Skip to main content


Drugs Inside Cells: How Hard Can It Be, Right?

One would imagine that we drug discovery and development types have a reasonable handle on how much of our latest candidate compound makes it into cells. And that we would further know how much of that is floating around freely in there, versus tied up to some protein or another. One could not be more wrong.

This is truly a major (the major?) unsolved problem in pharmacokinetics: intracellular concentrations and distribution. There are situations where we can get such data (often with a fluorescent compound or something of that sort), but those are far outnumbered by the ones where we really have only a fuzzy idea. Considering the number of drug targets that are inside the cell membrane, that’s a big gap – and we really should be saying “membranes” there, too, because once inside the cell you have the mitochondrial membrane, the nuclear membrane, and all sorts of other gates, walls, and velvet ropes that your compound might or might not pass.

This new paper (from Manchester) is a good look at the situation. As they say, understanding such things “would provide enormous clarity” which is now lacking. Unfortunately, the lack of clariy is not only due to the lack of data – we have our own misunderstandings to deal with, and the first one the paper calls out is the way that a lot of us think about compound binding (and free concentration) in the blood plasma:

A key driver of intracellular concentration in vivo must be the unbound drug present in the circulation. Unfortunately, this subject is greatly misunderstood in the scientific community. Prestige publications in high-impact international journals still carry a message that plasma protein binding determines the unbound concentration in the circulation. Anecdotal collection of publications and presentations would indicate that over one-half of the drug research scientific community has been misled.

Instead, it’s all about clearance – clearance of whatever got absorbed and escaped first-pass metabolism. Plasma protein binding is an equilibrium process, and the drain down at the bottom right of your mental picture of it is compound clearance. As the authors note, there are basically no examples of any drug-drug interaction that works through displacement of plasma protein binding. Doesn’t happen. If it really were a determining factor for unbound concentration, that wouldn’t be true.

There’s also confusion around the volume of distribution. Unconsciously (well, sometimes consciously) people assume that if a drug has a low distribution volume that it must also be reaching low concentrations inside cells, perhaps because it’s having difficulty reaching cells at all. But that doesn’t follow, either. The big influence on steady-state volume of distribution is compound binding (or, if you like, fraction of unbound compound, same thing). If your compound doesn’t bind to much, it’s going to show a high volume of distribution, since it’s going to be spread all over the aqueous compartment(s) of the body, and you can get all sorts of VoD numbers down from there as you get more or less lipophilic, charged, and so on, but these aren’t directly tied to unbound intracellular concentration.

The paper goes into several other pharmacokinetic assumptions of this sort; I definitely recommend it for clearing your head on these subjects. It then gets down to the question of intracellular concentrations, with examples from infectious disease (tuberculosis) and oncology, among others. Our data are just not all that good. As mentioned, many of the studies in the field depend on some sort of fluorescence, and tagging drug molecules with a fluorescent group profoundly changes their properties. You have cases like doxorubicin, where the drug itself has intrinsic fluorescence, but it’s pointed out that “dox” is likely a low-permeability drug to start with, and that the great majority of cellular permeability studies with it have been single-dose (with a relatively short time course) rather than multiple-dose steady-state conditions.

And getting such data without fluorescence is an even harder problem. The paper presents an analysis of published data on various tyrosine kinase inhibitors, and estimates that at best, most of them don’t have average steady-state concentrations even as high as four times the measured Ki or IC50 values, and if you assume a more physiological concentration of ATP, most of them probably have unbound plasma concentrations *below* those values. But they still work.

In the end, the paper presents two options to deal with all this uncertainty: we can try to come up with some new technology that can measure bound and unbound drug concentrations in different cell types (preferably in vivo). . .or we can just say the hell with it, assume that we need high (lipoidal) permeability to get through cell membranes, which will help to cancel out transporter effects in things like CNS, tumor cell, and bacterial penetration (and when possible, try to design away from those as well). The authors advocate the second choice! That’s largely because the technological barriers are extremely high. As they say, doing such measurements on single cells in homogeneous culture is still at or beyond the state of the art. And as soon as you move from single-cell measurements to samples that have different varieties of cells in them (that is, something approximating the real world), “the idea of technology helping to provide a global solution becomes almost fanciful“.

In the end, you’ll be trying to make compounds with the highest passive permeability you can, and avoiding the big efflux pumps like P-gP as much as possible. “But we already knew that”, is going to be the reaction, and I think that this paper is replying “Yes, you did, didn’t you? Or you should have.” The idea of measuring intracellular concentration (more particularly, unbound concentration, seems like something we would very much want, but “the more we ask the more complex it becomes“. Better, the authors say, to stick to first principles as much as you can.

I see their point! After all, those first principles are not going away, no matter how fancy our instrumentation gets. The tricky part is that these days we’re all pushing the boundaries of compound properties with exotica such as bifunctional degrader molecules, protein-protein inhibitors, drug-biomolecule conjugates and all sorts of other things. Passive permeability may be a challenge for some of those things, and the urge to know just how much of them are getting to the targets is not going to go away. But I think this paper performs a valuable service in showing just how hard a request that is, and in warning people to keep their thinking straight about pharmacokinetics in general.

34 comments on “Drugs Inside Cells: How Hard Can It Be, Right?”

  1. Patient says:

    Interesting topic, but I’m refreshing the page for your thoughts on the aducanumab comeback…

    1. kevin says:

      That is exactly why i’m here as well

    2. Derek Lowe says:

      Lots of (real day-job) work today, and lots of dust still settling. That’s tomorrow’s post, tough!

    3. anon says:

      Not sure if data look solid.

      1. former biogen employee says:

        It would be a home run if they threw in the towel too early, silly management.

  2. Engaged says:

    Or even confirming direct target interaction vs. inference from drug concentrations in blood or (now maybe) in cells? Important problem for chemoproteomics to solve. This recent paper shows quant. of drug-target engagement in small tissue or blood samples (PMID: 31591248 ). Need more like it.

  3. HTSguy says:

    Maybe I’m misunderstanding this, but HSA normally is in the regime called “receptor excess”. This is not surprising, since HSA in around 600 mM in the blood. So displacement is extremely unlikely. So I don’t understand the following statement:
    “As the authors note, there are basically no examples of any drug-drug interaction that works through displacement of plasma protein binding. Doesn’t happen. If it really were a determining factor for unbound concentration, that wouldn’t be true.”

    1. ScientistSailor says:

      Was going to say the same thing…up vote, if we had that on this site…

      1. Richard H. Ebright says:


        1. unbound in Seattle says:

          “there are basically no examples of any drug-drug interaction that works through displacement of plasma protein binding.”

          I was going to write… Is it possible this is so because the ratio protein/drug is so high that the system never comes even close to saturation? So, there’s no DDI because the second drug does not realize the first drug took away (say) 1% of the binding sites….

          1. That’s the same point the people you replied to were making, just in different language.

            As an additional point, if you take random drug A and random drug B, they’ll probably bind to different places on the serum proteins, so interaction wouldn’t be expected even if you were saturating all their binding sites.

    2. Someone says:

      I would prefer not to generalize here.
      In the strong binding limit (koff assumed low) you will saturate the HSA in couple of doses if koff < Cl. No matter what is the concentration of the HSA.

      1. HTSguy says:

        If koff is slow, the affinity is very likely to be high and thus the dose low. Affinity is not the only thing that matters here: the total amount of HSA does, as well (Druid says 1.8 mmoles). 2 X ug doses are not going to saturate HSA.

        1. HTSguy says:

          Oops, typed too fast: that should be ug/kg doses.

      2. HTSguy says:

        I just realized that “Someone” was referring to slow off-rate (high affinity) for HSA. Does anyone know of any examples? Even compounds with 99.9% protein binding turn out to have uM affinities and fast on/off kinetics (as measured by SPR). Any counterexamples (with references, if possible)?

  4. PUI prof says:

    I thought displacement of warfarin from HSA by other drugs was the classic DDI involving protein binding.

      1. PUI prof says:

        Thanks for the reference. I’d claim to be misled, but your enlightening reference is almost 20 years old!

      2. Druid says:

        Sulfonamides are a risk for causing displacement because the dose is large and the molecular weight is low. Sulfadiazine mw 250 – a 4g dose is 16mmoles. Albumin in circulation is about 3L x 600uM = 1.8mmoles, with a slightly larger amount in tissues. So, there is an excess of drug and a definite risk of displacement. A displaced drug has a higher free fraction and the total clearance is increased until equilibrium of the free concentration is restored. So, it is a transient effect. However, warfarin has a half-life of 40 hours (could be shorter under displacement conditions), leaving enough time to cause harm. In addition, sulfa drugs are inhibitors of the CYPs that metabolise warfaring, adding some options (and confusion) to the mechanism of action. Not a risk I would want to take if on warfarin. Other anticoagulants are safer.

      3. PM says:

        They state themselves that the new steady state is reached after 7-10 days. Warfarin is carefully titrated drug. In pactice it means that there is a high risk window of a adverse bleeding when displacer is added. On contrary there will be a high risk window of adverse coagulation when the displacer is withdrawn.
        The authors state “major problems have been infrequent”. Well that means that they are there and are just reflecting the probability of something will happen within those risky days. Also there is no mentioning of a “minor effects”.

    1. willie gluck says:

      Aside from the cited “Opinion” see a paper by Les Benet, a major figure in the PK world
      Benet LZ and Hoener BA (2002) Changes in plasma protein binding have little clinical relevance. Clin Pharmacol Ther 71:115–121.

  5. anon says:

    Just curious and because many have been working in the area of drug loaded cell penetrating peptides, nano delivery and other concepts, how much drug concentration are we talking about that is getting delivered?

  6. Nice PK! says:

    “Prestige publications in high-impact international journals still carry a message that plasma protein binding determines the unbound concentration in the circulation.”

    So, are we saying that unbound drug concentration is determined by the clearance and not by the PPB?
    If so, 2 drugs with the same clearance (say CL=2 mL/min/kg) but PPB=1% and 99%, respectively, will produce the same unbound concentration?

    Could someone please help me understand? Thanks.

    1. Med chemist says:

      I was confused about this similarly. Dogma I have always learned is that free drug is total drug*Fu, which based on the free drug hypothesis should be the concentration in tissues (provided there isn’t active uptake or efflux). Obviously Fu impacts the CL since only free drug can be cleared by the liver.

      By your point though- 2 drugs with the same CL but different fraction unbound will have very different CL,u – which is a better measurement for comparison than total clearance since it will reflect how much free drug is available for target engagement. Or at least, so I thought. This post seems to contradict this thinking, unless I’m mistaken. Any experts can weigh in here?

      1. Dmpk says:

        If two different compounds have same CLtot and same distribution etc but different fup, it’s possible for one to be more “favorable” estimated AUCu. But the fold changes required for fup to alter in vivo CLint may be pretty high.

        1. Nice PK! says:

          Then, What about two compounds with same unbound clearance (say CLunb= 20 mL/min/kg) but different PPB of 10% and 90%. Are their steady state unbound concentrations the same?

          1. Dmpk says:

            Possibly if CLtot and distribution is the same. CLu is useful for testing effect of pk properties like nonspecific binding but I think estimating active in vivo conc should be done with unbound concentration.

          2. dmpk says:

            Also see
            J Med Chem. 2014 Oct 23;57(20):8238-48. doi: 10.1021/jm5007935

          3. Nice PK! says:

            Thanks for your comments!
            Your JMC reference shows in equation 1 that unbound concentrations depend on PPB, for their model – which obviously is different than the one used by the authors of this new paper.

            Also, since CLunb=CL/fu, the only way that CLunb=CL is when fu=1.
            Is it possible that, in spite of what the authors assert, PPB has an effect on unbound drug concentrations?

          4. dmpk says:

            Equation 1 is a static in vitro system with no clearance.
            One thing that is hard to separate here is whether one is trying to compare different compounds or the same compound under different PPB conditions. If you could change PPB in vivo (e.g. using rat with lower albumin) with the same compound, you will increase CLtot so any gain in AUCu is offset. If you try to compare 2 different compounds with different PPB in vivo, then you need all the other properties to be the same to compare the effect of PPB on AUCu.
            So, not sure if I’m answering your question! Personally i think it is fine to estimate AUCu plasma using in vitro PPB, but if you show that one compound achieves higher AUCu than another, it’s probably not just because PPB is different. Yes, CLu=CL/fu, but if CLtot increases as fu increases, then when you calculate AUCu, it may be no different….

          5. Nice PK! says:


    2. The key qualifiers that the paper uses are “in steady state” and “following repetitive oral dosing of drugs”. Then it’s simply a mathematical proposition: drug in equals drug out. But that’s a bit sly; it says nothing about how long it’ll take to achieve that steady state. (A day? A week? A month?)

  7. Jakob says:

    Interesting post. Always nice when researchers are honest.

    The problem seems to be quite like the old question of how to measure circulating testosterone? Should we measure “total testosterone” or “free testosterone” or “bioavailable testoterone”. And do these terms have anything to do with what actually goes on inside the body, or are they just somebodys vision of what goes on in blood plasma?

  8. not a real scientist says:

    They can’t measure the concentrations of drugs inside cells? What? Why don’t you label the compound with C13, or something like that, wash the cells, and then “spectro-analyze”?

Leave a Reply

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

Time limit is exhausted. Please reload CAPTCHA.