Poll: fragment linking and growing

A seminal paper in the fragment field is the 1996 SAR by NMR report in which two fragments were linked together. In theory, linking fragments can give a massive improvement in affinity beyond simple additivity, but in practice this is rare. The challenges of linking were not obvious in the early days, and led to much hair-pulling. Indeed, partially for this reason, Teddy has asserted that the 1996 paper is not just the most impactful paper in the field but also the most destructive.

Nonetheless, there are successful examples of linking, particularly for challenging targets (such as here and here). So how often does it really work?

Our latest poll has two questions: one on fragment linking, the other on fragment growing (see sidebars on right side of page). Tell us whether, in your experience, fragment linking didn’t work at all, worked marginally (ie, perhaps a modest boost in potency), worked OK (perhaps additivity), or worked well (synergy). You can vote multiple times, so if you’ve worked on multiple projects with different outcomes, please vote early and often. We’re asking the same questions for fragment growing since these two strategies are often compared.

Admittedly the categories are somewhat fungible: one person’s “OK” may be another person’s “well,” and some may see merging where others see linking. Still, hopefully we’ll get enough votes to discern some trends.

You Probably Already Knew This...

Academics can spend time and resources doing, and publishing, things that people in the industry already "know".  This keeps the grants, the students, the invitations to speak rolling in.  It also allows you to cite their work when proposing something.  This is key for the FBHG community.  There are many luminaries in the FBHG field, and we highlight their work here all the time. Sometimes, they work together as a supergroup.  Sometimes, Cream is the result.

Brian Shoichet and Gregg Siegal/ZoBio have combined to work together.  In this work, they propose to combine empirical screening (TINS and SPR) with in silico screening against AmpC (a well studied target).  They ran a portion of the ZoBio 1281 fragment library against AmpC.  They got a 3.2% active rate, 41 fragments bound.  6 of these were competitive in the active site against a known inhibitor.  35 of 41 NMR actives were studied by NMR; 19 could have Kds determined (0.4 to 5.8 mM).  13 fragments had weak, but uncharacterizable binding; 3 were true non-binders. That's a 90% confirmation rate.  34 of 35 were then tested in a biochemical assay.  9 fragments had Ki below 10 mM.  Of the 25 with Ki > 10mM, one was found to bind to target by X-ray, but 25A from the active site.  They then did an in silico screen with 300,000 fragments and tested 18 of the top ranked ones in a biochemical assay.  

So, what did they find? 

"The correspondence of the ZoBio inhibitor structures with the predicted docking poses was spotty. "  and "There was better correspondence between the crystal structures of the docking-derived fragments and their predicted poses."

So, this isn't shocking, but it is good to know.  This is also consistent with this comment.  So, the take home from this paper is that in silico screening can help explore chemical space that the experimentally much smaller libraries miss.  To that end, the authors then do a a virtual experiment to determine how big a fragment library you would need to cover the "biorelevant" fragment space [I'll save my ranting on this for some other forum].  Their answer is here [Link currently not working, so the answer is 32,000.]

Getting misled by crystal structures: part 4

A picture is worth a thousand words, but words can mislead as easily as inform. So it is with crystal structures, as Charles Reynolds discusses in the July issue of ACS Med. Chem. Lett. We’ve touched on this issue before (for example, here and here), but this is a nice update.

He starts with a cringe-worthy catalog of horrors found in the protein data bank (pdb):

Just to give a few examples: 1xqd contains three planar oxygens as part of a phosphate group; 1pme features a planar sulfur in the sulfoxide; 1tnk, a 1.8 Å resolution structure, contains a nonplanar tetrahedral aromatic carbon as part of a substituted aniline; and 4g93 contains an olefin that is twisted nearly 90° out of the plane.

Of course, with 100,000 structures, it is inevitable some dross will slip through, but Reynolds argues that around a quarter of all co-crystal structures contain errors so severe that they could lead to misinterpretations.

Why is the situation so dire? Reynolds suggests a number of reasons. First, there’s the push for quantity over quality: fully refining a structure may not be as valued as solving a new one. Second, small molecules comprise only a small portion of the overall structure and thus make minimal contributions to the metrics crystallographers use to assess quality during refinement. Third, with the exception of very high resolution structures, the quality of the electron density maps are such that properly placing the small molecule requires a fair bit of modeling. This challenge is complicated by the fact that most crystallographers were not trained as chemists and thus may not immediately recoil from a tetrahedral aromatic carbon atom. Also, much of the off-the-shelf software used for refining structures is not optimized for small molecules.

Nonetheless, there is good software available that properly accounts for small molecules. Hopefully publicizing errors will encourage more crystallographers to use it. In the meantime, caveat viewor!

EthR revisited: fragment growing this time

A few months ago we described a fragment linking approach against the protein EthR, a transcriptional repressor from Mycobacterium tubercuolosis responsible for resistance to the second-line tuberculosis drug ethionamide. In a new paper in J. Med. Chem., a different team led by Benoit Deprez and Nicolas Willand (Université Lille Nord de France and Institut Pasteur) describe work on the same target using fragment growing and merging.

The researchers started with a fragment (compound 3) they had previously made as part of an in-situ click chemistry effort. A thermal shift assay revealed that this compound marginally stabilized EthR. More convincingly, it displayed mid-micromolar inhibition of EthR binding to DNA, with respectable ligand efficiency.

Interestingly, when compound 3 was cocrystallized with the protein, it bound at two different locations within the binding site. (In the work we highlighted previously this year, a different fragment also bound at two sites, and in that case the researchers linked fragments bound at each site to create a tighter binder.) In the current paper, the researchers focused on fragment growing.

Compound 3 is a sulfonamide that can be readily constructed from amines and sulfonyl chlorides, and the researchers started by constructing a 976-member virtual library of larger sulfonamides. These were then screened in silico against the protein, and many of the top-scoring hits resulted from an isopentylamine building block (such as compound 4). Ten of these were made and tested, and indeed, compounds 4 and 8 were more effective than compound 3 at stabilizing EthR in the thermal shift assay. Moreover, not only did compound 8 show low micromolar activity in the DNA-binding assay (IC₅₀ = 4.9 µM), it also showed low micromolar activity in sensitizing M. tubercuolosis to ethionamide (EC₅₀ = 5.7 µM).

Crystallography of compound 8 bound to EthR revealed that the isopentyl substituent was binding in a hydrophobic part of the pocket, and adding a few fluorine atoms (compound 17) gave a satisfying increase in potency as well as solubility. Replacing the sulfonamide with an amide (compound 19) further improved potency.

The researchers also made a couple compounds in which a second copy of compound 3 was merged with compound 19, and although this approach did produce a compound with nearly the same potency, it was also larger and less soluble.

This team has been pursuing EthR for some time, and they were able to use information from previous structures both in the computational screening as well as in the optimization. In that sense, this is an example of fragment-assisted drug discovery. It is also another nice example of fragment work in academia.

Halogen Hydrogen Bonding...Designable or Not?

The use of brominated fragments for X-ray screening is well known; it was the basis for former company SGX (now part of Lilly).  The purported advantage of brominated fragment is that you can identify the fragment unambiguously using anomolous dispersion.  In this paper, they are focused on using fragments to identify surface binding sites on HIV protease.  Prior work has focused on creating a new crystal form (complexed with TL-3, a known active site inhibitor) that has four solvent accesible sites: the exosite, the flap, and the two previous identified sites.  They took 68 brominated fragments and soaked these crystals: 23 fragments were found.  However, most of these actives were uninteresting.  Two compounds were found to be interesting, one bound in the exosite and one in the flap site. 

So, what's interesting in this paper?  Well, they (re)discover that brominated fragments can bind all over with a variety of affinities.  However, the bromine allows you to unambiguously identify those fragments through anomolous dispersion.  This is NOT interesting.  They discover that although it is a subject of much debate lately: specific interactions of the ligand with the target dominate the "bromine interaction".  This IS interesting.  They do not discuss this in much detail, but their grand extrapolations of this method to general applicability I don't buy. 

 I think the key take away from this paper is whether the halogen hydrogen bond undesignable and just a subject of serendipity? 

Fragments vs Dengue virus helicase and methyltransferase

Dengue fever is a disease whose nastiness is hinted at by its common name, “breakbone fever”. The eponymous virus relies on two viral proteins, a helicase (Hel) and a methyltransferase (MTase), for replication. In a recent paper in Antiviral Research, Karine Barral and coworkers at Aix-Marseille Université use fragment-based methods to tackle both of these proteins.

The researchers used the 500 compound Maybridge fragment library (2009 edition) and performed a biophysical cascade, starting with a thermal shift assay at 2 mM of each fragment. Hits were defined as fragments that stabilized the protein by at least 0.5 °C; there were 36 hits against Hel and 32 against MTase, with 6 in common.

Both Hel and MTase are amenable to crystallography, so the hits against each protein were soaked into crystals. Unfortunately, none of these yielded structures for Hel. Critics of thermal shift assays could argue that this is yet another example of false positives, a possibility the researchers consider. That said, 11 of the fragments inhibited Hel by at least 25% in biochemical assays (at 2 mM fragment), though none inhibited greater than 50%.

The results against MTase were more salubrious: 7 fragments produced structures, for a success rate of 22%. Interestingly, these fragments bound to 4 distinct sites on the protein. Two different enzymatic assays were used to test all the fragment hits, and many of them showed activity in one or both. A few fragments – including 5 of those characterized crystallographically – were sufficiently active that IC₅₀ values could be determined, though these were mostly millimolar.

My one quibble is that the authors state that “to our knowledge, only one example of random FBS has been conducted on an RNA virus target.” This ignores some beautiful work from both Roche and Astex on Hepatitis C, another virus that carries its genome as RNA. Still, it is fair to say that fragments could play a larger role in tackling infectious diseases.

At the last CHI FBDD conference, Rod Hubbard noted that, as more academics enter the fragment field, we will see more publications describing fragment hits against tough targets. The next steps – taking a fragment to a nascent lead – are often harder to resource in an academic environment. Still, it’s good to see these initial studies. At the very least, they go some way to addressing the question of whether the targets are ligandable.

Who's Reviewing this Crap?

Dan and I teach a short course on FBDD; next chance to catch a version in October.  My favorite part of Dan's section is when he goes off on PAINS and the continuing pollution of the literature.  Rhodanines in particular get Dan's dander up.  I often come off as anti-academic because many of their "drug discovery" papers are crap.  Or they claim something is a lead without it being one.   I think it is time to start PAINS shaming these papers.  For those of you not intimately familiar with "shaming" it is very popular with dog shaming.  Well, here is our first PAINS Shaming.  This paper (pointed out by Matt Netherton at B-I, thanks Matt!) unabashedly points out the offensive molecule as a rhodanine. In the article, they point out:

What is particularly interesting about the most active species investigated here is that it has a structure that is very similar to that found in the drug epalrestat, an aldolase reductase inhibitor that is used to treat diabetic neuropathy, and is approved for clinical use in Japan, China, and India. This is encouraging because rhodanines as a class are known to often have activity in widely different assays, and indeed computer programs such as PAINS categorize, [our compounds] (as well as epalrestat) as possible “pan assayinterference compounds”. This can mean that the compounds cause false positives in assays, or that they may be multitarget inhibitors. In some cases multitargeting may be undesirable; however, in the context of anti-infective development, multitargeting is expected to increase efficacy as well as decrease the possibility of resistance development, both very desirable features.

  I am sorry, but this is exactly the Underpants Gnomes business plan. All I can say after reading this is who's reviewing this crap and saying it is all right?

Metric poll results: plus ça change…

Of all the topics on Practical Fragments, none have generated as much Sturm und Drang as ligand efficiency (LE) and related metrics. But how are researchers actually using these? The results of our latest poll are in, and it appears that ligand efficiency is still widely used.

Lipophilic ligand efficiency (LLE or LipE) has made significant strides in the past 3 years, but despite the proliferation of other metrics none of them have really caught on. (Note that we didn’t ask about enthalpic efficiency or SILE in 2011.)

So at least for now – and while fully acknowledging that all metrics are simplifications – I think we can conclude that most readers of this blog find LE and LLE both useful and sufficient for their needs.

(Full technical disclosure: the 2011 poll was run in Blogger, but the polling feature is buggy and has a tendency to lose votes. This poll was run using Polldaddy, but unfortunately the free version does not reveal how many individuals voted. Since respondents could choose more than one metric, the number of voters is impossible to determine from the results. I tried to account for this upfront by including a “Please Click this Box” to tally individuals, but only 77 people did so, less than the 91 people who voted for LE. I thus had to guess at the total number of respondents. The most conservative choice was to add 91 (the largest single category) to the 8 respondents who use no metrics, giving 99 independent voters. Obviously this might overstate the apparent percentage of people who use any given metric, but as 86% of respondents used LE in 2011 the effect is probably small.)

Fragment Events in 2014 and 2015 - midyear update

The year is nearly half over, but there are still some great events ahead, and 2015 is already shaping up nicely!

2014

July 19-22: Zing conferences is holding its first-ever Fragment Based Drug Discovery in Punta Cana, Dominican Republic. In addition to the amazing location there are more than two dozen great speakers, so definitely check this one out - there is still time to register by June 30.

August 10-14: The 248th ACS National Meeting will be held in the cool gray city of San Francisco, and there will be several presentations on fragments as well as what I believe will be the first session on PAINS.  

September 21-24: FBLD 2014 will be held in Basel, Switzerland. This marks the fifth in an illustrious series of conferences organized by scientists for scientists, the last of which was in San Francisco in 2012.  I believe this will also be the first major dedicated fragment conference in continental Europe. You can read impressions of FBLD 2010 and FBLD 2009. Also, if you're new to the field (or looking for a refresher) Rod Hubbard, Ben Davis, and I will teach an introductory workshop on Sunday, September 21.

October 8-10: CHI's 12th Annual Discovery on Target takes place in Boston, where it looks like there will be several talks relevant to readers of this blog. And on October 7, Teddy and I will be teaching a short course on targeting protein-protein interactions, which naturally includes fragments.

2015

February 17-18: Newly added! SELECTBIO is holding its Discovery Chemistry Congress in Berlin, Germany, with a number of talks on fragment-based lead discovery.

March 22-24: The Royal Society of Chemistry will be holding Fragments 2015 in Cambridge, UK, the fifth in the series organized by RSC-BMCS. You can read impressions of Fragments 2013.

April 21-23: CHI’s Tenth Annual Fragment-Based Drug Discovery will be held in San Diego. You can read impressions of this year's meeting here and here, last year's meeting here and here, the 2012 meeting here, the 2011 meeting here, and 2010 here. As this will be their ten-year anniversary, I think they're planning something big!

June 9-12: NovAliX will hold its second conference on Biophysics in Drug Discovery in Strasbourg, France. Though not exclusively devoted to FBLD, there is lots of overlap; see here, here, and here for discussions of last year's event.

December 15-20: Finally, the first ever Pacifichem Symposium devoted to fragments will be held in Honolulu, Hawaii. The Pacifichem conferences are held every 5 years and are designed to bring together scientists from Pacific Rim countries including Australia, Canada, China, Japan, Korea, New Zealand, and the US. There is lots of activity in these countries, and since travel to mainland US and Europe is onerous this should be a great opportunity to meet many new folks - in Hawaii no less! Abstract submissions will open in January.

Know of anything else? Add it to the comments or let us know!

Irreversible fragments

Last year we highlighted a paper in which fragments were attached to a “warhead,” a chemical group that could form a reversible covalent bond with cysteine residues in proteins. These fragments were then screened against a kinase to identify nanomolar inhibitors. In a new paper in J. Med. Chem., Alexander Statsyuk and coworkers at Northwestern University describe a similar approach using irreversible fragments. (See here for In the Pipeline’s take.)

As we noted last year, one of the nice things about using reversible warheads is the fact that you can run experiments under thermodynamically-controlled conditions. Indeed, this was the idea behind Tethering (here and here), in which a library of disulfide-containing fragments is allowed to equilibrate with a cysteine-containing protein; if a fragment has inherent affinity for the protein, the disulfide bond will be stabilized towards reduction and can be identified using mass spectrometry. (Full disclosure: I am an inventor of Tethering, and my company, Carmot Therapeutics, Inc., has an exclusive license to the intellectual property covering this and other technologies.)

Irreversible warheads operate at least partly under kinetic, rather than thermodynamic, control. For example, if all the fragments are extremely reactive, the protein will react with whichever fragment it happens to encounter first, regardless of whether the fragment has any inherent affinity for the protein.

Acrylamide moieties are able to form irreversible bonds to cysteine residues and are even starting to be found in drugs. In 2012, researchers at Imperial College London tested an acrylamide-containing analog of one of the (disulfide) hits from the original Tethering paper. This successfully labeled the target protein thymidylate synthase (TS), while several other acrylamide-containing molecules did not, and the fragment was selective for TS over two other enzymes with active-site cysteine residues.

Regardless of whether your warhead is reversible or not, it is important that different members of the library have similar reactivities: if the inherent reactivities of the fragments are different, it will be difficult to distinguish inherent binding energies from chemical reactivities. One of the problems with acrylamides is that subtle changes in chemical structures can have dramatic effects on intrinsic reactivities. The new paper compares rate constants of two acrylamides reacting with a low molecular-weight thiol, and finds that one reacts 2,000-fold faster than the other. The researchers tested three other classes of electrophiles – vinylsulfonamides, aminomethyl methyl acrylates, and methyl vinylsulfones – and found that these had narrower ranges of reactivity. They chose acrylates and built a set of 100 acrylate-modified fragments. Happily, when 50 of these were tested for reactivity, there was only a 2.4-fold difference between the most reactive and the least reactive members.

These 100 fragments were tested in pools of ten against the classic cysteine protease papain using mass spectrometry, and three hits were identified. Enzymatic assays revealed that these three hits were also irreversible inhibitors of the protein with respectable activities (k_inact/K_i = 0.46-1.2 M^-1s^-1), while a member of the library that did not label in the mass spectrometry assay was a weaker inhibitor (k_inact/K_i = 0.037 M^-1s^-1). Moreover, the papain hits did not label three other enzymes containing active-site cysteines, though these enzymes could be modified with other fragments in the library.

In the case of Tethering, the disulfide linker was a means for finding fragments; when these fragments were developed further, the disulfide linker was replaced. However, given the renewed interest in covalent drugs, some warheads might be able to be retained. It will be fun to see how these types of strategies develop.

Liberte, egalite, fraternite! with Lemonade

Ligand efficiency has a long and glorious history in FBDD, discussed in depth.  Most recently, it was the subject of a LONG discussion.  Since its inception, FBDD has tried to fit in, be like the cool kids as it were.  "Regular" medchem had the Rule of 5, so Astex gave us the Rule of 3, aka the Voldemort Rule.  The principles of FBDD just make sense, but overturning entrenched dogma is hard.  So, simple metrics were devised to explain to people why smaller is better.  Groups tried to create better and more predictive metrics.  Well, we can add lipophilicity, or log D, or make it empirical.  The more complicated, obviously the better it is.  The search for a GUM (Grand Unifying Metric) is ongoing.  

Then Schultz had to come along and assault the Bastille.  So, to the walls Men!  Defend the King! And so he was, but there was no peace.  But trouble was brewing and now another assault has come.  Peter Kenny and his compatriots have taken another shot at the King, with the completely unprovocative title: "Ligand Efficiency Metrics Considered Harmful".  The paper reviews current LEMs (Ligand Efficiency Metrics), and proposes a better approach.  As Pete has said at length, one of his major problems is with the choice of arbitrary assumptions of standard state, as he laid out in this comment.  Of course, the real issue is the intercept and what that means.

What are the key problems the authors point out?

• Correlation Inflation
• Defining assumptions are arbitrary
• Scaling (dividing measured activity by physicochemical property) and offsetting (subtracting physicochemical property from measured activity) are used but no one has ever said why you do one or the other for a given LEM
• Scaling assumes a linear relationship between activity and property (zero intercept)
• Offsetting implies the LEM has a unit slope
• LEMs are not quantitative, but are presented as such.
• Units matter!  ΔG = RTln(Kd/C) where C= concentration of standard state.  Substitution of IC50 in this equation is easy to do, but not at all correct.  A more subtle example of the problem is provided by the definition of LLEAT as the sum of a (dimensionless) number and a quantity with units of molar energy per heavy atom.[Click to embiggen.]
  
  

 So, this is a fun paper to read.  I highly recommend it.  The authors go into exquisite detail of what each and EVERY factor that goes into a LEM means and how it is used, and more importantly how it SHOULD be used.  The first conclusion is that there is no basis in science for any of the assumptions related to scaling or offsetting.  They recommend that data be modeled to determine a trend.  They argue that only in this way can a given compound be determined to have beaten the trend.  The King is dead; long live the King?

They are right, of course, and IMNSHO it doesn't matter.  I still aver that ligand efficiency metrics are useful.  I can measure accurately with a meter stick that is only 95 cm, as long as I know it is 95cm.  The same thing with any LEM; understand its limitations and use it appropriately.  And remember, its a guide, not a hard and fast rule. 

Pete and colleagues set up the obvious acronym, LEMONS (Ligand Efficiency Metrics of No Substance).  And when life gives you LEMONS, you make LEMONADE (Ligand Efficiency Metrics with No Additional Determinate Evaluation). 

Fluorinated Fragments vs G-quadruplexes

Recently we highlighted an example of fragment-based ligand discovery against a riboswitch. Of course, RNA can form all kinds of interesting structures, and in a new paper in ACS Chem. Biol. Ramón Campos-Olivas (Spanish National Cancer Research Centre) and Carlos González (CSIC, Madrid) and their collaborators describe finding fragments that bind G-quadruplexes.

G-quadruplexes, as their name suggests, consist of groups of four guanine residues hydrogen bonding to one another in a planar arrangement. These individual tetrads then stack on top of one another. They can form in guanine-rich regions of RNA or DNA. Most famously, G-quadruplexes are found in telomeres at the ends of chromosomes. However, they are also found in telomeric repeat-containing RNA (TERRA), and are required for cancer cells to proliferate indefinitely.

The researchers used ¹⁹F-NMR screening to identify fragments that bound to an RNA containing 16 (UUAGGG) repeats (TERRA₁₆). ¹⁹F-NMR is a technique about which Teddy waxes rhapsodic, and in this incarnation involves examining the NMR spectra of fragments in the presence or absence of TERRA₁₆. Fragments that bind to the RNA show changes in ¹⁹F spin relaxation, resulting in broader, lower intensity signals. The library consisted of 355 compounds from a variety of sources, and although most of them were fragment-sized, a couple dozen had molecular weights above 350 Da.

The initial screen produced a fairly high hit rate (20 fragments), of which seven were studied in detail. Standard proton-based STD NMR confirmed the ¹⁹F-NMR results. The researchers then turned to a shorter RNA containing only two repeats (TERRA₂); this RNA sequence dimerizes to form a G-quadruplex. All seven fragments stabilized this complex against thermal denaturation, consistent with binding. Six of the fragments also induced changes to the ¹H NMR spectrum of TERRA₂, though one also caused general line broadening that could indicate aggregation. For the well-behaved fragments, dissociation constants (K_D) were determined by measuring changes in chemical shifts with increasing concentrations of ligand. K_D values ranged from 120 to 1900 micromolar, with modest ligand efficiencies ranging from 0.17-0.28 kcal/mol/atom.

Of course, selectivity against other nucleic acid structures is a major concern, so the researchers used ¹H and ¹⁹F NMR to assess compound binding to a tRNA, a DNA duplex, and a DNA analog of TERRA₂ also able to form a G-quadruplex. Aside from the putative aggregator, none of the seven compounds bound tRNA, and only two (including the aggregator) bound duplex DNA. However, all the compounds bound to the DNA G-quadruplex. Interestingly though, the DNA sequence used can form two types of G-quadruplexes in solution (parallel or antiparallel), whereas the equivalent RNA can only form a parallel dimer. In all cases the small molecules appeared to shift the equilibrium of the DNA to the parallel conformation, consistent with their initial identification as RNA binders.

Last year we highlighted another paper in which fragments were identified that may bind to a different DNA G-quadruplex. It would be interesting to functionally compare these two sets of hits. For example, do the hits identified initially against the DNA G-quadruplex also bind RNA G-quadruplexes? Of course, as with the riboswitch effort, there is a long way to go. It should be an interesting journey.

Fragments vs riboswitches

Most of fragment-based lead discovery – indeed, most of lead discovery – is directed against proteins. However, RNA is also an essential biomolecule, and in a new paper in Chem. Biol. Adrian R. Ferré-D’Amaré and colleagues at the National Heart, Lung, and Blood Institute, along with collaborators at the University of Cambridge and the University of North Carolina Chapel Hill, demonstrate that fragments can potentially make an impact here as well. This is the first example I know of where crystallography has been used to assess fragment hits against RNA molecules.

The story begins several years ago, when Chris Abell and colleagues became interested in the TPP riboswitch thiM. This is a bacterial stretch of RNA that binds to the essential cofactor thiamine pyrophosphate (TPP). This binding causes a change in conformation that regulates protein translation; small molecules that interfere with this process could lead to new antibiotics. In 2010 the researchers described a fragment screen using equilibrium dialysis, in which the RNA was added to one chamber along with radiolabeled thiamine, which binds with low micromolar affinity. This chamber was separated from another chamber containing fragments by a dialysis membrane permeable to small molecules and fragments but not to (larger) RNA. Fragments were screened in pools of five, and pools that caused displacement of radioligand were then deconvoluted to identify the active fragments. A total of 20 fragment binders were identified out of roughly 1300 tested.

WaterLOGSY NMR was used to confirm the binding of these 20 fragments to the riboswitch, and all of them were then tested using isothermal titration calorimetry, which yielded dissociation constants for 17 of them ranging between 22 and 670 micromolar. When tested against a different riboswitch, 10 of them appeared to be selective for thiM. The chemical structures of all of these were reported in 2011, along with some speculation as to how they might bind.

Of course, speculation is just that, and in fact fragment hits have been identified against RNA and DNA before. In the new paper the researchers use X-ray crystallography to actually determine the structures of several fragments bound to the riboswitch. This provides several interesting observations.

First, despite the different chemical structures of the fragment hits, all four of those whose structures were determined bind in the same region where the pyrimidine moiety of the natural ligand TPP binds. In fact, fragment 1 (magenta), which is essentially a fragment of TPP (green), almost perfectly superimposes on the corresponding moiety of TPP.

More strikingly, the co-crystal structures of each of the fragments bound to the riboswitch reveal that one of the guanosine residues (magenta stick in figure above) rearranges to fill the pocket that would otherwise be occupied by the pyrophosphate moiety of TPP (orange and red above). This occurs with fragment 1 as well as other fragments that do not resemble the natural ligand.

The researchers also took the useful step of solving the crystal structure of thiamine (cyan) bound to thiM. Since thiamine is intermediate in size between TPP and fragment 1, you might expect the structure to resemble one or the other, but as it turns out it binds in yet a third mode in which the pyrimidine ring no longer superimposes with the other two structures, nor does the guanosine residue rearrange to fill the pyrophosphate-binding pocket. This provides an interesting example of fragmenting natural products (TPP to thiamine to fragment 1). Although all of the molecules bind with high ligand efficiencies, it is unlikely that their binding modes could have been accurately predicted.

As the researchers note, the conformational shifts observed with these fragments could lead to antibiotics that selectively target an inactive form of the riboswitch. Although they’ve got a long way to go, it is fun to see folks applying FBLD to non-traditional targets.

Is that a Forest over there? The place with all the trees...

Sometimes, in the FBHG arena, we get deep into the weeds.  I am a simple man, with a simple understanding of FBHG.  There are a few tenets that I hold dear: Small, not very complex molecules have less of a chance of bad interactions.  You need high sensitivity (read:biophysical) methods to detect them robustly.  Libraries don't need to be bigger than a few thousand.  Now, this last one can be/has been/will be a subject of much debate.  There is no way to cover ALL of available chemical space and still run a screen in this lifetime.  However, I think most people agree that you can sample an "adequate" (that term being highly fungible) amount of available chemical space with a few thousand compounds.  Heck, if you have the resources to screen 10,000s of fragments, go right ahead.  

But, what molecules should be in your fragment collection?  To paraphrase Justice Brandeis, you know a good fragment when you see it.  I think every body could have the same 2000 or so fragments and each find their own success.  However, we are lucky that there are a metric boatload of commercial vendors of fragments, so we won't have to.  In this paper, the authors aim to determine what is currently available in commercial fragment space (emolecules) that covers known medchem space.  They use CHEMBL as the source of "medchem" space and analyzed the distribution of biologically active molecules and discovered property trends that differentiate active from inactive.  Here are there results: 

In their discussion, they state:

While many typical fragment libraries contain commercial fragments that mostly conform to the Ro3 physicochemical properties criteria, the chemical space these fragments represents may not entirely project to the leadlike and drug-like chemical space that are relevant to medicinal chemistry.

They realize that the overlap between fragment space and CHEMBL-medchem space do not overlap well.  In fact, almost half of the bioactive-derived substructures are not covered by commercially available fragments.  They propose that fragment libraries should focus on those structures represented by bioactive CHEMBL compounds.  They make some suggestions to expand fragment libraries, e.g. polar fragments that are not similar to commercial fragments and more 3D-arity.  However, there is a paucity of these molecules commercially available, so they suggest synthetic focus should be placed here.  

[Ed: Link to paper fixed.  Sorry about that.]

From substrates to fragments – or not

Recently we highlighted a paper in which enzyme substrates were deconstructed into component fragments and tested against an enzyme with unknown specificity. In a new paper in J. Am. Chem. Soc. a collaboration led by Karen Allen (Boston University), Frank Raushel (Texas A&M), and Brian Shoichet (UCSF) has performed a similar experiment to ask whether fragments could be used to identify substrates.

The researchers chose six enzymes from three different classes and collected various-sized fragments based on known substrates. These were then tested in functional assays to see whether they could be substrates or inhibitors. Stunningly, in most cases the fragments showed no activity against the enzymes; when activity was detectable, it was usually at least 100,000-fold lower than the natural substrate. Even subtle tweaks, such as removing a hydroxyl group, were enough to mess things up, as illustrated for adenosine deaminase (compare compounds 1 and 4). Breaking the substrate in two was sometimes better: compound 8 was turned over slowly by the enzyme, though its complementary fragment 9 had no effect on activity – positive or negative – when added to the assay along with compound 8 or the natural substrate.

Of course, functional assays are less sensitive than biophysical assays, but in the one case where the researchers tried soaking fragments into crystals of the enzyme they found that the fragments bound in a different manner than the substrate – echoing previous work deconstructing synthetic inhibitors of protein-protein interactions.

As the authors note, the remarkably sharp structure-activity-relationships (SAR) observed here could reflect a fact of nature: most enzymes need to be highly selective for their substrates to avoid mucking up cellular metabolism.

Moreover, the notion that two fragments, when properly linked together, can bind more tightly than the sum of their individual binding energies has been a primary motivator behind fragment-based lead discovery for more than 30 years. In a sense, this paper illustrates this principle in reverse. Indeed, it is possible for the energy gained by linking two fragments to exceed the binding energy of an individual fragment.

This is a nice study from which we can draw two lessons, one pessimistic, the other optimistic. On the down side, we are unlikely to be able to use fragments to predict the natural substrates of uncharacterized enzymes, at least on a general basis. As noted previously, this is not surprising: the concept of molecular complexity predicts that fragments should be fairly promiscuous, and we’ve seen time and again that fragment selectivity is not necessarily maintained during optimization.

On the positive side, this study beautifully illustrates that it is possible to achieve massive enhancements in affinity with relatively small changes. Beyond just the magic methyl effect, we’ve got the magic hydroxyl effect, the magic thiophene effect – heck – the magic fragment effect. Of course, these are retrospective analyses, and it’s easier to break things than make them. That said, folks at Astex demonstrated that it is possible to improve the affinity of a millimolar fragment a million-fold by adding just six atoms. Perhaps such opportunities are more general than we have previously dared to dream.

Enthalpy arrays revisited: PDE10A

Two years ago we highlighted enthalpy arrays: very tiny temperature sensors that measure the heat generated during the course of an enzymatic reaction. Molecules that compete with a substrate will alter the kinetics of the reaction and can thus be identified as inhibitors. In the original paper a number of fragment hits were identified against the phosphodiesterase PDE4A, but unfortunately none of these could be structurally characterized. In a new paper in J. Biomol. Screen. Michael Recht and colleagues at Palo Alto Research Center have teamed up with Vicki Nienaber and colleagues at Zenobia to apply enthalpy arrays to the neurological target PDE10A, and this time they were able to obtain numerous crystal structures.

The researchers started by confirming that literature reference compounds behaved as expected. Next, they screened all 16 of the PDE4A hits against PDE10A, including several that were quite weak against PDE4A itself. All of these were active in the enthalpy array assays, with Ki values ranging from 94 to 1400 μM and good ligand efficiencies. In fact, most of the fragments were more potent against PDE10A than the phosphodiesterase against which they were original screened – which perhaps touches on the question of fragment selectivity.

The researchers also screened an additional 85 fragments at a concentration of 2 mM, leading to 8 more hits. All 24 of the hits were then soaked into crystals of PDE10A, yielding 16 crystal structures of bound fragments – a respectable 67% success rate. Interestingly, fragments that produced structures were more potent (average K_I = 590 µM) than those that didn’t (average K_I = 1000 µM), and this difference was statistically significant.

All of the fragments bound at the active site, and fragment growing was used to improve the affinity of two of the fragments. This led to low or sub-micromolar compounds, albeit with a loss in ligand efficiency. These more potent compounds were also selective for PDE10A over PDE4A, though solubility limits precluded testing at very high concentrations.

The paper frankly discusses some of the limits of using enthalpy arrays. For example, since the fragment should be present at a higher concentration than enzyme, very tight binders would require unfeasibly low enzyme concentrations. This limits the practical range of the technique to inhibitors with K_Is ranging from ~500 nM to 2 mM. Also, as Morgen G observed in a comment to the last post, this is more of a biochemical assay (monitoring the heat of an enzymatic reaction) rather than what most people think of when you say the word calorimetry (monitoring the heat of binding, as in the case of isothermal titration calorimetry). Still, enthalpy arrays seem pretty cool; hopefully folks will warm to them.

Fragment Library Vendors (2014 Edition)

We have been updating a lot of lists recently.  One that I think has changed significantly, is the fragment library vendor list, last updated in 2010. As Dan said four years ago, FOB Chris Swain has done a great job of curating who is selling what.  Instead of duplicating efforts, I will just focus on what has changed and making some comments.  I am not going to list companies that have libraries you can access, only those that sell outright their libraries. 

What are the keys for purchasing a good library?  I think minimally, purity and aqueous solubility should be experimentally tested and guaranteed.  I note those vendors who specifically point this out, but one should not assume that those who don't also don't have this data.  Comments can be sent directly to me or made below, and I will update this list.

Some general thoughts: 

• There is no special sauce.  Every library is good and will work for you. It's the choice of screen and how you prosecute it after that makes the difference.
• You don't need no stinkin' IP. 

3DFrag Consortium (New 2014):  I think this ran its course.  While I think by and large it had great ideas I don't think it ever truly answered the question "Do 3D fragments work better (in some areas)?"

Analyticon (New 2014): This is another example of fragments from nature.  The utility of these types of libraries are still up for discussion. 

Asinex: "Inspired by Nature" is its tagline.  However, they do have focused libraries, for such targets as PPIs.  They have 3159 in this library.. 

Chembridge: The collection is now 7000+ compounds (was 5000).  The guarantee greater than 90% purity, but nothing about solubility. 

Chemdiv (New 2014): Their collection is almost 14,000 fragments.

Enamine: More than doubled in size, from 12,000 to more than 28,000. 

Iota: I think they were the first to regularly use nPMI in their compound assesment.  You can only look at their library under CDA. 

Key Organics: They have quite a few specialized fragment libraries: CNS, self-assembly, brominated, fluorinated, and chiral cyclic molecules, in addition to their main libraries.  They guarantee 95% purity, 1mM aqueous solubility, solubility up to 200 mM in DMSO, and with almost no overlap with the Maybridge collections (68 compounds).

Life Chemicals:They are now up to 47,500 fragment molecules (less than 300 MW), of course only 31,000 of these exist, the other 16,000 can be made upon request.  They have 3900 19F fragments.  In terms of those that have experimental solubility, there are 8200.  However, 75% are soluble at 1mM, and 60% at 5mM in PBS.   So, always read the fine print.  They are also the vendor for the Zen-Life library, another library based on nature. 

Maybridge: The grandfather of them all.  30,000 fragments in total.  The 2500 Diversity collection is guranteed soluble at 200 mM in DMSO and 1mM in PBS.  The NMR spectrum is available, but only in organic solvent. It is available in many formats, from powder to DMSO-d6 solution. 

Otava: 8800 fragments in general.  800 19F fragments.  And 575 chelating fragments, if you want a warhead and all the issues they bring with.

Prestwick: 2200 fragments.

Timtec: No number available, but also can be shipped in DMSO solution. 

Vitas-M: The least helpful website out there.  It's Voldemort Rule compliant and available in multiple formats: mg, mcmol, sets, DMSO solution, dry film.

Zenobia:  Several different collections of very small fragments.

In defense of ligand efficiency – and poll!

Last year we highlighted a provocative article from Michael Shultz in which he took aim at the concept of ligand efficiency (LE). As we noted at the time, he raised some good points, and I am the first to argue that there is value in questioning widespread assumptions.

However, in addition to questioning the utility of LE, Shultz also questioned its mathematical validity. He repeated the attack earlier this year by asserting that ligand efficiency was a “mathematical impossibility.”

This is incorrect.

To set the record straight, Chris Murray (Astex), Andrew Hopkins (University of Dundee), György Keserü (Hungarian Academy of Sciences), Paul Leeson (GlaxoSmithKline), David Rees (Astex), Charles Reynolds (Gfree Bio), Nicola Richmond (GlaxoSmithKline) and I have written a response just published online in ACS Med. Chem. Lett. demonstrating that ligand efficiency is mathematically valid.

One of the criticisms of LE is that it is more sensitive to changes in small molecules (such as fragments) than in larger molecules. However, this is a property of any ratio, and we show that the same behavior applies to more familiar examples such as fuel efficiency: a few blocks of stop-and-go traffic has more of an effect on the overall fuel efficiency of a short trip than a long trip.

Of course, that’s not to say that ligand efficiency and other metrics are perfect or universally applicable; we discuss a number of situations where they may be more or less useful.

In this spirit, Practical Fragments is revisiting a poll from 2011 to see what metrics you use – please vote on the right-hand side of the page, and share your thoughts here. Note that you can vote for multiple metrics, and please check the last box (Polldaddy does not tally individual responses, so this box will track total number of voters to allow us to calculate percentage of respondents who use a given metric).

Keep the comments coming, and check back to see the poll results.

Biofragments: extracting signal from noise, and the limits of three-dimensionality

What does this protein do? Now that any genome can be sequenced, this question gets raised quite often. In many cases it is possible to give a rough answer based on protein sequence: this protein is a serine protease, that one is a protein tyrosine kinase, but figuring out the specific substrates can be more of a challenge. In a recent paper in ChemBioChem, Chris Abell and collaborators at the University of Cambridge and the University of Manchester attempt to answer this question with fragments.

The bacterium Mycobacterium tuberculosis (Mtb), which causes tuberculosis, has 20 cytochrome P450 proteins (CYPs), heme-containing enzymes that usually oxidize small molecules. Although some are essential for the pathogen, it is not clear what many of them do. The researchers used an approach called “biofragments” to try to pin down the substrate of CYP126.

The biofragments approach starts by selecting a collection of fragments based on known substrates. Of course, the specific substrates are not known, so in this case the researchers started with a set of several dozen natural (ie, non-synthetic) substrates of various other CYPs, both bacterial and eukaryotic. They then computationally screened the ZINC database of commercial molecules for fragments most similar to these substrates and purchased 63 of them. Perhaps not surprisingly given their similarity to natural products, these turned out to be more “three-dimensional” than conventional fragment libraries, as assessed both by the fraction of sp³ hybridized carbons and by principal moment-of-inertia.

Next, the researchers screened their fragments against CYP126 using three different NMR techniques (CPMG, STD, and WaterLOGSY). Since they were primarily interested in hits that bind at the active site, they also used a displacement assay in which the synthetic heme-binding drug ketoconazole was competed against fragments. This exercise yielded 9 hits – a relatively high 14% hit rate.

Strikingly, all of the hits are aromatic, and 7 of them could reasonably be described as planar. In other words, even though the biofragment library was relatively 3-dimensional, the confirmed hits were some of the flattest in the library! The researchers interpreted this to mean that “CYP126 might preferentially recognize aromatic moieties within its catalytic site,” but there could be something more general going on – perhaps aromatics are simply less complex, and thus more promiscuous.

Examining the fragment hits more closely, the researchers found that one of them – a dichlorophenol – produced a spectrophotometric shift similar to that produced by substrates when bound to the enzyme. This led them to look for similar structures among proposed Mtb metabolites. Weirdly, pentachlorophenol came up as a possible hit, and a spectrophotometric shift assay reveals that this molecule does have relatively high affinity for CYP126. Whether this is a biologically relevant substrate for the enzyme remains to be seen.

This is an intriguing approach, but I do have reservations. First, in constructing fragment libraries based on natural products, it is essential to avoid anything too “funky”. The Abell lab is one of the top fragment groups out there, well aware of potential artifacts, and has a long history of studying CYPs, but researchers with less experience could easily populate a library with dubious compounds.

More fundamentally though, I wonder about the basic premise of biofragments. The whole point of fragments is that they have low molecular complexity and are thus likely to bind to many targets, so is it realistic to try to extract selectivity data from them? Indeed, as we’ve seen (here and here), fragment selectivity is not necessarily predictive of larger molecules.

That said, the approach is worth trying. Even if it doesn’t ultimately lead to new insights into proteins’ natural substrates, it could lead to new inhibitors.