About this Author
Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases.
To contact Derek email him directly: derekb.lowe@gmail.com
|
Category Archives
July 16, 2008
Posted by Derek
At various points in my drug discovery career, I’ve worked on G-protein-coupled receptor (GPCR) targets. Most everyone in the drug industry has at some point – a significant fraction of the known drugs work through them, even though we have a heck of a time knowing what their structures are like.
For those outside the field, GPCRs are a ubiquitous mode of signaling between the interior of a cell and what’s going on outside it, which accounts for the hundreds of different types of the things. They’re all large proteins that sit in the cell membrane, looped around so that some of their surfaces are on the outside and some poke through to the inside. The outside folds have a defined binding site for some particular ligand - a small molecule or protein – and the inside surfaces interact with a variety of other signaling proteins, first among them being the G-proteins of the name. When a receptor’s ligand binds from the outside, that sets off some sort of big shape change. The protein’s coils slide and shift around in response, which changes its exposed surfaces and binding patterns on the inside face. Suddenly different proteins are bound and released there, which sets off the various chemical signaling cascades inside the cell.
The reason we like GPCRs is that many of them have binding sites for small molecules, like the neurotransmitters. Dopamine, serotonin, acetylcholine – these are molecules that medicinal chemists can really get their hands around. The receptors that bind whole other proteins as external ligands are definitely a tougher bunch to work with, but we’ve still found many small molecules that will interact with some of them.
Naturally, there are at least two modes of signaling a GPCR can engage in: on and off. A ligand that comes in and sets off the intracellular signaling is called an agonist, and one that binds but doesn’t set off those signals is called an antagonist. Antagonist molecules will also gum up the works and block agonists from doing their things. We have an easier time making those, naturally, since there are dozens of ways to mess up a process compared to the ways there are of running it correctly!
Now, when I was first working in the GPCR field almost twenty years ago, it was reasonably straightforward. You had your agonists and you had your antagonists – well, OK, there were those irritating partial agonists, true. Those things set off the desired cellular signal, but never at the levels that a full agonist would, for some reason. And there were a lot of odd behaviors that no one quite knew how to explain, but we tried to not let those bother us.
These days, it’s become clear that GPCRs are not so simple. There appear to be some, for example, whose default setting is “on”, with no agonist needed. People are still arguing about how many receptors do this in the wild, but there seems little doubt that it does go on. These constituitively active receptors can be turned off, though, by the binding of some ligands, which are known as inverse agonists, and there are others, good old antagonists, that can block the action of the inverse agonists. Figuring out which receptors do this sort of thing - and which drugs - is a full time job for a lot of people.
It’s also been appreciated in recent years that GPCRs don’t just float around by themselves on the cell surface. Many of them interact with other nearby receptors, binding side-by-side with them, and their activities can vary depending on the environment they’re in. The search is on for compounds that will recognize receptor dimers over the good ol’ monomeric forms, and the search is also on for figuring out what those will do once we have them. To add to the fun, these various dimers can be with other receptors of their own kind (homodimers) or with totally different ones, some from different families entirely (heterodimers). This area of research is definitely heating up.
And recently, I came across a paper which looked at how a standard GPCR can respond differently to an agonist depending on where it's located in the membrane. We're starting to understand how heterogeneous the lipids in that membrane are, and that receptors can move from one domain to another depending on what's binding to them (either on their outside or inside faces). The techniques to study this kind of thing are not trivial, to put it mildly, and we're only just getting started on figuring out what's going on out there in the real world in real time. Doubtless many bizarre surprises await.
So, once again, the "nothing is simple" rule prevails. This kind of thing is why I can't completely succumb to the gloom that sometimes spreads over the industry. There's just so much that we don't know, and so much to work on, and so many people that need what we're trying to discover, that I can't believe that the whole enterprise is in as much trouble as (sometimes) it seems. . .
Comments (20)
+ TrackBacks (0) | Category: Biological News | Drug Assays
July 11, 2008
Posted by Derek
Here's an interesting idea: Merck, Lilly, and Pfizer are bankrolling a startup company to look for new technologies for drug development. Enlight Biosciences will focus on the biggest bottlenecks and risk points in the process, including new imaging techniques for preclinical and clinical evaluation of drug candidates, predictive toxicology and pharmacokinetics, clinical biomarkers, new models of disease, delivery methods for protein- and nucleic acid-based therapies, and so on.
It's safe to say that if any real advances are made in any of these, the venture will have to be classed as a success. These are hard problems, and it's not like there's been no financial incentive to solve any of them. (On the contrary - billions of dollars are out there waiting for anyone who can truly do a better job at these things). I wish these people a lot of luck, and I'm glad to see them doing what they're doing, but I do wish that there were more details available on how they plan to go about things. The opening press release leaves a lot of things unspoken, no doubt by design. (For instance, where are the labs going to be? What's the hoped-for balance of industry types to academics? How many people do they plan to have working on these things, and how will the companies involved plan to share the resulting technologies?)
Enlight is a creation of Puretech Ventures, a Boston VC firm that's been targeting early-stage ideas in these areas. Getting buy-in from the three companies above will definitely help, but their commitment isn't too clear at present. For now, it looks like they're getting to take a fresh look at some areas of great interest, without necessarily having to spend a lot of their own money. The press release says that Enlight will "direct up to $39 million" toward the areas listed on their web site, but those problems will eat thirty-nine million dollars without even reaching for the salt. Further funding is no doubt in the works, with the Merck/Pfizer/Lilly names as a guarantee of seriousness, and if any of these projects pan out, the money will arrive with alacrity.
Comments (11)
+ TrackBacks (0) | Category: Business and Markets | Drug Assays | Drug Development
June 3, 2008
Posted by Derek
We recently encountered a problem that’s (unfortunately) a rather common one. An enzyme assay turned up an interesting hit compound, with some characteristics that we were hoping to see for leads against our target. A re-test showed that yes, the activity appeared to be real, which was interesting, since this hit was a welcome surprise from a class of compounds that we weren’t expecting much from.
It was a comparatively old compound in the files, and all we could find out was that it had been purchased rather than made in house. Looking around, it seemed that there were very few literature references to things of this type, and only one commercial source: the Sigma-Aldrich Library of Rare chemicals, known as SALOR. That, though, was a potential warning flag.
Those compounds come from an effort started by Aldrich’s Alfred Bader many years ago, who started trolling around various academic labs looking for unusual compounds that no one wanted to keep around any more. Over time the company has accumulated a horde of oddities that are often found nowhere else, but there are several catches. For one, these things are usually available only in small quantities, tens of milligrams for the most part. That’s plenty for the screening files, but you’re not going to make a bunch of analogs starting from what comes out of a SALOR vial. Another catch is that the compounds are sold, very explicitly, as is: the university sources tell Aldrich what’s on the label, so that’s what they sell you and caveat emptor all the way, dude.
So often as not, you get what we got, a nice-looking white powder which, on closer analysis, turned out to only have a vague relationship to the structure on its label. We knew that we were in trouble as soon as the first NMR came out: way too much stuff in one region, nowhere near enough in some others. Mass spec confirmed that this thing weighed more than twice as much as what it was supposed to. We’ve since pretty much nailed down what the stuff really is, and our interest in it has decreased as each of the veils has been removed from the real structure.
We’re correcting the data in our own screening files, of course. And yes, we’re going to tell the folks at Aldrich to change their label, too, assuming they have any of this stuff left. At least the next person will know what they’re getting. For once. But there are more of these things waiting out there – in every large compound collection, in every catalog, in every collection of data are mistakes. Watch for them.
Comments (4)
+ TrackBacks (0) | Category: Drug Assays | Life in the Drug Labs
May 20, 2008
Posted by Derek
For those who were wondering, my copper reactions the other day worked out just fine. They started out a beautiful blue (copper iodide and an amino acid in straight DMSO – if that’s not blue it’s maybe going to be green, and if it’s not either one you’ve done something wrong). Of course, the color doesn’t stay. The copper ends up as part of a purple-brown sludge that has to be filtered out of the mix, which is the main downside of those Ullman reactions, no matter how people try to scrub them up for polite company.
And DMSO is the other downside, because you have to wash that stuff out with a lot of water. That’s one of the lab solvents that everyone has heard of, even if they slept through high school chemistry. But it’s not one that we use for reactions very much, because it’s something of a pain. It dissolves most everything, which is a good quality, but along with that one comes the ability to contaminate most everything. If your product is pretty greasy and nonpolar, you can partition the reaction between water and some more organic solvent (ether’s what I used this time), and wash it around a lot. But if your product is really polar, you could be in for a long afternoon.
That mighty solvation is something you need to look out for if you spill the stuff on yourself, of course. DMSO is famous for skin penetration (no, I have no idea if it does anything for arthritis). And while many of my compounds are not very physiologically active, I’d rather not dose myself with them to check those numbers. At the extreme end of the scale, a solution of cyanide in DMSO is potentially very dangerous stuff indeed. I’ve done cyanide reactions like that, many times, but always while paying attention to the task at hand.
Where DMSO really gets used is in the compound repository. That dissolves-everything property is handy when you have a few hundred thousand compounds to handle. The standard method for some years has been to keep compounds in the freezer in some defined concentration in DMSO – the solvent freezes easily, down around where water does (Not so! Actually, I've seen in freeze in a chilly lab a couple of times, now that I'm reminded of that in the comments to this post. Pure DMSO solidifies around 17 to 19 C, which is about 64 F C - a bit lower with those screening compounds dissolved in it, though).
But there are problems. For one thing, DMSO isn’t inert. That’s another reason it doesn’t get as much use as a lab solvent; there are many reaction conditions during which it wouldn’t be able to resist joining the party. You can oxidize things by leaving them in DMSO open to air, which isn’t what you want to do to the compound screening collection, so the folks there do as much handling under nitrogen as they can. Compounds sitting carelessly in DMSO tend to turn yellow, which is on the way to red, which is on the way to brown, and there are no pure brown wonder drugs.
Another difficulty is that love for water. Open DMSO containers will pull water in right out of the air, and a few careless freeze/thaw cycles with a screening plate will not only blow your carefully worked out concentrations, it may well also start crashing your compounds out of solution. The less polar ones will start decided that pure DMSO is one thing, but 50/50 DMSO/water is quite another. So not only do you want to work under nitrogen, if you can, but dry nitrogen, and you want to make sure that those plates are sealed up well while they’re in the freezer. (As an alternative, you can go ahead and put water in from the start, taking the consequences). All of these concerns begin to wear down the advantages of DMSO as a universal solvent, but not quite enough to keep people from using it.
And what about the compounds that don’t dissolve in the stuff? Well, it’s a pretty safe bet that a small molecule that can’t go into DMSO is going to have a mighty hard time becoming a drug, and it’s a very unattractive lead to start from, too. That’s the sort of molecule that would tend to just go right through the digestive tract without even noticing that there are things trying to get it into solution. And as for something given i.v., well, if you can’t get it to go into straight DMSO, what are the chances you’re going to get it into some kind of saline injection solution? Or the chances that it won’t crash out in the vein for an instant embolism? No, the zone of non-DMSO-soluble small organics is not a good place to hunt. We’ll leave proteins out of it, but if anyone knows of a small molecule drug that can’t go into DMSO, I’d like to hear about it. Taxol, maybe?
Comments (14)
+ TrackBacks (0) | Category: Drug Assays | Life in the Drug Labs
April 3, 2008
Posted by Derek
I was having a discussion the other day about which therapeutic areas have the best predictive assays. That is, what diseases can you be reasonably sure of treating before your drug candidate gets into (costly) human trials? As we went on, things settled out roughly like this:
Cardiovascular (circulatory): not so bad. We’ve got a reasonably good handle on the mechanisms of high blood pressure, and the assays for it are pretty predictive, compared to a lot of other fields. (Of course, that’s also now one of the most well-served therapeutic areas in all of medicine). There are some harder problems, like primary pulmonary hypertension, but you could still go into humans with a bit more confidence than usual if you had something that looked good in animals.
Cardiovascular (lipids): deceptive. There aren’t any animals that handle lipids quite the way that humans do, but we’ve learned a lot about how to interpolate animal results. That plus the various transgenic models gives you a reasonable read. The problem is, we don’t really understand human lipidology and its relation to disease as well as we should (or as well as a lot of people think we do), so there are larger long-term problems hanging over everything. But yeah, you can get a new drug with a new mechanism to market. Like Vytorin.
CNS: appalling. That goes for the whole lot – anxiety, depression, Alzheimer’s, schizophrenia, you name it. The animal models are largely voodoo, and the mechanisms for the underlying diseases are usually opaque. The peripheral nervous system isn’t much better, as anyone who’s worked in pain medication will tell you ruefully. And all this is particularly disturbing, because the clinical trials here are so awful that you’d really appreciate some good preclinical pharmacology: patient variability is extreme, the placebo effect can eat you alive, and both the diseases and their treatments tend to progress very, very slowly. Oh, it’s just a nonstop festival of fun over in this slot. Correspondingly, the opportunities are huge.
Anti-infectives: good, by comparison. It’s not like you can’t have clinical failures in this area, but for the most part, if you can stop viruses or kill bugs in a dish, you can do it in an animal, or in a person. The questions are always whether you can do it to the right extent, and just how long it’ll be before you start seeing resistance. With antibacterials that can be, say, "before the end of your clinical trials". There aren’t as many targets here as everyone would like, and none of them is going to be a gigantic blockbuster, but if you find one you can attack it with more confidence than usual.
Diabetes: pretty good, up to a point. There are a number of well-studied animal models here, and if your drug’s mechanism fits their quirks and limitations, then you should be in fairly good shape. Not by coincidence, this is also a pretty well-served area, by current standards. If you’re trying something off the beaten path, though, a route that STZ or db/db rats won’t pick up well, then things get harder. Look out, though, because this disease area starts to intersect with lipids, which (it bears saying again) We Don't Understand Too Well.
Obesity: deceptive in the extreme. There are an endless number of ways to get rats to lose weight. Hardly any of them, though, turn out to be relevant to humans or relevant to something humans would consider paying for. (Relentless vertigo would work to throw the animals off their feed, for example, but would probably be a loser in the marketplace. Although come to think of it, there is Alli, so you never know). And the problem here is always that there are so many overlapping backup redundant pathways for feeding behavior, so the chances for any one compound doing something dramatic are, well, slim. The expectations that a lot of people have for a weight-loss therapy are so high (thanks partly to years of heavily advertised herbal scams and bizarre devices), but the reality is so constrained.
Oncology: horrible, just horrible. No one trusts the main animal models in this area (rat xenografts of tumor lines) as anything more than rough, crude filters on the way to clinical trials. And no one should. Always remember: Iressa, the erstwhile AstraZeneca wonder drug from a few years back, continues to kick over all kinds of xenograft models. It looks great! It doesn’t work in humans! And it's not alone, either. So people take all kinds of stuff into the clinic against cancer, because what else can you do? That leads to a terrifying overall failure rate, and has also led to, if you can believe it, a real shortage of cancer patients for trials in many indications.
OK, those are some that I know about from personal experience. I’d be glad to hear from folks in other areas, like allergy/inflammation, about how their stuff rates. And there are a lot of smaller indications I haven’t mentioned, many of them under the broad heading of immunology (lupus, MS, etc.) whose disease models range from “difficult to run and/or interpret” on the high side all the way down to “furry little random number generators”.
Comments (8)
+ TrackBacks (0) | Category: Animal Testing | Cancer | Cardiovascular Disease | Diabetes and Obesity | Drug Assays | Drug Development | Infectious Diseases | The Central Nervous System
March 27, 2008
Posted by Derek
There’s an excellent paper in the most recent issue of Chemistry and Biology that illustrates some of what fragment-based drug discovery is all about. The authors (the van Aalten group at Dundee) are looking at a known inhibitor of the enzyme chitinase, a natural product called argifin. It’s an odd-looking thing – five amino acids bonded together into a ring, with one of them (an arginine) further functionalized with a urea into a sort of side-chain tail. It’s about a 27 nM inhibitor of the enzyme.
(For the non-chemists, that number is a binding affinity, a measure of what concentration of the compound is needed to shut down the enzyme. The lower, the better, other things being equal. Most drugs are down in the nanomolar range – below that are the ulta-potent picomolar and femtomolar ranges, where few compounds venture. And above that, once you get up to 1000 nanomolar, is micromolar, and then 1000 micromolar is one millimolar. By traditional med-chem standards, single-digit nanomolar = good, double-digit nanomolar = not bad, triple-digit nanomolar or low micromolar = starting point to make something better, high micromolar = ignore, and millimolar = can do better with stuff off the bottom of your shoe.
What the authors did was break this argifin beast up, piece by piece, measuring what that did to the chitinase affinity. And each time they were able to get an X-ray structure of the truncated versions, which turned out to be a key part of the story. Taking one amino acid out of the ring (and thus breaking it open) lowered the binding by about 200-fold – but you wouldn’t have guessed that from the X-ray structure. It looks to be fitting into the enzyme in almost exactly the same way as the parent.
And that brings up a good point about X-ray crystal structures. You can’t really tell how well something binds by looking at one. For one thing, it can be hard to see how favorable the various visible interactions might actually be. And for another, you don’t get any information at all about what the compound had to pay, energetically, to get there.
In the broken argifin case, a lot of the affinity loss can probably be put down to entropy: the molecule now has a lot more freedom of movement, which has to be overcome in order to bind in the right spot. The cyclic natural product, on the other hand, was already pretty much there. This fits in with the classic med-chem trick of tying back side chains and cyclizing structures. Often you’ll kill activity completely by doing that (because you narrowed down on the wrong shape for the final molecule), but when you hit, you hit big.
The structure was chopped down further. Losing another amino acid only hurt the activity a bit more, and losing still another one gave a dipeptide that was still only about three times less potent than the first cut-down compound. Slicing that down to a monopeptide, basically just a well-decorated arginine, sent the activity down another sixfold or so – but by now we’re up to about 80 micromolar, which most medicinal chemists would regard as the amount of activity you could get by testing the lint in your pocket.
But they went further, making just the little dimethylguanylurea that’s hanging off the far end. That thing is around 500 micromolar, a level of potency that would normally get you laughed at. But wait. . .they have the X-ray structures all along the way, and what becomes clear is that this guanylurea piece is binding to the same site on the protein, in the same manner, all the way down. So if you’re wondering if you can get an X-ray structure of some 500 micromolar dust bunny, the answer is that you sure can, if it has a defined binding site.
And the value of these various derivatives almost completely inverts if you look at them from a binding efficiency standpoint. (One common way to measure that is to take the minus log of the binding constant and divide by the molecular weight in kilodaltons). That’s a “bang for the buck” index, a test of how much affinity you’re getting for the weight of your molecule. As it turns out, argifin – 27 nanomolar though it be – isn’t that efficient a binder, because it weighs a hefty 676. The binding efficiency index comes out to just under 12, which is nothing to get revved up about. The truncated analogs, for the most part, aren’t much better, ranging from 9 to 15.
But that guanylurea piece is another story. It doesn’t bind very tightly, but it bats way above its scrawny size, with a BEI of nearly 28. That’s much more impressive. If the whole argifin molecule bound that efficiently, it would be down in the ten-to-the-minus nineteenth range, and I don’t even know the name of that order of magnitude. If you wanted to make a more reasonably sized molecule, and you should, a compound of MW 400 would be about ten femtomolar with a binding efficiency like that. There’s plenty of room to do better than argifin.
So the thing to do, clearly, is to start from the guanylurea and build out, checking the binding efficiency along the way to make sure that you’re getting the most out of your additions. And that is exactly the point of fragment-based drug discovery. You can do it this way, cutting down a larger molecule to find what parts of it are worth the most, or you can screen to find small fragments which, though not very potent in the absolute sense, bind very efficiently. Either way, you take that small, efficient piece as your anchor and work from there. And either way, some sort of structural read on your compounds (X-ray or NMR) is very useful. That’ll give you confidence that your important binding piece really is acting the same way as you go forward, and give you some clues about where to build out in the next round of analogs.
This particular story may be about as good an illustration as one could possibly find - here's hoping that there are more that can work out this way. Congratulations to van Aalten and his co-workers at Dundee and Bath for one of the best papers I've read in quite a while.
Comments (12)
+ TrackBacks (0) | Category: Analytical Chemistry | Drug Assays | In Silico
February 14, 2008
Posted by Derek
I’ve been reading an interesting paper from JACS with the catchy title of “Optimization of Activity-Based Probes for Proteomic Profiling of Histone Deacetylase Complexes”. This is work from Benjamin Cravatt's lab at Scripps, and it says something about me, I suppose, that I found that title of such interest that I immediately printed off a copy to study more closely. Now I’ll see if I can interest anyone who wasn’t already intruiged! First off, some discussion of protein tagging, so if you’re into that stuff already, you may want to skip ahead.
So, let’s say you have a molecule that has some interesting biological effect, but you’re not sure how it works. You have suspicions that it’s binding to some protein and altering its effects (always a good guess), but which protein? Protein folks love fluorescent assays, so if you could hang some fluorescent molecule off one end of yours, perhaps you could start the hunt: expose your cells to the tagged molecule, break them open, look for the proteins that glow. There are complications, though. You’d have to staple the fluorescent part on in a way that didn’t totally mess up that biological activity you care about, which isn’t always easy (or even possible). The fact that most of the good fluorescent tags are rather large and ugly doesn’t help. But there’s more trouble: even if you manage to do that, what’s to keep your molecule from drifting right back off of the protein while you’re cleaning things up for a look at the system? Odds are it will, unless it has a really amazing binding constant, and that’s not the way to bet.
One way around that problem is sticking yet another appendage on to the molecule, a so-called photoaffinity label. These groups turn into highly reactive species on exposure to particular wavelengths of light, ready to form a bond with the first thing they see. If your molecule is carrying one when it’s bound to your mystery protein, shining light on the system will likely cause a permanent bond to form between the two. Then you can do all your purifications and separations, and look at your leisure for which proteins fluoresce.
This is “activity-based protein profiling”, and it’s a hot field. There are a lot of different photoaffinity labels, and a lot of ways to attach them, and likewise with the fluorescent groups. The big problem, as mentioned above, is that it’s very hard to get both of those on your molecule of interest and still keep its biological activity – that’s an awful lot of tinsel to carry around. One slick solution is to use a small placeholder for the big fluorescent part. This, ideally, would be some little group that will hide out innocently during the whole protein-binding and photoaffinity-labeling steps, then react with a suitably decorated fluorescent partner once everything’s in place. This assembles your glowing tag after the fact.
A favorite way to do that step is through an azide-acetylene cycloaddition reaction, the favorite of Barry Sharpless’s “click” reactions. Acetylenes are small and relatively unreactive, and at the end of the process, after you’ve lysed the cells and released all their proteins, you can flood your system with azide-substituted fluorescent reagent. The two groups react irreversibly under mild catalytic conditions to make a triazole ring linker, which is a nearly ideal solution that’s getting a lot of use these days (more on this another day).
So, now to this paper. What this group did was label a known compound (from Ron Breslow's group at Columbia) that targets histone deacetylase (HDAC) enzymes, SAHA, now on the market as Vorinostat. There are a lot of different subtypes of HDAC, and they do a lot of important but obscure things that haven’t been worked out yet. It’s a good field to discover protein function in.
When they modified SAHA in just the way described above, with an acetylene and a photoaffinity group, it maintained its activity on the known enzymes, so things looked good. They then exposed it to cell lysate, the whole protein soup, and found that while it did label HDAC enzymes, it seemed to label a lot of other things in the background. That kind of nonspecific activity can kill an assay, but they tried the label out on living cells anyway, just to see what would happen.
Very much to their surprise, that experiment led to much cleaner and more specific labeling of HDACs. The living system was much nicer than the surrogate, which (believe me) is not how things generally go. Some HDACs were labeled much more than others, though, and my first thought on reading that was “Well, yeah, sure, your molecule is a more potent binder to some of them”.
But that wasn’t the case, either. When they profiled their probe molecule’s activity versus a panel of HDAC enzymes, they did indeed find different levels of binding – but those didn’t match up with which ones were labeled more in the cells. (One explanation might be that the photoaffinity label found some of the proteins easier to react with than others, perhaps due to what was nearby in each case when the reactive species formed).
Their next step was to make a series of modified SAHA scaffolds and rig them up with the whole probe apparatus. Exposing these to cell lysate showed that many of them performed fine, labeling HDAC subtypes as they should, and with different selectivities than the original. But when they put these into cells, none of them worked as well as the plain SAHA probe – again, rather to their surprise. (A lot of work went into making and profiling those variations, so I suspect that this wasn’t exactly the result the team had hoped for - my sympathies to Cravatt and especially to his co-author Cleo Salisbury). The paper sums the situation up dryly: "These results demonstrate that in vitro labeling is not necessarily predictive of in situ labeling for activity-based protein profiling probes".
And that matches up perfectly with my own prejudices, so it must be right. I've come to think, over the years, that the way to go is to run your ideas against the most complex system you think that they can stand up to - in fact, maybe one step beyond that, because you may have underestimated them. A strict reductionist might have stopped after the cell lysate experiments in this case - clearly, this probe was too nonspecific, no need to waste time on the real system, eh? But the real system, the living cell, is real in complex ways that we don't understand well at all, and that makes this inference invalid.
The same goes for medicinal chemistry and drug development. If you say "in vitro", I say "whole cells". If you've got it working in cells, I'll call for mice. Then I'll see your mice and raise you some dogs. Get your compounds as close to reality as you can before you pass judgment on them.
Comments (5)
+ TrackBacks (0) | Category: Biological News | Drug Assays | Drug Development
January 29, 2008
Posted by Derek
I've had some questions about animal models and testing, so I thought I'd go over the general picture. As far as I can tell, my experience has been pretty representative.
There are plenty of animal models used in my line of work, but some of them you see more than others. Mice and rats are, of course, the front line. I’ve always been glad to have a reliable mouse model, personally, because that means the smallest amount of compound is used to get an in vivo readout. Rats burn up more hard-won material. That's not just because they're uglier, since we don’t dose based on per cent ugly, but rather because they're much larger and heavier. The worst were some elderly rodents I came across years ago that were being groomed for a possible Alzheimer’s assay – you don’t see many old rats in the normal course of things, but I can tell you that they do not age gracefully. They were big, they were mean, and they were, well, as ratty as an animal can get. (They were useless for Alzheimer's, too, which must have been their final revenge).
You can’t get away from the rats, though, because they’re the usual species for toxicity testing. So if your pharmacokinetics are bad in the rat, you’re looking at trouble later on – the whole point of tox screens is to run the compound at much higher than usual blood levels, which in the worst cases you may not be able to reach. Every toxicologist I’ve known has groaned, though, when asked if there isn’t some other species that can be used – just this time! – for tox evaluation. They’d much rather not do that, since they have such a baseline of data for the rat, and I can’t blame them. Toxicology is an inexact enough science already.
It’s been a while since I’ve personally seen the rodents at all, though, not that I miss them. The trend over the years has been for animal facilities to become more and more separated from the other parts of a research site – separate electronic access, etc. That’s partly for security, because of people like this, and partly because the fewer disturbances among the critters, the better the data. One bozo flipping on the wrong set of lights at the wrong time can ruin a huge amount of effort. The people authorized to work in the animal labs have enough on their hands keeping order – I recall a run of assay data that had an asterisk put next to it when it was realized that a male mouse had somehow been introduced into an all-female area. This proved disruptive, as you’d imagine, although he seemed to weather it OK.
Beyond the mouse and rat, things branch out. That’s often where the mechanistic models stop, though – there aren’t as many disease models in the larger animals, although I know that some cardiovascular disease studies are (or have been) run in pigs, the smallest pigs that could be found. And I was once in on an osteoporosis compound that went into macaque monkeys for efficacy. More commonly, the larger animals are used for pharmacokinetics: blood levels, distribution, half-life, etc. The next step for most compounds after the rat is blood levels in dogs – that’s if there’s a next step at all, because the huge majority of compounds don’t get anywhere near a dog.
That’s a big step in terms of the seriousness of the model, because we don’t use dogs lightly. If you’re getting dog PK, you have a compound that you’re seriously considering could be a drug. Similarly, when a compound is finally picked to go on toward human trials, it first goes through a more thorough rat tox screen (several weeks), then goes into two-week dog tox, which is probably the most severe test most drug candidates face. The old (and cold-hearted) saying is that “drugs kill dogs and dogs kill drugs”. I’ve only rarely seen the former happen (twice, I think, in 19 years), but I’ve seen the second half of that saying come true over and over. Dogs are quite sensitive – their cardiovascular systems, especially – and if you have trouble there, you’re very likely done. There’s always monkey data – but monkey blood levels are precious, and a monkey tox screen is extremely rare these days. I’ve never seen one, at any rate. And if you have trouble in the dog, how do you justify going into monkeys at all? No, if you get through dog tox, you're probably going into man, and if you don't, you almost certainly aren't.
Comments (6)
+ TrackBacks (0) | Category: Animal Testing | Drug Assays | Drug Development | Pharmacokinetics | Toxicology
January 22, 2008
Posted by Derek
There’s been a big trend the last few years in the industry to try to build our molecules up from much smaller pieces than usual. “Fragment-based” drug discovery is the subject of many conferences and review articles these days, and I’d guess that most decent-sized companies have some sort of fragment effort going on. (Recent reviews on the topic, for those who want them).
Many different approaches come under that heading, though. Generally, the theme is to screen a collection of small molecules, half the size or less of what you’d consider a reasonable molecular weight for a final compound, and look for something that binds. At those sizes, you’re not going to find the high affinities that you usually look for, though. We usually want our clinical candidates to be down in the single-digit nanomolar range for binding constants, and our screening hits to be as far under one micromolar as we can get. In the fragment world, though, from what I can see, people regard micromolar compounds as pretty hot stuff, and are just glad not to be up in the millimolar range. (For people outside the field, it’s worth noting that a nanomolar compound binds about a million times better than a millimolar one).
Not all the traditional methods of screening molecules will pick up weak binders like that. (Some assays are actually designed not to read out at those levels, but to only tell you about the really hot compounds). For the others, you’d think you could just run things like you usually do, just by loading up on the test compounds, but that’s problematic. For one thing, you’ll start to chew up a lot of compound supplies at that rate. Another problem is that not everything stays in solution for the assay when you try to run things at that concentration. And if you try to compensate by using more DMSO or whatever to dissolve your compounds, you can kill your protein targets with the stuff when it goes in. Proteins are happy in water (well, not pure distilled water, but water with lots of buffer and salts and junk like the inside of a cell has). They can take some DMSO, but it’ll eventually make even the sturdiest of them unhappy at some point. (More literature on fragment screening).
And once you’ve got your weak-binding low-molecular weight stuff, what then? First, you have to overcome the feeling, natural among experienced chemists, that you’re working on stuff you should be throwing away. Traditional medicinal chemistry – analog this part, add to that part, keep plugging away – may not be the appropriate thing to do for these leads. There are just too many possibilities – you could easily spend years wandering around. So many companies depend on structural information about the protein target and the fragments themselves to tell them where these little guys are binding and where the best places to build from might be. That can come from NMR studies or X-ray crystal determinations, most commonly.
Another hope, for some time now, has been that if you could discover two fragments that bound to different sites, but not that far from each other, that you could then stitch them together to make a far better compound. (See here for more on this idea). That’s been very hard to realize in practice, though. Finding suitable pairs of compounds is not easy, for starters. And getting them linked, as far as I can see, can be a real nightmare. A lot of the linking groups you can try will alter the binding of the fragments themselves – so instead of going from two weak compounds to one strong one, you go from two weak ones to something that’s worse than ever. Rather than linking two things up, a lot of fragment work seems to involve building out from a single piece.
But that brings up another problem, exemplified by this paper. These folks took a known beta-lactamase inhibitor, a fine nanomolar compound, and broke it up into plausible-looking fragments, to see if it could have been discovered that way. But what they found, each time they checked the individual pieces, was that each of them bound in a completely different way than it did when it was part of the finished molecule. The binding mode was emergent, not additive, and it seems clear that most (all?) of the current fragment approaches would have been unable to arrive at the final structure. The authors admit that this may be a special case, but there’s no reason to assume that it’s all that special.
So fragment approaches, although they seem to be working out in some cases, are probably always going to miss things. But hey, we miss plenty of things with the traditional methods, too. Overall, I’m for trying out all kinds of odd things, because we need all the help we can get. Good luck to the fragment folks.
Comments (6)
+ TrackBacks (0) | Category: Analytical Chemistry | Drug Assays
December 11, 2007
Posted by Derek
Man, do we ever have a lot of assays in this business. Almost every drug development project has a long list of them, arranged in what we call a screening cascade. You check to make sure that your new molecule hits your protein target, then you try it on one or more living cell lines. There are assays to check its potency against related targets (some of which you may want, most of which you don’t), and assays to measure the properties of the compound itself, like how well it dissolves. Then it’s on to blood levels in animals, and finally to a disease model in some species or another.
Not all these assays are of equal importance, naturally. And not all of them do what they’re supposed to do for you. Some processes are so poorly understood that we’re willing to try all sorts of stuff to get a read on them. I would put the Caco-2 assay firmly in that category.
Caco ("cake-o")-2 cells are a human colon cancer cell line. When you grow them in a monolayer, they still remember to form an “inside” and an “outside” – the two sides of the layer act differently, and they pump compounds across from one side to the other. This sort of active transport is very widespread in living systems, and it’s very important in drug absorption and distribution, and from a practical standpoint we don’t know much about it at all. Membranes like the gut wall or the lining of the brain’s blood vessels do this sort of thing all the time, and pump out things they don’t like. Cancer cells and bacteria do it to compounds they judge to be noxious, which covers a lot of the things we try to use to kill them. Knowing how to avoid this kind of thing would be worth billions of dollars, and would give us a lot more effective drugs.
The Caco-2 cell assay is an attempt to model some of this process in a dish, so you don’t have to find out about it in a mouse (or a human). You put a test amount of your compound on one side of the layer of cells, and see how much of it gets through to the other side – then you try it in reverse, to see how much of that flow was active transport and how much was just passive leak-through diffusion. The ratio between those two amounts is supposed to give you a read on how much of a substrate your compound is for these efflux pumps, particularly a widespread one called P-glycoprotein.
I have seen examples in the literature where this assay appears to have given useful data. Unfortunately, as far as I can remember, I cannot recall ever having participated in such a project. Every time I’ve worked with Caco-2 data, it’s been a spread of numbers that didn’t correlate well with gut absorption, didn’t correlate well with brain levels, and didn’t help to prioritize anything. That may be unfair – after all, I’ve had people tell me that ‘s worked out for them – but I think that even in those cases people had to run quite a few compounds through before they believed that the assay was really telling them something. The published data on these things can turn out to be a small, shiny heap on the summit of a vast pile of compost - the unimpressive or uninterpretable attempts that never show up in any journal, anywhere.
You can think of several reasons for these difficulties, and there are surely more that none of us have thought of yet. These are colon cells, not cells from the small intestine (where the great majority of absorption takes place) or from the blood-brain barrier. They're from a carcinoma line, not a normal population (which is why they're still happily living in dishes). But that means that they’re far removed from their origins, to boot. (It’s well known that many cell lines lose some of their characteristics and abilities as you culture them. They’re not getting the stimuli they were in their native environment, and they shed functions and pathways as they’re no longer being called for). There’s also the problem that they’re human cells, but they’re often used to correlate with data from rodent models. Our major features overlap pretty well (most mouse poisons are human poisons, for example), but the fine details can be difficult to line up.
But people still run the Caco-2 assay. I think that now it’s mostly done in the hope, mostly forlorn, that this time it’ll turn out to model something crucial to this particular drug series. A representative list of compounds that have already been through the pharmacokinetic studies is tried, and the results are graphed against the blood levels. And, for the most part, the plots look like soup thrown against a wall – again. The quest to explain these things continues. . .
Comments (21)
+ TrackBacks (0) | Category: Drug Assays | Drug Development
October 11, 2007
Posted by Derek
Now we open the sedate, learned pages of Nature Methods, a fine journal that specializes in new techniques in molecular and chemical biology. In the August issue, the correspondence section features. . .well, a testy response to a paper that appeared last year in Nature Methods.
“Experimental challenge to a ‘rigorous’ BRET analysis of GPCR oligimerization” is the title. If you don’t know the acronyms, never mind – journals like this have acronyms like leopards have spots. The people doing the complaining, Ali Salahpour and Bernard Masri of Duke, are taking issue with a paper from Oxford by John James, Simon Davis, and co-workers. The original paper described a bioluminescence energy transfer (BRET) method to see if G-protein coupled receptors (GPCRs) were associating with each other on cell surfaces. (GPCRs are hugely important signaling systems and drug targets – think serotonin, dopamine, opiates, adrenaline – and it’s become clear in recent years that they can possibly hook up in various unsuspected combinations on the surfaces of cells in vivo).
Salahpour and Masri take strong exception to the Oxford paper’s self-characterization:
“Although the development of new approaches for BRET analysis is commendable, part of the authors’ methodological approach falls short of being ‘rigorous’. . .Some of the pitfalls of their type-1 and type-2 experiments have already been discussed elsewhere (footnote to another complaint about the same work, which also appeared earlier this year in the same journal - DBL). Here we focus on the type-2 experiments and report experimental data to refute some of the results and conclusions presented by James et al.”
That’s about an 8 out of 10 on the scale of nasty scientific language, translating as “You mean well but are lamentably incompetent.” The only way to ratchet things up further is to accuse someone of bad faith or fraud. I won’t go into the technical details of Salahpour and Masri’s complaints; they have to do with the mechanism of BRET, the effect on it of how much GPCR protein is expressed in the cells being studied, and the way James et al. interpreted their results versus standards. The language of these complaints, though, is openly exasperated, full of wording like “unfortunately”, “It seems unlikely”, “we can assume, at best” “(does) not permit rigorous conclusions to be drawn”, “might be erroneous”, “inappropriate and a misinterpretation”, “This could explain why”, “careful examination also (raises) some concerns”, and so on. After the bandilleros and picadors have done their work in the preceding paragraphs, the communication finishes up with another flash of the sword:
In summary, we agree with James and colleagues that type-2 experiments are useful and informative. . .Unfortunately, the experimental design proposed in James et al. to perform type-2 experiments seems incorrect and cannot be interpreted. . .”
James and Davis don’t take this with a smile, naturally. The journal gave them a space to reply to the criticisms, as is standard practice, and as they did for the earlier criticism. (At least the editors know that people are reading the papers they accept. . .) They take on many of the Salahpour/Masri points, claiming that their refutations were done under completely inappropriate conditions, among other things. And they finish up with a flourish, too:
"As we have emphasized, we were not the first to attempt quantitative analysis of BRET data. Previously, however, resonance energy transfer theory was misinterpreted (for example, ref. 4) or applied incorrectly (for example, ref. 5). (Note - reference 4 is to a paper by the first people to question their paper earlier this year, and reference 5 is to the work of Salahpour himself, a nice touch - DBL). The only truly novel aspect of our experiments is that we verified our particular implementation of the theory by analyzing a set of very well-characterized. . .control proteins. (Note - "as opposed to you people" - DBL). . . .In this context, the technical concerns of Salahpour and Masri do not seem relevant."
It's probably safe to say that the air has not yet been cleared. I'm not enough of a BRET hand to say who's right here, but it looks like we're all going to have some more chances to make up our minds (and to appreciate the invective along the way).
Comments (21)
+ TrackBacks (0) | Category: Biological News | Drug Assays | The Scientific Literature
August 20, 2007
Posted by Derek
As I've mentioned before, advances in molecular biology have continued to make all sorts of brute-force approachs possible - things that would have been laughed at (or, more likely, not even proposed at all) a few years ago.
Another recent example of this is a paper earlier this year in Nature from the group of Michael White at UT-Southwestern. The authors selected a lung cancer cell line that's know to be very sensitive to Taxol (paclitaxel), and looked for possible targets that might increase the drug's effectiveness. (It's a good compound to pick for a study like this, since it's simultaneously quite effective and quite toxic).
So, how do you go fishing for such combinations? These days, you set up 21,127 experimental wells, each one contained some cells and some silencing RNA molecules targeting, one at a time, 21,127 different human genes. And you look to see if knocking down expression of any of those genes increased the potency of a normally ineffective dose of the drug. (There were four different siRNAs per gene, actually, and each one was run in triplicate with and without Taxol, leading to a Whole Lotta 96-well plates. I'm glad I'm not paying for all the pipet tips, I can tell you that for sure.)
As you'd imagine, working up the data from this kind of thing takes as long, or longer, than setting one up. After comparing everything to the control wells and to each other several different ways, they ended up with 87 candidate genes whose knockdown seems to make the drug more effective. Gratifyingly, many of these make one kind of sense or another - there are several genes, for example, that are known to be involved in spindle formation, which is the target of paclitaxel itself.
Even more interestingly, not all the hits were obvioius. Another group of genes code for parts of the proteasome. That part of the cell is targeted by Millennium's Velcade (bortezomib), and it's recently been reported that the combination of Velcade and paclitaxel is more effective than expected. And there's another combination that seemingly hasn't been tried at all: the experiment suggests that inhibitors of vacuolar ATP-ase should synergize with Taxol, and (as it happens) a compound called salicylihalamide A has been looked at for just that target. They tried this experimental combination out on the cells, and it seems to work well - so, in humans?
As a commentary in the New England Journal of Medicine on this work dryly put it, "This hypothesis should be tested." And so it should. I've always had doubts about how far one can extrapolate cell data in cancer studies, but this kind of thing will tell us for sure. If something hits from this work, more such studies will come pouring out - they're getting easier to do all the time, you know. . .
Comments (8)
+ TrackBacks (0) | Category: Cancer | Drug Assays
July 10, 2007
Posted by Derek
There's a problem in the drug industry that people have recognized for some years, but we're not that much closer to dealing with it than we were then. We keep coming up with these technologies and techniques which seem as if they might be able to help us with some of our nastiest problems - I'm talking about genomics in all its guises, and metabolic profiling, and naturally the various high-throughput screening platforms, and others. But whether these are helping or not (and opinions sure do vary), one thing that they all have in common is that they generate enormous heaps of data.
We're not the only field to wish that the speed of collating and understanding all these results would start to catch up with the speed with which they're being generated. But some days I feel as if the two curves don't even have the same exponent in their equations. High-throughput screening data are fairly manageable, as these things go, and it's a good thing. When you can rip through a million compounds screening a new target, generating multiple-point binding curves along the way, you have a good-sized brick of numbers. But you're looking for just the ones with tight binding and reasonable curves, which is a relatively simple operation, and by the time you're done there may only be a couple of dozen compounds worth looking at. (More often than you'd think, there may be none at all).
But genomics/metabolomics/buzzwordomics platforms are tougher. In these cases, we don't actually know what we're looking for much of the time. I mean, we don't understand what the huge majority of the genes on a gene-chip assay really do, not in any useful detail, anyway. So the results of a given assay aren't the horserace leader board of a binding assay; they're more like a huge, complicated fingerprint or an abstract painting. We can say that yes, this compound seems to be different from that one, which is certainly different from this one over here but maybe similar to these on the left - but sometimes that's about all we can say.
Of course, the story isn't supposed to stop there, and everyone's hoping it won't. The idea is that we'll learn to interpret these things as we see more and more compounds and their ultimate effects. Correlations, trends, and useful conclusions are out there (surely?) and if we persevere we'll uncover them. The problem is, finding these things looks like requiring the generation of still more endless terabytes of data. It takes nerve to go on, but we seem to have no other choice.
Comments (22)
+ TrackBacks (0) | Category: Drug Assays
April 26, 2007
Posted by Derek
When I wrote about lousy animal models of disease a few days ago, there was a general principle at the back of my mind. (There generally is - my wife, over the years, has become accustomed to the sudden dolly-back panorama shots that appear unannounced in my conversation). It was: that a bad model system is much, much worse than no model system at all.
I've been convinced of that for a long time. When you have no model for what you're doing, you're forced to realize that you have no clear idea of what's going on. That's uncomfortable, to be sure, but you at least realize the situation. But when you have a poor model, the temptation to believe in it, at least partially, is hard to resist. Even if it's giving you the right answers at a rate worse than chance, you can still take (irrational) comfort in knowing that at least you're not flying blind - even as you do worse than the people who are.
There are many reasons to hold on to an underperforming model. Sometimes pride is the problem. I've seen groups that stuck with assays just because they'd invented them, even though the method was slowly wasting everyone's time. Never underestimate cluelessness, either. People will use worthless techniques for quite a while if they're not in the habit of checking to see if they're any good. But the biggest reason that useless procedures hang around, I'm convinced, is fear.
Fear, that is, of being left out in the middle of the field with no models, no insights, and no path forward at all. It's a bad feeling, rather scary, and rather difficult to explain to upper management if you're a project leader. Better, then, to hold on to the assays and models you have, to defend them even if you're not sure you trust them. With any luck, the project will end (although probably not happily) before the facts have to be faced. As Belloc advised children in other situations: "Always keep ahold of Nurse / For fear of finding something worse."
Comments (23)
+ TrackBacks (0) | Category: Animal Testing | Drug Assays | Drug Development
March 12, 2007
Posted by Derek
I wanted to link tonight to the "Milkshake Manifesto" over at OrgPrep Daily. It's a set of rules for med-chem, and looking them over, I agree with them pretty much across the board. There's a general theme in them of getting as close to the real system as you can, which is a theme I've sounded many times.
That applies to things like "Rule of Five" approximations and docking scores - useful, perhaps if you're sorting through a huge pile of compounds that you have to prioritize, not so useful if you've already got animal data.
He also takes a shot at Caco-2 cells and other such approximations to figure out membrane and tissue penetration. I've never yet seen an in vitro assay for permeability that I would trust - it's just too complicated, and it may never yield to a reductionist approach.
I'm a big fan of reductionism, don't get me wrong, but it's not the tool for every job. Living systems are especially tricky to pare down, and you can simplify yourself right out of any useful data if you're not very careful. The closer to the real world, the better off you are. It isn't easy, and it isn't cheap, but nothing good ever came easy or cheap, did it?
Comments (6)
+ TrackBacks (0) | Category: Drug Assays | Drug Development | In Silico
February 5, 2007
Posted by Derek
Here's an interesting press release on a potential new class of anticancer drugs. It has a nice hook ("Lab mistake leads to cancer finding!"), and the work itself isn't bad at all. It's an neat biochemical result, which might eventually lead to something. You have to know a bit about drug discovery and development to spot the problem, though - and not that many people do, which provides the ecological niche for this whole blog, frankly.
The discovery (from the University of Rochester) has to do with PPAR-gamma compounds, an area of research I've spent some time in. I didn't spend enough time there to understand it, mind you - no one has spent enough time to do that yet, no matter how long they've been at it. I wrote about some of the complexities here in 2004, and things have not become any more intelligible since then. The PPARs are nuclear receptors, affecting gene transcription when small molecules bind to them. There are, however, zillions of different binding modes in these things and they affect a list of genes that stretches right out the door. Some get upregulated, some down, and these vary according to patterns that we're only beginning to understand.
The Rochester group found that a particular class of compounds, the PPAR-gamma antagonists, had an unexpected toxic effect on some tumor cell lines. Their tubulin system was disrupted - that's a structural protein which is very important during cell division, and is the target for other known oncology drugs (like Taxol). The PPAR ligands seem to be messing with tubulin through a different route than anyone's seen before, though, and that definitely makes it worth following up on.
But the tone of the press release is too optimistic. (I should turn that line into some sort of macro, since I could use it twenty times a day). It mentions "high-dose" PPAR antagonist therapy as a possible cancer treatment, but take a look at the concentrations used: 10 to 100 micromolar. Even for cells in a dish, that's really hammering things down. And there's hardly any chance that you could attain these levels in a real-world situation, dosing a whole animal (or human). As blood levels go, those are huge.
But how about using more potent compounds? Of the three that are mentioned in the paper, BADGE is pretty dead, but the other two are actually quite potent. Tellingly, nothing happened at all with any of them up to 1 micromolar. These things will mess with other PPAR-gamma driven processes at much lower concentrations, so you have to wonder what's really going on here. And keep in mind that other PPAR compounds whose mode of action is roughly the opposite of these have been suggested as potential anticancer agents, too - this sort of thing happens all the time with nuclear receptors, and reflects their head-grabbing complexity.
This is still worth figuring out; don't get me wrong. There might be a new mechanism here that could lead to something, eventually, although it looks to be a tough problem. But that's the part of this work that's interesting - the level of activity seen here isn't. If I had a dollar for every compound that affects tumor cells at 50 micromolar, I wouldn't need to be sending my CV out these days.
Comments (5)
+ TrackBacks (0) | Category: Cancer | Drug Assays
January 10, 2007
Posted by Derek
After yesterday's post, several people brought up the issue of inverted screening cascades. What happens when your compound works better in the mice than it did in the cells? Worse, what if it would have worked in the mice, but you never put it in there because it was so weak in the cell assays?
These kinds of questions are worth worrying about, because we almost never get a chance to answer them. For obvious reasons, the vast majority of compounds that make it into animal models are ones that looked good in the earlier steps. You'd have to think that the hit rate in vivo would be much lower for the dud compounds, but you'd have to be pretty arrogant to think that it would be zero, too.
As I recall (and I was just down the hall when it happened), the discovery of Schering-Plough's cholesterol absorption inhibitor came out of a compound that made it into an animal model and worked well, even though it turned out later to be a loser at the project's original target. (I believe that the in vitro assay was down that week for some reason, but one of my former colleagues will probably set me straight if I'm wrong about that). This sort of thing is food for thought, all right, extreme example though it might be. Even if your compounds don't suddenly hit a new target, there's still room for plenty of surprises in pharmacokinetics and the like.
But it would be unethical just to shove everything into animals, tempting though it is sometimes. And it would cost an insane amount, too - let's not forget that. But I do advocate getting as close to the real disease as quickly as possible. You can really waste time and effort by over-optimizing in vitro, all the time convincing yourself that you're doing the right thing.
Then there's the ultimate question in this line: how many compounds are there that don't work well in the animal models, but would be good in humans? I've wondered about this for many years, and I'm going to go on wondering, because data points are mighty scarce. Human biomarkers might eventually lead to some companies crossing their fingers and going into man with a compound that they expect to outdo the animal models. But it's going to take a lot of nerve. (And here's another complication - those upside surprises that might show up in the animals? How many of those are going to translate to humans, do you think? Not all of them, clearly. . .)
I have no doubt that there are many potentially useful drugs that are abandoned early. False negatives are probably on the shelves all around us. I don't see that as a strong argument against animal use (what, after all, is the alternative?), but it sure isn't a big argument for it, either. It's just, for now, the way things are.
Comments (11)
+ TrackBacks (0) | Category: Clinical Trials | Drug Assays | Drug Development
Posted by Derek
I've noticed a few stories making the rounds recently about possible new cancer therapies. Johns Hopkins has press-released the work of a group there on, and several news outlets have picked up on a British study on the effect of vanilloid agonists (such as the hot-pepper compound capsaicin) on cancer cells.
And all this is fine, until the word "cure" starts being tossed around. It always is. The number of times you see it, though, is inversely proportional to how reliable your favorite news source is. I wish the Nottingham and JHU people all the best in their research, and I hope that their projects lead to something good. But they have a long way to go, which you might not realize from the "Johns Hopkins Patents Cancer Cure" and "Hot Peppers Can Cure Cancer" headlines.
You see, these studies are all on cell cultures. I've worked on several cancer research programs, and I'm sure that other readers who've done the same can back me up here: unless you've seen cancer drug discovery work at close range, you may have no idea of just how many compounds work against cancer cells in a dish. It isn't that hard. I have absolutely no idea of how many thousands of compounds I could dig up from our files that will just totally wipe out a lot of the common cancer cell lines - in culture, that is.
We don't even bother looking at a compound unless it goes through cultured cell lines like a flaming sword. Problem is, a good number of those compounds will go through normal cells in the same fashion, which isn't exactly what the oncology market is looking for. And of the ones that are left, the ones that aren't hideous toxins - well, a lot of those hit the skids when they go into a live mouse model. Drug candidates that rip through the cell assays but fizzle in the mouse are very easy to come by. Anyone who does oncology drug discovery can furnish you with piles of them, and you're welcome to the darn things.
Now comes the really ugly part. We've ditched the nonselective cell killers, and we've shaken out the compounds that can't cut it in a live animal. How many of these actually work in human beings? Nowhere near as many as we'd like, that's for sure. AstraZeneca's drug Iressa is always useful to keep in mind. That one was going to be a huge hit, back when it was in development. But in real patients, well. . .for the vast majority of them, it just doesn't do much at all. There are a few responders (some of whom we can screen for), but otherwise, you'd have to call the compound a massive failure in the real world. Oh, but you should see it kick through the cell assays, and watch what it'll do for the mice.
Our assays just aren't that predicitive. It's a big problem, and everyone in the field knows it, but so far (despite crazy expenditures of time, money, and brainpower), no one's been able to improve things much. Anyone who does cancer work knows not to celebrate until the human trials data come back, and you'd better be careful even then. So the next time you read about some amazing thing happening to cells in a dish, well - wish the researchers luck. And go back to what you were doing before. There's time.
Comments (11)
+ TrackBacks (0) | Category: Cancer | Drug Assays | Drug Development
October 22, 2006
Posted by Derek
"You like those scatterplots, don't you?", someone said to me the other day. And I can't deny it. On most projects that my lab has been assigned to, at some point I end up messing around with all the project data, plotting one thing against another and looking for correlations.
Often what I find is negative. Plotting liver microsome stability (a measure, in theory, of one of the major pathways for drug metabolism) against compound blood levels in animal dosing has rarely, in my perhaps unrepresentative experience, shown much of a correlation. In vivo blood levels are just too complicated, and influenced by too many other things. But I'm often surprised by how many people assume that there's a correlation - because, to a first approximation, it sort of makes sense that there might be - without actually having run the numbers.
That's a theme that keeps recurring: a fair amount of what people think they know about their project isn't true. I think it's because we keep reaching for simple explanations and rules of thumb, in hopes that we can get some sort of grip on the data. We give these too much weight, though, especially if we don't examine them every so often to see if they're still holding up (or if they ever did in the first place).
Another factor is good ol' fear. It's unnerving to face up to the fact that you don't know why your compounds are behaving the way that they are, and that you don't know what to do about it. It's no fun to plot your primary assay data against your secondary data and see a dropped-paintcan scatter instead of a correlation, because that kind of thing can set your whole project back months (or kill it altogether). One of the biggest problems in an information-driven field is that not everyone wants to know.
One time when I was giving the numbers a complete run-through, I noticed one of the plots actually seemed to have a fairly good shape to it. Y-axis was potency (plotted as -log), and there it was, actually increasing - broadly, messily, but undeniably - with the X-axis, which was. . .corporate compound number, the one assigned to each new compound as it was sent in for the assay. Oh, well. It showed that we were making progress, anyway. And at least nobody suggested that we attempt to give the compounds numbers from years in the future, in order to make them instant surefire winners. I've heard sillier suggestions.
Comments (7)
+ TrackBacks (0) | Category: Drug Assays | Drug Development
August 30, 2006
Posted by Derek
If you want to make your friends in the cell culture lab jump, just walk up behind them and shout "mycoplasma!" (What's that? You say you have no friends in the cell culture lab? Hmm. . .)
Mycoplasma is a scary word because they're scary little organisms. They're bacteria, just barely, running much smaller than usual and without any sort of cell wall. They also have the tiniest genomes you're likely to ever see - being parasitic allows them to get away with a pretty limited instruction set. They can cause diseases in humans and other animals (excellent review here), but they just love to hang out with your cultured cell lines, too.
From their (admittedly rather limited) perspective, what's not to like? Constant temperature, lots of food, and plenty of well-taken-care-of cells to mooch off of. Problem is, once they get in there, they alter the behavior of the cells they've infected, and you can't trust the results of your assays with them any more. Every cell culture lab tests for these things, and every one of them still has the occasional outbreak. It's the price of doing business. If the cells aren't precious, they're tossed - if they are, there are some antibiotics that will generally kill the little creatures off, but you still have to watch things closely for a while. (If you don't want to test them yourself, you can send samples to these guys, and they'll do it for you).
There have been periodic mycoplasma spasms in many research areas, as various groups have found that their results are suspect due to contamination. Since the little beasts pass right through filters that will strain out normal bacteria, and can't even be reliably seen under normal microscopy conditions in many cell cultures, a little paranoia is justified. Have you checked your cells recently?
Comments (4)
+ TrackBacks (0) | Category: Drug Assays
June 21, 2006
Posted by Derek
Here's a question for my readers in the research community: what assay have you dealt that turned out to be the biggest waste of time and effort? I can think of several strong nominees, but I'll lead off with one from quite a while ago.
This one happened in an antiviral group, and I believe that they were targeting a viral protease. Several chemists started cranking away on the lead compound, turning out analogs for the primary assay. But there was no decent SAR trend to latch on to. Things would look (briefly) sensible, then fall apart again, and there was only a scatter when you tried to correlate things with the secondary assay.
After some three or four months, the reason for all this became clear (it doesn't always, I have to note). Turns out, as it was told to me, that a biologist on the project had everything tested against the wrong enzyme. Who knows what it was, but it sure wasn't the protease of interest. What's more, he had apparently realized early on that it wasn't the right stuff, and was frantically working in the background trying to get the right stuff running. It never worked out. He ended up generated week after week of meaningless data, hoping that the project would go away. Instead, as it turned out, he went away (and not by choice).
So that's my entry. No doubt horrors will quickly emerge to beat it.
Comments (16)
+ TrackBacks (0) | Category: Drug Assays
June 18, 2006
Posted by Derek
Well, my last post on biological systems and their ingredients really touched a nerve (see here for an example). I guess I should, um, clarify my position before the leaky bottles of beta-mecaptoethanol start arriving by FedEx.
I already knew the reasons for several of the components I spoke about - EDTA, for example. And I realize that there's a reason for everything that's in there. But what throws me as a chemist is that some of these recipes seem to be handed on "just because they work" Does a particular enzyme prep need EDTA in it or not? Many times, no one checks, because it probably won't harm things and it's better to be on the safe side, so in it goes. It may be hard for a biologist to understand how odd that feels to a synthetic organic chemist, but I can tell you for sure that it does.
One of the commentors to the last post brought up an important point: biologists optimize for the function of a system. And that often means having a lot of buffers, chelators, cofactors, adjuvants, reducing agents, and chaperones floating around in there with your protein of interest, to keep it thinking that it's still in some kind of cellular environment, thus putting it in the mood to do what it's supposed to be doing. There's no point in trying to see how minimal you can make the system if it's working the way you want it to already.
But we chemists are minimalists. We optimize for the function of a system, too, but in our case, purity is usually a good first variable to tune up. The cleaner everything is in our reactions, the better it generally works. That means pure, distilled solvents, with no water in them. It means an inert gas atmosphere, so there's no reactive oxygen around. And it means that your starting materials and reagents should be as clean as you can practically get them, because when there's two percent of this or five percent of that in the flask, things often start to go wrong in unpredictable ways. When a reaction wipes out on us, the first thing we check is whether everything was clean enough.
So you can imagine how biology looks to an organic chemist, whose ideal reaction is a clear solution in a clear glass flask, with one pure solvent and two pure reactants cleanly converting to only one product. Biological systems, to us, look like trying to do science by adding squirts of barbecue sauce to bowls of beef stew. Of course, as the biologists know, the stuff in those bowls was derived from stew (and worse), and was born to the stuff. It won't work unless things achieve a certain level of stewiness, and the surest way to kill it would be to turn an organic chemist loose on it to clean it all up.
Comments (6)
+ TrackBacks (0) | Category: Drug Assays
June 15, 2006
Posted by Derek
You know, I mean no offense to all my pharmacologist friends and readers, but. . .do y'all really know why all those things are in your buffers and solutions? I've been wrestling with this the last few days, trying to straighten out my "vial thirty-three" problem, and it's been interesting.
There's some reducing agent in there, naturally. Can't have those thiols turning into disulfides and balling up the protein, I understand - but does something bad happen if it's not in there? Generally, no one finds out, because, hey, why mess with it? And there's some EDTA, and some salt, and their function is? Well, as far as I can tell, they're also in there because they've sort of always been. Same goes for the squirt of detergent (Brij-35 or some such), and the tiny bit of bovine serum albumin, of all things. It's just part of the old-fashioned recipe from Grandma's Protein Kitchen.
Now, organic chemistry has a little of this, true, but it hasn't reached quite the Ancient Runestone levels of enzymology. We like to use tetrahydrofuran (THF) for a lot of organometallic reactions, for example, but at least we know that that's because THF is a good co-ordinator to metal cations. At least we don't have six other trace constituents in there that we always use whether we need 'em or not. Another example is how we tend to stick to good ol' ethyl acetate and hexane to run TLC plates, rather than look into other solvent combinations that might do a better job - probably because there are just too many of them to investigate, and EtOAc/hexane works well enough.
And that, I think, is the problem that the biologists face. Biochemical systems are tricky. They have way too many variables, which means that their degrees of freedom have to be reduced just to get anything to work. So all sorts of recipes and rules of thumb are handed down. Not all of them are optimal, but they're mostly decent and will allow you to get on with the project without wasting too much time. Especially in the early part of a project, an immediate 70% effectiveness is worth a lot more than a 98% that would take you a month of work to tweak up to.
Comments (27)
+ TrackBacks (0) | Category: Drug Assays
August 18, 2005
Posted by Derek
One of the odd things about science is that you spend a good amount of time trying to prove that you don't know what you're talking about. At least, if you're doing it right, you should.
Take the first part of a drug discovery project, for instance. Most of them have a "primary assay", which is usually done against a purified protein in vitro, under fairly artificial conditions. Compounds that meet some standard of activity against that target then move on to the secondary assay, which is supposed to be aimed at the same process, but now it's done in living cells. That's a much tougher test. (It's a big leap from pure proteins to cells, about the same size as the leap from cells to whole animals.)
The hope is that the two assays will correlate with each other, but it's never a perfect fit. Generally, what you see is some of the active compounds dropping out for no apparent reason in the cell assay. If your target is in the cytoplasm, then there's always the possibility that these compounds don't penetrate into the cell as well as the others. Or they make it in, but are pumped right back out before they can get anything accomplished. Or perhaps they find some other (even tighter) binding site once they're inside, on some protein unrelated to the readout of your assay. There are always plenty of ways to explain these misfires.
And that's fine, as far as it goes. But if you don't double back and check these things out occasionally, you run the risk of fooling yourself. If your two assays don't correlate very well, it might be that cell penetration is lousing things up, sure - and it might also be that your assays aren't measuring the same thing. Or it could be that your target from the first assay isn't as important as you thought it was. These are the sorts of thing you really ought to be sure about.
So you need to keep yourself honest. Take some of your not-so-good compounds, the ones you'd normally discard after the first cut, and take them on to the cell assay regardless. They'd better not work! Test some of the compounds on a closely related cell line that doesn't have your target in it, if you've got some - is your target really the reason for the activity you're seeing?
Most of the time, you'll find that things are just fine. The inactive compounds really are inactive all the way through. But I've seen the exceptions occur, and more than once. You don't want to wait any longer than necessary to find out that your project is a dud. And worse yet, you really don't want someone else to find out for you. It leads to some of those awkward scenes we'd all rather avoid.
Comments (4)
+ TrackBacks (0) | Category: Drug Assays
|
