Updated polls: affiliation, methods, and library size

Practical Fragments has run polls on topics including readership, screening methods, favorite metrics (here and here), maximum and minimum fragment size, and the importance of structural information.

We’re interested in how things are changing, particularly in terms of our readership (last polled in 2010) and what fragment-finding methods are most popular (last polled in 2011).

Also, one topic that comes up repeatedly (for example here and here) but has never been actually polled is how many fragments make a good library – we’ve added a question about that too.

Please add your input for 2013; the more people who vote, the more representative the numbers will be. You can find all three questions on the right side of the page, below the "Links of Utility".

We’ll summarize the results and compare them to the previous polls in an upcoming post.

Also, please let us know if you would like us to repeat any of our other polls, and feel free to suggest new topics.

Happy voting!

Fragments against PPI Hot Spots

Protein-Protein interactions are important to so many physiological processes.  There is mounting literature examples of utilization of fragments to block PPIs.  In this paper, Rouhana et al. show how they approached the PPI of Arno and ARF1, ADP-ribosylation factor (part of the RAS superfamily). Arno is part of the brefeldin A-resistant GEFs and share a 200 amino acid domain called SEC7.  SEC7 interacts with ARF through insertion of ARF switch regions into hydrophobic regions of SEC7.  This interaction is interesting from a ligand design standpoint is very interesting because it does not involved an alpha-helix inserting into the partner's hydrophobic groove.  Rather SEC7 has a rather large interface denoted by "hot spots". 

The figure shows their "innovative" FBDD strategy.  First, a Voldemort Rule compliant library was screened in silico.  Since in silico screening is not typically used for fragment screening (but becoming more common) they imposed some initial rules: docking site is small (1-2 residues!), hot spots defined by interaction energy (>1kcal/mol from alanine scan), and very strict selection criteria.  3000 fragments from the Chembridge library were screened.  33 molecules were selected and 40 random fragments chosen as negative controls. 

This was followed by a fluorescence assay (2mM fragments) to test their computational results, just as I say you should do.  Promiscuous binders were removed, not by using detergent, but using protein polarization to directly detect interaction with the target.  This seems like over-complexation of an assay, but without knowing the details of system there may be a very good reason for this approach. 

Compounds 1-4 were identifed as inhibitors (35%, 16%, 38%, and 23% inhibition at 2mM respectively) from each of the "hot spots".  I think it is interesting that these compounds were predicted to have affinities of 10uM or better from the docking.  To me, that just illustrates that predicted affinites are rediculous.  Why do people even report them?  Compound 1 had a Kiapp of 3.7mM which is a LEAN of 0.12!  These were then compared to the PAINS list and 3 is "ambiguous".  Compounds 5 and 6 were chosen as negative controls.  SPR confirmed the binding of 1,2, and 4, but at less than stoichiometric binding levels (the assay was run at 250uM).  3 could not be confirmed as a binder.  Does this mean anything for ambiguous PAINS? 

STD NMR was then used to confirm binding.  In a nice departure, they actually talk about conditions they used: 10 and 30uM ARNO with 0.1mM and 1mM compounds at 32 and 12C.  30uM ARNO with 1mM fragments @12C was what worked (33x fold excess fragments). Confirming the SPR, compounds 1, 2, and 4 were shown to bind, while "ambiguous" 3 had some binding. Finally, compounds were soaked with fragments 1, 2 and 4.  This led to crystal structures which could then be used for more model building, compound design, etc.  This led to the following compound (1.61mM KiApp, LEAN = 0.13) (the methoxy derivative of 1) for further analysis:

By and large, this is a well done, thoughtful work.  They really understand how to setup and interpret STD-NMR. However, these compounds are really atom inefficient.  Is that a consequence of the type of interaction they are inhibiting?  As a fragment, there is nothing wrong with it. 

[Quibble: The authors claim that this is an innovative approach, but I am not seeing it.  They claim their in silico screen first then following up by biophysical techniques is the innovation. ] 

Supplemental Information here.

Natural products as fragments

Natural products were used as drugs long before there was a drug industry, and there is a case to be made that they make good starting points for lead discovery. For one thing, they tend to be more “three-dimensional” than many synthetic molecules. For another, the fact that some organism, somewhere, made them proves that they can bind to proteins. Practical Fragments has previously highlighted examples in which natural products were conceptually fragmented into smaller molecules or incorporated into a fragment library. In a new paper in ACS Chemical Biology, Ronald Quinn at Griffith University and a team of Australian and US collaborators describe a fragment library consisting entirely of natural products.

The researchers assembled a library of 331 natural products with the following characteristics:

• MW ≤ 250 Da (mean = 195.6)
• ClogP < 4 (mean 0.4)
• hydrogen bond donors ≤ 4 (mean 1.3)
• hydrogen bond acceptors ≤ 5 (mean 2.6)
• rotatable bonds ≤ 6 (mean 2.2)
• polar surface area ≤ 45% (mean 17.7%)

The maximum number of donors and acceptors allowed is slightly higher than typical in a fragment library, consistent with the fact that natural products tend to have more oxygen and nitrogen atoms than your typical Suzuki-derived biphenyl. However, the molecular weights are kept low, and despite the tolerance for more lipophilic molecules, the vast majority of the library has ClogP < 3 (with many molecules having ClogP < 0).

Having assembled the library, the researchers used native mass spectrometry to screen pools of eight fragments against the malarial enzyme Plasmodium falciparum 2′-deoxyuridine 5′-triphosphate nucleotidohydrolase (PfdUTPase). They found that a molecule called securinine binds to the enzyme, and six analogs also showed varying degrees of binding as assessed by mass spectrometry.

At this point things get a bit strange. Most of the molecules show some anti-plasmodial activity in culture, but they all seem to modestly activate PfdUTPase. It is unclear whether these two observations are mechanistically related: is the activation of PfdUTPase really what’s causing the anti-plasmodial activity, or are the molecules hitting a different target?

In fact, securinine comes up as a hit in a variety of different biological assays. Looking at the molecular structure this is perhaps not surprising: it contains a reactive electrophilic center that has previously been shown to react with amines under mild conditions, so presumably it can react with all sorts of biological nucleophiles in vivo. This is not to say that covalent inhibitors are unacceptable – dimethyl fumarate looks set to become a blockbuster drug – but it is nice to know if you are dealing with them, and the authors seem not to have considered the possibility.

In the end, I do think libraries of natural products such as these could be useful, but they will require care in their construction, use, and interpretation. Just as there are many synthetic compounds best left out of screening collections, the same goes for natural products. Toxoflavin, for example, is a notorious redox cycling PAIN that has (embarrassingly) been reported as an inhibitor for multiple targets with no evidence for specificity. I’m not ready to put securinine into this category, but I would urge caution.

Just because something is natural doesn’t mean it’s healthy.

WAC vs other methods: all roads lead to good fragments

Among the many ways to find fragments, one of the relatively inexpensive newcomers is weak affinity chromatography, or WAC (see also here). The technique works by immobilizing a target protein onto a column and flowing fragments over it; molecules that bind to the target will elute more slowly than those that don’t. WAC has a number of potential benefits, but as with any technique the question is how well it really works. In a paper published a few months ago in Analytical Chemistry, Sten Ohlson at Linnaeus University and collaborators at Vernalis compared WAC with more established methods.

The protein they chose, HSP90, is sort of the fruitfly of FBLD: just about every technique has been tested on it. It’s also an oncology target with which Vernalis has many years of experience. The researchers chose 111 fragments from the Vernalis library and screened these using WAC. They also screened most of the fragments using surface plasmon resonance (SPR), fluorescence polarization (FP), thermal shift, and NMR (using three techniques: STD, waterLOGSY, and relaxation filtered spectra; only fragments that confirmed in all three NMR assays were considered hits).

The top 27 hits from WAC were also investigated with isothermal titration calorimetry (ITC), and 32 hits were soaked into crystals for X-ray crystallography.

The results were quite encouraging, with good agreement between the different methods:

NMR performed the best, though this could be due in part to the fact that three separate NMR techniques were used. Thermal shift performed the worst, with both false positives as well as false negatives, but even here the agreement was always greater than 50%. It is also important to note that assay conditions varied from technique to technique (for example, the pH ranged from 6.5 to 7.5), which could account for many of the discrepancies.

These results are in sharp contrast to some other comparisons of fragment finding methods (such as here and here), which showed little or no correlation between hits. Why the difference? One possibility is that the folks at Vernalis have worked out all the kinks in their assays and are very adept at separating the true hits from the chaff. Of course, it probably doesn’t hurt that they were working with a well-behaved and extensively characterized target.

The main focus of the paper is WAC, which performed admirably. Compounds could be screened in pools of up to 16 fragments when mass-spectrometry was used as a detection method, and less than 2 milligrams of HSP90 was used to prepare all three of the WAC columns made. One worry with immobilizing your protein is long term stability, but the columns seemed to be stable for at least 6 months through multiple runs.

Of course, no technique is perfect, and one area where WAC gets whacked is in determining dissociation constants. The correlation between K_D values measured by SPR and ITC was excellent (R² = 0.91) but much worse for WAC versus ITC (R² = 0.38) and nonexistent for WAC versus SPR (R² = 0.016), though some of this could possibly be explained by differences in buffer conditions.

Overall it looks like WAC is a great way to find fragments, though you may want to use other methods to actually quantify binding. This paper provides a detailed guide for using WAC, as well as good descriptions of other fragment-finding methods.

Fragment to Lead

In this paper, Constellation and their partner Jubilant Biosys report on their FBHG effort that lead to BET (bromodomain and Extra C-terminal)  inhibitors (BRD4).  Since this is a letter details are short, so hopefully a longer, more detailed paper will be forthcoming.  What they report is a fragment screen that identified micromolar compounds.  These were then co-crystallized (not soaked) leading to several high resolution crystals.  Of particular interest was this fragment: This fragment should ring a bell it is part of the known inhibitor IBET151 from GSK (and is the known preferred binding motif for bromodomains).  Compound 1 binds in a similar fashion to JQ1, binding to the asparagine that recognizes the endogenous Ac-K.  This suggested to them that the isoxazole fragment could replace the triazole of JQ-1.  It's LEAN is 0.34 (33uM) and has a Binding Efficiency of 25.7.  This works describes the replacement of the triazole (Left) with the preferred isoxazole (Right).

Their SAR work is shown in the table below.  They were able to improve biochemical and cellular potency to that of the known inhibitors by replacing the sidechain with a carboxamate (Cpd 3).  the crystal structure of this compound showed similar binding to previously described isoxazoles.

They then went after the 4-chlorophenyl ring to see if they could modify the biophysical and three-dimensional properties of the molecule, while maintaining the potency seen with 3. 

A chloro scan around the ring showed that the o-Cl substitution was 10x less potent, but ortho-Me was tolerated, which led them to believe that it was a steric rather than electrostatic interaction. Overall, there was no better aromatic moiety for this position, and aliphatic moieties were definitely no good.  Compounds 3, 21 (phenyl), 22 (cyanophenyl), and 25 (aminopyridyl) were tested in in vitro ADME assays.  They showed good stability in human microsomes and generally stable (I am not ADME expert, so really what do they mean here?) in rat microsomes.  They showed high plasma protein binding in human plasma but negligible CYP inhibition.  Compounds 3 and 22 (cyano-phenyl) supported further profiling in rat PK experiments.  Compound 3 was superior to 22 and showed adequate exposure in mouse and showed excellent PK in dogs.  They were able to see a dose-dependent decrease of MYC.  MYC suppression was correlated with the amount of compound in the tumor and plasma.  

This is a really nice example of how fragments can be used to "scaffold hop", even if the entire scaffold is not changed.  Also, I think, based on the author list, this is a really good example of CRO-client collaboration.  There are many more out there I am sure, I just don't think we are aware enough of them.

The calm before the click in chitinase

In situ click chemistry is a topic we’ve covered before on Practical Fragments. Essentially, two ligands bind near one another on a target protein and react to form a linked molecule. There are several published examples, but it is not clear why it sometimes works and sometimes doesn’t. A new paper in Proc. Acad. Nat. Sci. USA by a team of Japanese and US researchers led by Satoshi Ōmura and Toshiaki Sunazuka at the Kitasato Institute in Tokyo addresses this question.

The researchers had previously discovered potent inhibitors of an antibacterial target enzyme called Serratia marcescens chitinase B, or SmChiB, using in situ click chemistry. In the presence of SmChiB, azide 2 reacts with alkyne 3 to yield triazole 4, which binds 26-fold more tightly than azide 2:

 

In the new paper, the goal was to use crystallography and computational chemistry to investigate how the reaction proceeds. To avoid azide 2 reacting with alkyne 3 in the crystal and so better visualize starting points, the researchers prepared the closely related alkene 5 mimic of alkyne 3. Unfortunately, due to its (unmeasurably poor) affinity, alkene 5 did not yield a co-crystal structure on its own.

The researchers were able to obtain a co-crystal structure of SmChiB bound to triazole 4 (green carbons below). Surprisingly, a co-crystal structure of azide 2 showed the molecule bound in a quite different orientation. However, a co-crystal structure of the ternary complex of azide 2 (cyan below) and alkene 5 (magenta) bound simultaneously to SmChiB revealed a close overlay of azide 2 with the corresponding fragment in triazole 4. Alkene 5 in the ternary complex adopted two conformations (the electron density is memorably described as resembling “a two-horned goat head”). As shown in the figure below, one of these orientations places the alkene moiety in close proximity to the azide moiety, primed for clicking.

Next, the researchers used this ternary structure to run high-level density functional theory calculations to determine the energetics of the click reaction and compared these with the same reaction run in water. The values were quite similar (if anything, the protein had a slightly higher activation barrier), suggesting that the protein was not directly catalyzing the reaction with specific amino acid side chains. Rather, the reaction was being accelerated simply by the preorganization of the azide and alkyne.

On the one hand, these results aren’t really a surprise: I think most people assumed that in situ chemistry works by bringing the reactants together rather than anything more exotic (with the odd exception). On the other hand, it is nice to see experiment match theory.

More generally, the results help to explain why in situ click chemistry is so challenging. The crystal structure of azide 2 and alkene 5 shows the relevant moieties quite close to each other, yet the reaction is still somewhat inefficient. Finding two fragments that not only bind near one another but are also oriented properly is likely to be a rare event.

Biophysics Conference (pt 3)

I have been giving my thoughts on the Novalix Conference on Biophysics in DD here and here.  Today's installment is on the "Emerging Technologies" and "Hits and Leads" section of the conference.  

Stefan Duhr- NanoTemper: Microscale Thermophoresis (MST) has been discussed here previously. Both Dan and I really like this technology.  This talk was an excellent overview of the theory.  Nanotemper claims that it has a dynamic range up to the mM range, however in their talk all of the examples were relatively, or very, tight binding complexes.  It has definite advantages in that it only uses 4 uL of sample/data point and it takes 40s/data point.  

There were a variety of talks on technologies that are definitely cool in a "Amazing they can do that" sort of way.  However, as an application to drug discovery, not so much.  There was a talk about Backscattering Interferometry (BSI), a switchable DNA chip (definitely cool tech, but with no discernible advantage over similar technology), Cryo-TEM (!), most of these talks I could not figure out how you would use in screening/FBHG.  However, the point of emerging technology is to emerge, so maybe in the near future there will be pretty boxes that have notable, robust discovery uses.

Chris Marshall -UToronto: This talk and Till's (below) were about GTPases.  This talk focused on a NMR-based GTPase assay.  What was particularly interesting was that they tethered their GTPase (Rheb) to a nanodisc, which should tumbling properties semi-independent of the nanodisc. This is a much more "biological" condition that many people typically use.  Other than that, this was a decidely academic talk.  In an organization with unlimited resources, and no time lines, you might follow the same approach as this group did.  In reality, I can't imagine you would.

Helena Danielson - Uppsala U/Beactica: This was a very interesting talk (per usual).  One key comment she made was: ease of use of a technology is NOT the same as ease of implementation.  In terms of Beactica's fragment library: 2000 compounds (from her slide) that are largely Voldemort Rule compliant.  It is enriched in known drug frameworks with diversity and scaffold representation (I am not sure what is meant by that).  For her first case study, they only used 930 fragments.  She didn't mention why a subset of the entire library was used.  She mentioned that they use an early biochemical screen as an orthogonal assay.  She spent a lot of time discussing the deconstruction of sensorgrams, in particular, if you have specific and non-specific binding contributing.  She also presented a case study against a GABA-A like receptor.  She then spent the rest of her talk discussing Chemodynamics: varying sample conditions, like temperature or pH. For BACE, for example, compounds need to bind at neutral and then acidic pH. 

Till Maurer- Genentech:  Till's talk was on k-RAS by NMR.  (As an aside, k-RAS has become a "hot" target largely due to this work.  Way back in 2003, we published a new method for NMR screening using k-RAS as one of our targets because it was so interesting we knew legal would let it go.)  Their fragment library had 3285 fragments (it is now 5000) biased towards high solubility for X-ray follow up in mixtures of 5.  Of 3285 fragments they found 1092 with a S/N >5.  Of these 266 confirmed (higher S/N threshold and other criteria) and were followed up by H-N HSQC.  Of 25 confirmed by HSQC, 6 produced crystals. 

Johannes Ottl- Novartis: The last talk of the conference was another really nice overview of the various biophysical methods and how they are applied in a few different case studies. 

So, what was the take home of this conference?  Biophysics is a rich and diverse toolbox.  However, in many cases we still don't know how to use these powerful tools prospectively, rather they are much more used retrospectively.