Fragments in China

The 2017 International Symposium on Fragment Based Lead Discovery (pdf here) was held in Shanghai, China last week. I was fortunate to be able to attend what I believe was the first significant FBLD meeting in Asia. Antimicrobials were a major theme, particularly against drug-resistant pathogens. The two days were filled with nearly 20 talks, so I’ll just try to capture a few impressions.

Ian Gilbert discussed the fragment-based efforts underway at the University of Dundee, focusing especially on library design. Among initially purchased commercial compounds, only 56% passed quality control, with 26% insufficiently soluble (at least 2 mM in water) and most of the rest either unstable or impure, similar to what has been seen by others. Ian has also enlisted undergraduate students to make “capped” fragments ready for optimization, as well as novel heterocycles.

Biophysics was a major theme of the conference, and Ian made a strong case for biolayer interferometry (BLI), one of the lesser-used fragment finding techniques. A screen can be completed in just a few days with less than a milligram of protein. In particular, BLI may be useful for assessing ligandability: Ian tested 31 targets, 13 known to be ligandable and 5 known to be not ligandable, and found good agreement with previous research. Ligandable targets generally gave primary hit rates >4.5%.

Ismail Moarefi (Crelux, now part of WuXi AppTec) highlighted microscale thermophoresis (MST) and differential scanning fluorimetry (DSF). NMR had identified ten hits against Pim1, but only six had yielded crystal structures, despite considerable effort. Of the four that didn’t, three had no activity by MST, while the fourth was very weak. Ismail also discussed the Prometheus nanoDSF instrument, which is sufficiently sensitive that it can resolve two-stage melting curves for a two-domain protein.

Another lesser used fragment-finding technique, affinity mass spectrometry, was described by Wenqing Shui (ShanghaiTech University). This uses ultrafiltration to separate protein-bound ligands from unbound molecules and mass spectrometry to identify hits; up to 1000 molecules can be screened in a single assay! Wenqing provided several success stories, including fragment hits with very weak (millimolar) affinity. She also demonstrated that the technique works against a membrane preparation of a GPCR.

Among more common biophysical methods, NMR was represented by Ke Ruan (University of Science and Technology of China). The challenge was characterizing a low-solubility ligand which caused extensive line-broadening of the protein due to intermediate exchange rates. This was solved by examining the distance between a fluorinated ligand and a paramagnetic label on the protein and using this to model the binding mode.

But by far the star of the show was crystallography. We’ve previously mentioned the high-throughput capabilities developed at the Diamond Light Source, and part of the impetus for this conference was to bring these technologies to China. Frank von Delft (Diamond and University of Oxford) noted that since the XChem platform launched in late 2015 more than 50,000 crystals have been screened against more than 40 targets, resulting in more than 1000 fragment structures. The group is committed to removing barriers and bottlenecks and today can process 1000 crystals per week through compound soaking, harvesting, data collection, and processing (using specially developed programs such as PanDDA). More than 30 external groups have used the facility, and every target has yielded at least one hit.

Of course, to collect data on 1000 crystals requires you to reproducibly grow lots of well-diffracting crystals that can handle the rigors of soaking, and Diamond has released a handy list of tips and tricks. Getting the right crystals was also the theme of two talks, one by Sheng Ye (Chinese Academy of Sciences) and the other by Carien Dekker (Novartis). Sheng emphasized the importance of optimizing the protein construct, which could include trimming flexible termini or disordered loops, mutating flexible surface residues, or considering different species. He also noted that adding heavy metal ions can actually improve the quality of the crystals as well as making the structures easier to solve. Carien also emphasized the importance of getting the construct right and discussed how seeding (crushing a hard-won crystal and using this to seed new drops) can be very useful. As we’ve noted, screening fragments at extremely high concentrations seems to be the current state of the art, with Novartis moving to 50 mM in the final soak and Diamond going beyond 200 mM! (In contrast to other types of screens at high concentrations, crystallography should not yield false positives, though hits might bind so weakly as to be undetectable by any other method.)

Such a wealth of structures can be daunting, and Anthony Bradley (Diamond) described the construction and use of a “poised library” for follow-up studies. The 768 fragments are (mostly) soluble to 500 mM in DMSO and are designed such that simple chemistry could generate 1.4 million analogs based on reagents currently in stock at Enamine. Potential analogs can be searched using the Fragment Network approach described here, and I was happy to see that Diamond has released their own open-source version (updated link as of 3 Jan 2018).

Jianhua He (Chinese Academy of Sciences) described the facilities at the Shanghai Synchrotron Radiation Facility (SSRF). This is the first third-generation synchrotron in China and has hosted more than 200 research groups since it opened in 2009. Feng Ye, who works at SSRF, gave a talk (in Mandarin) about screening a bacterial protein at XChem; the movies showing liquid handling and robotics would be impressive in any language. Renjie Zhang (Diamond), who also spoke in Mandarin, gave a talk describing (I’m told) not just XChem but how outside users can apply for access. Although there is currently a long waiting list, this should be addressed within the next year or so when SSRF gains Diamond status.

At the 2015 Pacifichem meeting there were only a few speakers from China. Given the level of interest and expertise I saw last week, I predict that the 2020 meeting will see many more.

Twelfth Annual Fragment-based Drug Discovery Meeting

CHI’s Drug Discovery Chemistry meeting took place over four days last week in San Diego. This was easily the largest one yet, with eight tracks, two one-day symposia, and nearly 700 attendees; the fragment track alone had around 140 registrants. On the plus side, there was always at least one talk of interest at any time. On the minus side, there were often two or more going simultaneously, necessitating tough choices. As in previous years I won’t attempt to be comprehensive but will instead cover some broad themes in the order they might be encountered in a drug discovery program.

You need good chemical matter to start a fragment screen, and there were several nice talks on library design. Jonathan Baell (Monash University) gave a plenary keynote on the always entertaining topic of PAINS. Although there are some 480 PAINS subtypes, 16 of these accounted for 58% of the hits in the original paper, suggesting that these are the ones to particularly avoid. But it is always important to be evidenced-based: some of the rarer PAINS filters may tag innocent compounds, while other bad actors won’t be picked up. As Jonathan wrote at the top of several slides, “don’t turn your brain off.”

Ashley Adams described the reconstruction of AbbVie's fragment libraries. AbbVie was early to the field, and Ashley described how they incorporated lessons learned over the past two decades. This included adding more compounds with mid-range Fsp³ values, which, perhaps surprisingly, seemed to give more potent compounds. A 1000-member library of very small (MW < 200) compounds was also constructed for more sensitive but lower throughput biophysical screens. One interesting design factor was to consider whether fragments had potential sites for selective C-H activation to facilitate fragment-to-lead chemistry.

Tim Schuhmann (Novartis) described an even more “three-dimensional” library based on natural products and fragments. Thus far the library is just 330 compounds and has produced a very low hit rate – just 12 hits across 9 targets – but even a single good hit can be enough to start a program.

Many talks focused on fragment-finding methods, old and new. We’ve written previously about the increasingly popular technique of microscale thermophoresis (MST), and Tom Mander (Domainex) described a success story on the lysine methyltransferase G9a. When pressed, however, he said it did not work as well on other targets, and several attendees said they had success in only a quarter to a third of targets. MST appears to be very sensitive to protein quality and post-translational modifications, but it can rapidly weed out aggregators. (On the subject of aggregators, Jon Blevitt (Janssen) described a molecule that formed aggregates even in the presence of 0.01% Triton X-100.)

Another controversial fragment-finding technique is the thermal shift assay, but Mary Harner gave a robust defense of the method and said that it is routinely used at BMS. She has seen a good correlation between thermal shift and biochemical assays, and indeed sometimes outliers were traced to problems with the biochemical assay. The method was even used in a mechanistic study to characterize a compound that could bind to a protein in the presence of substrate but not in the presence of a substrate analog found in a disease state. Compounds that stabilized a protein could often be crystallized, while destabilizers usually could not, and in one project several strongly destabilizing compounds turned out to be contaminated with zinc.

Crystallography continues to advance, due in part to improvements in automation described by Anthony Bradley (Diamond Light Source and the University of Oxford): their high-throughput crystallography platform has generated about 1000 fragment hits on more than 30 targets. Very high concentrations of fragments are useful; Diamond routinely uses 500 mM with up to 50% DMSO, though this obviously requires robust crystals.

Among newer methods, Chris Parker (Scripps) discussed fragment screening in cells, while Joshua Wand (U. Penn) described nanoscale encapsulated proteins, in which single protein molecules could be captured in reverse micelles, thereby increasing the sensitivity in NMR assays and allowing normally aggregation-prone proteins to be studied. And Jaime Arenas (Nanotech Biomachines) described a graphene-based electronic sensor to detect ligand interactions with unlabeled GPCRs in native cell membranes. Unlike SPR the technique is mass-independent, and although current throughput is low, it will be fun to watch this develop.

We recently discussed the impracticality of using enthalpy measurements in drug discovery, and this was driven home by Ying Wang (AbbVie). Isothermal titration calorimetry (ITC) measurements suggested low micromolar binding affinity for a mixture of four diastereomers that, when tested in a displacement (TR-FRET) assay, showed low nanomolar activity. Once the mixture was resolved into pure compounds the values agreed, highlighting how sensitive ITC is to sample purity.

If thermodynamics is proving to be less useful for lead optimization, kinetics appears to be more so. Pelin Ayaz (D.E. Shaw) described two Bayer CDK kinase inhibitors having either a bromine or trifluoromethyl substitution. They had similar biochemical affinities and the bromine-containing molecule had better pharmacokinetics, yet the trifluoromethyl-containing molecule performed better in xenograft studies. This was ultimately traced to a slower off-rate for the triflouromethyl-substituted compound.

The conference was not lacking for success stories, including MetAP2 and MKK3 (both described by Derek Cole, Takeda), LigA (Dominic Tisi, Astex), RNA-dependent RNA polymerase from influenza (Seth Cohen, UCSD), and KDM4C (Magdalena Korczynska, UCSF). Several new disclosures will be covered at Practical Fragments once they are published.

But these successes should not breed complacency: at a round table chaired by Rod Hubbard (Vernalis and University of York) the topic turned to remaining challenges (or opportunities). Chief among these was advancing fragments in the absence of structure. Multiprotein complexes came up, as did costs in terms of time and resources that can be required even for conventional targets. Results from different screening methods often conflict, and choosing the best fragments both in a library and among hits is not always obvious. Finally, chemically modifying fragments can be surprisingly difficult, despite their small size.

I could go on much longer but in the interest of space I’ll stop here. Please add your thoughts, and mark your calendars for next year, when DDC returns to San Diego from April 2-6!

Fragments deliver a chemical probe for BRD9

Bromodomains have nothing to do with bromine. Rather, they are small (~110 amino acid) domains that recognize acetylated lysine residues, a common modification on histones, and are thus key epigenetic “readers”. Humans have more than 60 of them, and as you can imagine selectivity is not assured. However, fragments have proven very useful in targeting these proteins. Since the first mention of bromodomains on Practical Fragments back in 2011 the number of posts has been growing rapidly, so for the first time ever we’ve decided to devote an entire month to the topic.

In other words, July is bromodomain month! We’ll start with two papers against the bromodomain BRD9, part of the SWI/SNF chromatin remodeling complex that seems to be important for acute myeloid leukemia.

The first paper, in J. Med. Chem. (and open access), is published by Laetitia Martin and collaborators at Boehringer Ingelheim, University of Oxford, and Cold Spring Harbor. The researchers used three orthogonal biophysical screening methods: differential scanning fluorimetry (DSF), surface plasmon resonance (SPR), and microscale thermophoresis (MST). A library of 1697 fragments was screened at 0.4 mM (DSF), 0.1 mM (SPR) or 0.5 mM (MST), and hits were then validated using ¹⁵N HSQC NMR. The 77 hits that confirmed were taken into crystallography, producing 55 structures.

Validation rates in the NMR secondary screen were excellent for DSF (94%) and SPR (84%) but less so for MST (31%). That said, of the 38 validated hits from MST, 29 were not found in either of the other techniques, and 14 of these produced crystal structures. This is a useful reminder that while screening cascades can whittle down many hits, they do run the risk of throwing out the proverbial babies along with the bathwater.

In parallel with the biophysical screens, a virtual screen of ~73,500 fragments was conducted using Glide to identify 208 fragments that were then tested using SPR and DSF. This led to 23 hits, 11 of which produced crystal structures.

Two of the more potent fragments were the structurally related compound 3 (from the biophysical screen) and compound 4 (from the virtual screen). Optimization started with compound 4 by adding electron donating groups to the phenyl ring to try to improve a stacking interaction observed in the crystal structure. This led to compound 10, and building out the other ring to make it more similar to fragment 3 led to BI-9564.

BI-9564 has low nanomolar activity in both a biochemical assay as well as isothermal titration calorimetry (ITC). It is also quite selective: among 48 other bromodomains, it only hits the closely related BRD7 and CECR, and it is >10-fold more potent on BRD9. None of a panel of 321 kinases were inhibited with IC₅₀ < 5 µM, and only 2 of 55 GPCRs were inhibited. The compound is also cell active, reasonably soluble, has good pharmacokinetics in mice, and orally bioavailable. In short, BI-9564 is an excellent chemical probe – and is in fact being offered as such.

While we’re on the subject of BRD7 and BRD9, it’s worth noting another recent paper, this one in ChemBioChem from Ke Ruan and colleagues at the University of Science and Technology of China. The researchers screened their library of 890 fragments against BRD7 using three different ligand-detected NMR techniques: STD, WaterLOGSY, and CPMG. Fragments were screened in pools of 10 with each fragment present at 400 µM. This yielded just 10 hits, of which 5 confirmed when tested individually. Protein-observed NMR was then performed on these, suggesting that they all bind in the acetyl-lysine recognition sites; they have similar affinities for both BRD7 and BRD9, with dissociation constants between 22 and 600 µM. Crystallography confirmed the binding mode for one of the fragments bound to BRD9. Interestingly, this showed quite a bit of plasticity in the protein compared to the un-liganded structure. Indeed, the BI researchers suggest that different degrees of protein flexibility between BRD7 and BRD9 could account for the selectivity differences observed for BI-9564.

Stay tuned next week for more fragment-screening against a different class of bromodomains!

Microscale thermophoresis revisited

One of the less commonly used fragment-finding methods is microscale thermophoresis (MST). This measures the movement of proteins in a temperature gradient; ligand binding changes the movement. When we first described MST in 2012, we noted that the technique seemed relatively low throughput. In a paper recently published in J. Biomol. Screen., Alexey Rak and colleagues at Sanofi teamed up with Dennis Breitsprecher and researchers at NanoTemper (which makes MST instruments) to try to increase this.

The researchers chose the kinase MEK1 and carefully developed assay conditions; their detailed description is a useful resource for those who decide to give MST a try. Adding nonionic detergent to the assay proved to be essential for reproducibility and to prevent the protein from sticking to the capillary or aggregating. Also, rather than relying on the weak chromophores (such as tryptophan) in native proteins, MEK1 was labeled with a fluorescent dye. The substrate ATP was used as a positive control, and the measured affinity was in good agreement with previous results.

The screen itself was performed on a set of 193 fragments that had been computationally preselected as potential ligands for the kinase MEK1 (work we blogged about here). These were serially diluted using automated liquid handling and tested in 12-point dose-response curves to try to determine dissociation constants (Kd values) for each fragment. All together this run of more than 2000 capillary tubes required only 90 micrograms of protein and took less than 7 hours. Retrospective analysis suggested that a single-point screen at 150 µM of each fragment would have caught most of the best hits and cut analysis time to 70 minutes, so it looks like MST is becoming competitive with other biophysical screening methods in terms of time and reagent consumption.

What about results? The overall hit rate was nearly 38%, which is high, though not outrageously so given that the fragments were computationally pre-selected. Of these, the best 25 fragments showed well-defined dose-response curves with
Kd < 200 µM and competition with ATP. One nice feature of the method is that pathological behavior such as aggregation or denaturation could be observed directly in the form of irregular or bumpy MST traces, thus allowing false positives to be rapidly weeded out. Similarly, a loss in fluorescence signal was interpreted as the protein unfolding and sticking to the wells or pipette tips.

It is always useful to cross-check hits in orthogonal assays. As we noted previously, these fragments had previously been screened against MEK1 using surface plasmon resonance (SPR) and differential scanning fluorimetery (DSF). Most of the best hits from DSF were rediscovered by MST, though MST found many hits DSF had missed. In contrast, most of the SPR hits did not confirm in MST. The rank order of hits was also similar for MST and DSF but not for MST and SPR.

A picture is worth a thousand words, and some of the best hits were subjected to crystallography. In fact, 7 of the top 15 MST hits had previously been characterized by crystallography, and 7 new crystal structures could be determined out of 11 additional MST hits for which crystallography was attempted.

Overall then it appears that MST is coming into its own. If you’ve tried it, please share your experiences.

Fragment finding smackdown: 2015 edition

Our current poll (right-hand side of page) asks about NMR. But of course, there are lots of other ways to find fragments, and the question often arises as to which ones are best. This is the subject of a recent paper in ChemMedChem by Gerhard Klebe and collaborators at Philipps University Marburg, Proteros, NovAliX, Boehringer Ingelheim, and NanoTemper.

Long-time readers will recall that the Klebe group assembled a library of 361 fragments, some of which violated strict “rule of 3” guidelines. These were screened in a high-concentration functional assay against the model aspartic protease endothiapepsin, resulting in 55 hits, of which 11 provided crystal structures. The authors wondered how other techniques would fare. In the new paper, they retested their entire library against the same protein using a reporter displacement assay (RDA), STD-NMR, a thermal shift assay (TSA), native electrospray mass spectrometry (ESI-MS), and microscale electrophoresis (MST). To the extent possible they tried to use similar conditions (such as pH) for the different assays, though the fragment concentrations ranged from a low of 0.1 mM (for ESI-MS) to a high of 2.5 mM (for TSA), while protein concentrations ranged between 4 nM (for the biochemical assay) to 20 µM (for ESI-MS).

All told, 239 fragments hit in at least one assay – a whopping hit rate of 66%. Actually, the number is even higher since, for various reasons, not all fragments could be tested in all assays. And yet, not a single fragment came up in all of the assays! Overall agreement was in fact quite disappointing, with most methods having overlaps of less than 50%, and often below 30%. This is in contrast to a study from a different group highlighted a couple years ago.

What’s going on? One clue might be the solubilities, which were experimentally measured for all library members. In general, hits tended to be more soluble than the library as a whole, emphasizing the importance of this parameter not just for follow-up studies but for identification of fragments in the first place.

Another possibility is that some fragments bind outside the enzyme active site, and thus would not be picked up in a biochemical assay or the RDA. Some evidence for this is provided by follow-up NMR studies in which hits were competed with ritonavir, which binds in the active site. Ritonavir-competitive binders shared greater overlap with biochemical and RDA hits, while there was more overlap between ritnovair-uncompetitive binders and hits from methods such as ESI-MS, TSA, and MST that rely solely on binding. (This could also explain similar observations made earlier this year.)

If a picture is worth a thousand words, how many of the 11 hits that had previously yielded crystal structures would have been identified had they been tested in other methods? Here the numbers vary significantly, from 27% for ESI-MS and MST to 100% for NMR, though these statistics should be taken with a grain of salt since – for example – only 7 of the 11 crystallographically-confirmed hits could actually be tested in the NMR assay. Also, it is possible that some hits from these methods might have generated new crystal structures for fragments not identified in the initial biochemical screen.

One admirable feature of this paper is that the authors provide all their data, including structures and measured solubility numbers for each component of their library. This should provide an excellent dataset for a modeler to use in benchmarking computational methods.

All in all this is a thorough and important analysis and a sobering reminder that, even if a fragment doesn’t hit in orthogonal assays, that doesn’t necessarily mean it’s not a useful starting point. On the other hand, artifacts are everywhere, and paranoia is often justified. The art is deciding which hits are worth pursuing – and how.

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.