Skip to Content

A New Diabetes Approach? Small-Molecule Screening Wins Again

The molecular biology/chemical biology tools we have now are quite something, and have opened up whole areas of research that previously wouldn’t have been feasible. But as a chemist, I’m glad to say that there’s often still nothing like a small molecule. That’s one of the things I take away from this recent paper in Cell, a multicenter collaboration between Dana-Farber, Harvard, Yale, the Broad Institute, and Scripps-Florida. They’re looking at a powerful protein involved in metabolic pathways, PGC-1-alpha. It’s a coactivator protein with the PPAR nuclear receptors, and is a central player in mitochondrial activity, adaptation to cold temperatures, exercise, lipid and glucose handling and more.

You’d think that this would be a fine target for pharmaceutical intervention, and to a first approximation you’re right.  Type II diabetes is the most obvious target, but it’s not easy to see how you’d get a drug to work. If you knock down the protein in a whole animal model, it manages to compensate, for the most part, but selective ablation in the liver (in otherwise diabetic animal models) has shown much more promising effects.

That’s because the liver is an inadvertent bad actor in Type II patients. It’s well known that the brain uses glucose as its fuel source, whereas other tissues can switch over to fatty acids if need be. This metabolic switch, which is activated during fasting, also tells the liver to start making glucose from scratch, for the brain to use. Insulin levels are generally believed to be the proximate signal for these events, but Type II patients are characterized by insulin resistance. Their tissues don’t respond to the (rather high) insulin levels they have, and that means (among other things) that their livers assume that they’re in a low-insulin fasting state, and continue cranking away on gluconeogenesis. The last thing a diabetic patient needs being dumped into their blood is a continuous supply of fresh glucose, but that’s just what’s going on.

Many are the mechanisms that have been tried over the years to shut down this inappropriate sugar source, and PGC-1a is on that list as well. But PGC has no small-molecule binding sites of its own – it’s not regulated in that fashion, and it does its work by a variety of protein-protein interactions. These are, of course, notoriously hard to target with small molecules. It’s not impossible, but if you have another potential mechanism, you should strongly consider it as an option. In this case, the possible way out is acetylation of PGC-1a. Multiple lysines on it are post-translationally modified in this way, and that’s an important regulatory pathway. Everyone’s old friend in this area, the sirtuin SIRT1, seems to be the deacetylation enzyme that activates PGC-1a in response to fasting (thus a big part of the connection of sirtuin inhibitors to diabetes therapy). If you could find a compound that keeps the protein acetylated, by one mechanism or another, you could have something interesting. This latest paper describes a screen for just such molecules:

Here, we designed and developed a cell-based high- throughput chemical screen using an AlphaLisa assay aimed at identifying chemical scaffolds that induce PGC-1a lysine acetylation. Subsequent secondary assays identified a subset of previously uncharacterized small molecules that were able to reduce glucose production in primary hepatocytes. As a proof of concept of the potential use of these compounds as anti-diabetic drugs, a single hit from our screen reduced fasting blood glucose, significantly increased hepatic insulin sensitivity, and improved glucose homeostasis ameliorating diabetes in dietary and genetic mouse models.

That hit is SR-18292, shown at right. The original hit was the analog with a 3-methyl on the indole ring, but that one (from the NIH libraries collection) wasn’t available in quantity, so the group synthesized the desmethyl shown, which seems equally active. (Several other close analogs with changes around the indole portion of the molecule were inactive, though). Interestingly, it retained that activity in cells even in the presence of sirtuin inhibitors and HDAC inhibitors, which suggests that it’s working independently of the deacetylation pathways.

As mentioned in that quote above, the compound has just the downstream effects, in cells and in whole animals, that you’d expect from a compound that is (somehow) increasing the amounts of acetylated PGC-1a versus its more active deacetylated form. But the exact mechanism still isn’t clear – tracking that down could itself lead to new sorts of diabetes targets, of course, since that could be screened on its own. This paper, then, is a good example of the sort of thing that a solid phenotypic screen can pull out for you – a useful small molecule that might (if everything goes well) lead to a therapy on its own, but at the very least a tool that lets you look into processes that you didn’t even know existed before. I will be glad to see what comes of both of these lines of research.

Update: the comments to this post raise a number of interesting questions about this compound – I’ve written to the authors for comment, and will report back with any news.

28 comments on “A New Diabetes Approach? Small-Molecule Screening Wins Again”

  1. David Borhani says:

    I am unable to find in this paper any mention of the long-acting, potent beta-adrenergic blocking activity of these compounds. See: Glushkov et al., “Search for long-​acting β-​adrenoblockers in the series of 4-​hydroxyindole derivatives”, Khimiko-Farmatsevticheskii Zhurnal (1993) 27(8):8-12. (Chem Abstracts CAN120:244527); Abstract: “New β-​adrenoceptor, e.g. 4-​hydroxyindole derivs. I (R [indole 3-position] = Me, H, R1 [indole 2-position] = H, R2 [phenyl 4-position] = H, Me, OMe, NO2, F, Cl, CO2Et)​, were synthesized. They differed from each other in the presence of alkyl substituents at the 3-​position, benzoates instead of hydroxyl, and p-​substituted Ph substituents on the free N of the side chain. The β-​adrenoceptor blocking potency of most substances is less than pindolol and bopindolol, but the duration of activity of p-​fluoro- and p-​nitrophenyl derivs. exceeds that of the known β-​adrenoceptor blocking drugs.” [Base compound I is 2-​Propanol, 1-​[(1,​1-​dimethylethyl)​[(4-​phenyl)​methyl]​amino]​-​3-​(1H-​indol-​4-​yloxy)​-, i.e. SR-18292 without the 4-methyl substituent.]

    Am I overlooking it—or is this a surprising omission?

    My understanding of the complex effects of beta-receptor blockade on hepatocyte gluconeogenesis, etc, is limited, but this is clearly something that has been has been investigated for decades. See, for example,, which indicates a slight decrease in the AUC in a glucose tolerance test after treatment with propranolol.

    1. milkshaken says:

      As this is a simple-to-make molecule and an active field, we will see soon enough if this new lead is for real or if it comes an artifact of beta-blocker activity – though even an indirect beta-blocker activity on diabetes would be interesting as these drugs are known to be reasonably safe in long-term use.

      Having said that, I actually worked under some people on this paper, and you have to take this as a preliminary result from academic group published in a biology journal. If they missed something glaringly obvious it wouldn’t be the first time it happened.

      1. tlp says:

        Given 3-step synthesis and medicinal chemist(s) on board, I suspect authors have hidden significant part of SAR study. Otherwise I can’t rationalize not even mentioning separation of enantiomers.

      2. Hates electron rich indoles says:

        I’m not so sure this is “simple-to-make.” Ever try to do chemistry with 4-oxyindoles? They are the type of substrate that oxidatively decomposes “while-you-watch.” Your “purified” product turns into a rainbow colored solution in only a few seconds as you are trying to transfer it from the rotovap flask to a storage vial.

    2. Derek Lowe says:

      Interesting observation – you’re right that it does look like an adrenergic compound. FWIW, the bioactivity data at PubChem shows it not to be active in a beta-adrenergic screen (rows 222 and 223), but it would be prudent to go back and check that, probably by running it through one of the commercial GPCR screening panels.

      1. milkshaken says:

        I would expect a rapid metabolism on the electron rich indole part – especially since there is no substituent in 2 and 3 position

        Also, I don’t see too many medicinal chemists there on the author list so it looks to me like a quickie, to satisfy a biology collaboration with other groups. Sometimes the amount of chemistry work at academic institution depends on whether they can get a grant (Scripps medchem group funding has been troubled in the past, and their overhead is huge)

        1. FBC says:

          There is at least one med chemist from the Broad on the list. Definitely not a team, though.

  2. Magrinhopalido says:

    Agreed with Dr. Borhani. I’ve worked on similar compounds and they were very promiscuous. CEREP counterscreens against CNS GPCRs lit up like a Christmas tree. Generally, a basic, tertiary amine between two aromatics will do this.

  3. Peter Kenny says:

    But will it hERG?

  4. Emjeff says:

    We need new medications to treat this disease, that’s for sure, but I’d hate to be in this space. The recent news that canagliflozin increases amputation risk (while lowering Hgba1c) is not only bad news for Janssen, it’s really bad news for diabetes drug developers, because Hgba1c is looking much less solid as a surrogate end-point. That means long, expensive outcome trials, and carrying significant risk into Phase 3. That’s not a deal that anyone is going to jump into.

  5. Barry says:

    Any company that worked in beta-blockers should have a few thousand compound in their archive that presents most of this pharmacophore, not all of which were (presumably) active against the intended target.

  6. David Borhani says:

    Do others also find the points that @milkshaken makes just a bit troubling? This is a Cell paper, with a bevy of very notable and highly intelligent scientists as authors (e.g., Stu Schreiber, Steve Gygi, others). What about Cell’s reviewers?

    Note that I’m not saying their proposed activity on PGC-1a isn’t real—I haven’t, and I don’t plan to, review the data at that level. However, not to test, or to test and not to mention, the *obvious* activity on adrenoreceptors is shocking. (To not be aware that the molecule is a very likely beta-antagonist, just by looking at it, is also surprising.)

    @milkshaken: Regarding “rapid metabolism,” pindolol, a beta-blocker of nearly identical structure, is metabolized somewhat but not too rapidly (half-life ~3 hours). The N-benzyl group in this compound will presumably confer increased cLogP; that, plus binding of the benzyl in the secondary pocket near helices H1/H7 presumably confers the (Russian) reported long duration of action (cf. indacaterol [and salmeterol] for the same idea in beta-agonists).

    @tlp: Why would they even separate enantiomers at this stage? (OK, why wouldn’t they? To me, this is just a sign of moving quickly as opposed to laziness or incompetence.) The lack of SAR is a bit interesting (though also very academic); maybe it’s in the patent application that will publish soon?

    @Derek: Both those PubChem assays are looking specifically for *agonists* (either by b-arrestin clustering [Assay #504454], or by b-arrestin-mediated signaling to tTA, leading to expression of (transfected) luciferase [Assay #492947]). I don’t see any adrenergic *antagonist* assays in the list. Also interesting is that the molecule is about 4 times more potent as an *inhibitor* of glucocerebrosidase (cf. Gaucher disease)—IC50 ~0.6 uM, “Complete curve; partial efficacy; poor fit”. Not a desirable off-target activity, I suspect.

    @Magrinhopalido: Yes, indeed. But more than that, RN(H)-CH2-CH2(OH)-CH2-O-Aryl is *the* pharmacophore of beta-antagonists. (Change O-Aryl to Aryl-OH and you have an agonist.)

    @Peter: you made me laugh out loud! 😉

    1. milkshaken says:

      Without knowing any specifics about this paper, I have seen quite a few academic collaborations that instead of work being paid with grant money, they were paid with co-authorships, contacts and favors. It goes like this: a boss asks you to synthesize few compounds “when you have a little time” because the department head has a biology friend who needs it done quickly. If you are a postdoc or staff scientist with a visa problem, you do it maybe over few weekends, and you earn co-authorship on a paper in a good journal, for making just couple of simple compounds, and maybe even with co-inventorship on a patent.

      Your boss is the coauthor too, and the boss of your department who set up or approved that collaboration of course is on the paper, and his name is also on a submitted grant application (to improve chances of it being approved) – so now you have an ever-growing list of authors that includes some notable names especially when you have a collaboration paper from academic groups in 3 separate institutes. In reality, the number of people actually doing the experiments in such academic collaborations tends to be much smaller, and some of the bosses will just read the paper before it is submitted. Again, I don’t know if this is the case, but I would not be surprised.

      1. Anonymous says:

        “Without knowing any specifics about this paper…” That is really what you should have said first.

        “Your boss is the coauthor too, and the boss of your department who set up or approved that collaboration of course is on the paper…” So kinda just like industry as well. This isn’t solely an academic issue.

        1. milkshaken says:

          You seem grumpy and defensive (is your name also on the paper?) so you may have missed that I have written this entirely in response to D. Borhani – when he is asking incredulously ‘How come that a Cell paper with so many notable people from 3 institutes can be so half-baked’, I am explaining that this is more than possible, in academic biology-driven collaborations publishing in a biology paper, especially in the preliminary stages of the project, grant funding for medchem in academia is not easy to get it and it takes time so the amount of medchem done at the beginning of the project would be limited, and that one should appreciate the medieval sociopolitical factors that result in long author lists in such collaborations while the actual work was produced perhaps only by handful of junior researchers…

          1. Anonymous says:

            Interesting response (no, I’m not an author nor part of that work), but nice to know when someone disagreed that’s right were you went (typical internet response). Neither grumpy nor defensive. Simply pointed out two things: 1. you admitted you didn’t know much about the work, and 2. having bosses on a manuscript is not an academic only practice. Name any mid- to upper-manager/scientist in industry and tell me how much they actually “do” to be part of a manuscript. Now grumpy and defensive could easily be used to describe your comments. An axe to grind would also come to mind.

          2. tangent says:

            Why are you making this an industry/academic thing? When a similar practice occurs in industry, a similar result follows, big brains on the authors line who didn’t apply commensurate brainpower to critiquing their paper. You happier now?

          3. Anonymous says:

            Not making it an academic/industry thing. The poster has an issue with the work and went on about how academic collaborations “work” and how junior scientists do everything. I’m simply pointing out that the same practice occurs in industry as well, which has nothing to do with the grant process. Not one vs. the other, it’s one and the other.

          4. milkshaken says:

            sometimes you can find weak or preliminary results from biotech startups, especially of the virtual variety (they do limited research at CROs, then hype the results for the benefit of he investors) but established companies usually do not publish the early lead stuff. The more common industry approach is to get a broad composition of matter patent coverage of entire series, and try to even obfuscate the identity of their best compounds in the patents, all the way to the moment the project goes to the clinic, or is cancelled. So what you typically see from larger pharma companies are publications on projects that are maybe 5+ years old and have large series compounds made.

            With regards to the co-authorship padding by the names of the bosses, this sort of patronage thing does frequently happen in the industry but industry tends to be less cash starved than academia so there is less incentive for “lets ask them if they can measure MALDI for us as a favor” which is paid for with 4 additional names on the paper

    2. tlp says:

      My whole point was that the published story is far from being complete for non-scientific reasons (patent application, unreported/unwanted activity, grant deadlines – who knows what they are). And of course med-chem component is not supposed to be of major significance for paper quality here. Yet how it’s done leaves a feeling of sloppiness.

  7. Nerd v geek says:

    It is unfortunate that a lot of high profile chembiol papers often fail to mention important prior art that a simple SciFinder search would pick up. My guess is the authors probably think acknowledging previous work detracts from the apparent novelty of their findings. This should have been picked up in review of course, but then maybe no chemist reviewed the manuscript.

    1. Thoryke says:

      The other possibility is that they didn’t _do_ a search to build a robust lit review, and just put in the references they were most familiar with. I have to rebuild lit reviews all the time because people have cited the papers from 20+ years ago, even when supporting sentences that start with the words “Currently” or “Recently”. Not all reviewers (internal or external) will catch those kinds of mistakes.

  8. tcor says:

    On another topic… Can’t see past colloidal aggregation on those hits (Fig S1) with steep Hill slopes and micro molar AC50 values. I wonder if those compounds are actual inhibitors.

    1. Derek Lowe says:

      Those are not very nice looking curves, to be sure. Most of them, though, are things that weren’t followed up on in the paper, but it still doesn’t look like this screen generated a lot of useful chemical matter. I would like to see what their SR-18292 compound looks like when run back through the primary assay, with a generous number of dose-response points.

      1. David Borhani says:

        As in many academic papers, presentation of dose-response data seems to be undervalued.

    1. Derek Lowe says:

      That’s true – glucose is preferred, but after several days of starvation, the use of acetone, hydroxybutyrate, etc. goes up in brain tissue. The liver is still making glucose, though.

  9. MFernflower says:

    Just for fun, I deiced to run this compound past a freely available mathematical model for target prediction and needless to say this compound tripped almost every GCPR there ever was:

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.