Superbugs: Resistance rising, part 2

0:00This BBC podcast is supported by ads outside the UK. This is summer at its peak. Whole Foods Market Summer Fruit Fest is your invitation to eat the season. Fresh, organic, and bursting with flavor. Start your day with peaches and organic blueberries and yogurt. Build a grazing board with fresh fruit, prosciutto, and artisanal cheese. Then fire up the grill with no antibiotics ever, proteins, and fresh produce. Savor the season. Shop Summer Fruit Fest at Whole Foods Market.

0:58TextNow.com From a personal aspect, I had the wonderful opportunity back in 2007 of actually meeting the very first child to take Linasolid ever. Steve Brickner is a medicinal chemist who had succeeded in bringing an entirely new kind of antibiotic to market in the year 2000, though he had already seen its power to cure. She was dying from a vancomycin-resistant enterococcus infection. You know, she was very seriously ill, and they knew that she wasn't going to be around. And there were no clinical trials then. She took this drug on a compassionate use basis. They made arrangements with the FDA and Upjohn to fly out, you know, the drug in powder form.

1:50They didn't even know how to give it to her because they couldn't get an IV in her. She was that sick. and this girl was two years old at the time she was treated with it but she recovered completely at the time I met her in 2007 she was 11 years old wonderful young lady and she thanked us for saving her life and it's very it's a very movie. I'm Roland Pease and I'd asked Dr. Brickner in 2014 what it was like to know you'd created a treatment with the power to save thousands of lives that would otherwise be lost. And I've never forgotten that reply. It had taken Dr. Brickner 13 years of testing and improving thousands of prototype compounds at Upton Pharmaceuticals before finding the optimal form that could be tried with humans.

2:46It started as a personal so-called Friday afternoon project before the company then took more interest and built his growing team. And when Linazolid was approved and launched in 2000, there was quite a fanfare. Perhaps I thought good medicinal chemists get used to that kind of success. Linazolid obviously is one of your successes. How many antibiotics have you been responsible that have come to market? Well, I feel very fortunate as I know all of my colleagues and my co-discoverers must feel that we got one. It is so hard. You know, my subsequent years were spent at Pfizer and Bay Pharma, and we worked very hard.

3:30But unfortunately, none of those efforts on many other classes of compounds came to fruition at the time. So finding new classes that have this new mechanism is extremely difficult. And so it remains. For this episode of Discovery from the BBC World Service, I'm hearing from some of the innovators of today, the chemists and their backers seeking new ways to keep the pipeline of new antibiotics alive so that life-threatening infections can still be treated even as the old ones become obsolete because of bacterial resistance, as I talked about last week. The hope back in 2000 for Lynethylid was that being an entirely new class of antibiotic and being entirely synthetic as opposed to penicillin, vancomycin and most treatments borrowed from nature, it would be harder for infections to find ways to resist its effects.

4:29But that underestimates microbial evolution. Resistance to linozolid has realistically been around since, you know, the first introduction of linozolid. We found it in the accumulation of mutations in the target gene. And now we find gene-based mechanisms for resistance as well in the form of these other genes called CFR, Pioxita, OPTRA. And those ones are quite problematic because these genes go in and out of these different bacteria and confer resistance immediately. Leah Roberts leads the Bacterial Genomics Research Group at the University of Queensland in Australia, where vancomycin-resistant Enterococci, the same infections as threatened the life of that first child patient are becoming a major threat and lenezolid is the fallback treatment.

5:16But the infections are increasingly fighting back with multiple genes, as we heard about last week, that can move on plasmid cassettes from bacterium to bacterium. We actually had a particular outbreak here in Brisbane and the combination of all these different genes that can cause resistance and the fact that these genes can then mobilise onto these various different genetic cargo ships that can jump between cells means that once we start to see this resistance, it can quite rapidly spread through different populations in different settings. When you're finding this linezolid resistance, is it just there on its own plasmid or are there other resistance genes alongside them?

5:58Yes, so very often we find clusters of resistance genes. So you have multiple resistance mechanisms all co-located inside a single genome. And so that's the situation that we saw in our particular outbreak. So we had both vancomycin resistance and linozolid resistance co-located on the plasmids in our samples. And so it creates this compounding problem of, well, what are we going to do now? We've got all these resistance genes. They're quite happily sitting in all these, you know, bacteria. And then we get these bacteria jump into an immunocompromised patient that we can now no longer treat. The slightly new adaptomycin has been a fallback for when linezolid fails, but mobile resistant genes to that are also now being reported, which is why every generation needs its Steve Brickners to grind out something new and powerful against infection.

6:47To grasp the nature of the challenge, first understand that antibiotics are essentially chemical weapons which disable vital molecular functions in a bacterium. This is effectively molecule-on-molecule warfare, as chemist Sharia Mobashari once explained to me. The antibiotic binds to a target in bacteria. These could be proteins, nucleic acids, or it could be membrane constituents of bacteria. And by so doing, they interfere with the life processes of the microorganism. The microorganism cannot sustain its existence because of that interaction of the drug with the target. and ultimately it dies because it can't cope with the problems that the drug caused for the microorganism.

7:34So any antibiotic is a carefully shaped, shall we say, chemical which interferes with a specific life process which is another chemical in the bacterium? Yes, they usually are. Penicillin binds to enzymes that are involved in assembly of cell walls. So the cell wall keeps everything intact, and it's absolutely critical for the bacterium to survive. So if the drug is put into the picture, the cell wall becomes defective, and the organism bursts open and dies. Vancomycin also disrupts the cell wall, though through a different mechanism. Linesolid prevents bacteria from making proteins without which there's no life.

8:25Those are the antibiotics. Resistance is a chemical countermeasure deployed by the bacteria. For example, mutations can disguise the target compound. Or often the bacteria will arm themselves with further molecules which neutralise the antibiotic. Betalactamases are one sort, enzymes that chew up and defuse penicillins and related compounds. They come in thousands of varieties. I think we'll go in here, it's reasonably quiet. At the INEOS Institute in Oxford, just across the road from where penicillin was first manufactured, Alistair Farley and his team are working on the next layer of counter-countermeasures to these enzymes, inspired by an invention of 50 years ago.

9:13What you can do is develop a combination therapy where you've got an antibiotic and you've got something that protects it, an inhibitor that protects the antibiotic. This was first demonstrated by researchers of Beecham, which is now GSK, on the development of clavulanic acid and amoxicillin, which is used clinically as Augmentin. Clavulanic acid is what's colourfully called a suicide inhibitor. Its core feature is an indigestible version of the molecular feature that penicillins and related compounds use to target bacteria. and which is in turn the target of those resistance enzymes. They're called beta-lactamases because they break down the beta-lactam ring, which is the core part of the most widely used class of antibiotics, the penicillins and so on.

9:57So an inhibitor in this instance then is one that gets in the way of the work that that enzyme, that beta-lactamase enzyme is doing. Exactly. So it just blocks the mechanism of the degradation. So it effectively inactivates the beta-lactamase and it allows the antibiotic to do its job. We've just celebrated the 50th year since its discovery, 40 years since its first clinical approval. It is still widely used in the UK. In many parts of the world, this drug is becoming less effective. And that's because bacteria can choose from many subtly different enzymes which all resist the beta-lactan antibiotics.

10:34There are more than 5 ,000 different beta-lactamases. They're all structurally different. So you might be hitting one but another one will just push through. That will come through. So you need strategies where you can hit as many of those as you can. And so countering the resistance countermeasures starts to feel as daunting a challenge as coming up with a completely new antibiotic. Though Alistair Farley and his team aren't daunted. So the challenge is not about hitting one resistance mechanism, it's hitting many of them. And in our lead programme at the moment developing this resistance block of the inhibitor, started out in this lab about 10 years ago.

11:09It was a biochemist working on developing an assay, understood what the enzyme of interest was, and then identified an interesting molecule, a fragment. It bound in quite an interesting way the first time that these molecules have been shown to bind to an enzyme in this way. And then we developed the medicinal chemistry, expanded their range of activity to improve this molecule, not just against the target, not just against bacteria, but they do work from isolates taken from sepsis infections, but also now is very much making sure that they are safe to be used. This is really important. You might have the best molecule, the best antibiotic, but if it's not safe...

11:54I mean, it's quite a challenge. To go back to the start, so you start off with a small bit of molecule that seems to bind really well, seems to lock onto really well the resistance protein that you're trying to tackle. But that's only the start. Then you'd make thousands of variants of that, is it? We're close to a thousand now on that. And yeah, it blocked tightly, but not well enough. It didn't hit the range of all the enzymes that we talked about. So breadth of activity, safety, make sure that they're not hitting any targets in humans. And you need to make something that is cost effective. So how efficiently can you make it?

12:30But you're saying you have actually managed to get one of your candidate compounds going into the next step. Exactly. So we have declared a candidate and we continue to profile that with the aim that we would do initial safety in humans within a couple of years. So promising, but not a shortcut, it seems, compared with the 10 plus years that establishing entirely new antibiotics seems to take. A reminder, you're listening to Discovery on the BBC World Service, wondering where antibiotics of the future are going to come from. Alastafali's approach, like Steve Brickner's with linezolid, is synthetic chemistry, cooking up compounds in the lab with glassware, heat and catalysts, and a lot of patience.

13:14Penicillin and most of the other antibiotics we use started out as natural products, molecules that bacteria concoct while just sitting there at room temperature. In a corridor of Warwick University's chemistry department, Lona Alkalaf called me over to a wall chart illustrating bacterial enzymes. So like A cell transferase always loads it, ketoreductase always turns carbonyl into an alcohol, there's also enoreductase which will always turn a double bond into a single bond. These are the organic catalysts that do the natural chemical cooking. Although all the easy natural antibiotics were found decades ago, Lona Alkalaft thinks nature still has a lot to offer.

14:01Bacteria have evolved the ability to create complex chemicals over billions of years. And modern genetic techniques gives us the chance to learn how. Quite a hive of activity. Oh, yeah, yeah. It's very busy. In Lona's lab, chemists and biologists work side by side. But the best chemists, the bacteria, are kept cosy in incubators. So all the streptomite strains like to grow at 30 degrees. So you've got jars of sloshing fluids in there. So they look like streptomites to me. Streptomites? So streptomites are soil bacteria that often make a lot of natural products. So when you scoop things up from the soil, bacteria in there are always engaging in this chemical warfare to try and kill each other.

14:46And so what we want to do is see what chemicals they're making and can we take those chemicals and use them as antibiotics for our own purposes to kill the bacteria that we want to kill. And there are well-known antibiotics already. There are many well-known antibiotics. So vancomycin, for example, there's many others. So what we want to do here is grow those strains. They're shaking away and hopefully the bacteria there are making the natural products. and then we can purify, isolate those and see what they're making. But not just what they're making, but how they're making them. Reverse engineering the chemistry, if you like.

15:19And last year, that tinkering struck gold with an old friend in the microbial world. This particular strain of streptomyces is very well known. It's been studied for 20 odd years. It's known a lot of the compounds that it makes. But what we're particularly interested in is how the bacteria make those compounds. just to be clear that's because they do it step by step a bit like you would in a lab but they're doing it all inside their cells so they have different enzymes that do a step at a time to go from a starter material or a number of starter materials to make these specialized molecules and we're trying to break down what those steps are so we go in and delete the genes that encode the enzymes that do those steps if you delete the gene you get things backed up in that assembly line process so in this case we deleted a series of enzymes and what we noticed is that in some cases we've got accumulation of this one molecule that we're calling pre-methylomycin C-lactone because it's got a functional group that's lactone.

16:13Say that slowly. Pre-methylomycin C-lactone. The interesting part is that although we don't routinely test these intermediates for bioactivity we're really interested in them as a means to understand the biosynthesis. The PhD student at this time thought no I'll test it see if it has any antibiotic activity and it turned out to be a hundred times more active than the thing that we were initially looking at in the first place. I mean that sounds crazy. Yeah it is quite crazy because I suppose the question is why does the bacteria do that? Why is it not making the thing that is the most active antibiotic?

16:46Maybe it hurts itself. Maybe it hurts itself, maybe it can't get rid of it fast enough if it's too active but it does beg the question should we be doing these kinds of experiments more routinely? And it's powerful against what sort of bacteria? So the The most interesting activity is powerful against enterococcus, which is a particular problem with vancomycin resistance that is becoming more of a problem. And our compound was both active against these types of strains and also we did an experiment where we tried to induce resistance so you can grow it in the presence of the antibiotic and see how much resistance occurred.

17:21And when we do a control of vancomycin, we can introduce resistance quite quickly. And we do that same experiment with our compound. it was much harder to introduce resistance. So that's really quite promising. That is a good sign, yes, and it's not one of these compounds where you'll introduce it and then in a year everything's resistance to it and it's no use. In terms of pre-clinical testing and so on, Lola's compound is even further away from proving itself than Alistofale's inhibitor compound and it could fail. But by reading the genomes of bacteria the team are finding all kinds of natural chemical pathways which may yield other compounds, reigniting an old approach to antibiotics.

18:00I think what's interesting is that so many of the antibiotics we use have come from nature, penicillin famously, but I sort of imagined that that approach had been exhausted. We'd tried everything that could be done. So no, that's what they thought. So after this kind of big explosion in the 50s and 60s of all the things that we could find, a lot of people did lose interest in natural products as a field. But I think the advent of genome mining has really told us there's a lot more in there than we originally thought there was. And there are ways that we can get those compounds out. And I think our philosophy is where you should look for new antibiotics is where you found the old antibiotics.

18:38Especially if you can see in the genome that it looks like it should direct something. So you've got a whole oyster, a whole world out there. Yeah, you've almost got too much data. Essentially, nature spent millions of years trying to fight other bacteria. They've done the work. We just have to work out what work they've done. If Lona Alkalaf's approach is to reinvent the past golden age of natural product discovery, others are keen to work with a new shiny tool on the digital horizon. Sure, I'm Jim Collins. I'm the Tamir professor here at MIT, and my lab focuses on using AI to discover and design new antibiotics.

19:13Jim Collins is an engineer who long ago turned his attention to living organisms, whose internal organization represent a kind of epitome of hard-evolved engineering principles. And from understanding bacteria, that now includes finding modern ways of killing them. The golden age of antibiotic drug discovery was in the 1940s, 50s, and 60s, before the microbiology revolution, before the biotech revolution, before the genomics revolution, and even now before our current AI revolution. And what excited us now going back just a few years was to what extent could we use AI to address this discovery gap?

19:50Could artificial intelligence dramatically expand our ability to explore chemical spaces, to find new chemical matter that could impact bacteria and or design new chemical matter? And we've made some decent progress in that direction over the last few years. Chemical spaces is the key phrase in that answer. Remember Alistair Farley describing the thousands of compounds they cooked up over a decade to find his resistance inhibitors, week upon week of bench work reshaping complex molecules in the chemistry lab. But that is only a fraction of all possible chemical permutations. People have estimated that there may be 10 to the 60 possible compounds or molecules out there, which is truly beyond astronomical.

20:33I mean, that's a million, million, million, million, 10 times over, I think. Yeah, you keep going, right? And the largest in silico libraries at present that are florable are really more of the order of 10 to 10, which again, tiny fraction of 10 to 60. Now, it still is empirically impossible to test in the lab 10 to 10, but we now have capabilities to screen those quite quickly with AI-based models. In effect, this is doing medicinal chemistry inside a computer. Step one? The initial step would be to gather a set of compounds, apply them experimentally to a pathogen, seeing which you're exhibiting with some level of efficacy, antibacterial activity.

21:13Though they may not be known antibiotics, they may not actually be very good antibiotics, but they have some antibacterial activity. That can be very useful data to help train a model. So based on, say, that screening data, let's assume, let's say it's 40 ,000 compounds that we apply to a bacterium of interest. What the models do in our hands is they'll look at each compound structure, bond by bond, substructure by substructure, scaffold by scaffold, and associate those bonds, substructures, those scaffolds with being associated with antibacterial activity or not antibacterial activity. And by looking across those 40 ,000 compounds, now the model is trained to take a new compound structure and based on its bonds, based on its substructures, based on its scaffolds, make a calculation.

21:58What's the likelihood that that new compound that it had never seen before would be antibacterial? Now I can look at billions of compound structures that I didn't test experimentally and make a calculation with this model quite quickly as to what's the likelihood that there's a really effective antibiotic in those billions of compounds. Searching computerized chemical databases, the approach quickly yielded good results and oddly could select compounds that, when tested in the lab, killed more bacteria than they'd originally thought about. So, you know, for example, in the first piece we did in this, which we published now six years ago in Cell, was focused on training data from E.

22:36coli, from which we uncovered a molecule called halicin that was remarkably broad spectrum. It killed almost everything we applied it to, Asinobacter bimani, E. coli, Clostridium difficile, TB. Interestingly, the next study we did, which was trained on Asinobacter bimani data, led to the discovery of a compound called the balsam, which was remarkably effective against Asinobacter bimani and not effective against any other pathogen that we tested against. Nothing more. So it was really an odd dichotomy. For what it's worth, clinicians need both. Some compounds that they can throw at any infection, others that will really hit the difficult one.

23:13But meanwhile, Jim Collins and his team have gone full generative AI, the medicinal chemist equivalent of a chatbot that can invent never before imagined molecules. Our latest efforts on generative AI was to say, OK, could we take as a starting point an atom, a carbon atom and oxygen or a small fragment of a compound and then begin to build it out using generative AI approaches, each step and or at critical steps being guided by one of these structural models that could predict, are you moving in the right direction to make a powerful antibiotic? Arriving then at a set of compounds that are predicted to be very good from an antibacterial standpoint, while also not being toxic to human cells, as well as potentially being synthesizable, and then advancing those toward being synthesized and tested in the lab.

24:01And you're right, we, I guess of note, actually synthesized, I believe it was 24 compounds, seven of which exhibited significant antibacterial activity, two of which we actually advanced all the way through to animal studies. And so I guess one question is, of those two that you then took through to animal studies, do they look completely different from anything that anyone's looked at before? Or did you end up with the same old, same old? I'm presuming And the answer is no. The answer is no. They don't look like existing antibiotics. And that was by design, meaning that we put in as an additional criteria to ensure they don't look like something that we already have.

24:37And not looking like existing ones is important because that means that there won't be obvious resistance mechanisms to these new antibiotics out there in the wild. For the most part, yes. So in the great majority of cases, that's a correct statement, meaning that a novel structure will likely have a novel mechanism and are at minimum a novel target or interact with a known target in a different way and thus would not be hindered by existing resistance mechanism. And it in part has initially driven at least our phenotypic discovery and design. Again, it's early days. Jim Collins' MIT lab has spun out a non-profit company that's now trying to develop the compounds he's discovered.

25:15And they're not the only ones hoping that AI could pump up the pipeline for future antibiotics. I think it's beginning to happen. I think that the interest in using AI for broadly drug discovery and specifically antibiotic discovery is ramping up. I note that there are a number of pharma companies that are launching partnerships and or internal discovery efforts utilizing AI, whether it's Genentech and Roche, Eli Lilly or Pfizer or GSK, all have announced programs in the last couple of years. So I'm encouraged by that. I think AI is, I think, dramatically expanding our ability to discover new compounds with interesting chemistries.

25:54At present, it's getting to promising hits. We need to advance it to get to meaningful leads and then to drug candidates. But I think we're going to get there. And I do think it's already serving to thaw, but we've got a lot more thaw. A lot more thawing to do to get us to compounds that could be advanced into clinical trials. A new spring, maybe, after a long winter of depressingly little invention in the world of antibiotics. But that alone won't be enough. Once they've passed through the clinical trials and been approved, antibiotics these days face deadly drought. In the past decade, companies have brought new antibiotics to market yet proved to be a commercial disaster, losing their investors billions.

26:37It's either bankruptcy or liquidation or a distressed sale or some other recapitalization event, which results in the original research and development investors losing everything. But these aren't every company that made it successfully to approval. So these are the successful companies. And it's not just one or two of the successful companies. It's every one of them. So behind each one of these companies that made it are 30 or 50 that failed along the way because the science is hard. So where's the positive example for an investor? Show me the antibiotic company that I could invest in and I would have made money.

27:15Unless the economic model can be fixed, the chemist pipeline will never reach the patients who need new antibiotics. And that's the topic for next week's Discovery with me, Roland Pease, from the BBC World Service, fixing the antibiotics market.

27:59it. Download Mistplay today and enter for your chance to win. How did a boycott Jimmy become a billionaire from posting videos? On Good Bad Billionaire, we're going to find out how the world's most popular YouTuber, Mr Beast, made his fortune. He's buried himself in a coffin for days. Counted to 100 ,000 on camera. And even recreated Squid Games, all in an attempt to go viral on the internet. But it all started when he gave a homeless man$10 ,000. So is he a philanthropist reshaping capitalism? Or is he just the king of the attention economy? Find out on Good Bad Billionaire. Listen on BBC.com or wherever you get your podcasts.