Skip to Content

Antibiotics From Scratch

There’s a lot more antibiotic news to cover this week, and I thought I’d start out with some very high-level organic chemistry. This paper from the Myers lab at Harvard is really nice stuff: they’re looking at macrolide antibiotics (such as erythromycin) and trying to find ways to make large number of variations of them from the ground up. Now, that can be a real challenge. I did total synthesis work on a macrolide antibiotic for my own PhD, and it was no stroll through the begonias. Many of these have been made over the years, but there are, to my knowledge, no really practical syntheses.

By “really practical” I mean what a guy like me, twenty-seven years in the drug industry, would mean: synthetic routes that can produce a wide range of structural analogs (to address potency, selectivity, pharmacokinetics, and all the other things that we have to worry about when developing a new drug), while at the same time being reproducible, scalable, and short enough to be industrially feasible. These criteria sort out the existing total syntheses in short order (including, I should most definitely add, my own PhD work). This new paper points out, correctly, that all the commercial erythromycin derivatives are semisynthetic: they start from erythromycin itself, and work what changes can be worked on the existing scaffold. We have organisms that can crank out erythromycin far, far better than any team of chemists can.

Semi-synthesis is a valuable technique, of course. But there’s another thing that you might think of: why not get these organisms to make you a bunch of variations of the original molecule? Well, that’s been tried, too (by researchers at Abbott and others) – the enzymes in the biosynthetic pathway have been dinked and tweaked in various ways, but doing so turns out, in many cases, to be about as hard as doing variations via organic chemistry. You can get molecules this way that no one’s ever had, but they come hard (and nothing commercializable has, to the best of my knowledge, come out of this work with respect to erythromycin).

Some years back, Myers reported a large synthetic effort in the tetracycline area, but this one is even harder. The group has targeted doubly-modified versions of erythromycin: the first change makes them ketolides, a class of compounds where a keto group has been introduced at C-3 (as in telithromycin, cethromycin, and solithromycin). The first of those is on the market, although it has been a wild ride to say the least (see that link for more), and it has significant side effects. The second seems to be safe to use, but has (as far as I know) not demonstrated any particular advantage over existing agents, and the third is in development and seems headed for the community-acquired pneumonia indication. The second modification is expansion of the macrolide ring with an extra nitrogen atom, as in azithromycin, which is well-known as an antibiotic in wide use. There’s really been only one erythromycin derivative reported in the open literature with both of these modifications, and it’s from 1998 and wasn’t particularly impressive.

The Myers group broke this system down into eight intermediates, all of which were prepared on 30 to 150 gram scale. The organic chemistry of these intermediates is very well done, as you’d expect, and as you go through the synthesis, you see the footprints of Woodward, Mukaiyama, and other giants of the field. The macrocyclization, using a technique that Boeckmann developed in the 1980s, is particularly effective, which is a good thing, because closing these large rings is (as we like to say) nontrivial.

So far, that’s how many total syntheses would work, except they’d probably have fewer building blocks (probably just two or three). But that’s by design here – breaking things down into smaller units allows you to make variations without digging way back into the synthetic scheme. For example, the group switches between 14-membered rings and 15-membered rings by making changes in three of the eight subunits. This strategy lets them do what very, very few total synthesis papers are ever able to do:

We prepared an initial library of >300 fully synthetic macrolide (FSM) antibiotic candidates by varying the building blocks in concert with modifying readily diversifiable elements (for example, an azido group, an amino group, a β-keto lactone function, an allyl group) that we introduced into and around the macrolide ring, a powerful tactical combination. In addition to members of the three primary macrocyclic scaffolds discussed in detail above, we prepared a number of other unique scaffolds by modifying the principal coupling components (left- and right-halves) and their modes of coupling using straightforward alternative chemical transformations (see Supplementary Information for a complete list of structures synthesized and Extended Data Figs 1, 2, 3, 4, 5, 6, 7,8, 9, 10 for exemplary schemes for their preparation). This strategy not only provided novel scaffolds to explore, but also permitted deep-seated variations of positions within these scaffolds, thus enabling access to molecules that could not be prepared using semisynthetic methods.

Three hundred analogs! Now that’s the sort of thing you can build a project or two around. And as that quote says, these aren’t just small changes around the edges of the molecules (although you can do, that, too) – they’re variants that are otherwise just completely unobtainable. And also to their great credit, the team went on to screen these in a good-sized panel of bacteria, including (in the later rounds) a number of resistant organisms. They obtained a number of scaffolds with promising activity – none of them are ready to leap into the clinic as is, but they’re definitely of interest. Some of them, in fact, show activity against some rather fiercely resistant strains. Having done some antibiotic drug discovery myself, I can barely imagine a screen of only three hundred compounds that returns a list of actives like this.

There are more to make – the same approach could produce thousands more compounds. It’s still a lot of work, but just getting into the “lot of work” category (and out of the “totally implausible” one) is a big accomplishment. I’m pleased to see the Myers has started a company (Macrolide Pharmaceuticals) to build on this work, and I wish him and his co-workers success.

31 comments on “Antibiotics From Scratch”

  1. anonymous says:

    I am not sure if there are any takers for these brutal/throw in all the workers to make it happen. Yea, good organic chemistry. I am one of those old timer and who spent several years working on EM analogs and I still loved the way Abbott or Pfizer did! Modify the existing antibiotics to arrive at semi-synthetic version of the same! Not impressed.

  2. anon3 says:

    I’m still confused how government funded research can result in patents/companies where value if fully owned by the university and the professor.

    1. SP says:

      The Bayh–Dole Act or Patent and Trademark Law Amendments Act (Pub. L. 96-517, December 12, 1980) is United States legislation dealing with intellectual property arising from federal government-funded research. Sponsored by two senators, Birch Bayh of Indiana and Bob Dole of Kansas, the Act was adopted in 1980, is codified at 94 Stat. 3015, and in 35 U.S.C. § 200-212,[1] and is implemented by 37 C.F.R. 401.[2]

      The key change made by Bayh–Dole was in ownership of inventions made with federal funding. Before the Bayh–Dole Act, federal research funding contracts and grants obligated inventors (where ever they worked) to assign inventions they made using federal funding to the federal government.[3] Bayh–Dole permits a university, small business, or non-profit institution to elect to pursue ownership of an invention in preference to the government.[4]

    2. DCStone says:

      To expand on @SPs reply, the practical effect of the act is two-fold:

      1. The Government is not on the hook for the costs of patenting, development, defending patents, etc., and has no additional financial loss from things that just don’t work out.

      2. The Unversity, non-profit, or small business gets to assume the financial cost of all these things, along with the risk of it ultimately not being commercializable. BUT, if things do work out, money returns to that organization (a) keeping folks employed (b) hopefully providing additional revenue streams independent of government financing and (c) thereby generating tax revenues for state and federal government.

      It’s not a perfect system, but you can see the calculations involved in the trade-offs. Bringing ideas through patenting to licensing to commercial products is an expensive business.

    3. anon says:

      While an interesting question, that has a few good responses already, I’d like to add that it seems like the bulk of this work was not done on government funding. The NIH fund cited refers to the antibacterial testing; the synthetic effort – which is the primary subject of the patents and company – was funded by private organizations.

    4. Morten G says:

      Also patents do not prevent research. If I patent a new gel for SDS-PAGE you can still make it yourself. You are barred from selling it, only the patent holder is allowed that.

  3. Vaudaux says:

    The company licenses the technology from the university

  4. annon too says:

    Consider the origins of Lyrica, originally made at Northwestern by a post-doc in the lab of R.B. Silverman. (see and

    The compound was licensed to Parke-Davis, and through several mergers/acquisitions the final development / marketing / sales was by Pfizer. Royalties paid by Pfizer are split between the school and the inventor(s).

  5. Anonymous says:

    One thing that I would say is that Derek’s definition of a practical total synthesis might be a bit too much of an ideal dream. Unfortunately, I doubt many molecules of sufficient complexity to be interesting would even be amenable to a synthesis where you could simultaneously use the same route for both medicinal chemistry and process development (with a few tweaks I’m sure). I just feel like the focus between the discovery phase and the process phase of development is too different to really lump that in as a huge ding against total synthesis as a field (there are already so many that you could level against it). Just one man’s opinion and I know a lot of people disagree with it.

    That being said, I do think the Myer’s group has really done something good here by breaking everything down into VERY small and easy to manipulate fragments. It’s something that not a lot of people would do for a molecule of the complexity of this. They’d think 3 or 4 key fragments at most but by doing a larger number you really do allow for significant flexibility in the synthesis and design of these analogues.

  6. Barry says:

    We all wish prof. Meyers and Macrolide Pharmaceuticals well. Of course, the hit-rate expected among 300 small molecules that all conserve large parts of a known drug will be higher than at the start of any screening campaign. The macrolide field has been productively plowed for sixty years now, and multiple avenues of resistance ( ribosomal modification, efflux of the antibiotic, and drug inactivation) have been characterized.
    I for one look forward to seeing x-ray diffraction structures of co-crystals with some of the mutated ribosomes. In the 21st century, that part of drug discovery needn’t be black-box.

  7. Magrinho says:

    High quality work, no doubt. But is this what we want Harvard research groups to be doing? Essentially, it is a feasibility study for a startup. A sad state of affairs.

    1. Road says:

      One of the main reasons that the federal government funds university research is to stimulate biomedical innovation and medical breakthroughs. If any academic research is to result in a new medicine, it must go through a biopharmaceutical company. That’s what’s happening here. If the company is successful, and the end-result is a new drug that can fight resistant bacteria, then society wins. I don’t find it sad at all, I think it’s an exemplar of public-private partnership.

    2. Andy says:

      Surely that’s preferable to making millions and millions of useless compounds based on what you can buy from the catalogues…?

  8. ScientistSailor says:

    A real chemistry tour de force, and a great way to make otherwise impossible structures. However, it’s not clear to me how this will enable them to achieve their goal of making Gram(–) active macrolides. Just being able to make new structures doesn’t mean they can adjust the properties of the scaffold to where they need to be for a Gram(–) drug…

    1. lynn says:

      Azalides have improved anti-Gram-negative activity [probably due to some self-promoted uptake through the outer membrane and possibly improved ribosome binding]. And adding a primary amine to the cladinose moiety of an azalide can increase Gram-negative activity (see Shankaran, K., R. R. Wilkening, et al. (1994). Bioorg. Med. Chem. Lett. 4(9): 1111-1116).

      1. ScientistSailor says:

        Hi Lynn, adding more amines sounds like a way to increase toxicity (aminoglycosides, polymyxins) along with potency, so the therapeutic window is not wider. You have to optimize in many parameters if you want to make a drug…

        1. anonymous says:

          @ scientist sailor : Years of research by us suggested that making molecule (such as erythromycins) slightly more basic (with amino group) resulted in increasing Gram negative penetration and as such a molecule will also have increased stability (to stomach acids) for an ideal oral dosing. Placing an amino group will not turn these macrolides into aminoglycosides (functionally) like gentamycin, kanamycin that you speak off and those I agree has some toxicity associated with them beyond the reasonable dose.

          1. Ed says:

            I agree with ScientistSailor but wonder whether a better understanding of self-promoted uptake, other than an additional positive charge exploits the negative membrane potential, is the way forward? For example, does(/can) the large molecule need to fold onto itself to shield the charge will permeating through the hydrophobic bilayer?

  9. Mikael says:

    Phil Baran’s commentary about same article in Nature:

  10. Me says:

    I suspect my old dept at GSK/Pliva made most of the analogues he’s churning out, since we had a massive, multi-year macrolide program staffed by ~25 chemists. I’ve personally done all of the chemistry described in the paper, or seen reports from it being done internally in the 80’s. Despite a slew of candidate molecules, I think we had massive trouble getting anything promising.

    Big problem with antibiotic stuff is that it’s been going on for decades in pharma

    1. chelator says:

      If only you had thought to publish, and you would have made the front page of Nature! Also saving us 30 years of wasted research.

      1. chelator says:

        And just to be clear, I think you’re grossly exaggerating what probably took place in the 80s, and hence this work was not in fact a wasted repetition of past chemistry as your claims would infer, and indeed is extremely fine work deserving much praise.

  11. halavan-fan says:

    I’m always a fan of these types of projects since they really do look to push the envelope of what can be accomplished with synthetic chemistry. Rather than pay lip service to exploring the biological activity, “yada-yada”, of the target an analogs, Myers should be commended for actually trying to do it. Good luck to all involved!

    Also, to “me” above, it is a shame much of this research is behind closed doors… I have no doubt that many reinventions have been made in our industry. Sigh.

  12. Anon says:

    Wouldn’t it be easier to figure out the weak part of the original which allows resistance, and then figure out how to tweak just that one part?

    1. Andy II says:

      To Anon:

      That is what the macrolide antibacterial development has done up to the 4th generation of solithromycin (hopefully to be approved shortly). The 1st generation is erythromycin which showed good antibacterial activity but unstable in acidic media forming a 6,9-hemiacetal ring. The 2nd generation are clarithromycin (6-OMe derivative of erythromycin: CRM) and azithromycin (9a-aza ring structure: AZM). CRM showed improved PK and acid-stability but not antibacterial profile. And, AZM had revolutionally changed macrolide antibiotic. It has great acid stability and extended half-life (>60hrs) and improved gram negative activity (H. influ). AZM is also shown to possess antiinflammatory activity. The 3rd generation is telithromycin, the first ketolide with 11,12-cyclic carbamate and side chain, which addressed macrolide-resistant gram positive pathogens. It was however linked to severe liver toxicity and CNS effects. The 4th generation is another ketolide with side chain and a fluoride at the C-2 position solithromycin, which shows better antibacterial activity against resistant pathogens to erythromycin/azithromycin.

  13. Thoryke says:

    Duplications of analogues already in pharma libraries might not be such a bad thing. Yes, some wheels might be reinvented, but if those turn out to be viable products, the owners of those entities will be motivated to commercialize them….in which case, we would still have a net benefit of meds not previously available.

  14. Dave says:

    Ok, this timing was almost perfect. 🙁

    There has been a case of an E. Coli infection, where the E. Coli is resistant to all known antibiotics. 🙁

  15. Helios says:

    You know that the definition of insanity is right? Doing the same thing and expecting different results. These new version of natural products will ultimately wind up with the same issue all antibiotics face which is resistance.

Comments are closed.