Pharmacokinetics – the study of how drugs are taken up, distributed, metabolized, and cleared – is obviously a key part of drug development. Every drug substance gets handled somewhat differently by the human body, and these differences can completely determine whether you’ve got an effective therapy or not. But the tools we have to get PK details are not always as fine as we’d like them to be.
The classic is, naturally, blood levels of the drug itself, and that’s the first key thing that you check. What’s the maximum concentration? How long does it take before that’s reached? How quickly does it come down, and how long is it before the drug drops back down below useful levels? Without this information, you’re flying blind, but there are a lot of detailed factors at work that add up give you these numbers. To get into the blood after an oral dose, a compound has to be absorbed through the walls of the tiny blood vessels in the walls of the intestine. And all of those drain (via the hepatic portal vein) right into the liver, which gets first crack at whatever the digestive system sends its way, and deals with them via a bewildering number of different mechanisms. Just circulating around in the blood isn’t a simple process, either – compounds can be bound to various plasma proteins or even sequester inside red blood cells themselves, while circulating enzymes get to work on the drug structures, too.
And we haven’t even reached the target cells yet! If your target is on the cell surface, then you can hope for your drug to just wash over it (as it diffuses back across the capillary wall and into the intercellular matrix), but that depends on what it’s bound to in the blood. If your target is inside the cell, then you have to make it across the cell membrane as well (which, to be sure, is behavior that you’re been presumably optimizing for during the development phase). OK, now you’re inside – but cells themselves are the very opposite of homogeneous, of course. Compounds can cross in and out of the various organelles and concentrate in different cellular compartments, and those may or may not be the ones that have your desired target in them.
Getting to this level of detail about a compound’s behavior is not easy. Blood levels are available, for sure, but actual real-time tracking down at the cellular (or intracellular) level? You can get tissue distribution with a radiolabeled compound or with careful mass spec work, but even with those techniques you have to take into account the vascularization of the tissues involved. Your compound, for example, probably isn’t accumulating in the spleen per se, but just showing up in the mass of blood vessels there. And those techniques tend to be static, since they involve (for example) freezing and sectioning an entire rodent carcass, or removing the organs of interest one by one for analysis. If you want more time points, you need more rodents.
So there’s a real need for technologies that can track compounds in living systems with higher spatial and temporal resolution, and many people have been working on them. Which brings me to this paper, a new one using Raman microscopy. The authors, from Bochum and Graz, are looking at the kinase inhibitor neratinib as a test case. The good thing about Raman is that a compound like this can (in theory) be followed without having to super-glue some sort of fluorescent label to it. Even when you label things are large as proteins, you’re always wondering what sort of perturbations you’ve induced, but labeling small molecules is even more problematic. The smallest fluorescent groups are still going to be a substantial chunk to add to your drug’s structure, so you’re now tracking a new species, which may or may not behave quite like the one you’re interested in.
The choice of neratinib was no accident. Its nitrile group has a Raman band in the 1800-2800 wavenumber region where cells have little or no Raman absorption. In the picture shown, panel A is a Raman band for CH deformation, which is a strong band in cells anyway. In that one, you can see various cellular structures. Panel B is the neratinib nitrile band (no real cellular background signal), and panel C is the overlay of the two. Panel D is a cluster analysis of the result of several different cell images and neratinib concentrations, and below each panel is a crossways image view, cut at the position of the white line in panel A.
That’s pretty good resolution, and especially for a label-free method. The paper also demonstrates far higher Raman sensitivity than has been achieved in past studies, and is in fact getting down to the real plasma concentration range for neratinib dosing (in the past, such work had to load in pretty substantial heaps of compound for it to be seen at all). There are some interesting effects when you look at other Raman bands as well: for example, there’s one in free neratinib that seems to be mostly due to the quinoline ring, but this one gets shifted to two new nearby bands as the compound accumulates in cells. The authors believe that this is due to a metabolic event, and the production of at least neratinib metabolites in cells can be confirmed by LC/MS – these are N-oxides, cleavage of the acrylamide, and hydroxylation of the quinoline ring, and these appear to be what are showing up as new Raman bands (DFT calculations of the expected bands agree).
The compound’s distribution can be monitored in even more detail by combining the Raman microscopy with fluorescent labeling of cell components. Specifically, neratinib treatment increases the number of lysosomes (an effect that many weakly basic drugs have), and the compound itself (and its metabolites) tend to accumulate in them. Labeling the drug’s targets (EGFR and HER2) shows another interesting effect: before treatment, these proteins are almost entirely on the cell surface, as they should be. But as neratinib hits them, its mechanism is a covalent bond formation through that acrylamide group. And what you see is labeled protein being internalized and in fact being taken into those lysosomes. Most of the lysosomal drug is free compound, but some of the signal is apparently the covalent protein adduct being pulled back in for degradation, since the cell’s systems can sense that something has gone haywire with their signaling.
This is great to see, a label-free method in live cells with real improvements in both spatial resolution and sensitivity. It’ll be particularly useful for compounds with functional groups that have a good Raman window, but with care, you could apply it to others as well. Such techniques can be used to prove target engagement and to follow pharmacokinetics in the kind of detail that we need in order to understand what’s really going on in living systems. And there’s a lot to understand!