Skip to Content

Chemical Biology

More Proteins Than You Ever Thought

When you take an NSAID (naproxen, ibuprofen, aspirin, etc.), how does it work? This is one of those questions that improves on further inspection – or deteriorates, according to your point of view, because it just keeps on getting more complicated. For decades, there was no good answer at all, but then there was “It reduces signaling in the inflammation pathways”, followed by “OK, it seems to do that by decreasing this prostaglandin signaling molecule” and “Ah, it does that by inhibiting cyclooxygenase, a key enzyme in the pathway to make those”. Then we moved to “Hold it, there’s more than one cyclooxygenase”, which is how we ended up with the COX-2 inhibitors like Vioxx, which led to “You know, the functions of these two are more complicated than we even thought”.

And on top of all these, there are clearly effects of NSAID drugs that have nothing to do with the COX enzymes at all. Blood clotting you can pretty much chalk up to the COX isoforms, but there are more things going on, and they’re very poorly defined. Which brings up a more general question: how do you ever know about all the things that a small-molecule drug might be doing in a living system? The answer is, you don’t. That’s not generally appreciated outside the biomedical professions, but it’s the truth. We don’t have some way to track things around and watch them interact at a molecule level through out a cell or an organism.

Well, maybe. Mapping compound-protein interactions is a big part of chemical biology, and this new paper from the Woo lab at Harvard shows an attempt to track small molecules in just this way. It uses photoaffinity labeling, via the now-nearly-standard combination tag of a diazirine and an acetylene, which lets you form a covalent bond to (most) of the thing the molecule of interest is spending time next to, and on the back end of the process, it uses isotope-labeled tags on the acetylene-click-conjugation step to give the mass spec detection a much higher signal/noise. Just digging through the whole cellular proteome without the kind of validation that isotope tagging gives you is a tall order, but the deliberately-weird-mass-combination effect of the stable isotope tags moves things up out of the noise.

Three NSAIDs are given the treatment: naproxen, celecoxib (Celebrex), and indomethacin. It’s a good spread of activity: naproxen hits both COX-1 and COX-2 pretty much equally, celecoxib is of course a COX-2 selective compound, and indomethacin is in a structurally distinct class that’s already known to have some other modes of action. (Update: here’s work from another group on an alkyne-tagged aspirin derivative, which also interacts with a long list of cellular proteins). All the photolabel-derivatized compounds were still active against COX enzymes (albeit with somewhat lower affinity), and all of them could be displaced by the parent compounds in competition assays. All of them photo-labeled COX-2 in an in vitro experiment, as they certainly should, and this experiment was used to validate the mass spec analysis methods for the whole-cell experiments. (For example, each of these compounds produces six or seven different photoadducts, depending on which nearby amino acid gets snagged by the reactive carbene that forms from the diazirene, and the group was able to see and account for all of these without difficulty).

So what happens in cells? They tried the dosing/photolabeling experiment in Jurkat cells, along with experiments with photoactive negative control (non-NSAID) compounds. About 700 proteins were labeled in all, with quite a bit of overlap between the three NSAIDs (40% of them are hit by all three), and that figure alone should make a person stop and think a bit: as simple and widely used a molecule as naproxen binds to hundreds of proteins well enough to photolabel them. No, we don’t know the details of what’s going on in there, do we? Admittedly, that’s at a high concentration (250 micromolar), but (1) patients take whopping doses of this stuff in the real world, and (2) the majority of these interactions also showed up at a 50 micromolar dose as well.

This same set of experiments was also run in K562 cells, a leukemia line. In this case, there were 513 enriched protein targets and only 206 of them overlapped with the set from the Jurkat cells. And that’s also something to think about – not only do small, well-known compounds like these interact with a long list of proteins, that list can change dramatically in different cell backgrounds. Only about 30% of the total list of proteins have ever been report to interact with any small molecules at all.

The distribution of these doesn’t show any particular compartmentalization effects in the cells; it pretty much tracks protein abundance across the cytoplasm, various organelles, nucleus and so on. A closer look at the proteins and their interaction sites can be found in the paper, but one thing worth mentioning is a “hot spot” in histone H2A, which is certainly not a target that anyone had been thinking about for NSAIDs, to my knowledge. Assigning all the specific sites of labeling across the proteome is still beyond current technology, although you can think of some experiments that would help narrow things down a lot (different protein digestion before the mass spec, and so on). But even at this level – which is more detailed than we’ve ever had for such compounds – there’s a lot to deal with here.

So if you needed a reminder about just what a complicated mixing bowl we’re throwing our compounds into, here you go. The hope is that such techniques and the ungodly huge piles of data that they generate will help up build up a clearer picture of drug action. But what a picture that’s going to be!

50 comments on “More Proteins Than You Ever Thought”

  1. Rs says:

    The sooner we realise that all rational drug design is a pretext to make compounds and don’t get fixated that our compound hits only the target we are testing for, the better off we will be.

    If your compound shows functional potency in an assay but shows relatively poor ‘on-target’ potency, do you care? Time may be up for targeted drug discovery sooner than you think.

    1. Advice giver says:

      You need to learn about grad students, if you tell them to make a compound for X protein, and you push them hard enough, they will produce a compound for X protein. Sure, it will fail in clinical trials, but by then you will be a rich dude.

  2. Uncle Al says:

    Immense international chronic consumption of OTC NSAIDs over decades leaves no epidemiological trail overall or subgroup-specific. Is this anomalous, observational laziness, or evidence of a very deep thermodynamic hole for life despite disruptive inputs? Apply the same question to exponential human exposure to radio and microwave radiation re snuggled cell phones to smart everything, WiFi, and Bluetooth
    … “more than 30 billion doses of NSAIDs are consumed annually in the United States alone” or a dose every three days spread over the US population.

    1. NSAID fan says:

      wait for a couple of generations and they’ll be rediscovered as essential nutrients

    2. Nomograph says:

      You used 85 words to say “they’re safe”.

    3. eyesoars says:

      Reye’s syndrome?

  3. 10 Fingers says:

    Regarding naproxen in the real world:
    More than half of it circulates as various acylglucuronides, and the parent is at least ~98% bound to plasma proteins (other major metabolites can be higher). So, the whopping dose most people take is distributed to a variety of depots, and free parent is (if memory serves) ~1% of total. I vaguely recall that total parent routinely is around 100-300uM, but the relevant concentration range for this kind of study is more likely to be ~1uM. Some compartments probably experience higher exposures due to the flux of the acylglucuronides, I suspect, but I haven’t looked into this.

  4. tlp says:

    So maybe most of those interaction don’t make much of a difference on systemic level (cf. silent allosteric ligands)? Or the whole idea of soaking cells in the solution of a small molecule is not a very relevant model after all.

  5. what says:

    people have been putting a diazirine/alkyne on probes for a decade and looking at targets. why are you highlighting “another one” of these?

    1. Derek Lowe says:

      Because we’ve never been able to do it to this depth before.

      1. word says:

        not much depth…histones H2A and H2B are just about the most abundant proteins in cells

  6. what says:

    sure you could. just do competitions to find specificity.

    1. Ugh says:

      exactly – you put a diazirine on a greasy molecule, toss it into cells at 250/50uM, you are gonna always pull out a bunch of proteins. Do the competition experiments to get stoichiometry/specificity.

  7. Chrispy says:

    The paper is too poorly controlled to draw any real conclusions. All they have shown is that greasy molecules tend to bind a lot of proteins. Yawn. If they had a wider selection of tested molecules or used a pair or active/inactive enantiomers then you could start to infer mechanism from what the drugs were binding. In the absence of that, this is little more than a noise/publication generator.

    Those of you blocked by a paywall can find the paper on SciHub.

    1. What says:

      My point exactly. It’s a data dump.

    2. hmmm says:

      ↑ this ↑. It is a list of the most abundant proteins in the mammalian cell.

      1. sgcox says:

        Second to that.
        CRAPome in action. Gives you JACS paper these days anyway.

        1. what says:

          gives you a JACS paper because of the correspondence line. that’s why I asked why this was being covered…there is no there, there.

          1. sgcox says:

            How rude.. It goes without saying.

        2. Ugh says:

          just also noticed. all of their probes are toxic at ~20uM….but they are looking for interactions at 250uM. so all the experiments were done in dead/dying cells….aka lysates?

    3. Jacs says:

      How did it get into jacs then? Is our peer review that broken?

    4. word says:

      Some sort of functional validation would have been nice. At least for one target. As others have mentioned, folks have been doing this for awhile – putting diazirine/alkyne tags on drugs, metabolites, bioactives, fragments, etc to ID targets and binding sites via proteomics. Im not sure how the authors can support the claim that they developed a “new platform.”

  8. annoying speelcheck says:


  9. Cathy Tralau-Stewart says:

    I fear we have been deluding ourselves for years and I suspect that most/all drugs modulate many targets and pathways. This reductionist view inhibits and clouds our understanding of key mechanisms. The future is optimizing pathways and maybe multiple pathways to get the required efficacy and safety.

    1. eub says:

      This is my pet theory, that the desired treatment effect often won’t come from modulation of the target we know about, it only comes if you hit that target plus a constellation of others. We make some compounds that hit our target, and then we see if they randomly hit a constellation that works, or one that doesn’t.

  10. tlp says:

    Here’s the second thought. If the results are true and relevant, PROTAC guys should be very worried. But I bet they are not.

    1. Anon says:

      They should slap the diazirine-alkyne tag on PROTAC. That’s probably another jacs.

    2. Fuh Dge says:

      I’m not extremely familiar with the field, but I disagree with you. If you read the first paper out of the Brander group about PROTAC, they control for that sort of phenomenon – check Figure 3 here: Winter, Georg E., et al. “Phthalimide conjugation as a strategy for in vivo target protein degradation.” Science (2015): aab1433.

      Since PROTAC alters protein abundance by design, proteomics is well-suited to see what off target effects the compounds that act through PROTAC are having. The Bradner group used quantitative mass spectrometry to compare the proteomes of cells treated with their PROTAC ligand, and control treated cells. They saw that the only proteins that changed drastically in relative abundance were the proteins their ligand was degrading, and proteins whose transcription was changed after their target protein was degraded (since their ligand primarily targets epigenetic regulators).

      I’m not sure if every compound that works through PROTAC described in the literature is characterized in this way, but the experiment (or a similar experiment with using SILAC) is so simple (for anyone with a halfway decent mass spectrometry facility) that I hope the field was doing that. In brief: at least the Bradner group’s PROTAC ligand dBET1 doesn’t suffer from the problems you’re implying.

      1. tlp says:

        So I win the bet 😉

        However, if one attached diazirine to a PROTAC ligand and tracked whatever the molecule was interacting with at high micromolar concentration, one could probably found the same promiscuity that the study in the post above found. My point is ‘not all interactions are created equal’. And I’m actually in favor of PROTAC functional data rather than ‘interactome’ hairball.

  11. john adams says:

    I would LOVE to see a study like this performed using a structurally diverse collection (3+) of compounds with low nM (or better) affinity for their designated molecular target….

  12. Carbenegeezer says:

    Long time lurker. Seldom comment but just wanted to point out that diazirines generate carbenes not nitrenes.

    1. Derek Lowe says:

      Brain lurch there – fixed, thanks!

  13. yawner says:

    YAWN. Harvard prof re-discovers greasy compounds bind the Crap-Ome. Even at Harvard!!!

    1. yawner2 says:

      yawn indeed

    2. Derek Lowe says:

      Indomethacin has a logD of 1, though. . .

      1. Notherethere says:

        Grasping for straws, Derek. Post-peer review is speaking.

  14. Barry says:

    That the experiment gave two such different results in two cell lines deserves more attention. We’ve long known that any immortalized cell-line has adaptations to life in plastic, unrepresentative of the parent tumor. If this paper stands up, we’ll have question their relevance to human (or even to murine) disease.

  15. What says:

    Barry, no. That’s just stochastic nature of these experiments. Nothing overlaps perfectly.

    1. Barry says:

      If one mammalian cell-line can’t predict results in another mammalian cell-line, it’s fanciful to think that it will teach us about what happens in a mammal.

  16. loupgarous says:

    Reminds me of “Radiation damage and radioprotectants: new concepts in the era of molecular medicine” by M.I. Koukourakis, Br J Radiol. 2012 Apr; 85(1012): 313–330, which not only described types of radioprotection, but had a YUGE table listing every drug and experimental compound known to exert one class or another of those ways of protecting cells from radiation damage.

    That last table looked like a screen dump of a pharmacy’s entire inventory (and the chemistry department’s storeroom in the university down the street), and shows how complex these processes are, and ought to be food for thought for radiation oncologists and others who have to deal with radiation injury.

  17. PJB says:

    Even if one calls a target any in vitro drug-target association below 10 micromolar, there is an average of over 11 known targets per drug according to our estimations.
    If we look at the targets predicted for celecoxib at 10 micromolar (instead of known target at 250 micromolar), its polypharmacology is still evident: You can compare the chemical structures of celecoxib with those most similar annotated with the predicted target by clicking on the number on the rightmost column (NB: those targets predicted with score 1 and a single hit with 100% similarity are not predictions, but known targets of celecoxib).
    A personal favourite of mine, sunitinib:

  18. What says:

    Barry, they are different lines. They were not ordained as supposed to be the same or similar. And these experiments are not controlled well. There are many, many good studies that you can go check out if you want to get into the cell line vs reality issue. This is not one of them.

  19. Bellero says:

    You track the phenotype of NSAIDS, and we already knew that already going back thousands of years didnt we, so molecular biology didnt do much, did it?

  20. steve says:

    The way the experiment should be done is to have several negative control molecules that are structurally similar but functionally inactive and then delete the proteins that are brought down in common. Too many papers just don’t have proper controls.

    1. yep says:

      yes. this shows that the group does not even know how to run properly controlled chemproteomic experiments. worrisome.

  21. Mrs CrabApple says:

    You seem to have used the word “arbeomplicated”.

    Are you sure that is a cromulant word?

  22. MFernflower says:

    Gotta say I think this paper should be retracted – there is no mention of NSAID glycoconjugates (this is what most of the dose becomes in humans!) and the alkyne diazridine labeling system is so reactive on it’s own that if you repeated this test with a simple cyclohexyl group replacing the NSAID in there molecules you would have near identical results!

  23. Barry says:

    I haven’t made the experiment, but I’d expect that if you were to hang a diazirine and an ethyne off a cyclohexane, incubate it with cells, then photolyze, what you’ll tag will be phospholipids and steroids, rather than proteins.
    The problem is not the reactivity of the diazirine (nor of the carbene generated).

  24. exGlaxoid says:

    If we know that naproxen is 98+% protein bound in plasma, is does not surprise me that it is found binding to many proteins. What would be interesting is to compare the proteins that were labelled verses their abundance and location, so see if it binds to something other than common plasma proteins. Those targets might have some real utility to know.

    But it also goes back to my theory that drug discover could be done a lot more productively by a human HTS of simply giving low doses of compounds that appear safe in simple animal models to volunteers and looking for actual effects. This is what people did with natural products for 1000’s of years and we discovered many useful ones (and some bad ones).

    But given that we can test very low doses and determine half life, metabolism, genetic effects, and much more on tiny doses first, it would not be that dangerous. Any compounds that show signs of useful effects could be carefully scaled up to larger amounts. People do this every day with unknown drugs they bought from China, dark web, and dealers, at least this would give them something more useful to do.

  25. Kevin says:

    “…because it just keeps on getting more arbeomplicated”
    and a new word was born.

Comments are closed.