Seventh Annual Fragment-Based Drug Discovery Meeting

CHI’s annual FBDD meeting took place in San Diego this week, and since this was the first time in a while both Teddy and I have been in the same place we’ve decided to make this a joint post. As with last year this does not aim to be comprehensive.

One of the highlights of the conference was a set of three talks on BACE1 inhibitors from Amgen (Ted Judd), Lilly (David Timm), and Pfizer (Ivan Efremov), the first two of which have been discussed here and here. It’s nice to see fragments playing a pivotal role in delivering advanced leads and – at least in the case of Lilly and Merck – clinical candidates against what has been one of the most difficult drug targets in industry.

Speaking of difficult targets, Till Maurer of Genentech gave a lovely presentation on using NMR-based fragment screening to discover inhibitors of the holy grail of oncology, Ras. They’ve recently published some of this story, which we’ll highlight in an upcoming post.

A common question is ‘how often do you find the same fragment using different methods?’ (see here for an ongoing discussion on LinkedIn). Cynthia Shuman from GE gave a nice case study in which she screened the protein PARP15 against 987 fragments using a Biacore T200. Of the 15 fragments with shapely, well-behaved sensorgrams, 14 were confirmed by NMR. On the other hand, only one of these hits was detected in a differential scanning fluorimetry (thermal melt) assay.

Marcel Verdonk at Astex described general trends from mining in-house and published data. After looking at 43 in-house targets, he found that 8000 compounds had been tested against 2 or more proteins, and after plotting by molecular weight found that, consistent with the original Hannian model, larger compounds are more selective. In a separate analysis of 53 fragments that had been advanced to leads, he found that in most cases the initial fragment maintained roughly the same position and orientation from start to finish.

Rod Hubbard, Teddy and I all ran round-table discussions, but only Teddy kept notes, which are summarized here.

The topic started as a discussion of 2D vs. 3D fragment libraries. In the recent Pfizer fragment build, a group of diverse chemists eyed every compound, and at least five had to agree to each molecule before it went into the library. 

The discussion went briefly to the old fight: Are nitros masked amines or noxious moieties?  The table ended up agreeing that if you would remove the nitro group or change it anyways, why put it in the first place?

We then dove right back into the 2D vs. 3D debate. Kinases seem to love 2D fragments, while other classes of targets seem to NEED 3D fragments. One idea discussed was 3D fragments as complements to 2D fragments. It was mentioned that 3D fragment libraries would need to be MUCH larger to cover equivalent chemical space. I thought the idea that 3D fragments would be exploring “vector” space rather than “chemistry” space would mean that you could go with a much smaller library, if you want to use it for vector space searching. It was also proposed that 2D fragments tend to be much smaller (~150-180Da-ish) and 3D fragments would be, by necessity, bigger (~250 Da).

The topic then changed to SBDD (structure-based drug discovery) as part of FBDD. Most people at the table were of the opinion that they wouldn’t use FBDD (on a normal priority target) without SBDD. And with SBDD, you don’t need 3D fragments to explore vector space since you will have the X-ray to guide you. 

The point was made that 3D fragments also HAVE to be scaffolds which would end up in the final compound. If you are simply using a scaffold to explore vector space without any hope of it ending up in the final product, why bother. [TZ Note: I think this is very shortsighted and people do not understand that targets will most likely NOT have SBDD to guide them, at least in the fragment-based lead generation stage.] 

The question was asked to the table: is FBDD a valid approach in the absence of structure? Three people said yes, the rest (~8) said no. All three who said yes were small company/ CRO people. Everyone agrees if you do go with FBDD without structure you need to have a HEAVY investment in biophysics to characterize the protein and the hits. 

One person asked about low confirmation hits following a fragment screen: the table agreed that this is most likely the result of the library being “wrong”. We finished the discussion by asking if 2D fragments are more pan-class (not targeted to a specific class of targets) and 3D fragments may be more target-focused. The resounding answer was “Who knows, but why would they be?”

As this post is already getting long we’ll stop here, but for those of you who were also at the conference please add your comments. And if you missed this one, there are still several exciting upcoming events this year!

Sometimes Your Compound Sucks

Folks, let me wax grouchy. But, first, let me explain. My kids (10 and 7) play sports. I try to inculcate them with a sense of fairplay and sportsmanship. From their first game, I have always said to them: two rules to playing sports. 1. Never blame the refs (umps). 2. Never blame the equipment. It has gotten me through a lot of losses, hurt feelings, and it generally makes me feel like a good dad. I then take the money I am putting away for when they need therapy later in life for all the bad things I will/have done and buy beer. Trust me, this will come into play in at the end.
I have been trying to figure out how to review this paper from Bastien et al. It hasn't been easy. The kinder gentler version of Teddy is just not a hit. Go figure. This paper is aimed at delivering J. Med Chem. quality data derived from fragments on a mutant DHFR. DHFR is a long time drug target and is currently used in many approved therapies.

The paper all started out so well, with description of the fragment library, you know, fragments, "generally comprise[d} cycles -- often aromatic-- and were generally of an elongated structure." Seriously, in J. Med. Chem., they described their fragment library that way. There is no supplementary information.
With this excellently characterized fragment library, the screened all 100 compounds (what a herculean effort) and found 7 compounds in the high micromolar to millimolar range.
This is what they found...oh, and no inhibition in the biochemical assay for follow up compounds 1a, 4a,b,c,5a, and 6a. The depth of the SAR startled me. They did extensive SAR which led to this:
Now, starting from fragments which had pretty decent ligand efficiency, there were able to generate these horrendous monsters, with absolutely revolting LE. They then point out that these are "poorly optimized compounds." DUH.
They then did docking of the molecules and it shows that these symmetrical molecules for a U shaped molecule and suggest that intermolecular stacking may occur in solution. They then point out that this may be due to limitation in the docking software. Then tried to soak (9) into crystals, but there was a change in the active site pore electron density. And, again another excuse, "The pore lies on the crystallographic symmetry axes, combined with the fact that the ligand did not appear to be present at full occupancy, it prove impossible to interpret the electron density." YET, in the very next sentence, seriously the very next one, "From what could be seen, the density was not consistent with the U-shaped conformation of compound 9 as suggested by the docking results, but the poor quality of the electron density PRECLUDES DRAWING CONCLUSIONS [emphasis mine] on the actual bound conformation of the inhibitor."
So, now at this point, my brain literally explodes like this:
One more highly quotable line is: "Notwithstanding the precise mode of binding, symmetry appears to play a key role in binding in selectivity. " I assure you all I am NOT making this up.

There may be some redeeming social value in the paper, but I give up in finding it. Bastien et al., stop blaming the refs and the equipment!!!

Library design, search methods, and applications of fragment-based drug design

This is the title of a new book, edited by Rachelle Bienstock, which comes out of two symposia she organized at recent ACS Meetings (one of which is summarized here).

The book starts with an overview of fragment-based drug design by Bienstock. This is a thorough summary of the talks in both symposia, including those that did not end up as full chapters. This chapter also includes a useful table of available software relevant for FBLD.

The longest section of the book is devoted to designing and searching fragment libraries. In chapter 2, Dimitar Hristozov and colleagues at Eli Lilly and the University of Sheffield describe an algorithm for the de novo design of new molecules based on known reactions. Chapter 3, by John Badger of DeltaG Technologies, addresses the question of how to design libraries for crystallographic screening, with some emphasis on software. Chapter 4, by Ammar Abdo and Naomie Salim at the University Teknologi Malaysia, describes a “Bayesian interference network” for virtual screening as an alternative to conventional similarity searching. And chapter 5, by François Moriaud and colleagues at MEDIT, describes the researchers’ mining of the protein data bank (PDB) to understand the relationship between ligands and their protein pockets. This information is used for generating bioiosteric replacements and to generate new compound libraries for specific targets.

The next section of the book is focused on docking. In chapter 6, Zsolt Zsoldos of SimBioSys describes their high-speed eHiTS (electronic High Throughput Screening) engine for docking fragments. This chapter delves deeply into a statistical scoring function, and should be of particular interest for the mathematically-inclined. Chapter 7, by Peter Kolb at UCSF, discusses DAIM (Decomposition and Identification of Molecules), a program designed to break larger molecules into fragments, as well as computational methods for docking these derived fragments. He describes the use of this software to discover high affinity ligands for the kinase EphB4 and other targets.

The last section is devoted to fragment growing and linking. Chapter 8, by Eugene Shakhnovich and colleagues at Harvard, discusses FOG (Fragment Optimized Growth), which grows molecules by adding fragments such that the resulting molecules resemble those in a training set. This allows one to focus on regions of chemical space that are believed to be particularly productive or drug-like. And finally, in chapter 9 Zenobia founder Vicki Nienaber describes how fragment-based approaches are ideally suited for discovering drugs targeting the central nervous system.

By my count this marks the fifth book completely dedicated to fragment-based lead discovery, but its focus on computational methods still sets it apart from the others. That’s the fun thing about fragments: there’s something for everyone.

FBLD 2012 - registration now open

We are delighted to announce that registration is now open for the FBLD 2012 meeting to be held in San Francisco from 23rd to 26th September this year. Please visit the conference site and click through to registration.

This promises to be THE major fragment-based discovery conference of the past two years (you can read impressions of FBLD 2010 and FBLD 2009). A great list of speakers has already confirmed, and we still have plenty of room for more talks and posters. We plan to have some themed sessions around particular targets and approaches - these will be announced on the website as they are arranged.

In addition to great science, we have worked hard to keep the costs down, both for registration and hotel.

So - visit the site http://www.fbldconference.org.

Early bird rates expire on the 1st July 2012.

For a complete list of fragment-based events please see here.

Looking forward to seeing you in San Francisco!

Universal fragments

Third world diseases are popular academic drug targets, perhaps none more so than malaria. In a recent paper in Bio-Medicinal Chemical Letters, researchers from the University of Durak report their efforts to discover inhibitors of falcipain-1, a cysteine protease important for the parasite that causes malaria.

The researchers performed a functional screen of a small fragment library. Among many hits, the most potent were compounds 1 and 2, both with high ligand efficiencies. When these were linked together the resulting compound 3 had an improved potency, and adding a small substituent to yield compound 4 gave a sub-micromolar inhibitor.

Remarkably, not only was compound 4 potent against falcipain-1, it was also potent against several other important disease targets. In recognition of this pan-potency, the researchers have named compound 4 “hitinane.” Whether or not this discovery will lead to a drug, it undoubtedly will fuel many papers, patents, and grant proposals.

Fragment docking: promiscuous but good enough?

Computational docking of fragments can be difficult, partly because the energetic differences between possible binding modes can be so small that it’s impossible to select the best one. In a recent paper in J. Med. Chem. Andrew Good and colleagues at Genzyme ask whether docking results are nonetheless good enough to act on.

The researchers were interested in the kinase Pim-1; we’ve previously highlighted their success using SPR-discovered fragments to generate nanomolar inhibitors of this target. In the current paper, they use the program Glide to virtually screen 13,888 fragment-sized molecules from their general collection against Pim-1. About 3% (462) of these were tested in a functional assay at 125 micromolar concentration, resulting in 46 fragments with IC50 better than 100 micromolar. Five diverse representatives of these fragments were then soaked into crystals of Pim-1, and their structures were compared with those from docking.

Overall, only two of the fragments showed a good correlation between the in silico and crystallographic models (rmsd 1.0 Å or better). However, two of the “failures” do make key hydrogen-bond interactions seen in the crystal structures, though some of the hydrophobic interactions are quite different.

As the researchers note:

Fragments that do not fill their binding pocket can exhibit promiscuous hydrophobic interactions due to the lack of steric constraints imposed on them by the boundaries of said pocket. As a result, docking modes that disagree with an observed crystal structure but maintain key crystallographically observed hydrogen bonds still have potential value in ligand design and optimization.

This seems reasonable, and is consistent with the notion that polar interactions are more directional – and thus perhaps easier to correctly dock – than more generic hydrophobic interactions.

But there may also be something more fundamental going on: the assumption seems to be that the observed crystallographic structures are definitive, but is this an oversimplification? After all, a crystallographically-derived model seldom provides more than one binding mode (though a notable exception led to the first approved fragment-based drug). Perhaps the reason it’s so difficult to score docked fragments is that fragments really can assume multiple binding modes, and our insistence on one single best model is the problem. If this is true, docking models are telling us more about reality than we are giving them credit for. NMR-based models are often presented as an ensemble of structures; perhaps the same should be done for docking? At any rate, NMR studies on Pim-1 with these fragments could prove interesting.

Practical Application of NMR

I am sometimes harsh on academic papers, especially those that purport to describe drug discovery. However, Isabelle Krimm and colleagues have continued their excellent work, previously discussed on this blog in a this paper. This paper reads like a "How To" on prosecuting an NMR-based screen. In this work, they have two goals in mind: studying fragments interacting with targets with multiple hotspots and determining the utility of fragments for allostery.

Glycogen phosphorylase is an interesting system to work in: it has an active site and six regulatory sites, including an allosteric site with a variety of positive homotropic and negative heterotropic effects between the various sites. They took 19 known inhibitors that bind to the active site (1-6), inhibitor site (7-9), allosteric site (10-12), and the "new" allosteric site (13-15) and deconstructed them.
Then they screened these fragments against GPa and GPb using both STD and WaterLOGSY. A compound was only deemed a hit if it was observed to bind via both methods. I find this approach very interesting. STD works via NOE from the saturated target to the bound compound. WaterLOGSY works via NOE from bound water to the compound. Each experiment has advantages and disadvantages, but are they truly orthogonal experiments. As a third experiment, the authors use transfer NOE to confirm binding. I would expect to see at least one truly orthogonal method to confirm binding, such as SPR. They then used competition screening against known inhibitors to bucket their fragments based upon the site they are binding to.

While I don't think the results here are not similar to results achieved in industry many times over, this is an excellent paper that shows the power of NMR in screening and how to apply that to drive answers to target validation and compound bucketing.

This paper leads I think to interesting academic musings. Does ontogeny recapitulate phylogeny for compounds derived from fragments? Does it matter if the fragment is 3D or not? Is there a floor below which a fragment will not bind? Does this floor move if you are using 3D fragments vs. highly planar ones?

LLE vs LELP

Besides ligand efficiency (LE), a slew of other metrics has been proposed to help evaluate what compounds to take forward in drug discovery. As seen in our poll and a recent round-table discussion, lipophilic ligand efficiency (LLE) is quite popular:

LLE = pIC₅₀ (or pK_i) – ClogP (or logD)

However, because this metric is not size-adjusted, it is not particularly useful for evaluating fragments, which often have low potency. In contrast, the metric LELP accounts for size:

LELP = logP / LE (where LE = ligand efficiency)

In a recent issue of J. Med. Chem., György Keserű and colleagues evaluate how these two metrics compare in a variety of settings.

The authors examine eight different compound sets: fragment hits and derived leads, HTS hits and derived leads, leads that subsequently became drugs (ie, “successful leads”), development candidates, compounds that entered phase II trials, and drugs on the market. Not surprisingly, drugs and phase II compounds had better LLE and LELP scores than other molecules. Also not surprisingly, fragments scored misleadingly poorly on the basis of LLE but well on the basis of LELP. What is perhaps unexpected, though, is that LELP was better at identifying successful leads than was LLE. Moreover, when compounds were evaluated for pharmacokinetic and safety parameters, LELP was more effective at predicting problems than was LLE. The authors state:

In summary, evaluation of pharmacokinetic and safety parameters revealed that LELP has benefits over LLE, as compounds with acceptable in vitro ADMET profiles are discriminated from compounds with significant liabilities.

Despite these potential advantages, LELP doesn’t seem to be widely used, perhaps because it is less intuitive than some of the other metrics. Indeed, it would be interesting to see these studies repeated using LLE_AT, which also takes lipophilicity into account but has the same scale as LE.

Fragment-based drug discovery and X-ray crystallography

I’m holding in my hands a book of this title, edited by Thomas Davies and Marko Hyvönen and published this year as part of Springer’s Topics in Current Chemistry series. I believe this is the fourth book entirely devoted to fragment-based drug discovery, which shows both the vitality and rapid development of the field.

The book starts with an introduction to fragment-based drug discovery by me. If you’re new to the field, this chapter should serve as a self-contained summary.

In the next chapter Thomas Davies and Ian Tickle describe how FBDD is practiced at Astex, paying particular attention to the use of X-ray crystallography. Notably, researchers from this company “do not consider a fragment hit to be ‘validated’ and suitable as a starting point for medicinal chemistry until it has been observed to bind by crystallography.” This chapter also contains a nice analysis of fragment library design and a couple case studies, including the discovery of the clinical-stage CDK2 inhibitor AT7519.

Rod Hubbard and colleagues at Vernalis and the University of York next describe their efforts to discover Hsp90 inhibitors using a combination of virtual and fragment screening. We’ve covered some of this before (here and here), but it’s nice to see the full story.

The next chapter also focuses heavily on a single target: Daniel Wyss and colleagues at Merck describe their success in discovering BACE inhibitors. This chapter also includes an excellent review of NMR methods for finding fragments.

Michael Hennig and colleagues at Roche (Basel) contrast the various biophysical methods used to discover fragments, with a heavy emphasis on SPR. Crystallography is also covered, in particular co-crystallization of fragments with protein. Co-crystallization is more time-consuming than soaking fragments into preformed crystals, so compound prioritization techniques such as SPR are especially useful.

One of the most promising applications of fragment-based methods is tackling tough targets such as protein-protein interactions, the subject of a chapter by Marko Hyvönen and colleagues at the University of Cambridge. The chapter contains a nice discussion of energetics and hot spots as well as a detailed analysis of methods to find fragments which complements some of the other chapters.

Eddy Arnold and colleagues at Rutgers discuss the use of crystallographic fragment screening against two HIV-1 targets, HIV protease and HIV reverse transcriptase (RT). We’ve previously discussed the former here. In the case of RT, fragments were soaked into crystals in the presence of a high affinity inhibitor, effectively blocking its binding site from fragments. More than 30 fragments were identified binding to multiple other sites on the protein – one fragment bound at 11 distinct sites! Interestingly, the fragments were enriched for halogen-containing molecules. Several also had functional activity with respectable ligand efficiencies. The authors also discuss other published fragment work on HIV RT.

Finally, Didier Rognan at the University of Strasbourg discusses computational approaches to library design, binding site determination, and predicting druggability. Fragment docking is extensively covered, along with a discussion of what factors contribute to success. It seems that docking is particularly good at identifying negatively charged, relatively buried fragments that make similar hydrogen bonds as the substrate. De novo ligand design, both the successes and challenges, is also covered.

Like last year’s book, all the chapters in this one are published online, but it is worth getting a bound copy as it is nicely put together, with color figures liberally integrated throughout rather than banished to plates at the back.

Molecular Medicine Tri-Con 2012

Molecular Medicine Tri-Con has just ended in San Francisco, and although it is one of those massive conferences with a very wide scope, from bioinformatics to stem cells, there were some talks relevant to FBLD.

Teddy recently summarized how fragment-based approaches have been used to develop bromodomain inhibitors at GlaxoSmithKline, and they’re certainly not alone: Mark Bunnage of Pfizer described how they used fragment-based methods to discover nanomolar inhibitors of BRD4 that are active in cells. This work was done in collaboration with the Structural Genomics Consortium (SGC), and one well-characterized compound, PFI-1, is being released as a chemical probe to the worldwide research community with no intellectual property entanglements.

There was a heavy emphasis on metrics, and as we saw in the poll last year ligand efficiency and LLE (sometimes also called LipE) seem to be dominant. However, the latter metric is used more for advanced leads than for fragments. Siegfried Reich from the Lilly Biotechnology Center (neé SGX) said of LLE that it is less important where you start than where you end up, though starting from a very polar fragment gives you the luxury of adding lipophilicity during optimization.

Slight changes to fragments can often cause them to bind in different orientations, and the same fragment may bind differently in closely related proteins, but Siegfried argued these multiple orientations could be advantageous by providing multiple opportunities for optimization. Siegfried also mentioned that deconstruction of HTS hits to fragments has been successful at identifying fragments with high ligand-efficiency that could subsequently be optimized to new series.

Richard Law of Evotec gave as clear account as possible of fragment molecular orbital (FMO) calculations, a high-level quantum mechanical method for understanding protein-ligand interactions. Although quantum mechanical calculations are notorious for taking hours, days, or even weeks to run, the calculations can be done much more efficiently by breaking larger molecules into fragments.

Richard also kept thorough notes at a break-out discussion I moderated, and was kind enough to share them; I’ve also made a few additions. There were nine people from both large and small organizations, all but one from industry.

Screening technologies

• As we’ve seen here, SPR seems to be the most common fragment-finding technique; in his presentation, Walter Huber said that it is the primary screening method at Roche.
• "Reverse SPR" - the Graffinity/Novalix technology we’ve discussed previously, has been applied to over 100 targets. It is reversed because the small molecules are immobilised on the chip, in multiple orientations to present different moieties to the target protein. This is also a larger library (~25,000).
• NMR was not being used as much as SPR, and no one at the table was using 19F NMR, though its use does appear to be growing, and compound suppliers are coming out with 19F fragment libraries.
• FCS++ is being used at Evotec; it has the advantage of being high-throughout but still highly sensitive and therefore accommodates a larger than average fragment library (~20,000).

3-D fragments

• There is increasing desire for sp3 (3-dimensional) fragments and a move away from planar fragments, though one participant had seen chemists shy away from too many chiral centers.
• Advantage of more vectors for SBDD, and additional solubility versus otherwise equivalent flat compounds.
• Despite these advantages, often planar fragments yield more hits - likely because planar compounds are more likely to form dispersive/non-specific interactions, whereas sp3-fragments must form very specific H-bond interactions in order to bind. Does sp3 therefore also enrich for enthalpic binders?

Use of indices
LE was used by everyone at the table, whereas LipE/LLE were not used until lead compound stage. BEI/SEI and other metrics were not really used.

FBLD vs HTS
Fragment screening is being used on most programs at many companies, in parallel with HTS and virtual screens. A subset of targets is addressed only with fragment screening either because of target-specific information or specific requirements to lower costs of screening.

Fragment screening of GPCRs
There was very limited experience using FBLD on G-protein coupled receptors, though one company is trying to use nanodiscs to stabilize GPCRs for fragment screening. As more crystal structures are solved people may become more comfortable tackling this target class.

Finally, there was widespread agreement that fragments are ideal for designing in the desired physico-chemical properties of molecules as fragments are developed. The parent fragment is not biasing, and could often be med-chemed away. Richard Law offered the analogy of a rock-climber:

An HTS gets you halfway up the cliff, but the route to the top may not be from where you are, or may just be too difficult to find from that position. Whereas a fragment hit is at the base of cliff but enables you to see and select the exact route to the top that you need.

If you attended the break-out discussion or the meeting please comment on your impressions, and if you missed this conference don’t worry – there are many more great events coming up throughout the year.

Upcoming Webinar on NMR in DD

I wanted to make everyone aware of an upcoming webinar: "You are smarter than you think: Applications of NMR in Hit-to-Lead Discovery". Here are the key learning points:

• Application of NMR in Fragment Screening using both target and ligand based methods
• Utilization of NMR Data, e.g. epitope mapping, in Hit to Lead efforts
• Structure determination of compounds by NMR

What I would love to have from the readers of this blog is suggestions as to what particular topics they feel would be most important to learn about. Obviously, the first two topics are broad and I wanted to give the readers here a chance to help set the specifics that will be discussed. Leave a comment or send me an email directly.

Slow-off, albeit tight, fragments

Practical Fragments recently discussed binding kinetics, and that got me wondering whether any fragments have slow off-rates. Turns out some do: a January 2012 review of protein-ligand energetics and kinetics in Drug Discovery Today by Sara Núñez and colleagues at Abbott summarized a paper published last October in Eur. J. Med. Chem. by Jos Lange et al. In it, Lange and colleagues extensively characterize six inhibitors of the enzyme D-amino acid oxidase (DAAO), a potential target for schizophrenia.

The researchers use biochemical assays, surface plasmon resonance (SPR), and isothermal titration calorimetry (ITC) to characterize the thermodynamics and kinetics of their inhibitors binding to DAAO. Although all six molecules are fragment-sized, these are not your typical fragments: the weakest has a Kd better than 1 micromolar, and all have ligand efficiencies of 0.79 kcal/mol/atom or better! Three of them are shown below, along with their dissociation constants (determined by ITC) and their dissociation rate constants (determined by SPR).

One interesting aspect of the kinetics is that compound 6 dissociates from the enzyme roughly 50-fold more slowly than compound 3, even though it binds only about 3-fold more tightly. As an interesting aside, compound 1 has a slower on-rate than any of the other molecules, a phenomenon the researchers attribute to tautomerization around the pyrazole.

The researchers go on to measure a number of other properties of these molecules, including Log P, pKa, thermodynamic and kinetic solubility, cell membrane permeability, and in vivo pharmacokinetics. There is a tremendous amount of data here, and it’s a lot of fun to dig into.

With a predicted half-life of more than 2 hours, compound 6 certainly classifies as a slow-off fragment. So is it a better drug? Well, it’s not orally bioavailable, in contrast to some of the other compounds, and it has faster clearance in rodents. Unfortunately the researchers do not report in vivo target modulation, but one has to assume that a schizophrenia drug would need to be oral. Chemists can optimize affinity, thermodynamics, kinetics, and drug-like properties all we want, but the body still has the final say.

Who's Regulating the Regulators?

In two recent ASAP papers in J. Med Chem. Chung et al. from GSK in Stevenage report on their recent efforts in discovering bromodomain inhibitors. Bromodomains are part of the "hot" target class largely lumped together in epigenetic targets. Bromodomains are the sole reader for Acetyl-lysine (AcK) and are thus important components of maintaining the "histone code". These domains are small (~110 aa) and have a common fold of four anti-parallel helices with the peptide recognition site in loops at one end of the helices. There are at least 56 bromodomains encoded in 42 proteins in the human genome.

The GSK group assembled a library that mimicked AcK: compounds had hydrogen bonding functionality and a small alkyl group. 1376 compounds were tested in a fluorescence anisotropy assay. Of these, 132 (~10%) showed >30% displacement of the fluorogenic ligand. After all actives were fully titrated, compounds were soaked into apo crystals. 40 structures were then analyzed by X-ray. The figure below shows how the native peptide is bound, analysis of the crystals showed that the H-bond interaction with the bridging water dominates that with N156, but both influence ligand positioning.
Below the structure of one bromodomain is shown, the yellow spheres are AcK sites. The other figure shows the preferred binding surface and low energy waters involved in AcK mimetic binding. These interacting residues tend to be conserved among bromodomains. The authors conclude the first paper by stating that their fragments are unlikely to discriminate between bromodomains. But, the real question is: is it possible to create compounds from these fragments which can?

And the answer is....[I did say there were two papers] maybe.

People have found compounds which work against bromodomains before (Cpds 1 and 2).
As can be seen they are relatively potent (nM) and decently ligand efficient. Cpd 3 and 4 are fragments found in the screen from GSK. Cpd 3 had ~30 % activity against two bromodomains. The authors admit it was far from the best fragment, but "it was small, efficient, novel, and chemically attractive." Who can argue with that.

Following up with this cpd, they found this compound binds in a manner consistent with its selection: one methyl of 3 mimics the terminal methyl of AcK, the second one overlaps the e-CH2 of AcK, and isoxazole N and O mimic the carbonyl. At 2A resolution, they couldn't differentiate which heteroatom was where, so they made 4 to prove their placement: it was.

As a start to the hit expansion, they used a 3D pharmacophore model to search for commerically avaailable analogs: analog by catalog. The found a series of phenyl-isoxazole with meta-sulfonamides on the phenyl ring. Compound 5a is ~100x more potent than the parent fragment and is very efficient (0.39). The xtal structure of 5a shows that it binds exactly as expected. SAR around this compound showed an increase in affinity, but with the expected loss of LE. This series also suffered from poor solubility. They could not add functionality to the sulfonamide (its pocket is rather hydrophobic), but instead where able to increase solubility through para-phenols (this points towards the solvent). They were able to increase solubility via this route, while keeping potency, but losing LE. This series also had cellular activity.
















This work shows that not all protein-protein interactions are flat and featureless. This work also shows that you can target PPI without having to change the rules: no need to relax the Lipinski Rules. These exciting new results show that the hot new targets in drug discovery play by the same rules as the same old targets.

Fragment linking: flexible rules

Linking two fragments together to achieve a boost in potency has been done a number of times (see here, here, here, and here), though it often doesn’t work as well as might be hoped (see here). To better understand the energetics of fragment-linking, Marc Nazaré, Hans Matter, and colleagues at Sanofi-Aventis Deutschland have analyzed ligands for the blood coagulation enzyme factor Xa (fXa) and published their results in a recent issue of Angew. Chem. Int. Ed.

The researchers “deconstructed” potent fXa inhibitors into component fragments, measured their inhibition constants (and thereby inferred their binding energies), and compared these binding energies with those of the original linked molecules. One of the first observations was that many of the component fragments bound so weakly as to show no measurable activity, a phenomenon that has been observed previously.

In an exemplary case, cleaving a single bond connecting the two component fragments of a 2 nM ligand (1a, below) yielded one fragment (1g) with 58 micromolar activity and another (1d) whose activity was worse than 10 millimolar. Because the second fragment has such low affinity, the binding energy of linking is really just a lower estimate, but it seems to be at least 3.3 kcal/mol, which is greater than the binding energy of fragment 1d itself. In other words, the affinity brought about by linking is greater than the affinity of the weakly binding fragment. The superadditivity provided by the linker in this case is about 300-fold, a similar value to that observed in the unrelated MMP-12 system. This is perhaps all the more remarkable given the fact that the fragments are connected by a linker containing several rotatable bonds, the entropy of which should partially counter the advantages of linking.

In fact, a common strategy to improve the potency of two linked fragments is to rigidify the linker. Often this doesn’t work: in a second case, the Sanofi-Aventis researchers cleaved one bond of a 3 nM ligand (2a, below) to yield two fragments with roughly equal potency. However, even though the linker is more rigid than in the previous example, the binding energy due to linking is less – just 2.0 kcal/mol, representing a boost of about 30-fold.

As the authors note:

The introduction of rigid aromatic moieties as a common approach to increase affinity does not necessarily maximize the benefit from the linker effect as detrimental affinity contributions might originate from suboptimal orientation and accommodation of specific binding elements.

There are many more examples in this paper than can be covered in a blog post; the authors dissect compounds 1a and 2a at a number of different points, and while the component fragments typically bind less tightly than simple additivity would suggest, there are lots of interesting details.

Finally, it is interesting to note that ligands 1a and 2a consist of a relatively hydrophobic fragment (1g or 2g) connected to a more polar fragment (1d or 2h). The fact that these show superadditivity is consistent with Mark Whittaker and colleagues' proposal last year that linking such fragments is likely to maximize additivity, although given the precise interactions made by both parts of the molecules the details get a bit messy. We’re not yet at the point where the universe of molecular interactions can be distilled to rigid rules.

Fragment events in 2012

2012

February 19-23: Molecular Medicine Tri-Con 2012 will be held in San Francisco, with a section on fragments on February 21.

March 13-14: Select Biosciences is holding its Discovery Chemistry Congress in Munich, Germany, with a full two days devoted to fragment-based lead discovery.

March 19-23: Keystone Symposium: Addressing the Challenges of Drug Discovery – Novel Targets, New Chemical Space and Emerging Approaches will be held in Tahoe City, CA. Although not exclusively devoted to fragments, there are many speakers I look forward to hearing.

April 17-18: Cambridge Healthtech Institute’s Seventh Annual Fragment-Based Drug Discovery will be held in San Diego, with short courses on SPR and FBLD taking place on April 16. This is a chance to meet both editors of Practical Fragments! You can read impressions of last year’s meeting here and 2010’s here.

April 27: Dr. Teddy Z will be giving a webinar on NMR in hit-to-lead discovery.

May 13-17: The 30th Noordwijkerhout-Camerino-Cyprus Symposium Trends in Drug Research will be held in Amsterdam, including a session on fragment-based drug discovery.

June 6-8: Cambridge Healthtech Institute’s Twelth Annual Structure-Based Drug Design will be held in Boston, with a session on FBLD.

August 19-23: The Fall 2012 ACS meeting will take place in Philadelphia, PA, and has at least one section on fragment-based drug discovery.

September 23-26: Finally, FBLD 2012, the fourth in an illustrious series of conferences, will be held in my fair city of San Francisco. This should be a biggy – the first such event in the Bay Area (and the weather in September is usually decent too). You can read impressions of FBLD 2010 and FBLD 2009.

2013

March 4-5: Fragments 2013, the 4th RSC-BMCS Fragment-based Drug Discovery meeting will be held at the Harwell Science and Innovation Campus near Oxford, UK.

Know of anything else? Organizing a fragment event? Let us know and we’ll get the word out.

Enthalpy Arrays

As regular readers know, there are lots of ways to find fragments, each with its own strengths and weaknesses. Isothermal titration calorimetry (ITC) is useful for being able to extract thermodynamic values from an experiment, but it tends to be low-throughput and is thus used more as a secondary rather than a primary assay. To make calorimetry more convenient, Michael Recht and colleagues at Palo Alto Research Center have constructed nanocalorimeters. They describe using these “enthalpy arrays” for fragment screening in a paper just published online in the Journal of Biomolecular Screening.

In a typical ITC experiment, a protein is mixed with a ligand and the tiny temperature change that occurs upon binding is detected. In the case of nanocalorimeters, up to 96 detectors are arranged on a plate, and sample volumes are typically a few hundred nanoliters. At this scale, the enthalpy of binding becomes very challenging to measure, but it is possible to measure the heat generated during the course of a reaction as an enzyme processes its substrate. This allows one to follow the reaction in real time without any dyes, labels, or artificial substrates. Also, since an inhibitor will change the reaction profile in a predictable manner, one can determine its mechanism of action.

Each detector in the enthalpy array is set up such that droplets are rapidly mixed together in sets of two. In the experimental set, one droplet contains the protein of interest while the other contains a substrate and a fragment. (Each detector also incorporates a control pair – one droplet containing the substrate/fragment mixture and another containing bovine serum albumin.) In the current paper, the researchers were looking for competitive inhibitors of PDE4A10, a phosphodiesterase implicated in inflammatory disorders. The protein was present at a final concentration of 5 micromolar, substrate was at 2 mM, and each fragment was present at up to 2 mM. 160 very small fragments (average molecular weight only 154 Da) were screened individually, resulting in 11 competitive hits with Ki < 2 mM; 2 other hits displayed more complex kinetics.

The 11 competitive hits were characterized in more detail; the most potent had a Ki of 0.32 mM and the most ligand efficient had LE = 0.43 kcal/mol-atom. In collaboration with Vicki Nienaber and colleagues at Zenobia, all 11 of these were taken into crystallography experiments. This proved challenging: unliganded PDE4A10 crystals suitable for fragment soaking could not be grown, necessitating more labor-intensive co-crystallography. Unfortunately, although some crystals were obtained, they did not diffract at high enough resolution to unambiguously fit the electron density of the fragments, though there was evidence for binding in the active site. The researchers were able to crystallize PDE4A10 with pentoxifylline, a known phosphodiesterase inhibitor. Since many of the fragments have structural features reminiscent of other phosphodiesterase inhibitors, this suggests starting points for modeling.

As described in this paper, enthalpy arrays could be used as a primary screen for fragment hits with defined modes of action before follow-up by slower methods. Although in this particular case crystallography was not successful determining co-crystal structures of the novel fragments, in a recent talk Michael described a related system which did yield good crystal structures. I look forward to seeing additional applications of this approach.