Skip to main content

Chemical Biology

Membrane Surprises

Drug discovery folks spend a good amount of time and effort dealing with cell membranes. Our drug candidates stick to them, get imbedded in them, might have to slip through them to get to their target proteins, or may target proteins that are localized in them, can get actively transported through them or actively pumped back out. . .there are a lot of possibilities. One of the simplest possible interactions is that first one on the list, when a compound just seems to bind nonspecifically. But it turns out that it isn’t always as simple as it looks.

That’s the message of this paper, anyway. The authors (a team from Queensland) are looking at a particular cyclotide protein, kalata B1. Cyclotides are interesting beasts. They’re natural products produced by a number of plant species, multicyclic proteins that feature six cysteine residues linked into a distinctive cystine-knot form. They’re known to target the cell membrane, and have a distinct polar region that interacts with the charged phosphates in the lipid bilayer, while a hydrophobic region buries down into the lipid chains in the middle. So far, so good.

The odd thing, though, is that if you make the enantiomeric form of kalata B1, the one with all D-amino acids, it’s less potent across all its biological activities (which include hemolysis, insecticidal and antiviral effects, etc.) “Well sure”, many people will be saying. “Why wouldn’t it be? Proteins are chiral and that’s part of their structure when they bind to their targets”. But think about that: we’re used to “targets” in that context meaning “other proteins”, and sure, those are chiral too and the enantiomer of your active agent is going to be a mismatch with those. But when your target is just the cell membrane? A lipid bilayer? Why should you see such an effect there?

One possibility is that kalata B1 is in fact targeting some (unidentified) membrane protein, and thus running into chiral interactions that way. That’s an easy way out, and a pretty plausible one. But this paper investigated the weirder possibility. Glycerol itself has a plane of symmetry, but glyceride esters can be chiral as you start decorating that core with phosphates, phosphocholines, and different fatty acid esters. The authors actually synthesized enantiomeric model lipid bilayers and checked to see if kalata B1 recognizes the intrinsic chirality of the membrane. And it does.

The polar region of the protein doesn’t seem to care; it binds to the phosphotidylcholine head groups just through good old polar interactions. But the second part of the binding process, the insertion of the hydrophobic part of the cyclotide into the lipid bilayer, that one prefers the natural chirality.  The team, as mentioned, already had both enantiomers of kalata B1 and a number of mutant forms of the various regions of the protein, and they were able to show all the match/mismatch possibilities: the unnatural protein has greater affinity for the unnatural membrane. When matched up, the proteins both bind better and perform their membrane-disrupting activities better. Meanwhile, when either enantiomer was tested in an achiral lipid-like model membrane, their activities were identical.

So we’re going to have to get used to paying attention to the chirality of the phopholipids themselves. It’s not clear how many other compounds recognize these sorts of interactions, but you can’t rule anything out. These effects could well be one of the subtle factors that go into the broad category of membrane permeability. It’s going to be of particular relevance to bioactive peptides, not least because the D-forms of those are often used in control experiments. The authors point out that early studies on helican antimicrobial peptides showed basically identical activity, and the assumption since then has been that membrane-targeting mechanisms don’t have to worry about this sort of thing. But there are outlier data points in the literature, which were generally explained away as some sort of off-target effect (just like the D- and L-kalata B1 results could have been). It ain’t that simple.

10 comments on “Membrane Surprises”

  1. Barry says:

    We have used the argument that ‘any drug whose two enantiomers show different potency must have a protein* target’ for decades. Notably in trying (failing!) to explicate how general anaesthesia works. But if kalata can discriminate phospholipid chirality, we can’t rule that out for fluoroalkane anaesthetics, either.

    *or polynucleotide in the case of ribosome-binders

    1. eub says:

      Isoflurane enantiomers show different potency, but isoflurane is known to have a protein target, the GABA-A receptor, where it acts as an allosteric modulator. (“Allosteric modulator” admittedly is a bit of a way to say , but the protein binding is known).

      1. eub says:

        Er, to say “<lo, here a miracle occurs>” which got eaten as a putative HTML tag.

  2. tlp says:

    I’d say that effect size doesn’t look too dramatic (Fig 3) for drug developers to start really worrying about that stuff. Phenomenon does look real but 2-3-fold difference in cell experiments for full reversal of protein chirality probably would translate into unnoticeable effect for small molecules.
    Maybe one could contemplate about role of this effect in origin of chirality.

    1. tlp says:

      Oh and they have melittin as positive control in one experiment (fig 6), which is also quite chiral but doesn’t care about membrane chirality.

    2. Derek Lowe says:

      Agreed, it would be hard to spot this in most cases. I’m just surprised it shows up at all!

  3. Cb says:

    I am surprised that scientists are surprised about the influence of chirality of a membrane composed of all sorts of enantiomerically pure glycerides, phospholipids etc on interaction with another chiral entity: indeed matched and mis-matched pairs . Many publications deal eg with saponification of racemic 1,2 diacyl glycerols by Lipases to give a single mono-acyl- enantiomer. Nobody is surprised and again nobody would be surprised if we could make the D-lipase and this produces the enantiomer of opposite chirality…….but….of course great work, in particular making the Lipase with D amino acids.

  4. Steve says:

    Chiral interactions with membranes have been noted for a while. Here is a 2001 paper on cholesterol interactions

  5. Anonymous says:

    Math vs Chem: I don’t think that the cylotides and other cysteine cross-linked peptides are, mathematically, knots. If you cleave the disulfides, you just have a plain planar circular peptide. (Oh … a simple circle IS sometimes considered to be the simplest or trivial knot but othertimes it is referred to as the unknot.) If you take the simplest NON-trivial knot, the trefoil knot, and cut the ring anywhere, you can untangle / untie the knot and glue to ends back together to get a plain, planar circle. I think that most knots (not a mathematician! not certain of this!) can be unknotted with a single cut and reannealed back into a trivial circle.

    If you cut the peptide backbone of a cyclotide, you cannot untie the knot because of the cysteine disulfides.

    Many of the cyclotides are “topologically non-planar.” Synthetic chemists were introduced to non-planarity by Simmons’ K_5 molecules (1981) and then by Walba’s Moebius ladders (K_3,3 non-planar) (1982). There were, already, many non-planar peptides and proteins in the literature but they had not been recognized as such. See a series of papers by Mislow and Liang on non-planar peptides and an algorithm to search for them in structural databases.

    Lipid chirality in membranes (and liposomes) has been know for many decades. I don’t have access to the lit and numerous comprehensive reviews to find examples of the interactions of chiral molecules (e.g., cholesterol, an important membrane component; small molecules, e.g., anesthetics, as mentioned above) and chiral lipid membranes. I would think that observation of such effects would have to back to the 1960s or earlier.

    What about the cardiolipins?

  6. Shane says:

    How fun to see my area of doctoral study turn up on my favourite blog. It is great to see gradual progress on the mechanism of cyclotides being made, with interesting more general insights along the way. The structure of the active form of the peptides in the membrane is still elusive, a reminder that the very idea of structure in biochemistry is limited to specific scales of time and space that leaves out a large amount of reality. For me the big remaining question is how do the plants that create cyclotides manage to prevent them from damaging their own membranes? They aren’t stored in a pre-folded and inactive state and seem to accumulate in the vacuoles if I recall correctly. Do cyclotide bearing plants have something peculiar about their membranes or is some other trick at play? One link might be that many cyclotide producing plants are mineral hyperaccumulators (nickel, zinc etc) so that might be a deactivation mechanism.

Comments are closed.