Fragments of the future - part 2 (The intersection of chemical and biological space)

In a previous post about heterocycles that appear chemically feasible but have not been reported, we wondered whether these molecules would show biological activity. The structure of biologically relevant chemical space – that fraction of possible molecules that will exhibit some biological effect – is of great interest, but as yet unknown. Brian Shoichet and coworkers at UCSF have just published a thought-provoking analysis in Nature Chemical Biology that is also relevant to developing new fragment libraries. 

The researchers ask why it is that HTS collections of a million or so compounds, vanishingly small in comparison to the roughly 1,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 possible small drug-sized molecules, nonetheless so often succeed in identifying hits. A lovely paper by Tobias Fink and Jean-Louis Reymond had previously computationally enumerated all possible compounds with up to 11 C, N, O, and F atoms. Of these 26,429,328 molecules, 25,810 are commercially available. 

Shoichet and colleagues compared the structures of these compounds with the structures of metabolites and natural products (all of which have by definition been processed by at least one protein) and found that the commercially available compounds were much more similar to natural products and metabolites than were non-commercially available compounds. Indeed, the more similar a molecule is to a known natural product or metabolite, the more likely that it is available for purchase; 2918 of the commercially available compounds are in fact natural products or metabolites. 

The bias also increases exponentially with molecular size: a random 11-atom commercial compound is almost 1000-times more likely to resemble a natural product or metabolite than is a non-commercially available molecule, whereas the bias is only about 2-fold for 6-atom molecules. Similar results were observed with other libraries. 

The authors conclude that: A major reason why the screening of synthetic compounds ever finds notable hits is that our libraries are biased toward the sort of molecules that proteins have evolved to recognize.  

This resemblance is reasonable. After all, most commercially available compounds are ultimately derived from naturally occurring starting materials, so their similarity to natural products isn’t surprising. Moreover, historically much of chemistry was devoted to natural product synthesis, so many of the intermediates built up over the years resemble natural products. And of course, once you learn how to do chemistry on one moiety, you will tend to stick with it unless you have a good reason to do otherwise; each heterocycle behaves (often frustratingly) differently, so if a natural-product-like molecule does the job, why look for trouble? 

But does this “biogenic bias” mean that the rest of chemical space is a biological desert? Not necessarily. I can imagine at least two alternative models of chemical-biological diversity space. 

Let’s call one model “lamp posts in dark fields.” Consider a vast field of some crop that can only be harvested by night. There are lights scattered haphazardly throughout the field. One might expect that the crops immediately under the lamp posts would be harvested more intensively than crops in darker parts of the field, even if other areas are equally productive. In this scenario, the lamp posts reveal natural products and similar molecules, but much – or even most – of (unlit) chemical space may also be biologically active, it just hasn’t been sampled yet. 

Another possibility is the “oil-field model.” As we are all too aware, petroleum is distributed very unevenly across the globe. In some areas, such as Texas, oil was easy to find and easy to extract. In others, such as the deep ocean or the high arctic, oil is harder to find and more technically demanding to access. In this scenario, there are vast pockets of chemical space that are relevant to biology, they just haven’t been identified (let alone accessed) yet. 

These are fun, speculative questions, but the paper provides some practical data. Specifically, 83% of core ring scaffolds found in natural products are absent from commercial libraries. In fragment or lead-sized molecules of MW < 350 with less than three stereocenters, 1891 rings scaffolds found in natural products are not commercially available. These could be useful additions to fragment libraries, and the paper lists 18 examples. 

In fact, at least one company, deCODE, is explicitly enriching its fragment collection with molecules based on natural products and metabolites. This might be a good strategy. After all, even if the “lamp posts in dark fields” model is correct, there are plenty of brightly illuminated, unharvested chemotypes. At least for now, picking these may be more productive than venturing into the twilight-zone of uncharted chemical space.

Hidden Talent Pool

I have a question for the followes/readers here. Is there a hidden talent pool of recent Ph.Ds who have training in FBDD and can immediately walk into a pharmaceutical company? I am aware of two companies that had openings for FBDD people. Both companies wanted only freshly minted Ph.Ds or just post-doctoral experience. There is a big and deep (and obviously more expensive) pool of FBDD out there who would jump for these jobs.

I was also under the impression that industrial post-docs were largely a thing of the past. So, two questions: 1. are there academic programs/labs who are specifically training NMR-FBDD people, and more in general FBDD practitioners at all? 2. If not, are jobs like these a way for management to dip their toe in the water without actually resourcing FBDD efforts?

Fragments of the future

We recently provided a list of suppliers of commercial fragment libraries. A problem with buying compounds from these sources, of course, is that all your competitors can buy the same compounds. And while talented medicinal chemists are adept at turning common fragments into novel clinical candidates, it’s awfully nice to start with fresh fragments (see here for examples of both).

Researchers at UCB Celltech have recently taken this to an interesting extreme: they’ve computationally enumerated all neutral mono- and bicyclic 5 and 6 membered heteroaromatic rings containing carbon, nitrogen, oxygen, sulfur, and hydrogen. The resulting VEHICLe (virtual exploratory heterocylic library) is a set of 24,847 ring systems, of which only 1701 have been reported.

Of course, as the authors note, many of the remaining molecules “are outlandish and would obviously be either very difficult or impossible to make.” To address this, they used a machine learning approach to gauge synthetic tractability. This resulted in over 3000 molecules, some of which look quite reasonable:



Interestingly, the researchers estimate that only 5 to 10 of these heterocycles are being made each year, which leaves hundreds of virgin synthetic targets.

Are these a rich source of new fragments? Or, as the authors also speculate, do many of these lie outside biological activity space?

Size Doesn't Matter...Complexity Does...or does it?

With 16 days left to vote (Use you franchise!!!) 17 of 20 respondents have said that FBDD is integrated in the hit finding stage at their companies. WOW! and Awesome!
Lorenzo Williams posted a discussion topic on one of the LinkedIn groups (I of course forget which one) asking how many chemists does your company use for going from Hit to Lead. Below are the answers he shared with me...

Aurigene 8 FTEs
Arqule 1-5FTEs
Roche up to 10-13 FTEs
Bayer 1-10 FTEs
Shanghai Medicilon 6 FTEs
Chiron/Novartis 4-6 FTEs

Of course, the proviso was that it all depends on the complexity of the project. My question to this group, since we know so many of your have integrated FBDD in the hit finding stage, is: "Are FBDD based hits resourced differently than non FBDD hits?"

Commercial fragments – part 2

Teddy recently observed that it’s not the size of your library that matters, but how you use it. But how do you get a fragment library in the first place? You can of course build one up from scratch, but it may be easier to just buy one. Last year we put out a call for sources of commercial fragments and received some good comments. There are now some new suppliers of which we’ve recently become aware, so an updated list follows.

Enamine
1190 fragments w. strict “Rule of 3”
11,717 fragment extension set

Iota Pharmaceuticals
Focused on fragment-based discovery
1500 fragments available for purchase
4000 additional fragments in collaboration with Vitas-M

Life Chemicals
22,000 fragments w. MW < 300, clogP < 3
9000 fragments with rotatable bond, PSA, HBA limits

Maybridge
30,000 fragment library (MW < 350)
1000 fragments w. “Rule of 3” and solubility > 1mM
1500 Br- and 5300 F- containing fragments

Zenobia Therapeutics
352 very small fragments (Avg. MW 155)
Verified solubility at 200 mM

Does anyone have any experience using any of these? Are there others we’re missing? Let us know or post comments – anonymously if need be!

It's Not How Big it Is, It's How You Use It.

Boys will be boys, won't they. There is some gene buried on the Y chromosome that makes them think if something works as is, the bigger you make it the more success you will have. I bring this up because as we all know, many companies are slow to pick up FBDD; it is typically the "Hail Mary" pass, so success is slow in coming, as is uptake of the method as a primary tool. But once there is success, everybody jumps on the bandwagon. So, then the "crappy" little library you had needs to be increased in size (because obviously that's why it wasn't deliver superdrugs before); thank god for the medchemists!!!

So, what is the optimal size of a fragment library.
This is a table put together by the Good Folks® at Evotec for this fabulous book, with some added data bny the Editor of said Tome. What does this show us? FBDD libraries range all over in size. They obviously know something. What do they know?
No matter your screening paradigm, fragment hits need to be confirmed by an orthogonal method. So, even though you can blow through 50-100k compounds by a biochemical screen, you are still limited by your ability to confirm the hits by NMR or SPR or whatever.
If you don't think you need orthogonal methods (and really you do?) then your fragments need to be whistle-clean, because even low level impurities screened at 1-10mM can negatively impact the results of the screen. So, you think you have 10k-50k whistle-clean compounds?
Then, comes this [Shoichet and Austin (J Med Chem. 2008 Apr 24;51(8):2502-11)]. Aggregators!!!
Egads, it's like the Vikings crashing across the North Sea. So, to filter them out, you still need an orthogonal method(s).
So, creating massive (relatively) fragment libraries because you can screen them biochemically cheaply and quickly, doesn't buy you anything, because you still need to confirm them (typically by the same methods you chose not to use in the first place).

In conclusion, it's not the size, but how you use it.

Fragment linking: too strained, too flexible, or too positive?

One of the most rewarding fragment-based outcomes is to link two fragments together and get a pop in activity, as predicted by William Jencks almost three decades ago. Unfortunately, all too frequently linking two fragments gives a less than additive boost in binding energies, and often it doesn’t work at all. A new paper from James Stivers and colleagues published in Nature Chemical Biology investigates why.

The authors had previously discovered bipartite inhibitors of the DNA repair enzyme uracil DNA glycosylase (UNG): one piece of the inhibitor was fixed as the substrate uracil, and the other piece was chosen empirically from a library of fragments; the two fragments were each attached by rigid oxime connectors to a flexible linker. A crystal structure of one of these inhibitors bound to UNG showed that the uracil fragment does indeed bind in the uracil-binding pocket, while the other fragment binds in a nearby phosphodiester-binding pocket. However, the linker exhibited an unusual kink, suggesting it might be strained. The researchers have now followed up on this observation to explore the effects of varying the linker.

Four of the molecules made and tested are shown below; these vary only in how fragments connect to the linker. Importantly, the authors were able to determine co-crystal structures of all of these molecules bound to UNG. The most potent molecule, MA1, has a (flexible) secondary amine connection to the uracil fragment and a (rigid) oxime connection to the benzoate fragment; the linker appears unstrained in the crystal structure. In contrast, although the two fragments of DO bind in the same manner as the two fragments of MA1, the linker assumes the apparently strained “kinked” conformation previously observed, and the molecule binds with a thirty-fold lower affinity. In the case of the other two molecules in this set, the uracil fragment still binds as expected, but the benzoate is either not visible in the electron density (MA2) or binds to a nearby molecule of UNG in the crystal (DA). Both of these molecules are weak inhibitors, with IC50s comparable to that of uracil connected to the linkers alone, without the benzoate fragment (700-750 micromolar).



The authors argue that the greater linker flexibility in MA1 compared to DO reduces linker strain when the two fragments assume their optimal positions. They also suggest that the effect of the linker on the tighter-interacting uracil fragment will be less pronounced than on the more “loosely interacting” benzoate fragment; uracil’s specific binding interactions are able to overcome linker strain, while the weaker benzoate requires more precise positioning. In other words, the flexible connector to the uracil and the rigid connector to the benzoate give a Goldilocks-type situation for MA1.

This is reasonable. But there may be more than strain and flexibility at work here. In particular, introducing a (positively charged) secondary amine near the benzoate may have a considerable electrostatic effect on binding. The authors consider this possibility unlikely, and provide some evidence against it, but given that this enzyme binds to negatively charged DNA, I can’t dismiss it. Although the linkers are solvent-exposed, the electrostatic surface of UNG is quite positive, and in fact the benzoate linkage is not far from a histidine residue. It would be interesting to see whether replacement of the oxime linkages with similarly uncharged ethers or methylenes has the same effects as the amines.

Still, this type of systematic analysis is a valuable experimental addition to the field of fragments. I hope someone will do high-level computational modeling on these complexes to try to further dissect the origins of the differences in affinities.

More on DOCKing fragments and sampling chemical space

A few weeks ago, we highlighted a paper from Brian Shoichet’s group at UCSF demonstrating that computational screening could successfully identify fragments binding to a protein target, and that the binding modes predicted were actually observed experimentally. A companion paper just published online in PNAS now extends these results, and also beautifully illustrates that it is possible to cover much more chemical space with fragments than with lead-like molecules.

Denise Teotico, Shoichet, and colleagues used the program DOCK 3.5.54 to screen 137,639 fragments against AmpC beta-lactamase, a bacterial protein responsible for antibiotic resistance. The protein had previously been the target of HTS and computational screens of drug-like like molecules. The computational screens had modest success rates (2-7%), but the HTS screen was a total bust: of the more than 1200 hits from the 70,000+ compound screening collection, more than 95% of these turned out to be false positives, mostly aggregators, with just a few dozen true inhibitors, all of which turned out to be covalent (irreversible).

In contrast, of the 48 high-scoring fragments that were experimentally tested, 23 had Ki values better than 10 mM, for a hit rate of 48%. The authors also assessed potential for false negatives by choosing 20 random fragments and testing these for inhibition; only one showed inhibition (with a Ki value of 3.1 mM), and this molecule had scored in the top 5% of docked fragments.

The paper presents a fascinating empirical test of the Hannian chemical complexity hypothesis. Starting with the 23 active fragments, the researchers calculated how many lead-like molecules (up to 25 non-hydrogen atoms) could contain these fragments. Of the roughly 47,000,000,000 to 430,000,000,000 possible lead-like molecules, only 675 are commercially available. By repeating this analysis with fragment-sized molecules (up to 17 non-hydrogen atoms), the size of the haystack was reduced by six orders of magnitude: only about 10,000 possible molecules contain these fragments, of which 93 are commercially available. Moreover, many of the active fragments represent unique chemotypes not previously observed in AmpC inhibitors. As the authors note:

The chances of discovering interesting chemotypes for biological targets is many orders of magnitude higher when targeting molecules in the fragment weight range than even at slightly higher size ranges.

But, as the paper asks, “are the docking predictions right for the right reasons?” The researchers solved the crystal structures of 8 fragments bound to AmpC. Four of these reproduced the docking predictions well, two were somewhat different, and two were way off. In these last two cases, the protein itself adopted different conformations than had been used in the docking studies.

Protein conformational flexibility is remarkably common, and likely to be a persistent difficulty for computational methods. Clearly, current computational methods can’t identify all possibilities, particularly with fluxional proteins. Still, especially with relatively rigid proteins, computational fragment-screening may reveal chemotypes that HTS won’t.

A notable feature of the fragments is their relatively poor ligand efficiency: with one unusual exception (a phosphinate), all of the active fragments have ligand efficiencies less than 0.3 (kcal/mol)/atom. AmpC has a large, open active site, and the authors suggest that the failure of other hit-ID methods against this target may reflect issues such as solubility.

It remains to be seen whether these fragments can be advanced to low nanomolar inhibitors, but at least fragment-screening has provided many new starting points. And the paper demonstrates, once again, that triaging a fragment set computationally can be an effective means for concentrating the needles in a haystack.

Updated Again: Fragment Events in 2009

It seems the fragment calendar was front-loaded this year, but there are still a few upcoming events.

June 8-9: GTCbio is holding its “Fourth Assay Development and Screening Technologies” conference in San Francisco, and there is one session on fragment-based screening. (Full disclosure: I’ll be speaking at that one - stop by and introduce yourself!)

June 26: Select Biosciences has a “Fragment-Based Lead Discovery Summit” in London. It’s only one day, but there are a number of great speakers.

September 21-23: Last but much-anticipated, FBLD 2009 will be held in York, UK.

Regarding previous events, RSC’s Fragments 2009 has been covered both here as well as on FBDD-Lit. There is also an excellent eBriefing of the NYAS symposium on molecular diversity that can be found here.

As always, let us know if we’ve missed anything and we’ll get the word out.

LELP fragments reach their potential

In last month’s issue of Nature Reviews Drug Discovery, György Keserü of Gedeon Richter and Gergely Makara of Merck published a thought-provoking analysis of recent trends in lead discovery. Their results illustrate the potential of fragment-based methods, but also point out the still sizable gap from current practice.

The authors assembled a database of 335 hit-lead pairs derived from high-throughput screening (HTS) that were published between 2000 and 2007. They also assembled a database of 84 non-HTS hit-lead pairs published between 2000 and February 2008, consisting of fragment-based, virtual screening, natural product, and miscellaneous examples. They then compared properties – such as potency, molecular mass, logP, logS (a calculated measure of solubility), and ligand efficiency – of the initial hits with the resulting leads.

The results for HTS hits are not pretty: the lipophilicity as assessed by logP was considerably higher on average for HTS hits than for hits from any other methods, and this only increased as the hits were progressed to leads. The same goes for (in)solubility (as measured by logS). Even more alarming, the average properties of even the hits are worse than those of a collection of 541 approved drugs.

Fragment hits start out with the lowest lipophilicity and highest predicted solubility, but during hit-to-lead optimization these properties deteriorate to the point where they are similar on average to leads derived from HTS. Also surprisingly, the ligand efficiency of fragment hits and leads are, if anything, lower than their HTS counterparts, contrary to expectations. Even the average molecular weight of fragment-derived leads is not much lower than HTS-derived leads.

So what’s going wrong? The authors point out that, at most larger companies, fragment-based approaches are often only attempted after HTS has failed, suggesting that the targets tackled by fragment-based methods may be inherently more difficult. But they also suggest that in the early stages of hit-to-lead optimization the primary measure of success is how many compounds are delivered to lead optimization, which could encourage rapid hit expansion with simple chemistries to rapidly boost potency by adding grease, leading to more hydrophobic, less soluble molecules that will ultimately struggle in the clinic.

The authors suggest a new metric, ligand-efficiency-dependent lipophilicity, or LELP, to help avoid this trap:

LELP = (log P / LE)

Since a desirable logP range is between 0 and 3, and a desirable ligand efficiency is above 0.4, one should strive for LELP values between 0 and 7.5. There are already lots of metrics out there for evaluating molecules: see, for example, discussions of %LE, antibacterial efficiency, and fit quality (also here and here). Is a new one really necessary? Perhaps, if it gets people to focus on non-lipophilic means of increasing potency.

The authors end on a positive note for fragments:

Bearing in mind the sampling of chemical space, hit properties and synthetic accessibility, we consider that fragment hits are the optimum starting points for lead discovery and optimization.

There is, however, a burden on the team transforming a fragment hit into a viable lead: it is important to focus not merely on improving potency, but on maintaining as many of the fragment-like properties that make fragments attractive starting points in the first place. Although this goes without saying, analyses like this one suggest that it still needs to be said – and heard.

Book Review on FBDD Book

There is a review just publish ASAP in JACS of the Zartler and Shapiro edited book on FBDD. Overall, it is a nice review (Thanks to Andrew and Phil). There is one error (chapters 4-11 are EIGHT chapters, not seven) and one critique that I want to address.

Although there is a paragraph in Chapter 3 covering X-ray methods, which are mentioned in the introductory chapters, it would have been nice to have an entire chapter dedicated to these methods as several groups in industry have applied them successfully.

Mike and I made a conscious choice NOT to include any chapters on X-ray. We thought that of all the methods for FBDD X-ray has been done; there was nothing new to the field that our book could contribute. The focus of this book was on practical applications and newer techniques. There are a plethora of nice reviews out there, X-ray focused companies, and three chapters entirely or mostly about using X-ray in the Jahnke and Erlanson book.

What are the feelings of others? Did we swing and miss by leaving that topic out of the book or is X-ray the "mature" FBDD method? I would argue (and did in my editorial choice) that there is little left to say about X-ray that hasn't already been said.

[Update]: Zartler and Shapiro:Amazon.com Sales Rank: #1,595,845 in Books.

Jahnke and Erlanson: Amazon.com Sales Rank: #1,238,437 in Books

Nuclear Magnetic Crystallography part II

In what can only be seens as a cosmic convergence, a second paper has appeared on NMR-X-ray hybridization. This one is a collaboration from Medivir, The University of Florence, and Bruker Biospin. This method is aimed at generating structural information for a family of related proteins (in this case MMPs). The authors argue that the cost of 13C and 15N labeling is so low that such samples should be readily available, making this method widely applicable. The thrust of their method is the use of X-filtered NOESY spectroscopy to generate distance constraints, then use Autodock to determine binding. I expect that this will get covered on our FriendBlog, the FBDD-Lit Blog, so I won't go into many details.
Instead I would like to make this a discussion of the perceived value of methods such as this to the FBDD community. I, despite being an NMR jock by trade, don't feel that labeled protein methods, give enough bang for the buck, compared to ligand-based methods. Will methods such as described Isaksson et al. change that cost-benefit analysis?
What do other people think about the value and the proper role of NMR?

Nuclear Magnetic Crystallography

There is often a competition, explicit or implied, between NMR spectrometrists and X-ray crystallographers. A new technique merges the best of both worlds

Researchers at the University of Shutka, Russia, have constructed a unique NMR spectrometer with a hole bored all the way through the magnet and probe, perpendicular to the main sample cavity. This hole allows the researchers to send an X-ray beam directly into a crystal mounted in a specially designed sample chamber, allowing them to screen for compound binding by NMR while simultaneously obtaining crystal structures. Of course, due to the solid state (crystalline) form of the protein, the researchers can’t actually detect the protein itself by NMR, but by mounting the crystal in a flow cell and testing pools of fragments, they can use target-based NMR to determine which fragments bind to the protein. Once they find a fragment that binds, they can then immediately obtain the crystal structure. The researchers are planning to bring their NMR to a synchrotron to have access to a brighter X-ray source.

Will the technique become widely accepted? If so, this could be the start of a beautiful friendship.

Fragments and kinases at Abbott

Of the fragment-derived drugs that have entered the clinic, roughly half target protein kinases. Phil Hajduk (of SAR-by-NMR fame) and Irini Akritopoulou-Zanze provide a nice overview in a recent paper of how Abbott applies fragment-based approaches to this target class.

As anyone who works in the field of kinases can attest, there’s a lot of competition: the authors state that, since 2001, more than 10,000 patents or patent applications describing protein kinase inhibitors have been published. With so many inhibitors already identified, developing novel molecules is a particular challenge. This is somewhat due to the fact that most kinase inhibitors bind at least in part to the conserved purine binding site, or hinge-region. This is a thermodynamic hot-spot, usually accounting for 40-60% of the total binding energy for fully elaborated molecules. It is straightforward to carve out fragments that bind in this site from existing molecules, as shown in the figure.


But as also shown in the figure, this approach can create complicated IP, with dozens or even hundreds of overlapping patents covering the fragments. Of course, this doesn’t mean that clever medicinal chemistry can’t navigate the IP minefield, as examples described on this site from SGX and Astex have shown.

Nonetheless, the Abbott researchers decided to steer away from this hazard. They have designed and synthesized about 50 novel hinge-binders and over 5000 more elaborated molecules, and used these in their screens against kinases as well as other targets. The results are quite interesting.

First, many of the fragments show quite a bit of selectivity, hitting only one or two kinases out of a panel of 11 different kinases. In other cases, selectivity could be achieved through subsequent optimization, though in these cases the process was more often driven empirically than by structure-based design.

Another interesting observation is that the same fragment sometimes bound to different kinases in different fashions, or changed its binding mode during elaboration. This phenomenon has been previously reported by researchers at Vernalis.

The third observation is that, while these molecules exhibited a 10-fold enrichment against 13 different kinase targets compared to the generic Abbott screening collection, they also exhibited a 3-fold enrichment for 20 non-kinase targets. The authors suggest this is due to the fact that they may be acting as adenine mimetics, but it is also consistent with the Hann model of less complex fragments being able to bind to more targets (as discussed here and here).

Of course, fragment-based methods are not the only way to identify kinase inhibitors. As the authors note, “there has been almost universal success in the design and identification of potent kinase inhibitors.” Still, this brief review provides some practical advice on how fragment-based philosophies can complement more traditional lead discovery approaches.

Fragments in silico, confirmed by X-ray

I’ve always been something of an empiricist, and have therefore been wary of computational fragment screening. It’s not that I think it’s impossible, just that the algorithms and parameters developed to date have not often shown themselves up to the task. A paper just published in Nature Chemical Biology from Brian Shoichet’s group at UCSF has caused me to reconsider my skepticism.

Shoichet and Yu Chen used the program DOCK to screen 67,489 commercially available fragment-sized molecules contained in the database ZINC against the active site of the beta lactamase CTX-M, a bacterial enzyme responsible for resistance to penicillin and cephalosporin. Of 69 top hits, 10 actually inhibited the enzyme when tested experimentally. In contrast, of 37 high-scoring hits from a similar computational screen of 1,147,326 larger lead-like molecules, none showed any inhibition up to the limit of their solubilities.

Interestingly, each of the ten active fragments contained an anionic group: 3 carboxylates, 2 sulfates, and 5 tetrazoles among the set. A reexamination of the docked lead-like molecules revealed a relatively high-scoring tetrazole, which exhibited an experimental Ki value of 21 micromolar (see figure). Although this was an in silico hit, it was swamped by the number of (inactive) hits and so had not been selected for experimental follow-up until the fragment results revealed tetrazoles to be privileged pharmacophores. Additional similarity searching of the lead-like molecules led to two additional low micromolar inhibitors.



Five of the inhibitory fragments and one of the lead-like molecules were characterized crystallographically, and the results were remarkable: all of them bound in a similar manner to that predicted by docking.

Chen and Shoichet also investigated the specificity of the fragments compared to the lead-like compounds, and the results agreed well with those predicted by Hann and colleagues (as discussed on our sister blog FBDD-Lit here). Namely, while the fragments had relatively low specificity against a mechanistically distinct beta lactamase (AmpC), the lead-like molecule exhibited roughly 100-fold tighter inhibition of CTX-M. In other words, fragments likely have a higher hit rate (and correspondingly lower specificity) due in part to their simplicity, but as fragments are elaborated, specificity can be readily built into the molecules.

So does this mean the era of computational fragment-based screening has arrived? While these results are impressive, it is important to keep them in perspective. CTX-M has a relatively rigid active site, while many proteins of interest show a level of flexibility that confounds modeling. Moreover, Chen and Shoichet were working with an ultra-high resolution (0.88-Angstrom) crystal structure of CTX-M in which they could actually see density for hydrogen atoms on some polar groups. Needless to say, this is atypical. Still, the paper does give hope that the computational tools are ready, as long as they are applied to appropriate systems.

Fragments on Glass

I’m always a fan of new fragment technologies, and Hioryuki Osada and colleagues at RIKEN have just published a very intriguing one which they call a “fragment combination array,” or FCA.

The approach involves immobilizing fragments onto a specially prepared glass slide using a photogenerated carbene reaction; this can be done as an array of microscopic spots. Next, the glass slide is treated with a fluorescently labeled protein and washed. If the protein sticks to the small molecule, it will show up as a fluorescent spot. Appealingly, if the protein of interest is genetically fused to a fluorescent protein, crude cell lysates can be used, simplifying the assay. The researchers previously demonstrated that natural products and drugs could successfully be immobilized to a treated glass chip using this method, and that the molecules retained their ability to bind to protein targets, despite the covalent linkage to the chip.

Of course, binding interactions between drugs and their targets are generally much stronger than between fragments and their targets. Also, because fragments are so small, there is a higher probability that the part of the fragment used to attach to the glass will be critical to but inaccessible for binding. Fortunately, the carbene chemistry is fairly non-selective, inserting into C-H and O-H bonds at random; thus, the likelihood is that at least some of the fragments will bind in a productive fashion. The technique is conceptually similar to the SPR-based methodology used by Graffinity, though to my knowledge Graffinity screens individual fragments as opposed to binary pairs.

But does it actually work? An initial proof-of-concept used FKBP12 ligands, the same ones previously described in the first famous “SAR by NMR” paper. In that example, a low micromolar pipecolinic acid derivative was linked to a high micromolar benzanilide derivative to generate a nanomolar binder to FKBP12. In the current case, the immobilized pipecolinic acid derivative was able to capture fluorescently-labeled FKBP12, while the benzanilide was not. However, co-spotting the two fragments led to a much stronger signal (more fluorescence) than the pipecolinic acid spot alone, suggesting synergy between the two immobilized fragments.

Having shown this, the researchers next turned to the protein carbonic anhydrase II (CAII), which has a predilection for sulfonamides. They created an array from four aromatic sulfonamide-containing fragments (and one negative control) and ten diverse non-sulfonamide-containing fragments. A screen of these 50 different mixtures against fluorescently-labeled CAII revealed a number of hits, and by merging elements of one of the diverse compounds onto a sulfonamide “anchor” fragment, the researchers were able to improve the potency of the sulfonamide from 435 nM to 29 nM.

Of course, there are limits: clearly the technique is not as sensitive as many other fragment-detection methods, as illustrated by the inability to detect benzanilide binding to FKBP12, an interaction with a Kd in the mid to high micromolar range. In fact, both test cases involve protein targets that have been shown to be highly amenable to fragment-based methods, and both start with known fragments with relatively high affinities. Moreover, the covalent immobilization methodology won’t work for all fragments; acetazolamide, a fragment-like high-affinity binder of CAII, didn’t work in this assay, likely due to poor geometry or sterics of the immobilized fragment.

Still, FCA is a neat and potentially very rapid method for finding a second fragment once a first has been identified. It will be fun to watch how the technique evolves.

Fragments 2009

Just a few quick thoughts on RSC BMCS Fragments 2009, which I had the pleasure of attending at AstraZeneca’s beautiful new conference center in Alderley Park, UK. The quality of the talks and posters was very high, and in many cases speakers presented unpublished and exciting research, so I don’t want to steal their thunder here (though see the FBDD-Lit blog for some nice summaries).

One striking observation was the number of speakers from big pharma. Of the 16 oral presentations, almost a third were from AstraZeneca, Pfizer, GlaxoSmithKline, or F Hoffmann-La Roche. Biotech represented about half the talks, with the remainder from academia.

Attendees were similarly diverse, both by employer as well as geography. Besides the UK and USA, many European countries were represented, as were China, Japan, and Korea. I remember recently it was rare to find anyone exploring fragments outside of the US and the UK.

Rod Hubbard observed in his closing talk that just a few years ago fragment-based drug discovery was seen as the domain of “exotic eccentrics.” No longer. The concept has gone mainstream, there has been a convergence as to the methods (particularly a rapid adoption of surface plasmon resonance), and large pharmaceutical companies are investing substantial resources in FBDD. I think the field can look forward to a wealth of new discoveries.

And for those of you who missed it, feedback was sufficiently positive that there will likely be a sequel: Fragments 2011.

Guest Blogger: Brian Stockman

[DrZ: Most of you probably know Brian and his excellent work in NMR and drug discovery, especially fragments. I have asked Brian to summarize his most recent paper for us. Below is his contribution. The Editors would welcome others to do the same if they are so inclined.]


A recent paper from Pfizer [Chemical Biology & Drug Design 73, 179-188 (2009)] described the concerted use of NMR screening, competition binding, TROSY-based binding site mapping, and NMR-based activity assays to identify allosteric fragment activators of 3-phosphoinositide-dependent kinase-1 (PDK1). This protein kinase presented an interesting challenge since, in addition to the ATP site typically targeted by structure-based drug design efforts, it was known to have an allosteric site that could activate (or potentially inhibit) activity.

An STD-based NMR screen resulted in 372 fragment hits out of 10,237 fragments screened. Testing the compounds in an activity assay would normally eliminate the many false-positive artifacts of the STD assay. A first pass of the hits through a Kinase-Glo assay revealed that many were in fact inhibitors. Fragments without activity in this assay, however, could not be discarded since this assay was not capable of monitoring events at the allosteric site and could not distinguish ‘non-inhibitors’ from activators. Thus fragments that did not inhibit in the Kinase-Glo assay were also run in a Caliper assay. This assay uses a shorter peptide substrate and is capable of detecting inhibition and activation. Ultimately, a subset of the original fragment hits that were either inhibitors with high ligand efficiencies, activators, and/or had very novel chemical structures were chosen for further studies.

STD competition binding experiments using the known ATP-site binder staurosporine or a short peptide known to bind in the allosteric pocket were very useful to distinguish these two binding sites. TROSY-based binding site mapping, using 15N-labeled PDK1 expressed in baculovirus, was used to confirm the binding site for several key compounds. Finally, the biochemical assay data was complemented with 19F NMR-based activity assays. These assays used the 2-fluoro-ATP method described in a previous paper from Pfizer [Journal of the American Chemical Society 130, 5870-5871 (2008)].

NMR-based activity assays proved very valuable since they could easily handle high fragment concentrations, and, since they directly monitor conversion of substrate to product, were capable of detecting both inhibition and activation. NMR-based activity assays are single-enzyme assays. As such, they are quite useful as both primary fragment screening assays and as orthogonal HTS-triage assays. NMR-based activity assays have been characterized as the ‘uncola’ of biochemical assays because, as opposed to many HTS and bench top assays, they do not rely on any coupling enzymes for their detection. NMR-based activity assays should prove very valuable for accurately evaluating compounds in the 10 uM to 1 mM dynamic range of activity typical of fragments.

Fragments in the clinic: How many?

At the Tri-Conference last week, Maria M. Flocco of Pfizer stated that a search of the IDDB3/Prous databases yielded 30 examples of compounds that had made it into the clinic from fragment-based approaches, of which 23 are still active, and 4 are in Phase II testing.

And not just in the clinic. According to her, tipranavir, an HIV protease inhibitor approved by the FDA in 2005, was derived from a 30 micromolar hydroxycoumarin fragment, back before people really thought in terms of fragments. This leads to the question, how many approved drugs could be considered the result of fragment-based drug discovery? I have argued that sorafenib fits the bill, having started from a relatively weak (17 micromolar) fragment-like screening hit. Any others?

BTW: Readers of Practical Fragments had previously identified 17 clinical compounds discovered through fragment-based methods. You can read that discussion here.

Hopping to selective nNOS inhibitors

Richard Silverman, the discoverer of pregabalin, has published a trifecta of papers describing the use of his “fragment hopping”strategy to discover selective, potent, and pharmacologically active inhibitors of neuronal nitric oxide synthase (nNOS).

nNOS is itself part of a triad of closely related enzymes: nNOS, iNOS, and eNOS. All synthesize the radical signaling molecule NO from arginine, but their tissue localizations and functions differ considerably. Both nNOS and iNOS have been pursued as drug targets, with nNOS implicated in a variety of diseases of the central nervous system. However, as eNOS is involved in maintaining blood pressure, identifying inhibitors selective for the desired NOS is essential. Doing so is difficult because all three enzymes share very similar active sites: of the 18 amino acids close to the substrate binding site, 16 are identical, and one of the two residues that vary has its side chain facing away from the substrate binding site.

The Silverman group has been working on selective nNOS inhibitors for over a decade, and succeeded in generating remarkably selective inhibitors, such as the compound shown on the left. But with its peptidic nature and dramatic CLogP, this compound is a long way from a drug. Enter “fragment hopping.” First, a known inhibitor is computationally deconstructed into its minimal pharmacophores; these can be as rudimentary as a positive charge or a hydrophobic spot. Next these virtual fragments are rebuilt into new virtual molecules; further computation weeds out molecules that are likely to be metabolically unstable or toxic. Finally, the best molecules are synthesized and tested. The process is computationally intensive, and I’m simplifying it greatly here, but the results, first published in J. Am. Chem. Soc. last year, are impressive: the dipeptide was transformed into a smaller, more drug-like molecule (center in figure), with increased ligand efficiency. It also retained selectivity over iNOS and eNOS.



In the latest issue of J. Med. Chem., Silverman and colleagues report using the technique again, this time to not only improve potency, but to improve the drug-like properties as well. The resulting molecule (right in figure) still maintains good selectivity, but now also boasts a respectable CLogP. Consistent with this, the molecule shows efficacy in a rabbit model of cerebral palsy. For a great review summarizing the long march to these compounds, check out this just released Acc. Chem. Res. paper.

What I really like about this work is that the researchers use fragment-based methods to actually improve the pharmaceutical properties of their molecules. Much of the focus of FBDD has been in discovery of early leads; these papers show that the approach can be useful downstream as well.

And finally, we can’t let today pass without sending out a Happy 200th Birthday to Charles Darwin (as well as Abraham Lincoln). As Teddy noted in an earlier post, “it’s evolution baby.”