Skip to main content

Chemical Biology

Sneaking Proteins Into Cells

Now here’s a weird and rather startling paper. One of the things that people in this line of work spend a lot of time on is getting things into living cells. Small molecules often slide in, one way or another (although, to be honest, our detailed understanding of how they do that could use some work). But full-sized proteins? Not so much. There are active transport pathways that can bring such things in, after presenting the right molecular password, but they’re picky and not always very reliable (people hang various cell-penetrating peptide sequences off of other species and hope for the best). As for just passively soaking in through the membrane (the way that we figure many small drug molecules do), that’s considered pretty much out of the question for larger proteins – after all, what would cells be like if proteins could just wander in and out on their own? So to get such species in, we resort to brute-force techniques like micro-injection and electroporation, physically penetrating or altering the cell membrane enough to get larger cargo inside.

Even for small molecules, things can be hard, especially if you have negatively charged groups like carboxylates on them. Cell membranes tend to reject negative charge in general, which is one reason that drugs in this category are often modified into prodrugs, usually some sort of ester that will give them better properties and eventually get cleaved to the active species. (Another reason is that carboxylates and the like are often also marked for fast clearance even when they do make it out into the bloodstream). This is a pretty common strategy, although it does complicate your life, since you’re now effectively dosing two drugs instead of one, and you have to investigate both species (as well as deal with the likely situation that different animal species will cleave your ester at different rates than humans will).

This new paper (from the Raines group at MIT) continues their work at extending this idea to whole proteins. They’d previously shown that good ol’ green fluorescent protein could get into cells if its surface carboxylate groups were esterified (via the diazo derivative of an arylglycine amide), and now they’re demonstrating that more stringently with a ribonuclease enzyme. That’s not something that the cell lets run around loose – there’s a ribonuclease inhibitor protein that has a ridiculously high affinity (sub-femtomolar). But ribonuclease variants with more anionic character don’t bind (although they’re less efficient as enzymes), so the group chose one of these, a species with five Asp and Glu residues swapped in, to test their idea. Neither the wild-type enzyme nor that DDADD variant get into the cell to any degree. Treating HeLa cells with 100 micromolar ribonuclease did nothing to them – but when that wild-type was esterified it had an IC50 of about ten micromolar. Meanwhile, the DDADD form also was totally inactive against cells, but when esterified it had an IC50 of 6 micromolar, despite being a less active enzyme in general. Control experiments with a ribonuclease whose catalytic site was modified (to make it inactive) showed no effect on cells either way, which ruled out nonspecific effects such as disrupting the cell membrane, etc.

Similar effects were seen in H460 lung-cancer cells, so this isn’t something that just happens in HeLas, which can be rather weird cells, to be sure. A final experiment involved a FLAG-tagged ribonuclease that was delivered into cells, and then after 24 hours the cells were lysed and the protein recovered by use of an anti-FLAG antibody column. (FLAG, for those outside the field, is a short peptide sequence that has had excellent antibodies developed for it – you can often append a “FLAG-tag” to a protein without disrupting its function significantly, which gives you an instant handle for experiments like this). The recovered ribonuclease enzyme had indeed had its ester groups stripped off during its time in the cell, as hypothesized (and it had also had about 160 daltons of phosphates stapled onto it as well).

So we might be looking at a new and rather simple way to deliver functional proteins into cells. It’ll be quite interesting to see how many other proteins this works on, and indeed whether the idea can be developed into a way to administer them therapeutically. Hey, it works with small molecules – perhaps there are esters that allow for blood stability (no easy feat) and cellular entry, but are cleaved intracellularly? Folks running in vitro experiments on cells, though, might do well to start paying attention to the idea right now, because it looks ready to be tried out.

21 comments on “Sneaking Proteins Into Cells”

  1. Isidore says:

    Very interesting, perhaps pointing to a strategy for getting ADCs to penetrate solid tumors.

    1. Greg says:

      If ADCs passively entered cells, you might lose the targeting function. They’d float down a concentration gradient into any old cell, creating an off-target problem.

      1. johnnyboy says:

        Indeed. The problem with ADCs appears to be getting them to the tumour in sufficient concentration compared to off-tumour binding/diffusion. Once they get to the tumour cells, getting them internalised is not the issue, if the antibody has been designed correctly.

  2. Joseph Novetsky says:

    160 daltons is around 1.7 phosphates. So are you saying that there were only about 2 phosphates on the molecule, or that it was 160 daltons heavier, which could mean more phosphates once you deesterify your protein?

    1. Isidore says:

      160 Da is the mass of exactly two phosphates. For each phosphate you add net [PO3H] to a side-chain hydroxyl group.

      1. Sig fig newtons says:

        Yeah, what Isidore said is pretty much correct. If you go into chem draw and draw acetic acid, the mass is 60 Da. If you add a phosphate (you have to lose and OH), the mass becomes 140. (2*140-2*60=160). The same mass difference is observed if you put a diphosphate instead of 2 monophosphates. The mass spectra for this are in figure S6, but I’m not sure that they provide the most convincing evidence for phosphorylation. The mass accuracy in their whole protein mass spec isn’t ideal, though I believe mass accuracy at such large masses by ESI is difficult. It might have made more sense to have done a bottom up proteomics experiment by digesting with trypsin and looking for peptides. Phosphoproteomics has its difficulties, but this bottom up experiment might say where the phosphorylation occurs and be more definitive if it actually occurs.

        1. Isidore says:

          Actually for small proteins like RNase with the instrument they are using getting mass accuracy by ESI-MS of 1 Da is perfectly reasonable, in fact expected. Addition of 160 u could, of course, be explained by a combination of other modifications, just pick any two that add to 160 from this list: However, phosphorylation at two sites appears the most logical choice given the experiment. It would have been nice to confirm this by digestion and identification of the peptides modified, but I suppose that was not the main focus of this paper.

  3. Hele says:

    I was thinking about this for a long time.
    Glad someone did it.
    Thank yout for pointing it out Derek

  4. Clark says:

    IC50 of 6 micromolar for a protein is still too high. I would be more excited if they can do it 6 nanomolar or lower.

  5. Troy Boy says:

    If you want to deliver biologics this way to cells, wouldn’t you still have to worry about an immune response? Wouldn’t 6 uM protein cause all sorts of bad things?

  6. Anoni says:

    “esterified it had an IC50 of about ten nanomolar.”

    10uM no?

    1. Derek Lowe says:

      Oy, that’s a bad one. Fixed!

  7. biologist says:

    blog request – 3D genome structure?

  8. Tomas says:

    Could anyone please recommend a cheap, commercially available diazo compound for protein esterification if possible? Any alternatives to 2-diazo-2-(p-methylphenyl)-N,N-dimethylacetamide? E.g. to esterify miligrams of protein with high efficiency at pH ~5.5 in aqueous buffer. Thanks!

  9. MrXYZ says:

    It would be very interesting to try this with a VHH (llama nanobody) or an scFv against an intracellular target. Nanobodies against the intracellular face of GPCRs might be a good place to start.

  10. Anon says:

    Has anyone managed to get an antibody or antibody fragment into cells with any great efficiency? On its own, without using some other kind of carrier molecule?

    Are there any other known proteins or protein fragments (either natural or engineered) that can wizz through the cell membrane very easily?

      1. Anon says:

        I mean, are there any globular protein domains that easily cross the cell membrain without these funny peptides tagged on?

        1. Derek Lowe says:

          Now that’s another question! No examples spring to mind, at least to my mind. . .

          1. Mo says:

            EF-Tu leaves bacteria, but that could be due to active export. Somehow. But it could also just be randomly and then the quantities are high (because of the abnormally high intracellular concentration) enough to evolve extracellular functions.

  11. Chicago Discovery Solutions scaled up the synthesis of this molecule……Proud to be associated with this molecule..

    Publication mentions…….This Diazo compound 1 (2-diazo-N,N-dimethyl-2-(p-tolueyl)acetamide) was synthesized by Chicago Discovery Solutions (Plainfeld, IL) as described previously……

    If anybody needs more quantities of this amazing but challenging to synthesize molecule 1 (2-diazo-N,N-dimethyl-2-(p-tolueyl)acetamide), please contact us at (

Comments are closed.