About this Author
College chemistry, 1983
The 2002 Model
After 10 years of blogging. . .
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: email@example.com
August 19, 2014
How many ways do we have to differentiate samples of closely related compounds? There's NMR, of course, and mass spec. But what if two compounds have the same mass, or have unrevealing NMR spectra? Here's a new paper in JACS that proposes another method entirely.
Well, maybe not entirely, because it still relies on NMR. But this one is taking advantage of the sensitivity of 19F NMR shifts to molecular interactions (the same thing that underlies its use as a fragment-screening technique). The authors (Timothy Swager and co-workers at MIT) have prepared several calixarene host molecules which can complex a variety of small organic guests. The host structures feature nonequivalent fluorinated groups, and when another molecule binds, the 19F NMR peaks shift around compared to the unoccupied state. (Shown are a set of their test analytes, plotted by the change in three different 19F shifts).
That's a pretty ingenious idea - anyone who's done 19F NMR work will hear about the concept and immediately say "Oh yeah - that would work, wouldn't it?" But no one else seems to have thought of it. Spectra of their various host molecules show that chemically very similar molecules can be immediately differentiated (such as acetonitrile versus propionitrile), and structural isomers of the same mass are also instantly distinguished. Mixtures of several compounds can also be assigned component by component.
This paper concentrates on nitriles, which all seem to bind in a similar way inside the host molecules. That means that solvents like acetone and ethyl acetate don't interfere at all, but it also means that these particular hosts are far from universal sensors. But no one should expect them to be. The same 19F shift idea can be applied across all sorts of structures. You could imagine working up a "pesticide analysis suite" or a "chemical warfare precursor suite" of well-chosen host structures, sold together as a detection kit.
This idea is going to be competing with LC/MS techniques. Those, when they're up and running, clearly provide more information about a given mixture, but good reproducible methods can take a fair amount of work up front. This method seems to me to be more of a competition for something like ELISA assays, answering questions like "Is there any of compound X in this sample?" or "Here's a sample contaminated with an unknown member of Compound Class Y. Which one is it?" The disadvantage there is that an ELISA doesn't need an NMR (with a fluorine probe) handy.
But it'll be worth seeing what can be made of it. I wonder if there could be host molecules that are particularly good at sensing/complexing particular key functional groups, the way that the current set picks up nitriles? How far into macromolecular/biomolecular space can this idea be extended? If it can be implemented in areas where traditional NMR and LC/MS have problems, it could find plenty of use.
+ TrackBacks (0) | Category: Analytical Chemistry
July 18, 2014
There's a new report in the literature on the mechanism of thalidomide, so I thought I'd spend some time talking about the compound. Just mentioning the name to anyone familiar with its history is enough to bring on a shiver. The compound, administered as a sedative/morning sickness remedy to pregnant women in the 1950s and early 1960s, famously brought on a wave of severe birth defects. There's a lot of confusion about this event in the popular literature, though - some people don't even realize that the drug was never approved in the US, although this was a famous save by the (then much smaller) FDA and especially by Frances Oldham Kelsey. And even those who know a good amount about the case can be confused by the toxicology, because it's confusing: no phenotype in rats, but big reproductive tox trouble in mice and rabbits (and humans, of course). And as I mentioned here, the compound is often used as an example of the far different effects of different enantiomers. But practically speaking, that's not the case: thalidomide has a very easily racemized chiral center, which gets scrambled in vivo. It doesn't matter if you take the racemate or a pure enantiomer; you're going to get both of the isomers once it's in circulation.
The compound's horrific effects led to a great deal of research on its mechanism. Along the way, thalidomide itself was found to be useful in the treatment of leprosy, and in recent years it's been approved for use in multiple myeloma and other cancers. (This led to an unusual lawsuit claiming credit for the idea). It's a potent anti-angiogenic compound, among other things, although the precise mechanism is still a matter for debate - in vivo, the compound has effects on a number of wide-ranging growth factors (and these were long thought to be the mechanism underlying its effects on embryos). Those embryonic effects complicate the drug's use immensely - Celgene, who got it through trials and approval for myeloma, have to keep a very tight patient registry, among other things, and control its distribution carefully. Experience has shown that turning thalidomide loose will always end up with someone (i.e. a pregnant woman) getting exposed to it who shouldn't be - it's gotten to the point that the WHO no longer recommends it for use in leprosy treatment, despite its clear evidence of benefit, and it's down to just those problems of distribution and control.
But in 2010, it was reported that the drug binds to a protein called cereblon (CRBN), and this mechanism implicated the ubiquitin ligase system in the embryonic effects. That's an interesting and important pathway - ubiquitin is, as the name implies, ubiquitous, and addition of a string of ubiquitins to a protein is a universal disposal tag in cells: off to the proteosome, to be torn to bits. It gets stuck onto exposed lysine residues by the aforementioned ligase enzyme.
But less-thorough ubiquitination is part of other pathways. Other proteins can have ubiquitin recognition domains, so there are signaling events going on. Even poly-ubiquitin chains can be part of non-disposal processes - the usual oligomers are built up using a particular lysine residue on each ubiquitin in the chain, but there are other lysine possibilities, and these branch off into different functions. It's a mess, frankly, but it's an important mess, and it's been the subject of a lot of work over the years in both academia and industry.
The new paper has the crystal structure of thalidomide (and two of its analogs) bound to the ubiquitin ligase complex. It looks like they keep one set of protein-protein interactions from occurring while the ligase end of things is going after other transcription factors to tag them for degradation. Ubiquitination of various proteins could be either up- or downregulated by this route. Interestingly, the binding is indeed enantioselective, which suggests that the teratogenic effects may well be down to the (S) enantiomer, not that there's any way to test this in vivo (as mentioned above). But the effects of these compounds in myeloma appear to go through the cereblon pathway as well, so there's never going to be a thalidomide-like drug without reproductive tox. If you could take it a notch down the pathway and go for the relevant transcription factors instead, post-cereblon, you might have something, but selective targeting of transcription factors is a hard row to hoe.
+ TrackBacks (0) | Category: Analytical Chemistry | Biological News | Cancer | Chemical News | Toxicology
July 8, 2014
There all all sorts of headlines today about how there's going to be a simple blood test for Alzheimer's soon. Don't believe them.
This all comes from a recent publication in the journal Alzheimer's and Dementia, from a team at King's College (London) and the company Proteome Sciences. It's a perfectly good paper, and it does what you'd think: they quantified a set of proteins in a cohort of potential Alzheimer's patients and checked to see if any of them were associated with progression of the disease. From 26 initial protein candidates (all of them previously implicated in Alzheimer's), they found that a panel of ten seemed to give a prediction that was about 87% accurate.
That figure was enough for a lot of major news outlets, who have run with headlines like "Blood test breakthrough" and "Blood test can predict Alzheimer's". Better ones said something more like "Closer to blood test" or "Progress towards blood test", but that's not so exciting and clickable, is it? This paper may well represent progress towards a blood test, but as its own authors, to their credit, are at pains to say, a lot more work needs to be done. 87%, for starters, is interesting, but not as good as it needs to be - that's still a lot of false negatives, and who knows how many false positives.
That all depends on what the rate of Alzheimer's is in the population you're screening. As Andy Extance pointed out on Twitter, these sorts of calculations are misunderstood by almost everyone, even by people who should know better. A 90 per cent accurate test on a general population whose Alzheimer's incidence rate is 1% would, in fact, be wrong 92% of the time. Here's a more detailed writeup I did in 2007, spurred by reports of a similar Alzheimer's diagnostic back then. And if you have a vague feeling that you heard about all these issue (and another blood test) just a few months ago, you're right.
Even after that statistical problem, things are not as simple as the headlines would have you believe. This new work is a multivariate model, because a number of factors were found to affect the levels of these proteins. The age and gender of the patient were two real covariants, as you'd expect, but the duration of plasma storage before testing also had an effect, as did, apparently, the center where the collection was done. That does not sound like a test that's ready to be rolled out to every doctor's office (which is again what the authors have been saying themselves). There were also different groups of proteins that could be used for a prediction model using the set of Mild Cognitive Impairment (MCI) patients, versus the ones that already appeared to show real Alzheimer's signs, which also tells you that this is not a simple turn-the-dial-on-the-disease setup. Interestingly, they also looked at whether adding brain imaging data (such as hippocampus volume) helped the prediction model. This, though, either had no real effect on the prediction accuracy, or even reduced it somewhat.
So the thing to do here is to run this on larger patient cohorts to get a more real-world idea of what the false negative and false positive rates are, which is the sort of obvious suggestion that is appearing in about the sixth or seventh paragraph of the popular press writeups. This is just what the authors are planning, naturally - they're not the ones who wrote the newspaper stories, after all. This same collaboration has been working on this problem for years now, I should add, and they've had ample opportunity to see their hopes not quite pan out. Here, for example, is a prediction of an Alzheimer's blood test entering the clinic in "12 to 18 months", from . . .well, 2009.
Update: here's a critique of the statistical approaches used in this paper - are there more problems with it than were first apparent?
+ TrackBacks (0) | Category: Alzheimer's Disease | Analytical Chemistry | Biological News
July 7, 2014
Catalysts are absolutely vital to almost every field of chemistry. And catalysis, way too often, is voodoo or a close approximation thereof. A lot of progress has been made over the years, and in some systems we have a fairly good idea of what the important factors are. But even in the comparatively well-worked-out areas one finds surprises and hard-to-explain patterns of reactivity, and when it comes to optimizing turnover, stability, side reactions, and substrate scope, there's really no substitute for good old empirical experimentation most of the time.
The heterogeneous catalysts are especially sorcerous, because the reactions are usually taken place on a poorly characterized particle surface. Nanoscale effects (and even downright quantum mechanical effects) can be important, but these things are not at all easy to get a handle on. Think of the differences between a lump of, say, iron and small particles of the same. The surface area involved (and the surface/volume ratio) is extremely different, just for starters. And when you get down to very small particles (or bits of a rough surface), you find very different behaviors because these things are no longer a bulk material. Each atom becomes important, and can perhaps behave differently.
Now imagine dealing with a heterogeneous catalyst that's not a single pure substance, but is perhaps an alloy of two or more metals, or is some metal complex that itself is adsorbed onto the surface of another finely divided solid, or needs small amounts of some other additive to perform well, etc. It's no mystery why so much time and effort goes into finding good catalysts, because there's plenty of mystery built into them already.
Here's a new short review article in Angewandte Chemie on some of the current attempts to lift some of the veils. A paper earlier this year in Science illustrated a new way of characterizing surfaces with X-ray diffraction, and at short time scales (seconds) for such a technique. Another recent report in Nature Communications describes a new X-ray tomography system to try to characterize catalyst particles.
None of these are easy techniques, and at the moment they require substantial computing power, very close attention to sample preparation, and (in many cases) the brightest X-ray synchrotron sources you can round up. But they're providing information that no one has ever had before about (in these examples) palladium surfaces and nanoparticle characteristics, with more on the way.
+ TrackBacks (0) | Category: Analytical Chemistry | Chemical News
May 22, 2014
C&E News has a story today that is every medicinal chemist's nightmare. We are paid to find and characterize chemical matter, and to develop it (by modifying structures and synthesizing analogs) into something that can be a drug. Key to that whole process is knowing what structure you have in the first place, and now my fellow chemists will see where this is going and begin to cringe.
Shown at left are two rather similar isomeric structures. The top one was characterized at Penn State a few years ago by Wafik El-Deiry's lab as a stimulator of the TRAIL pathway, which could be a useful property against some tumor types (especially glioblastoma). (Article from Nature News here). Their patent, US8673923, was licensed to Oncoceutics, a company formed by El-Deiry, and the compound (now called ONC201) was prepared for clinical trials.
Meanwhile, Kim Janda at Scripps was also interested in TRAIL compounds, and his group resynthesized TIC10. But their freshly prepared material was totally inactive - and let me tell you, this sort of thing happens all too often. The usual story is that the original "hit" wasn't clean, and that its activity was due to metal contamination or colorful gunk, but that wasn't the case here. Janda requested a sample of TIC10 from the National Cancer Institute, and found that (1) it worked in the assays, and (2) it was clean. That discrepancy was resolved when careful characterization, including X-ray crystallography, showed that (3) the original structure had been misassigned.
It's certainly an honest mistake. Organic chemists will look at those two structures and realize that they're both equally plausible, and that you could end up with either one depending on the synthetic route (it's a question of which of two nitrogens gets alkylated first, and with what). It's also clear that telling one from the other is not trivial. They will, of course, have the same molecular weight, and any mass spec differences will be subtle. The same goes for the NMR spectra - they're going to look very similar indeed, and a priori it could be very hard to have any confidence that you'd assigned the right spectrum to the right structure. Janda's lab saw some worrisome correlation patterns in the HMBC spectra, but X-ray was the way to go, clearly - these two molecules have quite different shapes, and the electron density map would nail things down unambiguously.
To confuse everyone even more, the Ang. Chem. paper reports that a commercial supplier (MedKoo Biosciences) has begun offering what they claim is TIC10, but their compound is yet a third isomer, which has no TRAIL activity, either. (It's the "linear" isomer from the patent, but with the 2-methylbenzyl on the nitrogen in the five-membered ring instead).
So Janda's group had found that the published structure was completely dead, and that the newly assigned structure was the real active compound. They then licensed that structure to Sorrento Therapeutics, who are. . .interested in taking it towards clinical trials. Oh boy. This is the clearest example of a blown med-chem structural assignment that I think I've ever seen, and it will be grimly entertaining to see what happens next.
When you go back and look at the El-Deiry/Oncoceutics patent, you find that its claim structure is pretty unambiguous. TIC10 was a known compound, in the NCI collection, so the patent doesn't claim it as chemical matter. Claim 1, accordingly, is written as a method-of-treatment:
"A method of treatment of a subject having brain cancer, comprising: administering to the subject a pharmaceutical composition comprising a pharmaceutically effective amount of a compound of Formula (I) or a pharmaceutically acceptable salt thereof; and a pharmaceutically accepted carrier."
And it's illustrated by that top structure shown above - the incorrect one. That is the only chemical structure that appears in the patent, and it does so again and again. All the other claims are written dependent on Claim 1, for treatment of different varieties of tumors, etc. So I don't see any way around it: the El-Deiry patent unambiguously claims the use of one particular compound, and it's the wrong compound. In fact, if you wanted to go to the trouble, you could probably invalidate the whole thing, because it can be shown (and has been) that the chemical structure in Claim 1 does not produce any of the data used to back up the claims. It isn't active at all.
And that makes this statement from the C&E News article a bit hard to comprehend: "Lee Schalop, Oncoceutics’ chief business officer, tells C&EN that the chemical structure is not relevant to Oncoceutics’ underlying invention. Plans for the clinical trials of TIC10 are moving forward." I don't see how. A quick look through the patent databases does not show me anything else that Oncoceutics could have that would mitigate this problem, although I'd be glad to be corrected on this point. Their key patent, or what looks like it to me, has been blown up. What do they own? Anything? But that said, it's not clear what Sorrento owns, either. The C&E News article quotes two disinterested patent attorneys as saying that Sorrento's position isn't very clear, although the company says that its claims have been written with these problems in mind. Could, for example, identifying the active form have been within the abilities of someone skilled in the art? That application doesn't seem to have published yet, so we'll see what they have at that point.
But let's wind up by emphasizing that "skilled in the art" point. As a chemist, you'd expect me to say this, but this whole problem was caused by a lack of input from a skilled medicinal chemist. El-Deiry's lab has plenty of expertise in cancer biology, but when it comes to chemistry, it looks like they just took what was on the label and ran with it. You never do that, though. You never, ever, advance a compound as a serious candidate without at least resynthesizing it, and you never patent a compound without making sure that you're patenting the right thing. What's more, the Oncoceutics patent estate in this area, unless I'm missing some applications that haven't published yet, looks very, very thin.
One compound? You find one compound that works and you figure that it's time to form a company and take it into clinical trials, because one compound equals one drug? I was very surprised, when I saw the patent, that there was no Markush structure and no mention of any analogs whatsoever. No medicinal chemist would look at a single hit out of the NCI collection and say "Well, we're done - let's patent that one single compound and go cure glioblastoma". And no competent medicinal chemist would look at that one hit and say "Yep, LC/MS matches what's on the label - time to declare it our development candidate". There was (to my eyes) a painfully inadequate chemistry follow-through on TCI10, and the price for that is now being paid. Big time.