of them. So people who didn’t have a bacterial infection or were going to get better soon anyway. Where is this heading?
Jim: Well, Professor Violi pointed to a study out of the United Kingdom that said by 2050, deaths due to infections could outpace those due to cancer. And those are currently 10 million per year. To put that in perspective, global deaths due to antibiotic resistant bacteria are about 700,000 per year today.
Nicole: Okay. What are the pharmaceutical companies doing about this? Are they looking at this new demand and trying to develop new drugs?
Jim: Now that’s a quirk of capitalist economics. Let’s hear some more from Professor Violi on this.
Angela Violi: If you think about, you know, like a pharmaceutical company, it’s more interesting to put money in a blood pressure drug or cholesterol drugs rather than an antibiotic that people take for like 10 days. The idea is that it’s more convenient to develop a drug that someone takes for the lifetime, the whole lifetime rather than for 10 days.
Nicole: It’s not worth it for them to make new antibiotics because bacterial infections aren’t chronic conditions.
Jim: Yeah, that’s the idea. The economic incentive isn’t there. For me to develop something like an antibiotic for something that’s rare or is not affecting a huge percentage of the population, at best I’m gonna make this much money off of that. At the same time, I can be investing all of my research and development money in Viagra, which apparently is a very popular drug, I’ve heard. So that’s where pharma puts its eggs.
Nicole: You know, it’s hard to get money to treat rare diseases. It’s hard to get money to research them. Maybe the answer is to fund research into innovative ideas like we’re doing right now.
Jim: I also spoke with one of our collaborators, Scott VanEpps. He’s a professor of emergency medicine. Not only does big pharma not have a strong incentive to make new antibiotics, they also don’t have a good technique.
Scott: I don’t think having big pharma make a little tweak around the edges of the current antibiotics that are out there or even just screening these huge libraries of chemical compounds that they have [would solve the problem]. They’ve been doing that. We’ve been doing that for a long time. We’ve pretty much dried up that well in my personal opinion. And so it really does need a different approach to it, and I think that this group is sort of the answer to that.
Nicole: He’s referring to this Blue Sky project?
Jim: Yes. The nanobiotics Blue Sky team.
Jim: When you talk about a different approach, you mean something drastic. I think when you and I spoke a year or two ago, you were talking about the minuscule changes to drugs that can have a big impact, but it doesn’t take much then for the pathogen to make its own changes, right?
Scott: Yeah. They evolve very, very fast, right? It’s a new generation of E. coli every 20 minutes, right? That’s an evolutionary speed that humans don’t move at at all. If you’re just taking a class of antibiotics we already know and give it a little umph here to try get around a resistance mechanism, the next resistance mechanism is equally as fast around the corner. In fact, since the advent of penicillin, the time from the discovery of a new antibiotic to the discovery of its resistance has been getting shorter.
Nicole: That is terrifying.
Jim: Well, it’s not good. This explanation is really illuminating in that it’s not just the way we use antibiotics that’s the problem, it’s the way we make them. We’re always in this evolutionary arms race, and we’re losing. Here’s Angela again.
Angela Violi: The weird thing is that people get excited about drugs for diabetes, for example—like insulin. Insulin can allow you to survive, or a blood pressure drug allows you to survive. But there are only a few drugs that actually can cure you and antibiotics can institute a cure within days. They are amazing. And what people don’t understand is that if we lost antibiotics, we lose a lot of things including a lot of surgery. They require prophylactic, you know, taking of antibiotics [when it involves] anything that can install a foreign object in your body. But I guess more than anything, we lose the ability to, I say, live daily life. Because you know, in the moment you can imagine that even if you scratch yourself, you can get an infection that can not be solved. Then I think your confidence towards life will change.
Jim: Already our deaths in the United States due to antibiotic resistance are about as high as those from car accidents.
Jim: Antibiotic resistant [infections] are projected to increase about tenfold.
Nicole: Or you could get into a car accident, get taken to the hospital and then die of an antibiotic resistant infection.
Jim: You sound like you could use some good news.
Nicole: Tell me how nanobiotics can save the day.
Jim: Well, antibiotic development is usually a trial-and-error process. Researchers test various molecules to see if they can kill bacteria without harming us. Once we have a kind of molecule that works, we make variations on that. That’s how most of our new antibiotics are made. That’s the strategy that is no longer good enough.
Nicole: Like Professor VanEpps was saying about the pharma companies.
Jim: Exactly. Now what his Blue Sky team wants to do is design a way to keep producing new classes of antibiotics. Antibiotics that work in a whole new way. Before we had to find those antibiotics in part through luck. And Professor Violi is trying to make that serendipity intentional.
Angela Violi: The idea is to intelligently design a nanomaterial. The reason why we do a nanoparticle instead of the molecule is that when you have a nanoparticle, your degree of freedom—the ability to design—it’s much bigger. You can look at the shape. You can look at chemical composition. You can look at charges. There are a lot of parameters that you can tweak.
Jim: It’s basically a blank slate for you.
Angela Violi: Yes. That basically you can tweak or tune for your goal.
Nicole: So they have more options for designing nanoparticles than they do with molecules?
Nicole: Too many options can be overwhelming though.
Jim: They’re getting some computational help. It can be overwhelming to try to choose a movie on Netflix but then they have that algorithm that sifts through the galaxy of films for recommendations.
Nicole: Oh, they’re doing machine learning?
Jim: Yes. Yes. And this is one of the really interesting things about it.
Angela Violi: We are also creating a new database, collecting data from different resources. So we are trying to gather as much information as possible from whatever is available in the literature but also in collaboration with other institutions.
Nicole: Help me understand why they need the other institutions if they also have the literature.
Jim: Well, because what they really need is the failures. A failed antibiotic doesn’t get a paper written about it. So you want to basically learn from the mistakes that’ve been made.
Nicole: Yeah. I’ve heard about a movement in science to publish more about what doesn’t work. People have even launched some journals specifically for negative results.
Jim: I’m sure those make thrilling reading.
Nicole: Well those journals tend to fail too.
Jim: Well in this case, Professor Violi’s team does want those boring results. They can feed the data into the nanobiotics computer model. Here’s Scott again.
Scott: It is a hungry monster for data, right? The more data it has, the smarter it gets. And there’s such a dearth of that information out there. We only publish what we’re successful with. If you look at it that way, then the computer just thinks that everything works ’cause it only ever sees positives. We actually benefit from our failures a lot more in this particular case ’cause every failure is a place where the computer goes: this is something I shouldn’t do.
Jim: So we should be rooting for you to fail?
Scott: To some degree, yeah. Yeah, I mean.
Nicole: So we’re hoping for failure?
Jim: Well, just enough failure so that the computer model knows what not to do.
Nicole: Okay. But what do they want the computer model to do? Why do they think that nanobiotics could be better?
Jim: Earlier, we had Professor Violi talk about how they have more degrees of freedom to play with when they’re designing nanoparticles. And they do with conventional drugs. Conventional drugs attack bacteria on just one front but nanoparticles could work in more ways at once.
Nicole: I see. So they could target more of the bacteria’s potential weak points all at the same time.
Jim: That and also one of their big, strong points.
Angela Violi: Biofilm is the structure, the house if you want. They’re bacteria built together.
Scott: And what’s interesting about that is that bacteria that are in a biofilm are pretty resistant to antibiotics even when they are susceptible to them.
Jim: Is that the sort of the strength in numbers aspect for them?
Scott: Yeah. They sort of build, like, a fortress, right? It’s the best way to think of it. And that fortress allows them to sort of go to sleep and not be very active and most of our antibiotics target the things they’re active against.
Nicole: So is it that if the bacterium isn’t active, then the thing that the antibiotic is targeting basically isn’t exposed so there’s nothing for it to target?
Jim: Well, that’s what I got from it but the biofilms have a couple more tricks.
Scott: The other one is they do build a fortress that limits the amount of antibiotic they can diffuse to them.
Nicole: The antibiotic might not even be able to reach the bacteria, let alone hit the target once it gets there.
Jim: That and the biofilms actually have one more trick.
Scott: There are these what we call “persister cells” inside of biofilms that are—they are like the stem cell of a biofilm. So you eradicate 90% of the biofilm and this one stem cell, the persister cell can then start to repopulate it. Those things in biofilms make them very difficult to treat even with conventional antibiotics that an individual cell in the population would normally be susceptible to.
Nicole: Even if we do kill most of the bacteria, if there’s this biofilm fortress thing, and you’ve got one of these nasty persister cells left alive, your whole infection can come raging back.
Jim: Yeah. That’s the challenge. And it’s worse for implants like pacemaker electrodes and artificial knees. Biofilms—they seem to be able to build safe havens around these non-living structures where immune cells can’t come from all sides.
Nicole: So this is what Professor Violi was talking about when she said that eventually we could not be able to perform surgeries. Do I have that right?
Jim: Yeah. This is part of it. Surgeries with implants are gonna be at particular risk.
Nicole: So what can you do about a biofilm?
Jim: You break it up. That’s one of the things that the team is trying to do, and they’ve even had some early success. You’ll hear more about that in a future episode. For now, I wanna talk more about the big picture for this project: the computer model. They wanted to have that up and running at the end of the first year of the project.
Nicole: Wow. They thought they could get that going inside a year?
Jim: Professor Violi is not to be underestimated. The job basically entails collecting whatever data they can find about what chemicals and nanoparticles affect what bacteria, and then making the machine learning computer program that can trawl the database and looks for trends. What features seem to be best for attacking a particular type of bacteria?
Angela Violi: So basically, we wanna be able to classify data. What that means is relate, somehow, the nanoparticle characteristics and their efficacy. Also, we are interested in identifying the hypothesis of the interactions between nanoparticle and bacteria so that eventually the final goal is to have a target profile of nanoparticles.
Jim: Basically, a checklist for what you want a nanoparticle to include.
Angela Violi: So I would like to say to this…engine… if you wanna call it an engine.
Jim: I’m just gonna note here that our producer Kate wanted to call the engine an Oracle. Angela was definitely not comfortable with that.
Nicole: Some kind of prophecy maker?
Jim: A prophet of antibacterial activity.
Nicole: Intelligent designer of nanoparticles.
Jim: I can see you rolling your eyes Kate. Professor Violi went with “computational engine.”
Angela Violi: Engine. If you want to call it an engine, a computational engine to tell me: your ideal nanoparticle should have about this shape, this charge, this composition, so that we can intelligently design that. So the expensive part—like the first step, when we talk about 10,000 molecules—is going to disappear at this point because with this kind of a trick, if you want, we can use only 500 nanoparticles. That’s what we hope. And so 500 experiments are doable. It’s a paradigm shift going from screening compounds to intelligently design. What I mean by “intelligently” is using machine learning to provide a hypothesis of what could be the best design.
Nicole: Wait, they have to start with 10,000 molecules?
Jim: That’s how antibiotic discovery works at pharma companies right now. They’ll make 10,000 molecules because molecules, they just aren’t that hard to make. Then they’ll see how well they work against whatever bacteria they’re targeting. Nanoparticles, on the other hand, those are a bit harder to make. So the Blue Sky team can’t do the brute force route. They need to be smart about which nanoparticles they produce.
Nicole: So they are making this computational engine to filter out the possibilities and find their 500 best candidates.
Jim: Right. Then they’ll make those.
Nicole: They’re still going to have to do some trial and error with their 500 nanoparticles. But it’s like 5% of the trial and error that they were doing with the molecules.
Jim: Angela, you used the word paradigm, the words paradigm shift.
Angela Violi: Yeah.
Jim: When you’re finished here, what do you hope that the pharmaceutical industry would take away from what you’ve done?
Angela Violi: I’m hoping to get the pharmaceutical company involved much earlier. So I’m not thinking about the end of the Blue Sky, but [eventually] if we are able already to come up with this computational engine that can demonstrate the ability to, first of all, create classification or correlation between structures and activity in terms of bacteria, that will be a great advancement for the pharmaceutical company also—because basically what we are saying is [we can] reduce the timeline to develop a product.
Nicole: That’s really exciting that the nanobiotics group is trying to get this to industry before Blue Sky is even over.
Jim: Well that’s their hope and one thing she stresses is that this isn’t about making the next nanobiotic, it’s about the method that will help us keep winning the arms race against bacteria.
Angela Violi: If we develop the nanobiotic right now, in the next, you know, hundred years, they will not be useful anymore. But with this Blue Sky what we are aiming is to just develop this computational machine. Is a framework that can be—it’s like spinning the wheel and then maybe something else comes up and then we can still use the same approach to come up with a solution.
Nicole: Are they also making some antibiotics?
Jim: They are. We’ll hear about their biofilm buster made from quantum dots next time along with an update on the computational engine. Till then I’m your host, Jim Lynch.
Nicole: And I’m Nicole Casal Moore.
Jim: See you next time, everybody.