Honesty is the Best Policy

Continuing our theme of computational FBDD, in this paper Brožič et al. present their results of virtual screening using fragments. Aldo-ketoreductases control the activity of androgens, estrogens, and progesterone modulating the occupancy and transactivation of their receptors.  Selectivity is key because the different AKR1C isoforms (1-4) have different physiological outcomes.  Specifically, you want to  inhibit 1 and 3, while having no activity at 2 and 4.  Crystal structures exist for all four isoforms and there is a good history of drug discovery against these targets.  Salicylates are selective for isoform 1, while phenylanthranilates are potent but unselective for all isoforms. 3-bromo-5-phenylsalicylic acid is 4.1 nM inhibitor against 1 and 20x selective against 2 and 100x for 3 and 4.  Phenylaminobenzoates are potent and selective for 3.  

A virtual screen followed by biochemical evaluation.  Compounds from Asinex, Chembridge, Maybridge, and NCI were pulled from ZINC, yielding 1.9M cpds.  After applying filters (fragment like properties, reactive groups, problematic groups, and predicted and known aggregators) they were left with 143000 cpds.  Active site was defined as portion of the enzyme within 6 A of crystallized ligands.  Docking (using FlexX) of these compounds in the active was performed with the mandated interaction that an H- bond acceptor no more than 3A from Y55 was present.  37 cpds for AKR1C1 and 33 for AKR1C3 were found. Of these 70 compounds, 11 were insoluble.  Single point inhibition at 400 uM was performed on both enzymes, 1 and 3.  Compounds with >55% Inh had the IC50 determined, as well as selectivity vs. 2.  25 actives were found against isoform 1 or 3, 11 of which are salicylates or aminobenzoates (known scaffolds) .  Compounds 1-15 represent new chemical space (based upon similarity calculations) with known scaffolds against this slate of targets. 

 Compounds 16-28 represent structures from new chemical classes. 16-21 show no selectivity, 22 and 23 (ketone and aldehyde) show no selectivity and selectivity for Isoform 3 respectively.  Compound 23 brings up the question, if you filtered to eliminate reactive groups how did an aldehyde make it through?

Compound 25 is selective for isoform 1, while 25-28 are selective for isoform 3.  Compound 26 was the most interesting, as it was potent and selective for isoform 3 (based upon a fluorescence assay). 

However, this selectivity could not be explained by docking (isoform 1 and 3 are 88% identical), which predicted the inhibition of isoform 1, but completely missed isoform 3 (docking rank 21509).  The authors then did MD studies (10ns only) to see what could possibly happen.  What follows is honest to goodness handwaving.  What they conclude is that binding of 26 could induce conformational changes in both inhibitor and enzyme to make it bind really well.  Isoform 3 has a larger SP1subpocket. 

In a rare case of brutal (if unintended honesty), the authors state: 

"It is obvious from these results that our virtual screening protocol is capable of finding potent AKR1C inhibitors but is unable to predict the isoform selectivity."

Atoms are like apples...

Many, many moons ago Mike Hann at GSK published a paper that is either the most cited paper in FBDD, or should be (Hann et al.  J. Chem. Inf. Comput. Sci. (2001) 41: 856).  In it, he presented a simple, but quite useful, model of binding.  

The whole point of this model is it illustrates how less complex ligands (fragments)have a better probability of binding productively, but more importantly have less chance of having a detrimental binding event (the dreaded +/+ or -/- in this model).So, in this case, atoms are like apples, one bad one ruins the whole bunch.  

As has been pointed out previously on this blog, the magic methyl is fleeting and may not even exist.  But it has unitary power that makes everyone seek it, like the One Ring.  So, in my neverending quest to generate new terminology (e.g. Fragonomics) and be entertaining, if not informational, I would like to coin the term "Sauron atom" for that one bad atom that can cause a fragment not to bind.  

 

Not to get too meta on you

If you read this blog, you should be aware that I  think you do not need structure to prosecute fragments.  So, when a paper comes along titled "Toward Rational Fragment-Based Lead Design without 3D Structures" you would think I would plotz.  After reading it, plotz I did, but not from excitement.  

Henen and colleagues (in Robert Konrat's lab in Vienna) present their vision for a structureless future of FBDD.  This future (dystopian for many, utopian for me) is predicated on their meta-structural analysis.  Meta-structural analysis relies on "higher level of abstraction" and "incorporating 3D structural information."  This seems like unfocusing the lens to get a sharper image.  I won't go into the fine details I am sure Peter Kenney has already or would do a much better job if he hasn't.  The general gist is (and trust me I may be misunderstanding this) for a target you predict its secondary structure (because the primary sequence has too low homology to build a reasonable model) and then search for proteins (in DRUGBANK) with known structure and similar secondary structure motifs.  You then get a list of "meta-structurally" similar targets and you make sure they have experimentally-verified ligands. To demonstrate this, the authors show the results of the meta-structure search for lipocalin Q83.

Red structure shows similarity from Q83 while orange shows similarity from the homologs, in this case the beta-barrel structure (A)Strepavidin; (B) FABP, (C) chorismate lyase.  The authors note that chorismate lyase only shares half of the beta-barrel structure.  The further demonstrate this approach, they then chose to go with chorismate lyase and its ligand vanillic acid.  It shares structural similarity with the known Q83 ligand enterobactin.  

A comparison of the solution structure of Q83 (where is my utopian strucutureless future?) and chorismate lyase shows that the two proteins bind their ligands in similar fashion.  From this they concluded that Q83 binds vanillic acid in a 2:1 stoichiometry.  They then used 1H-15N HSQC and ITC to verify the binding of vanillic acid to Q83.  Not surprisingly, it binds. And because they have the assignments (and structure) they found that it binds in the enterobactin binding site.  ITC confirmed that it binds in a 2:1 ratio.  Using the plethora of structural information they have, they wanted to test their ability to link two vanillic acid moieties and went with a "Analog by Catalog"  using the structural insights they have from the meta-structure (and NMR structure) approach with this beast ->.   [To their credit, the authors admit this is not in anyway a reasonable molecule.  I think its most appealing property is that it was available from Sigma-Aldrich.]  They confirmed that this bound to Q83, showing similar chemical shift perturbations as vanillic acid and confirmed by fluorescence quenching.  As their conclusion to this part of the paper, the authors state

Most importantly, it demonstrates that the information needed to rationally improve molecular fragments, found in a first iteration of an FBLD program, is eventually solely provided by meta-structural data without the requirement of a highly resolved crystal structure.

To quote the greatest movie ever, "Opinions vary." [Warning VERY NSFW and very potentially offensive].  As far as I can tell, they didn't have a crystal structure, but a NMR structure.  In my, admittedly biased eyes, that is still structural data. 

In the second part of this paper, the authors repeat this approach with beta-catenin.  But here, they want to show the AFP-NOESY (adiabatic fast passage-NOESY) method they developed can be used for epitope mapping.  The concept here is that:

sizable spin diffusion effects, as a result of the existence of dense hydrogen networks or hydrophobic clusters, lead to measurable shifts of the zero passage toward larger tilt angles.

This seems like "Another (impractical) NMR Method".  STD can deliver the exact same data, has been accepted in the hit discovery  community for a long time, and so on.  I don't see why AFP-NOESY is better.  Anybody care to change my mind?  They end with an example of dynamic combinatorial chemistry to select for a combination of two fragments with improved binding. 

So, did the authors deliver on the promise of their title?  Not in the least.  This is homology modeling by another name.  They still are using structures, just not of the target of interest.  Could it be useful?  Maybe. 

Picomolar beta-lactamase inhibitors with help from fragments

In 2009 Practical Fragments highlighted a computational fragment screen against AmpC beta lactamase, a challenging antibacterial target. That paper described several millimolar fragments, and at the time we noted, “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.” In a new paper in PNAS, Brian Shoichet and colleagues at UCSF, along with collaborators in Italy and France, have used the fragments to discover not just nanomolar, but picomolar inhibitors.

The researchers had previously identified reversible covalent inhibitors containing boronic acids, some of which were quite potent. However, they had reached an affinity plateau, and the compounds had only modest activity against bacteria in vitro. By examining the crystal structures of some of these molecules bound to AmpC, and superimposing these on the crystal structures of some of the fragments bound to the protein, they were able to come up with new ideas. For example, adding the tetrazole of fragment F3 to their previously reported compound 9 led to a low nanomolar inhibitor, and adding a small hydrophobic substituent improved the potency to an impressive 50 picomolar. Similar strategies worked with other fragment-inhibitor combinations.

AmpC beta lactamase degrades cephalosporin antibiotics, and the researchers demonstrated that many of their new molecules were effective at restoring sensitivity to otherwise resistant bacterial strains in vitro and, in the one case tested, in a mouse survival model.

This paper also addresses Teddy’s recent question, what is FBDD? Although the fragments described in 2009 were not themselves advanced, the information they provided was essential to moving the project forward. As the researchers note:

Whereas fragments are widely used to nucleate early discovery, this study suggests that they also may be used to guide late-stage optimiziation into chemotypes and geometries that would be hard to systematically sample by other methods.

This is all the more impressive given the relatively low ligand efficiency of the initial fragments. So though purists may argue whether this work is technically FBDD, it is certainly a nice example of fragment-assisted drug discovery, in this case merging parts of fragments with other molecules.

WACing GAK and thrombin (faster)

Last year we highlighted a fragment-finding technique called weak affinity chromatography, or WAC. Those of you who were at FBLD 2012 saw some nice updates on the approach, but if you missed the meeting you can read two new papers. In the first of these (in Anal. Bioanal. Chem.), Elinor Meiby and Sten Ohlson at Linnaeus University, in collaboration with researchers at Oxford University, describe the use of WAC against a kinase.

Recall that WAC works by immobilizing a protein onto an HPLC column and then flowing ligands through the column; ligands that bind to the protein will have longer retention times compared to their retention times in a column without bound protein. Small amounts of fragments can be injected (2 picomoles in this case), and the fragments are at low concentration, minimizing potential artifacts.

Here the researchers were interested in cyclin G-associated kinase (GAK), a potential target for Parkinson’s disease. They used less than 1 mg of protein to prepare their column and used adenosine as a positive control to show that at least 22% of the protein was still active. They then performed a virtual screen of 3200 fragments (from TimTec), looking for molecules that would bind to the ATP site of five different proteins; the 170 highest-scoring fragments were then tested by WAC in mixtures of 13 compounds. Despite the use of mixtures, run times were long (140 minutes per injection), so this was not a high-throughput approach, although more compounds could potentially be injected simultaneously (see below).

Mass-spectrometry was used as a detection method to identify fragments. The change in retention time between the protein column and reference column is related to the dissociation constant, and the researchers found that 78 of the fragments had an estimated Kd of better than 0.2 mM, a fairly high hit rate. (30 of the 170 molecules could not be detected on both the GAK and reference column, so over half of the assessed molecules tested positive.) However, it is possible that some of the fragments bind to but don’t inhibit the protein. In fact, 23 fragments eluted more slowly from the reference column than from the protein-containing column, illustrating that non-specific effects are certainly possible (see also below). That said, many of the best fragments are structurally similar to known kinase inhibitors.

One of the coolest features of WAC is that a couple of the fragments were racemic, and these gave double peaks on the GAK column but only single peaks on the reference column, suggesting that one of the two enantiomers binds more tightly than the other.

As noted earlier, this campaign was not high-throughput, and in the second paper (in J. Biomol. Screen) Minh-Dao Duong-Thi, Sten Ohlson, and collaborators at Linnaeus and AstraZeneca sought to speed things up. They took 590 fragments from their larger collection and pooled them into 11 groups of 35-65 members in DMSO, with final concentrations of each fragment as low as 0.023 mM. Fragments likely to be positively charged were pooled together, and negatively charged fragments were pooled separately to facilitate mass spectrometry analysis. Also, fragments in each pool were chosen to have unique molecular weights to facilitate unambiguous identification.

These researchers looked for hits against thrombin, a protein they had previously used in developing WAC. The mixtures were screened in 20-minute runs, so all 590 fragments were screened in under 4 hours. 60 fragments were not observed in the mixtures, but 36 of these could be detected when injected singly under more optimized conditions.

The 30 best hits were then confirmed by injecting them individually. To assess whether they bind to the active site of thrombin, the enzyme was inactivated with an irreversible inhibitor, and the fragments retested. Remarkably, only a single fragment showed a significant reduction in binding to the inactivated protein, suggesting that the other fragments were either promiscuous or bound to regions outside the active site. The selective fragment contains an amidine moiety, a known thrombin-binding motif.

This paper demonstrates that WAC can be done in a fairly high-throughput manner. Although the number of non-selective binders does raise concerns, WAC could still be a valuable primary screening method, particularly given that it can be done using standard laboratory equipment.

What is FBDD, and how pure is your library?

The most recent post on the newest biophysical technique for fragment screening coincides with a talk I heard yesterday and some thoughts I have been having recently.  Barry Morgan of GSK gave a talk on DNA Encoded Libraries, which is a Screen for Drug paradigm, as opposite from FBDD as you can go.  The problem I have with such highly elaborated molecules (4-5 synthons) is that they are Screening for Drug, rather they are Screening for Drug Affinity and unless they get extraordinarily lucky, they will still have to medchem to optimize everything else, and maintaining affinity.  I had some interesting discussions at the breaks with Dr. Morgan and a scientist from Vipergen as to the utility of this.  Dr. Morgan agreed that they now mostly focus on libraries of 2-3 synthons (which by and large would fit the definition of fragments (especially if you ignore the linker and humongous amount of DNA sticking off the molecule).  For the GSK work, their method detects signals on as little as 1000 molecules in 25 uL (~66 attoM) and the effective concentration of each molecule in each reaction is much less than 1 pM.  So, this got me to thinking about how pure this combi-chem libraries are (and combi-chem by any other name is still combi-chem)? But, it doesn't matter, aborted chemistry (e.g. 2 synthons instead of 3 synthons) will still be detected.  In fact, this is exactly where some of the power of this technique comes in.  So, is DEL technology FBDD by with an extraordinarly sensitive detection technique?

  

Don Huddler of GSK gave a talk right before me (and I swear we didn't coordinate our titles) about the FBDD work in his group.  When discussing the GSK library, Don mentioned that their purity cutoff is 95%.  Of course, this is pretty low for a High Concentration Screen (a 5% impurity would be present at 50uM!).  I would prefer to see a 99% cutoff for purity, but how realistic is that?  Additionally, some people use methods that don't detect strong binders or use a HCS % inhibition cutoff for their first triage.  So, a 50uM impurity (worst case for most libraries) would probably not even register as a hit in the primary screen.  But, I am curious what levels of purity people use for their fragment libraries (if any).  I somehow can't seem to get a poll to look right with the new blogger interface, so please post your comments (anonymously if you need to) and tell me what you think.

Microscale Thermophoresis (MST)

Practical Fragments has a soft spot for new biophysical methods to identify fragments, many of which are given unfortunately non-descriptive initialisms. To a list that includes SPR, ITC, STD, MS, TINS, CEfrag, and WAC, we can now add Microscale Thermophoresis (MST), described in a new paper in Angew. Chem. Int. Ed. by Philippe Baaske and colleagues at NanoTemper Technologies as well as academic collaborators.

Thermophoresis, also referred to as the Soret effect, occurs when particles move in response to a temperature gradient. In this case, the “particles” are proteins, whose movements depend on size, charge, conformation, and solvation, and can be altered by factors such as ligand-binding.

In MST, a fixed concentration of protein is incubated with varying concentrations of ligand in small capillaries. An infrared laser rapidly heats a spot on the capillary, and an ultraviolet light source excites aromatic residues within the protein. The fluorescence in the heated spot changes as the protein moves along the temperature gradient. This movement is affected by ligand binding, and so measurements at different ligand concentrations can be used to construct a binding curve.

The researchers used MST to study the binding of ligands to several proteins, including ionotropic glutamate receptors (iGluRs), p38-alpha MAP kinase, thrombin, and even the calcium sensor Syt1. The dissociation values determined by MST were mostly comparable to literature values, and the researchers could also perform competition studies in which adding an excess of one ligand blocked a different ligand for the same site.

A nice feature of the technology is that, since it uses native protein, one doesn’t need to worry about the effects of immobilization or conjugation, factors that researchers using SPR, TINS, and WAC must consider. On the other hand, the fluorescence signal relies on native amino acid residues (tryptophan in the examples here), which can be obscured by many compounds. Also, in its current incarnation MST doesn’t appear particularly high-throughput, though it also doesn’t use much protein

Still, this seems like a pretty cool approach. I’ve started seeing NanoTemper at more conferences (such as FBLD 2012), so hopefully you will have a chance to check them out and let us know what you think.

Fragment linking, enthalpy, and entropy: not quite so simple

The strategy of fragment linking dates to the origins of fragment-based lead discovery. The idea that two low affinity binders can be linked to produce a more potent molecule is based on the theory that the binding energies of linked fragments will at least be additive. Indeed, sometimes superadditivity can be observed; in those cases, the binding energy of the linked molecule is considerably better than the sum of the binding energies of the separate fragments. The most common explanation for this is that linking two fragments “pre-pays” the entropic cost of binding to the protein; rather than two fragments locking into fixed binding modes, only a single linked ligand pays this entropic penalty. This makes sense intuitively, but is it correct?

An early example of fragment linking was reported by Abbott researchers in 1997: two fragments that bound to the matrix metalloproteinase stromelysin were linked together to give a molecule that bound about 14-fold more tightly than the product of the affinities of the two fragments. Thermodynamic analyses were conducted to explore the roles of entropy and enthalpy, but these were complicated by the fact that one of the fragments contained an acidic phenol that was removed in the course of linking. In a new paper published in Bioorg. Med. Chem. Lett., Eric Toone and colleagues at Duke University have re-examined this system.

The researchers dissected several of the originally reported linked molecules into component fragments and examined their thermodynamics of binding using isothermal titration calorimetry. All of the experiments produced similar results; a particularly illustrative example is shown in the figure, in which a single bond in compound 1 was conceptually broken to yield component fragments 5 and 8.

As the researchers note, weirdly, the “favorable additivity in ligand binding – that is a free energy of binding greater than the sum of those for the constituent ligand fragments – is enthalpic in origin,” not entropic. It is not clear why this is the case, but what is clear is that the results are completely different from those obtained by Claudio Luchinat and colleagues on another matrix metalloproteinase. In that report, the enhanced affinity of the linked molecule was entirely entropic in origin, as might be expected. So what’s going on here?

One clue is provided by Fesik and colleagues in their original analysis of their stromelysin inhibitors. They noted that, when fragment 8 (acetohydroxamic acid) was added to the protein, biphenyl ligands similar to fragment 5 bound considerably more tightly than when fragment 8 was not present. In other words, the ligands displayed cooperative binding even when they were not covalently linked, probably due to non-covalent interactions between the two bound ligands or possibly to changes in protein structure and dynamics.

It is easy to assume that two ligands bind independently to two sites on a rigid protein, when in fact proteins are anything but rigid, and the addition of one ligand to a protein can dramatically change its properties. Thermodynamics measures changes in the entire system, not just the ligands, and if the protein changes upon ligand binding things can quickly get complicated. As Fesik and coworkers noted:

The observed cooperativity between the two ligands is a factor that should be considered when optimizing compounds for binding to nearby sites, since a portion of the binding energy is due to the cooperativity rather than interactions between the ligands and the protein.

All of which is to say that we remain woefully ignorant of the forces driving ligand binding, let alone fragment linking. But assessing how much better (or worse) a linked molecule binds than its component fragments can still be a useful exercise to guide optimization, even if the thermodynamic origins of the effects are unclear.

Fragments vs Hepatitis C NS3 protein

Hepatitis C is the target of numerous drug discovery programs, so it was only a matter of time before fragments were used to tackle it. In a paper just published online in Nature Chemical Biology, Harren Jhoti and colleagues at Astex Pharmaceuticals describe a particularly elegant example of fragment optimization against an unusual allosteric site.

The NS3 protein contains two functional domains, both of which are essential for viral function. Two drugs have recently been approved that target the serine protease domain, and the helicase domain has also been extensively studied. However, much of the effort has focused on truncated proteins consisting of only one domain without the other; in the cell, the protein remains intact and also complexes with another viral polypeptide, NS4a. It was this full-length protein complex that the Astex researchers went after.

In the full-length form of the protein, the protease is auto-inhibited by the C-terminus of the helicase domain, which binds in the active site of the protease domain. A crystallographic fragment screen identified fragments such as Compound 2 (below) that bind in a pocket near the protease active site, and the researchers wondered whether these might trap the protein in the inactive state. Compound 2 binds in a hydrophobic pocket and weakly inhibits the protease activity of the full-length protein. Initial optimization focused on improving hydrophobic contacts and restricting the conformational mobility of the molecules, leading to compound 4, with low micromolar activity. Further optimization to pick up additional polar contacts led to compound 6, with nanomolar biochemical and cell-based activity.

As is often (though not always) the case in fragment optimization, the optimized compound 6 shows a similar binding mode to the initial compound 2.

If the compounds truly act by keeping the NS3-NS4a protein complex in the “closed,” or auto-inhibited state, this should be detectable by various biophysical measurements, and in fact sedimentation velocity analysis, size exclusion chromatography, and dynamic light-scattering were all consistent with this mechanism.

The binding energetics of the identified molecules were also studied by isothermal titration calorimetry. In general, enthalpy played the major role in improving free energy of binding, with entropy playing an increasingly deleterious role as affinities improved. Though interpreting thermodynamic data is tricky, the contribution of enthalpy versus entropy is consistent with the molecules locking the protein into a single conformation, thereby decreasing its conformational freedom (and entropy).

This paper is a beautiful illustration of anti-reductionism: the compounds are not active against the isolated protease domain commonly studied; only by looking at the full-length protein complex could the allosteric site be identified. Molecules such as compound 6 should prove useful reagents for exploring, and ultimately preventing, hepatitis C replication.

Fragment events in 2012 and 2013

Hard to believe that 2012 is three-quarters behind us, but if you find yourself in the Southern Hemisphere there is still one more notable event this year, and 2013 is starting to take shape.

2012

November 15-16: Fragment-Based Drug Design Down Under will be held in Melbourne, Australia. They say that the surface plasmons are reversed down there!

2013

March 4-5: Feeling bummed that you couldn't make it to FBLD 2012? Can't wait till FBLD 2014? Then check out Fragments 2013, the 4th RSC-BMCS Fragment-based Drug Discovery meeting, which will be held at the Harwell Science and Innovation Campus near Oxford, UK.

April 16-18: Cambridge Healthtech Institute’s Eighth Annual Fragment-Based Drug Discovery will be held in San Diego. You can read impressions of this year's meeting here, last year’s meeting here and 2010’s here.

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

How Big is the Library

One of the things I did at FBLD2012 was count how many fragments were in the libraries people were reporting on.  Now, we all know that libraries are dynamic and not static, but I thought the numbers would give an idea of how big libraries are across the industry (and academia).  I may have missed some, but these are the ones I caught:

Pfizer: 3000-5000 cpds @30mM in DMSO with 50-100 mg of powder in supply
GSK: 2400 cpds @100mM in DMSO.  ~14 HA average for their "regular" library
          ~1000 cpds for NMR purposes
         ~1500 19F cpds
Selcia (Not sure if this is their library or the one they ran for a client, somebody with better notes?)
          1500 cpds @30mM DMSO, purity >95%
Roche: 6900 cpds
Elan: 5260 cpds
Heptares: 4500 cpds, average HAC 16
Amgen: 1200 19F cpds
ZoBio: 1500 cpds
CSIRO: 480 cpds
Emerald ~2000 cpds
U Cambridge: 1300 cpds
Genentech: 2500 cpds
Ariad 735 Ro3 compliant cpds
Vanderbilt: 11000 cpds
Abbott: ~10000 cpds
Merck: ~10000 cpds
Plexxikon: 20000 scaffolds (explicitly differentiated from fragments in the talk)

FBLD 2012

Fragment-based Lead Discovery 2012 concluded today in San Francisco. This is the fourth in an illustrious series of conferences that started in San Diego in 2008, moved to York in 2009, and then to Philadelphia in 2010. These meetings set a high bar in terms of quality, and I think it’s fair to say that FBLD 2012 has lived up to its predecessors. With 39 talks, 16 vendors, over 50 posters, and close to 250 attendees I won’t attempt a comprehensive summary, so please weigh in if you were there.

Pete Kenny is always ahead of the curve, and so despite his absence from the conference he did put together a nice preview. He worried that the big pharma talks might be “strategy-heavy and results-light,” but happily this was not the case. In particular, Jane Withka gave a very detailed talk on Pfizer’s carefully constructed fragment library (see also here). One statistic that caught my eye is that, after screening this library against 21 proteins, 44% of the 2500+ fragments have hit at least one target – considerably higher than the 33% that has been seen in several organizations. Whether this is because of a better library or simply more screens remains unclear.

Fragment validation – or the lack thereof – and fragment promiscuity were also frequently recurring topics. Peter Kutchukian from Novartis observed that frequent hitters are not always problematic: fragments that hit multiple targets were actually more likely to produce co-crystal structures than more selective fragments.

There were lots of cool approaches for finding fragments; one particularly impressive advance was presented by SensiQ in a workshop before the main conference. Their surface plasmon resonance (SPR) instrument operates as expected; what sets it apart is a cunningly designed injection method in which the sample compound flows through a long capillary before entering the flow cell. A concentration gradient forms within the capillary, and by controlling this gradient the user can run a full dose-response curve over several orders of magnitude without having to pre-dilute the sample. Instead of doing a primary screen at a fixed concentration and then following up on active compounds, dissociation constants can be determined directly from a primary screen.

Protein flexibility is a subject dear to my heart, but not typically observed in biophysical fragment-finding techniques besides crystallography and protein-detected NMR. Josh Salafsky described Biodesy’s second-harmonic generation technology to specifically find molecules that cause conformational changes. More surprisingly, Beactica’s Helena Danielson argued that SPR could be used to observe conformational changes in membrane proteins. Although I’m no SPR expert, I've only seen the technique being used to detect changes in mass, so it will be fun to see how this develops.

Fluorine NMR looks like it’s finally coming into its own; Brad Jordan discussed how this has become a standard screening technique at Amgen, and both Nino Campobasso (GlaxoSmithKline) and Stephan Zech (Ariad) mentioned that their companies are using it too.

In addition to the method talks there were plenty of hit-to-lead and success stories, some of which have been covered on Practical Fragments, and some of which will be as the publications come out.

If you weren’t able to make it this year, FBLD 2014 is tentatively planned to be held in Basel, Switzerland. And if you can’t wait that long for your next fragment fix, there is at least one more relevant event this year and several already taking shape for 2013 – details to come shortly.

Fragments vs DDAH – covalent and noncovalent

Using biochemical assays to find fragments sidesteps the need for expensive biophysical instruments, but is fraught with difficulties. In a recent paper in Bioorg. Med. Chem., Thomas Linsky and Walter Fast at the University of Texas, Austin, describe how they used functional screening to discover fragment inhibitors of dimethylarginine dimethylaminohydrolase (DDAH), an enzyme involved in nitric oxide production.

The researchers used a 4000-member fragment library from ChemBridge, which they screened against two different isoforms of DDAH at 0.1 mM in a biochemical screen. This resulted in 79 hits against the human enzyme and 44 hits against a bacterial (P. aeruginosa) DDAH, 101 in total. 66 of these were then repurchased along with 41 analogs, and tested in a completely different biochemical assay against human DDAH-1; 31 showed >20% inhibition at 0.4 mM, including only 22 of the original hits, suggesting that many of the initial hits were indeed false positives. Further studies showed that 5 of the 31 compounds interfered with this secondary assay (ie, they showed “inhibition” even in the absence of enzyme), suggesting that they were false positives in both the primary and secondary biochemical assays.

DDAH contains an active-site cysteine, and the researchers wanted to exclude molecules that might be generically reactive, so they incubated the remaining 26 compounds with the low-molecular weight thiol glutathione and then retested them; this eliminated another 21 compounds.

Finally, the remaining 5 compounds were examined by mass-spectrometry, and one of these turned out to be a compound other than what was listed on the bottle! This left just 4 legitimate fragment hits.

Two of these compounds were 4-halopyridines, which, although not generally reactive with thiols, could covalently modify the active site cysteine of DDAH (see here for more details). The other two compounds were reversible, competitive inhibitors of human DDAH-1. Although low affinity (Ki values of 0.8 and 1.7 mM), they had respectable ligand efficiencies (0.38 and 0.29 kcal/mol/atom, respectively).

Interestingly, when Linksy and Fast retrospectively analyzed where the four validated fragments came from, they found that only the two halopyridines were detected in the primary screen; the two reversible fragment hits had been purchased for the secondary round of screening as analogs of primary hits.

This is a richly detailed and well-executed example of fragment-based screening in academia. It demonstrates once again that high-concentration biochemical screens can be used to find fragments, but be prepared to wade through a lot of junk: only about 2% of the original hits proved to be legitimate (see here for similar results from Vernalis on a different target). It also illustrates the utility of exploring analogs of initial fragment hits; in this case, even though most of the primary hits didn’t hold up, they nonetheless led to new fragments. Of course, this does raise the question of how spurious primary hits can lead to genuine inhibitors – what do you think?