Skip to main content

Chemical News

Antibiotic Progress – And Not a Moment Too Soon

Anyone who’s done antibiotic research can tell you about what a slog it is. Just looking at the rate of approval of new ones will tell you that, too – it really is like breaking rocks, except breaking rocks is a lot more straightforward and rewarding most of the time. As I’ve said before, when I look back at all the mammalian cells that I’ve killed with my molecules over the years, and compare that to the experience I had working against gram-negative bacteria, it’s pretty sobering. Killing gram-negative pathogens is hard. And killing them with a compound that (A) hasn’t already been discovered, in one form or another and (B) doesn’t kill everything else it touches is a challenge indeed.

There are two new papers, though, that give a person some hope. And we need some, because resistant bacteria, as everyone has been saying for years now, could really give our industrial civilization fits. Here’s some work by the Hergenrother group at Illinois, though, that sheds light on one of the biggest problems in antibiotic drug discovery: what kinds of structures should we be looking at?

That’s a real puzzle, because antibiotic compounds in general tend to have pretty wooly structures, especially the ones derived from natural products. They tend to be outliers in most any rule-of-thumb property screen, and often break several of them simultaneously. Yet they work, and that’s impressive, since to “work” in this context means to penetrate a lipopolysaccharide outer membrane, survive on the other side of it without being pumped right back out of the bacterial cell entirely, and penetrate that a second inner membrane in quantities sufficient to serve as a drug.

This paper looks through a set of compounds, carefully measuring the degree to which each accumulate in E. coli, and tried to draw some general structural lessons. This sort of thing has been done before, but this effort concentrates on simpler molecules rather than known antibiotics and their derivatives. What they come down to is this:

On the basis of these analyses, the following guiding principles for compound accumulation in E. coli were developed: compounds are most likely to accumulate if they contain a non-sterically encumbered amine, some non-polar functionality, they are rigid and have low globularity.

Simple enough, and you might read that and say “Thanks a lot”, because of the apparent vagueness. But the paper goes on to another key insight: if you already have a compound active against gram-positive bacteria, it likely already has the rigidity and shape needed. What if you just attach an open primary amine to it? They demonstrate this idea on the gyrase inhibitor deoxynybomycin, which is only a gram-positive compound. Hanging a plain methylamine group off it, though, converts it into a compound that accumulates in E. coli and shows antibacterial activity in a strain that approximates wild-type (that is, no efflux pump mutations or other half measures). The authors note that such unhindered primary amines are actually quite rare in screening collections (they’re right), and that biasing new collections towards this functionality and towards the structural features they’ve suggested may well improve screening efforts. The first order of business, though, will be more of those gram-positive conversion jobs, and I think that we will see the fruits of those efforts pretty soon. I hope that this trick works across a number of pathogens (and I hope we find that out soon as well!)

The second paper I referred to is the latest in the series from the Boger group at Scripps, which I last blogged about here. In this latest work (which should, by the time you read this, be here at PNAS) they’ve been modifying vancomycin, the famous “antibiotic of last resort”, which is quite the synthetic challenge (see that link if you haven’t looked at its structure; it’s a beast). They’ve managed to introduce variations that improve the potency of the compound substantially (up to 1000x over native vancomycin!), and in this new work they’ve got a compound that combines greatly improved potency with three separate mechanisms of action. This should make it very hard indeed for bacteria to evolve resistance. That’s already a slow process with vancomycin, due to its odd mechanism of action. It binds to D-Ala-D-Ala residues and interferes with cell wall formation, and the fact that it doesn’t go after an enzyme or active site lets it dodge the evolutionary pressures that can lead to resistance against such direct attacks. Throwing in more mechanisms of action can only slow that down even more, which is a very good thing.

I found it quite interesting that the latest modification is a side chain with an unhindered quaternary (tetramethyl)ammonium group, which seems to have profound (and as yet unexplained) effects on membrane permeability. This makes me wonder several things: are the primary amines seen in the Hergenrother work protonated under the conditions where they’re passing through the bacterial membranes? What would happen if they were also tetramethyl quaternary salts? Can compounds that are active against gram-positive enzyme targets but not against live organisms be rescued with a side-chain quat salt to improve their membrane permeability, in the same way that the Hergenrother paper rescues a gram-negative drug candidate?

You might also wonder if this latest modified vancomycin might start showing some gram-negative activity. Unfortunately, vancomycin in general is not the best candidate mechanism for those species, since its target, cell wall synthesis, is rather different in gram-negative organisms, as mentioned above. Gram-positive organisms have the single thicker cell wall, and disrupting it is a strong approach. The double-layer gram-negative organisms also have different compositions in those layers, making vancomycin’s primary mechanism less effective. But still. . .

This latest vancomycin analog will still (as I mentioned in this press writeup) have to go through the usual whole-animal models of efficacy and toxicity before heading into humans, but preliminary cell-based tox assays are promising. I very much look forward to seeing it progress. The activity it shows, though, now puts the ball right back in the organic synthesis court. The Boger group’s work on these structures is very high level stuff, a major synthetic challenge, and they deserve congratulations for just getting these compounds made in the first place. Finding ways to make them on scale will be another challenge of a similar order.


37 comments on “Antibiotic Progress – And Not a Moment Too Soon”

  1. anon says:

    Both super interesting work but for g- not always a requisite to get inside. Polymixins can be efficacious even without getting inside the bacterial cell. Didn’t see it in paper, but wonder if MCR1 plasmid resistance (a major issue due to horizontal transfer from E coli) could be reduced by Hergenrother modifications to allow increased access to Lipid A to compensate for target site mutation?

  2. Mister B. says:

    I’m a PhD Student in organic chemistry with an high interest in antibiotic (from a synthetic chemist point of view). The challenge behind the need of new antibiotic is really appealing !

    After this article, I have two questions:
    – Is there still a room for small molecules in antibiotics area ? (Some nice examples are shown in the Nature paper quote in this article, but Vancomycin is not really a “small” compound to me)
    – Would any of you have some references in books to read in this topic and companies that may look for chemists ?

    Thank you for any answer you can offer me !

    1. David says:

      Is there room? If you can find a small molecule that works, especially against those gram negative beasts, there’s definitely room!

    2. tlp says:

      I think making antibiotics is (one of) the best area to apply synthetic chemistry to drug discovery. You won’t get competition from antibodies or engineered T-cells anytime soon. ‘Bugs-as-drugs’/CRISPR-cas9/phages – all quite unlikely to compete with small molecules, too.
      The downside is of course you won’t be likely to discover a drug. But it’s more like a baseline, not a downside.

    3. DLunsford says:

      Sure there’s room! And link them together to generate single molecules with multiple MOAs and PD properties.

  3. Curious Wavefunction says:

    Is a tetramethyl ammonium ion really “unhindered”? However I can see why it might be fairly unique; it’s charged, small and spherical, but the charge appears more “shielded” from outside, which might explain the improved permeability. And given the significant desolvation penalty involved in having charges passing through lipophilic membranes, it could be reasonable to assume a greater proportion of the uncharged species. Of course, unhindered primary amines would probably be rare in screening decks because of obvious concerns about permeability (and downstream concerns about hERG etc.). What this study demonstrates is that especially in the antibiotics field the rules might get inverted; what’s “druglike” for your everyday screen isn’t suitable for an antibiotic, and vice versa.

    1. Peter Kenny says:

      I’d expect a quaternary ammonium group to be a lot easier to pull out of water than a protonated amine (although I don’t have figures to back this view). Measuring logD doesn’t really work for quaternary ammonium species because the effective logD is a function both of the nature and concentration of counter-ion. However, the need to balance charge is, arguably, less stringent for diffusion across a membrane.

      It is some time since I’ve worked in the antibacterial field so I don’t know if they have acidic intracellular compartments (e.g. lysosomes) that are present in mammalian cells and some parasites. If the target is within the acidic compartment then the accumulation is likely to be beneficial. As a general point, physiological concentration is a factor that more account needs to be taken of in drug design.

      1. Vaudaux says:

        Bacteria do not have intracellular membrane-bound organelles like those of eukaryotic cells.

        Note that for gram-positive bacteria, the target of vancomycin is not intracellular at all. The target is on the outer surface of the cytoplasmic membrane. I don’t understand Derek’s comment that the increased potency of the analogs is the result of increased permeation through the membrane (but I haven’t yet read the paper, just the abstract).

        1. rdgcooper says:

          I do not think it to be proven yet, but these vancomycin derivatives we found [25yrs ago] to have more than one mechanism of action. There was the normal vanco one of binding to d-ala-d-ala, and a suspected second one of inhibiting the trans glycosidase, both targets involved in cell wall synthesis. Possibly due to this oritavancin had a much greater cidal effect than vancomycin.

  4. Lars says:

    Speaking of vancomycin, Wikipedia used to say that its structure was so complicated that no IUPAC name could be assigned to it. I wondered about that claim at the time, and recent versions of the Wikipedia article do give an IUPAC name. So, a question for you chem guys: Is it the case that there are molecules which IUPAC cannot accomodate, and which features might be responsible?

      1. Anonymous says:

        Schwartz described the family of chiralanes on his website years before publishing the [6.6]-chiralane with Petitjean. (Truncate the URL provided by Mark by removing everything after uncleal/ )

        However, I need help from Derek’s readers again, this time to locate a paper by Andre Dreiding. Dreiding published a paper or book chapter or long abstract that I once had in my files in which he drew and described the [6.6]-chiralane (without calling it chiralane) before Schwartz published his own work on the web. Can anybody help me find that old Dreiding paper? (In some interesting emails about it ~8 years before he passed away in 2013, I even asked Dreiding but he couldn’t recall the specific work.) Thank you.

  5. Kazoo Chemist says:

    I can’t get past the PNAS portal to view the article and the structure, but I am guessing it is a quaternary TRImethyl ammonium compound, and not a tetramethyl ammonium salt.

    1. Derek Lowe says:

      That’s correct – I’ll try to reword this.

      1. Kazoo Chemist says:

        How about “tetraalkyl ammonium”. that allows for the three methyls and the fourth group from the vancomycin core.

        Doncha just hate nitpickers? 😉

  6. ROGI says:

    Vancomycin, as an aside to this post, has horrible down stream effects which rear their ugly head(s) years later. My wife and my son (35 years apart) were in life threatening situations with systemic staph infecionc where vanco was the drug of last resort. Both survived – well if she didn’t, obviously…..- but the ototoxcicity has rendered both functionally deaf due to targeting by vanco to the inner ear cilia. Good that both are still here, but there are sequellae to antibiotic use.

  7. Mach4 says:

    You bet amines are protonated in the acidic milieu, and its long been known and patented that polyamines inhibit translocation of antibiotics out of efflux dependent resistant cells.

    While adding a primary amine to a known scaffold may make it accumulate I’d like to know if it is by enhanced uptake or slowed translocation through efflux proteins.

  8. Sebastian says:

    If it is so difficult find antibiotics by looking for structures, given their lack of adherence to many general rules, wouldn’t we be better off just doing phenotypic screens? Seems like structural property screening is almost a waste of time, when phenotype screening might catch things that a computer can’t predict at this point.

    1. Derek Lowe says:

      The problem with general phenotypic screens against bacteria is that you have an ocean of false positives – mechanisms we already know about and things that are too toxic.

      1. Dr. Manhattan says:

        Derek is correct. Been there, did that fr many, many years. With simple bacterial phenotypic screens, you need a very robust dereplication system to weed out many toxic compounds (kills bacteria at the same rate as mammalian cells), already known antibiotics and a boatload of other stuff that shows up.

        1. Pennpenn says:

          Or to paraphrase XKCD- “It kills cells in a petri dish, so does a handgun”.

  9. tlp says:

    Great demonstration of (no) significance of cLogD(7.4) and MW for accumulation in the first paper. Funny, however, that they end up with other set of ‘rules’ and demonstrate their utility on n = 1 sample size of compounds.

  10. I did my PhD working on aminoglycoside-inactivating enzymes and often got the question of how these polycationic antibiotics get across cell membranes. Usually, removing amine groups or disrupting the membrane potential reduced their activity, suggesting the charge wasn’t something to overcome in antibiotic design, but a feature that helped get them across the membrane (though also helping bind the ribosome).

    This work with unhindered amines seems to line up with that, though we could definitely use some better understanding of how the antibiotics get across the membrane than we do! Most reviews on aminoglycosides still cite research from the 1970/80s on membrane transport and there is still a lot of work to be done to better understand how it really happens.

  11. drsnowboard says:

    So, not having access, did they make the non-amidine, bis peripherally modifed analogue?
    Because putting the amidine in the core looks like a ball-ache

    1. Kazoo Chemist says:

      I am not seeing the amidine in the core of vancomycin.

      Nit-picking once again 😉

      1. drsnowboard says:

        “So, not having access” was the clue.
        Dale Boger in press comments refers to 3 modifications, tetralkylammonium & biphenyl are 2 , I assumed the third was the previously described core amidine ie

        Hence my question.

        1. drsnowboard says:

          …but I now see it is the methylene analogue, thanks to Sci Hub.

        2. Kazoo Chemist says:

          I can see what you were thinking. They have probably made all the combinations of the various modifications and will trickle them out over many more publications.

  12. Chrispy says:

    The paper is on SciHub, for those who are blocked.

    SciHub is actually more convenient to use than my University’s access portal.

    1. Anon says:

      How long do you think you’ll be able to use it before publishers can’t afford to put out more papers for them to steal?

      1. zero says:

        Napster didn’t kill the RIAA. SciHub isn’t going to kill publishers. Change, yes, but not end.

  13. gippgig says:

    What will probably eventually kill vancomycin are escaped degenerate forms of the biosynthetic CYP genes that indiscriminately hydroxylate and degrade it.

  14. gippgig says:

    By the way, quaternary ammonium groups are common, but what about the flip side – tetraalkylborane anions? Have they gotten much study?

    1. tlp says:

      I’d bet these are protodeboronated fairly easily

  15. rdgcooper says:

    This comment is about 25 yrs out of date, the Lilly labs developed Oritavcancin 25 yrs ago by modifying vancomycin and showed for the first time that in doing this one could add another mechanism and get new antibiotics that were 1000 x more potent and highly potent against vancomycin resistant bacteria. The result of this research is now in clinical use.

  16. Paul McKeown says:

    “Most small molecules are unable to rapidly traverse the outer membrane of Gram-negative bacteria and accumulate inside these cells, making the discovery of much-needed drugs against these pathogens challenging. Current understanding of the physicochemical properties that dictate small-molecule accumulation in Gram-negative bacteria is largely based on retrospective analyses of antibacterial agents, which suggest that polarity and molecular weight are key factors.”

    How about a form of binary antibiotic, in analogy to binary chemical weapons? Two small molecules, harmless in themselves, but able to penetrate gram-negative bacterial membranes, and then react to form an antibiotic?

    1. Derek Lowe says:

      That would be a tough one to realize – they’d both have to penetrate, of course, but they’d also have to not react before they got to the same compartment, and the only thing I can see pushing them to do it then is concentration, which means that they’d not only have to penetrate, but do it really well. Not impossible, but definitely tricky.

Comments are closed.