Skip to Content

AMPK: Time To Think Hard

AMP-activated protein kinase, now that’s an enzyme for you. It sits at the center of a cell’s metabolic switchboard, and if you’re talking about anything to do with the response to levels of fuel like glucose or fatty acids and determination of their downstream fates, then AMPK is almost certainly crucial. Activating the enzyme sets off (among other things) downstream effects that cause fatty acids to get oxidized in the cell, glucose to be taken up from the bloodstream, cholesterol synthesis to be inhibited, breakdown of lipid stores to slow down, inhibition of triglyceride synthesis, and increased sensitivity in insulin release. A great many of the beneficial cellular effects of exercise are thought to involve activated AMPK. All sorts of metabolic drugs have been thought to have effects on its activity (metformin in particular, although that story is definitely complex and far from well understood), and as you’d imagine from that activity profile, modulating it somehow is of very great interest in the metabolic disease area.

That’s not easy. The enzyme has three subunits, each of which comes in different variations, so there are twelve varieties of the enzyme that are distributed in different tissues in ways that are incompletely understood. There are mutations known in these subunits, and some of them rev up the enzyme constantly. The problem is, these gain-of-function mutations seem to be partly compensated for during development on the metabolism end, while also leading to cardiac hypertrophy. So the question of “what happens when you activate AMPK out of the blue with a drug” is still open – is that heart effect something that happens just during development, or would you get that, too? The whole “human knockout” idea for disease and target understanding is a powerful one, but (like all the others) it has its limitations. Sometimes the only way to find out what a certain type of drug will do is go find yourself one of those drugs.

That’s what this paper from Merck has done. It describes an interesting-looking small molecule, MK-8722, that activates all twelve varieties of AMPK. I like that isosorbide on the right-hand side – it’s a cheap polar building block, and I’ve stuck it onto all sorts of compound classes over the years, just because it doesn’t always seem to be very well represented in compound screening libraries. (Here’s an earlier generation compound from this project, lacking that group). It’s potent and selective, and has good cell penetration and good PK properties in general – an excellent tool compound to start answering questions with. In rodents, acute dosing shows strong glucose-lowering (which is what you’d expect) in both lean normal and diabetic animals. In fact,  it showed these effects in every single metabolic-disorder rodent model they tried, which is pretty impressive. Checking skeletal muscle tissue showed the expected changes consistent with increased glucose uptake and fatty acid metabolism. Chronic dosing in the classic db/db mice showed effects that were just as good as the PPAR compound rosiglitazone, but without any of its weight-gain effects. The same effects were seen in a diabetic rhesus monkey study (a difficult and expensive one to run, I should add). These all took place without raising insulin levels, as well.

So far, so good! But long-term dosing also revealed. . .cardiac hypertrophy. Just as in the (known gain-of-function mutations. It showed up in both the rats and the monkeys, and when dosing of the compound was discontinued, the heart tissue gradually went back to normal. Cardiac function didn’t seem to be impaired, but still. . .this could be a show-stopper. The paper notes that it’s similar to the heart effects seen in elite athletes, which might well also be AMPK-driven, but it’s going to take some hard thinking to decide what the long-term effects of causing this pharmacologically might be. It’s going to be interesting to see what Merck and the other companies working in this area decide to do about this. Safety (particularly cardiovascular safety) is a huge concern in the diabetes field, since patients basically take these drugs for the rest of their lives and are already at increased risk. This looks to me like a hard sell, both for physicians and for the FDA, but we’ll see. . .

17 comments on “AMPK: Time To Think Hard”

  1. steve says:

    Metformin acts at least partly through AMPK and is a highly effective and safe drug; after 50 years it’s still probably the best diabetes drug out there. It activates AMPK indirectly, through inhibition of mitochondrial complex 1, which lowers intracellular ATP levels (and also activate reactive nitrogen species). With regard to the earlier thread on target-based discovery versus phenotypic, I would hazard a guess that no current pharma company would ever have discovered metformin and, if they did, it would have been tossed early on due to its pleiotropic effects.

    1. jerome says:

      The study in the article touches on the inhibition of mitochondrial complex 1, and it found inhibition only at supra-therapeutic dosages(≥ 5 mM , where therapeutic concentrations in the study were stated to be 40–70 μM). It gives a laundry list of other mechanisms of action for metformin; it may be the most pleiotropic drug I know of.

      1. steve says:

        The history of how it was discovered is informative
        https://www.news-medical.net/health/Metformin-History.aspx

      2. john adams says:

        The portal vein concentrations are higher, and metformin is actively transported into liver, so the hepatic concentrations are substantially higher. That said, the glucose lowering efficacy of metformin is NOT, as proposed first in 1991, via AMPK activation in the liver.

      3. Norbert Schtumpf says:

        Complex I inhibition (with a different drug) sufficient to activate AMPK provided happy results in mice with a lot of mitochondrial reserve but did not go well in humans.

        Clin Pharmacol Ther. 2015 Nov;98(5):551-9. doi: 10.1002/cpt.178. Epub 2015 Aug 4.
        Toxicity of a novel therapeutic agent targeting mitochondrial complex I.

        Low Wang CC, Galinkin JL, Hiatt WR

        Abstract
        R118 is an experimental compound that completed preclinical development as a potential medical therapy for the exercise limitation in peripheral artery disease. Animal studies established that R118 provided partial and reversible mitochondrial complex I inhibition with consequent increases in adenosine monophosphate (AMP) kinase activation in liver and skeletal muscle. After demonstration of improved exercise performance in a mouse model and safety in rodent and primate models, a phase I trial was performed in 24 subjects randomized to R118 vs. placebo (5:1) in escalating doses. Plasma lactic acid levels were transiently elevated in 20% of subjects at the lowest dose and in 100% of subjects using a different formulation at the highest dose, which was associated with hospitalization in all subjects and severe metabolic acidosis requiring prolonged intubation in two subjects. Thus, inhibition of mitochondrial complex I with R118 resulted in severe lactic acidosis, representing unacceptable toxicity from this mechanism of action.

        1. john adams says:

          The dose makes the poison – compare the toxicity of metformin vs buformin vs phenformin

    2. Derek Lowe says:

      I have to agree – at the very least, the odds would be heavily against it. And metformin’s structure alone stands as a rebuke to normal med-chem ideas of druglike chemical matter (mine included).

      1. Ben says:

        Preach brother!!!!!!@!@!!!

  2. Lane Simonian says:

    Be careful what you try to activate and when. Part of the benefits of AMPK activation are through the phosphatidyinositol 3-kinase/Akt pathway but when this pathway is blocked or inhibited by nitration, for instance, the result can be cell death instead. Under these conditions, metformin may increase the risk of Alzheimer’s disease.

    https://www.ncbi.nlm.nih.gov/pubmed/21623793

    https://www.diabetesdaily.com/blog/study-metformin-linked-to-higher-risk-of-alzheimers-and-parkinsons-393789/

  3. Peter Kenny says:

    The tautomer drawn is likely to be higher in energy than the alternative one in which the NH can interact favorably with the ‘lone pair’ of the nitrogen in the 6-membered ring.

  4. b says:

    Science allowed that structure to be published? Missing some H’s there… or some charges.

    1. AC says:

      Science and Nature seem to not care about chemical nomenclature/structures much. A couple of years ago Nature had an article on superconducting H2S (at extremely high pressures), and all through the article called it sulfur hydride. I understand that physicists don’t care about oxidation states and nomenclature, but don’t they have editors?

      1. Ben says:

        While it’s true that physicists in general don’t care that much about chemical nomenclature, I wouldn’t use that particular story as an example: at the pressures involved in that study, H2S first becomes metallic — at which point arguing about oxidation states becomes far less meaningful — and then decomposes to H3S + dissolved S, before it becomes superconducting. So I would say that the generic “sulfur hydride” to refer to the system is far more appropriate than the misleading assertion of oxidation number implied by “hydrogen sulfide”.

  5. gippgig says:

    There seems to be a general principle with these stress resistance systems – activate them a little and nice things happen but activate them too much and you fall off a cliff and really bad things happen (i.e., neurodegenerative diseases). It would be very interesting to compare low & high doses of this compound.

  6. Confused chemist says:

    Unless I’m getting my stereochemistry confused, MK-8722 actually contains an isomannide group, not isosorbide. I’ve never thought about using these as polar head groups in compounds before, but I’ll definitely look for opportunities to do it in the future.

  7. Bill says:

    Gossip is harrassment. That should not be censored

  8. GutDecipher says:

    Any thoughts on the Pfizer work that came out around the same time? They also have a panspecific molecule and one that…doesn’t hit musculoskeletal AMPK. Next time try to do it the other way around guys! http://www.cell.com/cell-metabolism/fulltext/S1550-4131(17)30215-2

Leave a Reply

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

Time limit is exhausted. Please reload CAPTCHA.