Skip to main content
Menu

Drug Assays

The Good Sides and Bad Sides of Polar Compounds

Yesterday’s mention of carbohydrates brings up another topic, one that was raised in the comments and in some email correspondence. Most drug companies with an internal screening collection are concerned, to some degree, about how greasy that collection has turned out to be. The concern comes from the general perception that the more hydrophobic a compound is, the harder it will be to progress to a drug. It would be safer to add a “past a certain limit” in that phrase somewhere, and it would also be safer to say that this rule is not as solid as some people believe it to be. At one point there was a lot of thought that polarity correlated pretty well with later PK and toxicity, but (as those links show) this is a more complicated story, although not everyone’s gotten that memo.

Rather than reopen that discussion, though, let’s come at the polar compound issue from another direction. In general, I think that the argument for having some more polar chemical matter is to be able to address a wider variety of binding sites. A look at (for example) antibiotics suggests that there are whole classes of useful drugs that are on the polar fringes of what medicinal chemists consider acceptable. We’re probably missing out by not having enough chemical matter in that region. (It’s worth remembering that even back in the early days of Lipinski’s “Rule of Five” that he explicitly noted that it did not seem to apply to antibiotics).

Naturally enough, one reason that there are so many comparatively hydrophobic species in the screening deck is that it’s been partially filled by all the med-chem optimization programs of years past, so you have a lot of later-stage analogs in there that may not be quite as “lead-like”. Another reason, which sounds like just an excuse but is pretty practical, when you get down to it, is that the more water-soluble compounds are harder to work with in the lab – harder to purify, harder to get concentrated down out of all that water, etc. So why not fix this by just buying a lot of more hydrophilic stuff? That’s not an unreasonable idea, on the face of it, but how well does it work?

Perhaps not as well as you’d want. I’m willing to bet that similar-sized collections of “rather polar” and “rather nonpolar” screening compounds will show quite different hit rates across most screens, with the advantage going to the greasier ones. Hydrophobicity is one of the most reliable ways to get potency, after all – if your compound wants to flee the solution state for the comfort of a binding site, that’s what potency is, and a lot of binding sites are friendlier than bulk solvent. That’s not always the best way to get said potency, since your compound may not be so particular about what ports it sail into, but it is straightforward.

And the hydrophilic compounds have some extra issues. If they’re happier out there in the aqueous solvent than usual, they have comparatively less reason to go anywhere else. To go somewhere else, they’re also going to have to shed some of those solvating water molecules, and that’s a complicated process. You have the enthalpy of favorable hydrogen bonds to give up, balanced by the possible entropic gains in letting those waters go (but that all depends on how ordered they are in the bulk solvent versus around your compound). On the receiving end, your polar compound is going to have to make some polar interactions with the binding site for the whole process to have any thermodynamic chance at all – the worst thing of all is to poke a hydrogen-bonding group into an unreceptive binding region. But those polar interactions are pickier than the hydrophobic ones. Hydrogen bonds in particular are quite directional, and one that just misses is probably going to be worse than one that never was there to form in the first place. It’s not that placement doesn’t matter with hydrophobic interactions, but on the whole, they do tend to be “blobbier” than the polar ones.

So while I think that adding a bunch of polar chemical matter to the screening deck is a good idea, it’s not going to be any kind of instant fix. Especially if you’re going to have to have (as I suspect) two, three, four times as many of those compounds as you do of the “regular” kind in order to have a real effect on your screening hits. That’s my hypothesis, anyway – I’m pretty sure that no one has managed to put so many of these into their collections yet, though. I note that the “dark chemical matter” paper, which talked about compounds that look fine but never seem to hit in assays, found that such compounds tended to be noticeably more polar and have fewer aromatic rings. This, I’d say, is not at all accidental. But as that paper noted, if you do get a hit from this part of the collection, it’s worth prioritizing, because chances are lower that it’s leading you on. And I’d say the same for any unusually polar compound (polyhydroxylic, etc.) So long as it’s not full of quinones or the short list of things that (most of us) would immediately run from, it’s worth prioritizing.

This kind of thinking has implications for the “unnatural natural product” field. Natural products themselves have had evolutionary optimization; that’s why they’re so special (for some definitions of “special”). But there have been moves to more randomly replicate those kinds of structures, with those kinds of properties, as a way of getting more diverse chemical matter. I’m sympathetic, but as mentioned, I think that if you’re going to put together a collection like this, you’d better be prepared to make it a lot larger than you’ve probably planned for.

43 comments on “The Good Sides and Bad Sides of Polar Compounds”

  1. Andy says:

    I sometimes wonder if we aren’t somewhat constrained, when exploring chemical space, by the fact that many building blocks are derived from or related to materials considered useful to man. It’s their raison d’etre. Their existence is already a function of selection, because the chemistry that allowed their synthesis was aimed at producing something that’s useful to man.
    The problem with this line of thinking is it’s hard to think of a conclusion or a possible solution to this limitation. A second problem is it makes my brain ache….

    1. loupgarous says:

      Good point. And a possible application for “deep computing” – exploring chemicals which aren’t in the libraries because no one’s worked with them yet, but are known biological entities – and ringing structural changes on them, and exploring those, too.

      Arsenicin A was found in New Caledonian marine sponge Echinochalina bargibanti and has an adamantane-like cage structure with four arsenic atoms – the first polyarsenic organic compound to be isolated. It’s been found to be active against acute promelocytic leukemia cells at concentrations 27 times lower than Trisenox, the arsenic (III) oxide compound used to treat that condition.

      I’m not a med-chem guy, so I don’t know – are there a lot of compounds like this being looked at in drug discovery compound libraries? This is just one example of an odd natural compound which seems to have potential for clinical use.

      1. loupgarous says:

        Here’s the link to the paper showing Arsenicin A’s activity against acute promelocytic leukemia cell lines: http://pubs.acs.org/doi/abs/10.1021/om201180d

      2. Me says:

        From looking at the sorts of stuff we had in our screening collections throughout my time in med. chem., I’d say no, those sorts of molecules aren’t being looked at by ‘traditional’ pharma companies.

        But then again, niche is king these days – wouldn’t be surprised to find a small biotech somewhere hawking such a discovery platform. And good luck to them if they are.

    2. Alex says:

      What about Salvarsan ?

  2. PoFI says:

    I’m sure that Mothers Against Molecular Obesity (MAMO) are already working on their Polar Forecast Index, based upon the Hogwarts School of Thermodynamics.

    1. Some idiot says:

      Classic! My coffee just exited my nose… 🙂

    2. Peter Kenny says:

      For years, Mothers Against Molecular Obesity(MAMO) have recommended the auto-da-fé as the most effective correction for heretics choosing to ignore The Sacred Rule Of Five and The Holy Forecast Of Properties. Repent now under PAINS of Enthalpy and heed the wise teachings (probably more intelligible, even to English speakers, in the original Hungarian) of the Hogwarts School Of Thermodynamics.

      1. Matthew 4:17 says:

        “Repent, for the kingdom of holy PFI has come near!”

  3. b says:

    To go along with your “electron-donating substituent the size of fluorine” on the chemist’s wish list, I’ve always wished for a highly polar solvent that was easy to evaporate.

    1. Derek Lowe says:

      Now that would be a big seller, for sure. Acetonitrile is the closest thing, but it’s just not all that polar.

    2. will says:

      methanol?

      1. b says:

        I’m talking more like a volatile DMSO

        1. Anon says:

          I wonder what hexafluoro-DMSO would behave like? It has been made, probably not that stable.

          1. will says:

            I can’t read the article, but I wonder if it is polar

          2. Anon says:

            Hmm, stable to water bp of 36C, partial decomposition after 2hrs at 200C.

            Kind of a pain to make though lol…

    3. a. nonymaus says:

      Have you considered anhydrous HF or neat HCN? Both are more polar than water, and have boiling points just below room temperature.

      1. b says:

        Good idea. I’ll try them out tomorrow. Hey, it’s Friday after all.

      2. Barry says:

        neat HCN isn’t particularly polar as a solvent and rapidly produces (non-volatile) trimer and polymer. 2,2,2-trifluoroethanol and TFA are both polar, volatile solvents. Polar, aprotic, and volatile is the yawning gap.

        1. Morten G says:

          How would a mix of ammonia and 2,2,2-trifluoroethanol or TFA fly? Only for a few applications obviously (I’m not an organic chemist).

          1. Peter Kenny says:

            A mixture of ammonia (stronger HB acceptor than donor) and TFE or TFA (stronger HB donor than acceptor) may well be less polar than either component.

    4. Carl Feynman says:

      How about ammonia? It’s highly polar and quite volatile. In fact it’s too volatile to be a convenient solvent.

  4. Barry says:

    I’ve never seen an estimate of how large the effective screening deck has been over millenia from which biologically active non-peptide natural products (e.g. morphine, quinine, erythromycin, streptomycin, Taxol, artemisinin…) have emerged. Organisms that make these secondary metabolites presumably gained competitive advantages over cousins that made something else or nothing at all. But I wouldn’t be surprised if the answer dwarfs the 10e5 that we often throw at an HTS campaign.
    As Derek noted, polar contacts (H-bonds and salt bridges) are often an enthalpic wash–you spend as much energy to desolvate the ligand and the host as you get back in the docking energy. But that polar potential drug also has to spend that energy of desolvation every time it (passively) traverses a phospholipid bilayer, with no docking energy in compensation.

    1. Anononon says:

      It would be of tremendous size, but unfortunately most screening programs don’t allow for millions-billions of years during the HTS phase. Investors and all that.

      1. Barry says:

        There you have it. Lipinksi’s “rules” aren’t inviolable law. But they guide us towards compounds that can be optimized and developed before the funding runs out.

  5. Kelvin says:

    It seems that the best drugs adapt to the hydrophilicity/phobicity of their environment, e.g., by folding up or associating into dimers, etc.

    1. Barry says:

      Lipinski noted that H-bonds that can be satisfied internally don’t count against you for transport properties. But I think the entropic advantage of pre-organization (rigidity) is more important than any benefit you can explain by refolding for transport. Got examples in mind?

      1. Kelvin says:

        Cyclodextrins and macrocyclic antibiotics? Great solubility and bioavailability despite many hydrogen bonds and high MW, in great contrast to Lipibski’s rules and most biologics.

        1. Kelvin says:

          I also meant to include cyclosporins.

        2. Andy II says:

          “Cyclodextrins and macrocyclic antibiotics? Great solubility and bioavailability despite many hydrogen bonds and high MW,”

          I’m afraid that these are not “orally available.” And, you will be amazed to see the aqueous solubility differences in cyclodextrin of different size. Alpha, beta, gamma. Beta-CD has been utilized to solubilize hydrophobic drugs but has some tolerability issues and thereby the drug cant be dosed high. Posaconazole is a good example. Merck has made it from a suspension into a tablet without b-CD. Macrocyclic abxs? Are you talking about daptomycin, colistin? These are not orally available and have to be dosed via iv.

          Yep. Antibiotics are very unique classes, especially for gram negative pathogens. as pointed out, they do not follow Lipinsky’s rule very well. Polarity works to get the molecule in but sometimes it prevents molecule from binding to the target site.

  6. loupgarous says:

    The link “did not seem to apply” is actually to a Canadian who cites Lipinski’s paper on the “rule of five” not applying to antibiotics

  7. Nate says:

    Lots of prodrugs drop some non-polarity after conversion to active forms ( stick a methyl-, ethyl-, propyl-) and let esterases do the work after you make it across lipid membranes.

  8. Peter Kenny says:

    ‘Polar’ can be defined in a number of ways (gas to water transfer free energy; alkane/water partition coefficient; octanol/water partition coefficient) and one needs to be careful when using the term ‘hydrophobic’. I have linked a recent article (The influence of hydrogen bonding on partition coefficients) that might be of interest to those participating or following this discussion as the URL for this comment.

  9. milkshaken says:

    antibiotics often do not have to have cell permeability – and the injectable ones do not even need oral availability. Ro5 is just a rule of thumb when discussing drug-likeness of a screening hit. The presumption being you are trying to get orally active small molecule targeting something inside human cell. But if it is just supposed to hit a receptor that is on the surface of epithelial cell, or a blood coagulation cascade, then it is maybe better if it does not follow Ro5. Similarly, if you want to get a molecule passively into brain (without relying on active transport), it better be Ro3, and preferably basic.

    Too greasy compounds: I worked on anti-tumor alkylphospholipids (N-acylated phosphoethanolamines). These things were so bad they did not dissolve in DMSO, and did not form true solutions at any concentrations. They did not even form liposomes. The way to get them into “solution” was to add diluted ethanolic stock gradually into diluted BSA, with sonication, then freeze, lyophilize, re-constitute and test. Pray that the double bonds did not auto-oxidized on air in the process. (Oxidized lipids when injected IV into animal are intensely immunogenic. ) The compounds had number of additional fun properties (formation of emulsions, formation of phosphate salts by ion exchange, gel formation and sudden transition into light “snoflakes” flying into rotovap or highvac manifold.)
    So the trouble seems to be that extra greasy compounds need careful formulation work and extra attention during testing, to make sure the obtained results are not from some artifact.

    1. Barry says:

      it’s true that antibiotics targeting the cell wall (beta lactams, vancomycin…) don’t need to get into the cell. But others (macrolide antibiotics, streptomycin…) that target e.g. the bacterial ribosome certainly do. And some of that latter class display features (mw, H-bond donors, charge) far outside the mainstream of our screening collections.

    2. Morten G says:

      Miltefosine is a hell of a drug.

  10. tlp says:

    >> Hydrophobicity is one of the most reliable ways to get potency, after all – if your compound wants to flee the solution state for the comfort of a binding site, that’s what potency is, and a lot of binding sites are friendlier than bulk solvent. That’s not always the best way to get said potency, since your compound may not be so particular about what ports it sail into, but it is straightforward.

    I think this passage deserves some kind of prize for the most clear explanation of potency and selectivity to lay audience.

  11. Tyr says:

    Why isn’t this called “The Positives and Negatives of Polar Compounds”? It’s so perfect

  12. We work with our clients to meet their unique needs, processes, and requirements, no matter how extensive. The owners of Horizon and its employees know that one size does not fit all.
    Oil well testing

  13. john_boy says:

    Great post Derek. My institution is one that is going down the polar screening deck route for various reasons so we may start to see more data on the comparitive hit rates. When discussing the merits of polar screening molecules I always go back to the molecular complexity view and remind people that a poorly placed polar group will reduce affinity more than a poorly placed hydrophobic group.

    It annoys me that we cant accept that a screening set is not a drug set – the issues that reduce discovery efficiency are much more varied than time taken from Hit to Lead.

  14. John Smith says:

    Dr Lowe, These grifters rased 10 million for a snake oil device. They need a light shone on them.

    http://www.geekwire.com/2017/nativis-raises-10m-device-uses-electromagnetic-fields-mimic-affects-cancer-drugs/

  15. franko says:

    When you say ‘wanting to flee the solution’ are you thinking of an entropy term? Call it ET(desolvation). I guess that would be different from the entropy term for displacement of water (call it ET(displacement) from the protein binding site. So then would the IC50 be related to Binding Energy + ET(displacement) + ET(desolvation)? Or would ‘wanting to flee the solution’ just be related to logP? We have LiPE = pIC50 – logP but that’s something different, isn’t it? Can anybody recommend some papers on the contribution of this term ‘wanting to flee the solution’ to the overall binding energy?

    I love the way you are raising, on the web, the level of discourse in science!

Comments are closed.