Review of 2019 reviews

The year ends, and with it the awkward teenage phase of the twenty-first century. As we have done since 2012, we're using this last post of the year to highlight conferences and reviews over the previous twelve months.

There were some good events, including CHI’s Fourteenth Annual Fragment-based Drug Discovery meeting in San Diego in April, their Discovery on Target meeting in Boston in September, and the third Fragment-based Drug Design Down Under 2019 in Melbourne in November, which also saw the launch of the Centre for Fragment-Based Design. Our updated schedule of 2020 events will publish next week.

Turning to FBLD reviews, Martin Empting (Helmholtz-Institute for Pharmaceutical Research Saarland) and collaborators published a general overview in Molecules. This is a nice up-to-date summary, covering library design, methods to find, confirm, and rank fragments, and optimization approaches. It’s also open access so you can read it anywhere.

Targets

Protein-protein interactions can be particularly challenging drug targets, and these are covered in a Eur. J. Med. Chem. review by Dimitrios Tzalis (Taros Chemicals), Christian Ottmann (Technische Universiteit Eindhoven) and colleagues. The focus is on clinical compounds, and several of these – including venetoclax, ASTX660, mivebresib, onalespib – are discussed in detail. The article is particularly useful in discussing late-stage optimization of pharmacokinetic and pharmacodynamic properties. It also provides a nice summary of physicochemical properties for fragment hits and derived candidates.

Target selectivity is always important, and this is the focus of a review in Exp. Opin. Drug Disc. by Rainer Riedl and collaborators at the Zurich University of Applied Sciences and the Università degli Studi dell’Insubria. Although the broader topic is de novo drug design, fragment-based methods are prominent, and include case studies we’ve discussed on nNOS, pantothenate synthetase, and MMP-13.

In terms of specific targets, Fubao Huang, Kai Wang, and Jianhua Shen at the Shanghai Institute of Materia Medica provide an extensive review of lipoprotein-associated phospholipase A2 (Lp-PLA₂) in Med. Res. Rev. This serine hydrolase has been studied for four decades but – as the researchers note – “divergence seems to be ubiquitous among Lp-PLA₂ studies.” At least this is not for lack of good chemical tools, fragment-derived (see here, here, and here) and otherwise.

Methods

Although NMR has fallen behind crystallography in our latest poll, that is certainly not reflected in terms of reviews. In particular, ¹⁹F NMR is covered in three papers. CongBao Kang (A*STAR) manages to pack a lot (including 261 references!) into a concise review in Curr. Med. Chem. Topics include protein-observed ¹⁹F NMR, in which one or more fluorine atoms are introduced into a protein genetically, enzymatically, or chemically, as well as ligand-observed methods, in which fluorine-containing small molecules are directly observed or used as probes that are displaced by non-fluorine-containing molecules.

Protein-observed ¹⁹F NMR (PrOF NMR) is covered in Acc. Chem. Res. by William Pomerantz and colleagues at the University of Minnesota. Although the first example was published 45 years ago, only in the past few years has the technique been used for studying protein-ligand interactions. The researchers note that introducing fluorines into aromatic residues is ideal because they are relatively rare, simplifying interpretation, and overrepresented at protein-protein interactions, maximizing utility. Several case studies are described, and even proteins as large as 180 kDa are amenable to the technique.

Ligand-based fluorine NMR screening is simpler and more common than techniques that focus on proteins, and this topic is thoroughly reviewed by Claudio Dalvit (Lavis) and Anna Vulpetti (Novartis) in J. Med. Chem. After a section on theory, the researchers discuss library design, including a long section on quality control (which involves assessing solubility, purity, and aggregation of the molecule in a SPAM filter). Direct and competition-based screening approaches are covered in detail; for the latter, a new method for determining binding constants is provided. The paper concludes with more than a dozen case studies. Clearly much has changed in the ten years since I wondered “why fluorine-labeled fragments are not used more widely.” This perspective is a definitive guide to the topic.

Moving to less common methods for characterizing fragments, György Ferenczy and György Keserű (Research Center for Natural Sciences, Budapest) cover thermodynamic profiling in Expert Opin. Drug Disc. After discussing several case studies, they conclude that “thermodynamic quantities are not suitable endpoints for medicinal chemistry optimizations” due to the complexity of contributing factors. This is consistent with another recent paper on the subject (see here), though the information provided is still interesting for understanding molecular interactions.

And although you might have thought the 2017 VAPID publication was the last word on the limitations of ligand efficiency (LE), Pete Kenny has published a splenetic jeremiad on the topic in J. Cheminform. (see also his blog post on the topic, which includes a sea serpent). This is largely a retread of a 2014 article on the same topic (reviewed by Teddy in his inimitable manner here). Pete also describes a more complicated alternative to LE involving residuals, though unfortunately he provides no evidence that it provides more useful information. Pete is of course correct to remind us that metrics have limitations, but assertions that LE “should not even be considered to be a metric” are overwrought.

Chemistry

Two articles discuss virtual chemical libraries. In J. Med. Chem., W. Patrick Walters (Relay Therapeutics) describes efforts to measure, enumerate, and explore chemical space. He notes that false positives could quickly overwhelm a virtual screen of a hundred million molecules, but as we saw earlier this year, progress is being made. Indeed, Torsten Hoffmann (Taros Chemicals) and Marcus Gastreich (BioSolveIT) focus on navigating the vastness of chemical space in Drug Disc. Today. They note that the Enamine REAL Space is up to 3.8 billion commercially accessible compounds, more than double the number of stars in the Milky Way. But this pales in comparison to the 10²⁰ potential compounds in Merck’s MASSIV space. Just storing the chemical structures of these in compressed format would require 200,000 terabytes – and searching them exhaustively is beyond current technology.

Ratmir Derda and Simon Ng (University of Alberta) discuss “genetically encoded fragment-based discovery” in Curr. Opin. Chem. Biol. This involves starting with a known fragment that is then coupled to a library of peptides and screened to find tighter binders. The researchers provide a number of case studies, though adding even a small peptide to a fragment will generally have deleterious effects on ligand efficiency. And – Rybelsus not withstanding – oral delivery of peptides is challenging.

Finally, Vasanthanathan Poongavanam, Xinyong Liu, and Peng Zhang, and collaborators at Shandong University, University of Bonn, University of Southern Denmark, and K.U. Leuven review “recent strategic advances in medicinal chemistry” in J. Med. Chem. Among a wide range of topics from drug repurposing to antibody-recruiting molecules is a nice, up-to-date section on target-guided synthesis. As I opined a couple years ago, I still doubt whether this will ever be generally practical, but from an intellectual standpoint I’m happy to see work continue on the approach.

And with that, Practical Fragments says goodbye to the teens and wishes you all a happy new year. Thanks for reading and commenting. May 2020 bring wisdom, and progress.

Fragments in the clinic: S64315 / MIK665

Earlier this year we highlighted the discovery of AZD5991, a phase 1 compound from AstraZeneca that inhibits the anti-apoptotic cancer target Mcl-1. Those efforts made use of a fragment previously published by a different research group. Mcl-1 has been a popular target for some time; the first mention on Practical Fragments dates to 2010. The story behind another investigational drug is described in a couple papers from earlier this year.

The first, in ACS Omega by Rod Hubbard and colleagues at Vernalis, University of York, and Servier, describes fragment screening efforts against both Bcl-2 and Mcl-1. The proteins are related both structurally and functionally, and Bcl-2 is the target of venetoclax – the second fragment-derived drug approved. Some of the early fragment hits bound to both proteins, but selective and potent inhibitors were ultimately developed. In the interest of space only those against Mcl-1 will be discussed here.

Both proteins required considerable protein engineering, which is described in detail. Ultimately one form of human Mcl-1 was used for crystallography, while mouse protein was used for NMR screening due to its better stability. A total of 1064 fragments were screened at 0.5 mM each (in pools of eight) using ligand-observed NMR; 39 confirmed using STD NMR, WaterLOGSY, and CPMG. Additionally, fluorescence polarization, 2-dimensional (HSQC) NMR, ITC, and SPR were used to validate hits. Crystallography proved challenging in the beginning but ultimately helped drive optimization of more potent molecules. The large number of different assays employed is consistent with our recent poll results.

Protein-observed NMR was particularly useful in providing information on the quality of both the ligand and protein (reminiscent of the “validation cross” discussed here). Before crystallography was able to play a meaningful role, “NMR-guided models,” combining partial protein assignments with flexible docking, were used to drive SAR.

While the first paper focuses on protein optimization and biophysics, the second (in J. Med. Chem.), by András Kotschy and collaborators, focuses on chemistry. Fragment 1a was one of several hits pursued, initially by looking for analogs, but most of these had comparable (weak) activity. In the absence of a crystal structure a systematic chemistry campaign was conducted, varying elements of the core and sidechains. Many of these molecules had comparable activity against both Mcl-1 and Bcl-2, but replacing the nitrogen linker with an oxygen led to selectivity against the former. The addition of hydrophobic substituents led to compound 10c, with submicromolar activity.

Anticipating poor cell permeability for a negatively charged, lipophilic molecule, the researchers introduced a positively charged methylpiperazine moiety at various positions around the molecule, ultimately leading to compound 18a. In addition to potent Mcl-1 binding, this molecule is active in cells and shows reasonable pharmacokinetic properties in mice. Further optimization to S64315 does not appear to have been published yet, though the structure was disclosed earlier this year, and the fragment origins remain clear.

Together these papers provide a thorough description of drugging a difficult target. They also provide insights into the investment required. The Mcl-1 project began around 2007, and it took a decade before S64315 entered the clinic. Enabling drug discovery against protein-protein interactions required multiple biophysical techniques in addition to all the standard components of pharmaceutical research. The researchers note that “establishing such a platform can take some time and resource – a tool compound is usually needed to validate the assays, but the assays are needed to identify the tool compound.” In the end they have succeeded, and Practical Fragments wishes them – and the patients being treated – the best of luck.

A new library of fluorinated Fsp3-rich fragments

Among fragment-finding methods, ligand-based NMR ranks near the top in terms of popularity. Of its many variations, fluorine (¹⁹F) NMR appears to be gaining in popularity. Fluorine NMR has several advantages, including high sensitivity and the fact that many fragments can be screened simultaneously because of the wide chemical shift range for fluorine. Although more commercial fluorine-enriched libraries are available now than when we first wrote about the approach a decade ago, the diversity of these libraries is still somewhat limited. This problem has been tackled by Mads Clausen at the Technical University of Denmark and an international team of collaborators in a new Angew. Chem. Int. Ed. paper.

The researchers wanted to create a fluorinated fragment library that would be not just diverse but also contain a high fraction of sp³-hybridized carbons (high Fsp³). Some of the early claims around “three dimensional” fragments have been questioned, and there seems to be little if any correlation between the shapeliness of fragments and that of derived leads, but if you’re going to make new fragments in academia it makes sense to explore interesting molecular architectures.

Starting from just six simple building blocks, each containing a trifluoromethyl group, the researchers generated nine different cores which were further derivatized at multiple positions to yield 115 diverse fragments. Consistent with diversity-oriented synthesis, no more than five synthetic steps were used for any molecule. All molecules were made as racemates in order to further increase the diversity of the library.

The resulting “3F Library” is mostly rule-of-three compliant, though given that the trifluoromethyl moiety alone adds 69 Da the fragments do tend to be larger, with an average molecular weight of 284 Da. They are, however, less lipophilic than two commercial fluorinated fragment libraries. And with an average Fsp³ = 0.7 and 3.3 chiral centers they are also quite shapely as assessed by principal moment of inertia.

Building a library is nice, but will it provide hits? To find out, the researchers screened the 102 fragments that passed quality control against four targets. They used a transverse (T₂) relaxation assay (specifically, CPMG) in which fragments bound to a protein tumble more slowly, causing a reduction in ¹⁹F signal intensity. Hit rates ranged from 3% to 11%, and about two thirds of these confirmed in STD or WaterLOGSY assays. As seen by the examples shown here, the fragments are quite diverse.

Whether these hits will lead to more potent molecules remains to be seen. Laudably the paper ends with the statement: “we hope that the 3F library will find use for other researchers and we encourage anyone interested in screening the fragments to contact us.” If you are looking for interesting new fragments that are tailored for follow-up chemistry, I encourage you to take the team up on their offer.

Poll results: affiliation and fragment-finding methods in 2019

The fourth iteration of our fragment-finding methods poll has just closed. If you want to jump right to the results feel free to skip the next paragraph, which focuses on methods.

The poll was run using Crowdsignal, the successor to Polldaddy, and ran from 20 October through 30 November. This free polling software tabulates total number of votes for a question but not the number of individual respondents. To determine individual respondents, we included a question on “workplace and practice.” Of the 137 individual respondents to this question, 116 identified themselves as practicing FBLD, and we assumed they also answered the second question. The overall number of responses is slightly higher than in 2013 but a bit lower than in 2016.

Readership demographics have shifted from previous years, with about two thirds of respondents hailing from industry, up from just over half historically. The fraction of respondents who actively practice FBLD is also up modestly, to 85%.

But the question probably of most interest is on screening methods, summarized here.

As we also saw in 2013 and 2016, nearly all fragment-finding techniques are being used more, with the average respondent employing 6 methods today compared with 4.1 in 2016, 3.6 in 2013, and 2.4 in 2011.

X-ray crystallography has leapt to first place, likely driven in part by increasing speed and automation as well as by studies suggesting that crystallography can give impressively high hit rates.

As in 2016, ligand-detected NMR, SPR, and thermal shift assays are all very popular. Use of computational approaches has increased, though perhaps not as much as might be expected given recent advances. Functional screening is the only technique for which use has remained constant, or perhaps even declined very slightly from 2013.

For the first time we asked about use of literature to identify fragments, and nearly a third of respondents said they incorporate previously published fragments into their work. As the amount of publicly available information continues to increase it will be interesting to see whether this number grows.

More niche methods such as mass spectrometry, MST, affinity selection, and biolayer interferometry are gaining adherents; 30 respondents reported using mass spectrometry, for example. While fewer than 20% of respondents are using affinity chromatography (including WAC), CE, or ultrafiltration, that proportion has nearly quadrupled from our previous three polls, though we can’t say which of these related methods accounts for the increase.

Finally, only four respondents reported using “other” methods, such as SHG. Perhaps we’ll ask about this and other emerging methods explicitly next time.

Do the results surprise you, or are they consistent with what you are using at your organization?

Reverse micelle encapsulation for measuring low affinities

NMR is among the more sensitive fragment-finding techniques: the starting point for clinical compound ASTX660 had low millimolar affinity at best. Now, three papers by A. Joshua Wand and colleagues at University of Pennsylvania have taken sensitivity to a new level, enabling the detection of fragments that bind hundreds of times less tightly. (Derek Lowe recently wrote about one of them, and I highlighted a talk last year.)

All three papers focus on a method called reverse micelle encapsulation, in which an aqueous solution of protein and ligand is encapsulated in nanoscale reverse micelles measuring less than 100 Å in diameter. At this size, each micelle will contain at most just a single protein and a few thousand water molecules. Because of the small volume, the protein concentration – and that of any fragments – will be extraordinarily high. The micelles have polar groups pointed inwards towards their watery interior, and their hydrophobic tails point out towards solvent, typically pentane. The overall water content of the sample is typically around 2%.

Various NMR techniques can be used to study the proteins. Although the reverse micelles are larger than the proteins themselves and thus would be expected to tumble more slowly, the low viscosity of the pentane solvent makes up for this, providing high-quality spectra.

The primary paper, in ACS Chem. Biol., focuses specifically on fragments. To establish that the technique can detect weak interactions, the researchers show that they can measure the 26 mM dissociation constant of adenosine monophosphate to the enzyme dihydrofolate reductase.

Next, they turned to the protein interleukin-1β (IL-1β), an inflammatory target with no reported small-molecule binders. One challenge of the method is that hydrophobic fragments could partition into the micelles or even diffuse into the pentane, thus reducing their concentration. To avoid this, the researchers assembled a library of 233 very polar, water-soluble fragments with cLogP values < 0.5. A 2-dimensional NMR screen (¹⁵N-TROSY) using standard conditions (100 µM protein and 800 µM fragment) yielded no hits.

In contrast, NMR screening using reverse micelles with the protein at an effective concentration of 5 mM and fragments at 40 mM yielded 31 hits. Chemical shift perturbations (CSPs) were used to determine where they were binding. Ten of the fragments didn’t show clear binding to specific sites on the protein, but the remaining 21 did, with all but one binding to multiple sites. Of these, 13 also showed non-specific interactions with other regions of the protein. Altogether, the fragment binding sites covered 67% of the protein surface, with the receptor-binding interface particularly well-represented.

Concentration-dependent CSPs were used to determine dissociation constants, which ranged from 50 mM to over 1 M. An SAR-by-catalog exercise was able to improve the affinity of one fragment from 200 mM to 50 mM at one site, though it also binds three other sites with slightly weaker affinity.

The second paper, also in ACS Chem. Biol., uses IL-1β but focuses on the interaction of even smaller molecules such as pyrimidine, methylammonium, acetonitrile, ethanol, N-methylacetamide, and imidazole. Not surprisingly, the dissociation constants are even weaker, averaging 1.5 – 2.5 M.

Finally, a Methods in Enzymology paper goes into depth on how to actually run the experiments, including details on choosing detergents and making the micelles. At high fragment concentrations, for example, pH needs to be carefully controlled.

Five years ago we asked “how weak is too weak” for a fragment. In terms of practicality, I’d say that these fragments qualify. Indeed, the ligand efficiency for the best fragment mentioned above is just 0.15 kcal mol^-1 atom^-1.

But the findings do raise the almost philosophical question of what exactly constitutes a small molecule binding site. Astex researchers reported several years ago that most proteins have more than one, and their more recent work with MiniFrags suggest on average 10 sites at high enough concentrations. Similar results were also reported earlier this month from Monash. Whether or not the fragments from such screens turn out to be immediately useful, they could certainly advance our understanding of molecular recognition.

Fragment-based Drug Design Down Under 2019

The last major fragment meeting of 2019 took place at the Monash Institute of Pharmaceutical Sciences, Monash University, in marvelous Melbourne last week. This was the third Australian meeting devoted to fragments; you can read about the first, in 2012, here. With some 125 participants from four continents, two dozen talks, and nearly as many posters I’ll just try to capture major themes.

Biophysics played a starring role – if you haven’t already voted (right side of page) on which fragment-finding techniques you use please do so. Sarah Piper (Monash) discussed cryo-electron microscopy and showed some lovely high-resolution structures of proteins with bound ligands, though not yet with fragments. Sally-Ann Poulsen (Griffith University) described using native-state ESI mass spectrometry to discover new carbonic anhydrase binding fragments (see here). She uses a 96-well “nanoESI” chip to generate 5 µm droplets as opposed to the ~100 µm droplets typically fed into the instrument. Smaller droplets contain fewer molecules of salt and buffer, and thus generate cleaner spectra.

NMR screening is the go-to method for screening at Monash University, as highlighted by Martin Scanlon and multiple other speakers. Indeed, Monash has built their own version of Astex’s MiniFrag library – their MicroFrags include 92 compounds with 5-8 non-hydrogen atoms. Rebecca Whitehouse has screened these at 300 mM (yes, millimolar) by ¹⁵N-¹H HSQC against the E. coli protein DsbA (EcDsbA) and found numerous hits, including at an internal cryptic site previously identified by Wesam Alwan (Monash). Encouragingly, the results were consistent with a crystallographic screen of the same library done at 1 M.

SPR was highlighted by Nilshad Salim (ForteBio) and in a separate Biacore user day, and is an essential tool for off-rate screening (ORS). ORS facilitates screening of crude, unpurified reaction mixtures, since the off-rate of a compound bound to a protein is not dependent on compound concentration (see here). Compound purification is a major time-sink, and avoiding it is a key component of REFiL, or Rapid Elaboration of Fragments into Leads.

As Bradley Doak (Monash) discussed, REFiL entails the parallel synthesis of compound libraries around a selected fragment in 96-well plates using diverse reagents and high-yielding chemistries such as amide bond formation, alkylation, and reductive amination. Reaction mixtures are evaporated, resuspended in DMSO, and screened using ORS; this has led to affinity improvements of ten-fold or better compared with the original fragment for four projects tested thus far.

Beatrice Chiew (Monash) presented a case study against the oncology target 53BP1. Screening 1198 fragments led, after catalog-mining and rescreening, to 25 hits, all quite weak. Applying REFiL improved affinities by up to 15-fold, with the best molecules around 10 µM. Beatrice noted that because SPR provides “on-chip purification,” active compounds could be identified even when the reaction yields were less than 10%. She did note that examining the raw data (sensorgrams in SPR-speak) is important to recognize and avoid false positives.

Similarly, Luke Adams (Monash) applied REFiL to the bromodomain BRD3-ET. After two cycles, he was able to improve a 230 µM fragment to a 1.5 µM binder. Importantly, the off-rates were similar for the purified molecules and the crude reaction mixtures.

And Mathew Bentley (Monash) is exploring the potential of REFiL using crystallography, or REFiLX. This led to a 60 µM binder against the notoriously difficult EcDsbA. That affinity is more impressive given that the previous structure-based design and synthesis of more than 100 compounds – aided by 25 crystal structures – had failed to break 250 µM.

Vernalis pioneered off-rate screening, and Alba Macias described the company’s latest developments in this area. In the case of tankyrase, a 700 µM fragment was used to generate 80 compounds, which took one chemist a couple days. This yielded a 350 nM binder, the structure of which bound to the enzyme was solved using the crude reaction mixture for soaking.

Following up on this success, Vernalis is exploring the limits of crude reaction mixtures for high-throughput crystallography. Although promising, Alba noted caveats for the two proteins tested. Unlike off-rates, crystallographic success is dependent on compound concentration, so low-yielding reactions can lead to false negatives. And as anyone who has spent time working with fragments can attest, a beautiful co-crystal structure is no guarantee of high affinity, so false positives (ie, no improvement in affinity over the starting fragment) can be a problem too.

Alba also gave a brief summary of the discovery of S64315/MIK665, a fragment-derived MCL-1 inhibitor discovered by Vernalis, Servier, and Novartis that is currently in phase 1.

MCL-1 is a member of the BCL-2 of family proteins, and BCL-2 itself is targeted by the second fragment-derived drug to be approved. Guillaume Lessene (Walter & Eliza Hall Institute) spoke about both of these proteins, as well as BCL-x_L. Long-time readers may remember this selective BCL-x_L inhibitor, discovered using second-site NMR screening. Blocking this protein leads to platelet cell death, but AbbVie researchers are ingeniously side-stepping this liability by conjugating a related small molecule to an antibody to reduce systemic exposure. The resulting ABV-155 may be the first antibody drug conjugate derived from fragments, and was said to be in phase 1.

There was quite a bit more, though in the interest of time (and readers’ patience!) I’ll stop here. But I must note before closing that this meeting launched the Australian Research Council-funded Centre for Fragment-Based Design. This is in some ways an Antipodean version of FragNet, though with a longer (five-year) funding period and the opportunity to include a few postdocs as well as graduate students. If you’re interested, please contact them.

A new tool for detecting aggregation

Historically the most popular method for finding fragments has been ligand-detected NMR. Preliminary results of our current poll (to the right) suggest crystallography has pulled ahead. (Please do vote if you haven’t already done so.) However, NMR has many uses beyond finding fragments, as illustrated in a recent J. Med. Chem. paper by Sacha Larda, Steven LaPlante, and colleagues at INRS-Centre Armand-Frappier Santé Biotechnologie, NMX, and Harvard.

Among the many artifacts that can occur in screening for small molecules, one of the most insidious is aggregation. A distrubing number of small molecules form aggregates in water, and these aggregates give false positives in multiple assays. Unfortunately, determining whether aggregation is occurring is not always straightforward. The new paper provides a simple NMR-based tool to do just that.

All molecules tumble in solution, but small fragment- or drug-sized molecules tumble more rapidly than large molecules such as proteins. The “relaxation” of proton resonances is faster in slower tumbling molecules, and in the NMR experiment called spin-spin relaxation Carr-Purcell-Meiboom-Gill (T₂-CPMG) various delays are introduced and slower tumbling molecules show loss of resonances. Indeed, this technique has frequently been used in fragment screening: if a fragment binds to a protein, it will tumble more slowly, resulting in loss of signal.

The researchers recognized that an aggregate could behave like a large molecule, and they confirmed this to be the case for known aggregators, while non-aggregators did not. The experiment is relatively rapid (~30 seconds), and has been used to profile a 5000-compound library to remove aggregators.

One of the frustrations of aggregators is that it is currently impossible to predict whether a molecule will aggregate, and indeed, the researchers show several examples of closely related compounds in which one is an aggregator while the other is not. Even worse, the phenomenon can be buffer-dependent: the researchers show a fragment that aggregates in one buffer but not in another, even under the same pH.

Many fragment screens are done with pools of compounds, and the researchers find that molecules can show a “bad apple effect”, whereby previously well-behaved molecules appear to be recruited to aggregates.

The limit of detection for T₂-CPMG is said to be single-digit micromolar concentration of small molecule, though the researchers note that double- or triple-digit micromolar concentrations are more practical, which is more typical of fragment screens anyway. And some compounds may show rapid relaxation due to non-pathological mechanisms, such as tautomerization or various conformational changes.

Still, this approach seems like a powerful means to rapidly assess hits, and pre-screening a library makes sense. Another NMR technique using interligand nuclear Overhauser effect (ILOE) has also been used to test for aggregation, though not to my knowledge so systematically. For the NMR folks out there, which methods do you think are best to weed out aggregators?

Second harmonic generation (SHG) vs KRAS

Practical Fragments is currently running a poll on fragment-finding methods used by readers – please vote on the right-hand side. One biophysical method that perhaps we should have included is second harmonic generation (SHG). A recent paper in Proc. Nat. Acad. Sci. USA by Josh Salafsky, Frank McCormick, and collaborators at Biodesy, University of California San Francisco, and elsewhere describes the technique and its application to find fragments that bind to the oncogenic protein KRAS.

In SHG, two photons of the same energy are absorbed by a material which then emits a single photon with twice the energy. In the commercial instrument developed by Biodesy, a powerful 800 nm laser irradiates a dye, and the 400 nm photon it emits is detected. The intensity of the signal is exquisitely sensitive to the precise orientation of the dye. If a protein is labeled with an SHG-active dye and then immobilized on a glass surface, even subtle changes in conformation will be detected.

The researchers chose the G12D mutant form of KRAS, which is one of the most common variants and is associated with particularly aggressive tumors. They labeled the protein with a lysine-reactive SHG dye under conditions in which each protein would, on average, have one covalently-bound dye molecule (though some would have none and others would have more than one). Proteolysis and mass-spectrometry analysis revealed that the dye molecule labeled three different lysine residues, which the researchers viewed as a feature since a ligand causing a conformational change to any of the lysine residues would generate a signal. The researchers also demonstrated that the dye modification did not interfere with the ability of KRAS to bind to the RAS-binding domain of RAF.

Labeled KRAS was then immobilized and tested against several proteins known to bind it, including antibodies and the nucleotide exchange factor SOS. These produced SHG signals, presumably by causing conformational changes to KRAS, while non-binders such as tubulin did not.

Having established that the assay could detect binders, the researchers screened 2710 fragments at 250 and 500 µM, and obtained a whopping 490 hits. These were then triaged by screening at lower concentrations and performing dose-titrations, and 60 were then characterized by SPR.

Fragment 18, 4-(cyclopent-2-en-1-yl)phenol, showed binding by both SHG and SPR, and was further studied by 2-dimensional NMR (¹H-¹⁵N HSQC). This technique allowed measurement of the weak 3.3 mM dissociation constant. More importantly, it allowed the researchers to establish the binding location as being near the so-called “switch 2” region where SOS normally binds. This is the same region where a previous NMR screen had identified the slightly more potent fragment DCAI. The current paper confirmed that finding, though the researchers found evidence that DCAI may bind to other sites too. Docking studies using SILCS suggested that fragment 18 likely binds in a similar orientation as DCAI. Not surprising given the low affinity, the new fragment did not show functional activity in a biochemical screen.

SHG is an interesting approach, and the ability to rapidly assess protein conformational changes distinguishes it from other biophysical techniques. Site-specific labeling would produce more informative data on which regions of a protein move. However, I wonder if SHG is perhaps too sensitive, as evidenced by the large number of hits. Indeed, the researchers demonstrated that the promiscuous lipophilic amine mepazine also generated a strong SHG signal with KRAS. It would be interesting to do a head-to-head comparison with other similarly rapid techniques such as DSF or MST. Have you tried using SHG, and if so, how did it perform for you?

New poll: affiliations and methods

It has been three years since we last asked about fragment-finding methods, and a lot can change in that time – just compare the world today to the world in 2016. Our new poll has two questions (right-hand side, under Links of Utility). Please answer the first, and answer the second if you practice FBLD.

The first question asks your affiliation and whether you actively practice FBLD or whether you are interested in the topic (though hopefully the latter also applies to the former!) We’ve simplified the question from prior years to include just four categories: For-profit practice, For-profit interest, Non-profit practice, and Non-profit interest. For-profit includes pharma and biotech as well as venture capital and consulting. Non-profit includes academia, government labs, disease foundations, and retirement.

Please answer this question as it is the only way we can count the number of respondents, which is essential for determining how many fragment-finding methods people are using on average.

If you do practice FBLD, the next question asks which method(s) you use to find and validate fragments. Please click every method you use, whether as a primary screening technique or for validation. You can read about these methods below, and if you select “other” please describe in the comments.

• Affinity chromatography, capillary electrophoresis, or ultrafiltration
• BLI (biolayer interfermotry)
• Computational screening
• Functional screening (high concentration biochemical, FRET, cell-based, etc.)
• ITC (isothermal titration calorimetry)
• Literature or known fragments
• MS (mass spectrometry)
• MST (microscale thermophoresis)
• NMR – ligand detected
• NMR – protein detected
• SPR (surface plasmon resonance)
• Thermal shift (or DSF)
• X-ray crystallography
• Other

Please forward this so we can get as many responses as possible.

Let the voting begin!

Fragments vs Mycobacterium TrmD – and native ESI-MS revisited

The Mycobacterium genus of bacteria contains more than 190 species, including the pathogens that cause leprosy and tuberculosis. Mycobacterium abscessus (Mab) is not as famous, though it is a growing concern for people living with cystic fibrosis. The enzyme tRNA (m¹G37) methyltransferase (TrmD) is essential for Mycobacterium growth. In a recent open-access paper in J. Med. Chem., Chris Abell, Tom Blundell, Anthony Coyne, and collaborators at University of Cambridge, Royal Papworth Hospital, and the US NIH describe potent inhibitors of this target.

The researchers screened 960 fragments using differential scanning fluorimetry (DSF). The 53 hits were soaked into crystals of TrmD, and 27 yielded electron density – all in the cofactor (S-adenosyl methionine, or SAM) binding site. One weak but particularly ligand-efficient fragment was advanced through fragment growing to low micromolar affinity, but further improvements proved challenging.

A powerful feature of FBLD is that screens often yield multiple starting points, and this proved useful here. Specifically, compound 16 bound in the pocket where the adenine of SAM normally sits, while compound 20 bound nearby in the ribose pocket. Merging these led to compound 23,  and a crystal structure suggested that the indole nitrogen would be a good vector for fragment growing, resulting in compound 24c. Addition of a positively charged moiety to engage a couple glutamic acid residues led to compound 29a, a mid-nanomolar compound as assessed by isothermal titration calorimetry (ITC).

Several of the molecules were characterized by native electrospray ionization mass spectrometry (ESI-MS) – an interesting but somewhat controversial technique in which protein-ligand complexes are gently ionized and detected in the gas phase. For large molecules such as proteins, even the gentlest ionization will generate highly charged species; in this case the dimeric TrmD protein had between 13 and 17 positive charges. Each of these charge states represents a different species, or perhaps several since the positive charges could be on different regions of the protein.

Compound 24c has modest affinity, but encouragingly, native ESI-MS at 100 µM ligand revealed two bound ligands, as expected for a protein dimer. However, for the much more potent compound 29a, native ESI-MS at 100 µM ligand revealed two bound ligands for a high charge state but largely unbound protein for a lower charge state. The researchers speculate this could be due to dissociation of the positively charged ligand in the gas phase.

Overall this paper is a nice exercise in fragment merging and growing, guided by crystallographic data. Some of the molecules elaborated from 24c (including 29a) inhibited growth of Mycobacteria species, though further improvements in affinity will be needed, and no DMPK properties are provided. Still, it's good to see solid work being done on new antibacterial agents.

The disconnect between observed affinity by native ESI-MS and that observed by ITC does make one cautious about the utility of the former technique. We’d welcome comments on your experiences with it.

The story of erdafitinib – abridged

In April we celebrated the FDA approval of erdafitinib for certain types of urothelial carcinomas with genetic alterations in the FGFR2 or FGFR3 genes. This marked the third launch of a fragment-derived drug. However, although erdafitinib was mentioned in a 2016 review, the fragment origins have remained obscure. The full story has yet to be published, but Chris Murray (Astex), David Newell (Newcastle University) and Patrick Angibuad (Janssen) have recently published brief highlights in MedChemComm.

The story begins all the way back in 2006, with a collaboration between Astex and the Northern Institute of Cancer Research at Newcastle University. The four fibroblast growth factor receptors (FGFR1-4) are kinases whose aberrant activation was known to be associated with multiple cancers. In those days most FGFR inhibitors also inhibited VEGFR2, which was linked to unacceptable side effects. A fragment screen in 2008 led to a series of potent, selective molecules and enticed Janssen to join the collaboration. This particular lead series was ultimately discontinued, though, as discussed below, it turned out to be important.

Meanwhile, fragment 1 was identified, characterized crystallographically bound to FGFR1, and improved to low micromolar compound 2. (Note: the structure of Fragment 1 has been corrected.) Virtual screening and medicinal chemistry led to compound 4, with improved selectivity for FGFR3 over VEGFR2. Superimposing the crystal structure of this compound with the previous series suggested building off the aniline nitrogen to improve potency (and, one might assume, solubility). This ultimately led to erdafitinib.

The researchers highlight the cooperative nature of this project. “Each team brought their specific expertise and most importantly, wisely and collaboratively capitalized on each other’s strengths.”

The publication also illustrates that success can take time – thirteen years in this case. This is not unusual for drug discovery, and is in fact halfway between the remarkably fast six years for vemurafenib, the first fragment-based drug approved, and the two decades required for venetoclax, the second.

But in the end good things are worth waiting for, and the 15-20% of metastatic bladder cancer patients with an FGFR alteration now have a new treatment option. Erdafitinib is being tested in at least a dozen other clinical trials for various cancers. Practical Fragments wishes all the participants the best of luck.

Combining fragments and HTS hits to target PHGDH

Boehringer Ingelheim has been on something of a tear reporting new chemical probes for difficult targets – see here for their NSD3 inhibitor and here for their RAS inhibitor. This is part of an ambitious effort to develop probes for the entire human proteome by 2035. In a new paper published in J. Med. Chem., Harald Weinstabl and collaborators at BI and Shanghai ChemPartner describe the discovery of BI-4924, a potent inhibitor of phosphoglycerate dehydrogenase (PHGDH).

The enzyme is the rate-limiting step in serine synthesis, and has been implicated in multiple types of cancers. However, metabolic enzymes such as PHDGH are particularly challenging drug targets for several reasons: cofactors such as NADH are present at high concentrations in cells, the substrate binding pocket is both shallow and polar, and one often needs near complete inhibition to see an effect. Thus, the researchers chose multiple approaches.

An STD NMR fragment screen was conducted against the apo form of the protein (250 µM fragment and 20 µM protein) to find compounds that would bind in the NAD⁺-binding site. Of 1860 fragments screened, 60 hits were identified, and 19 of these gave measurable dissociation constants in an SPR assay and were selective against two other proteins. Compound 9 was found crystallographically to bind in the adenine pocket of the NAD⁺-binding site. Fragment growing was challenging due to the “kinked shape” of this pocket: elaborated molecules tended to point out of the pocket into solvent. Careful design led to modest improvements in potency (compound 11), and adding a negatively charged moiety led to potent molecules such as compound 43. To avoid problems with permeability, the researchers tried various uncharged bioisosteres, but these were not tolerated. Interestingly, crystallography revealed that the carboxylic acid does not seem to make specific interactions with the protein; its necessity may be due to long-range electrostatic interactions with multiple nearby basic residues.

In parallel, a biochemical HTS screen of more than a million molecules yielded 27,000 hits, which were whittled down to 11,250 that confirmed and didn’t interfere with the assay. Removing PAINS and large, lipophilic molecules narrowed the set to 4750 compounds. Further rigorous assessment included biophysical methods, as recently recommended. Aware of the potential for metal contaminants to give false positives, the researchers examined select samples with inductively coupled plasma mass spectrometry and found that some contained mercury or copper, which inhibited the enzyme. Ultimately 77 hits were validated with dissociation constants better than 300 µM, including compound 8, which crystallography revealed binds in a similar manner to fragment 9.

Combining information from both campaigns and growing to engage an aspartic acid side chain ultimately led to BI-4924. This compound is soluble, stable, and selective against other dehydrogenase enzymes. Unfortunately, the carboxylic acid moiety does indeed impart low permeability, and perhaps because of this the molecule has only low micromolar activity in cells. However, the ethyl ester (BI-4916) transiently accumulates in cells and modulates serine levels.

Unfortunately, the researchers appear to have been scooped; as they politely note, “subsequently, these findings were independently confirmed….” As it stands BI-4916 is too unstable for use in vivo. Still, it could be useful for further unraveling the biology around serine biosynthesis and its role in cancer cells, and the paper itself stands as a nice example of structure-based lead design combining information from multiple sources.

Seventeenth Annual Discovery on Target

Just as April brings CHI’s Drug Discovery Chemistry in San Diego, September brings CHI’s Discovery on Target in Boston, and last week saw more than 1100 attendees attend 14 tracks over three days, along with associated training seminars and short courses. Fragments made appearances throughout.

Perhaps the most notable new development was an entire session devoted to covalent fragments. We spent three blog posts in June covering five papers on this topic, so this was a timely addition.

Eranthie Weerapana (Boston College) described proteomics methods for analyzing cysteine modification in cells. (Some of this is similar to Ben Cravatt’s work, which we discussed here.) She noted that many mitochondrial proteins have low abundance and are thus hard to detect, but by isolating the mitochondria she has been able to observe about 1500 cysteines on 500 proteins. Selenocysteine-containing proteins are even less common and can thus be lost in the noise, but by lowering the pH these rare beasts can be labeled selectively.

Alexander Statsyuk (University of Houston) gave two wide-ranging talks, and he noted that (beyond acrylamides) there are about 50 warheads that have not been widely explored. We’ve previously covered his work with fragments containing the 4-aminobut-2-enoate methyl ester, and Katrin Rittinger (Francis Crick Institute) discussed her recent use of a small library of just 104 of these to find a covalent inhibitor of LUBAC. And on the subject of very recent papers, I presented work from Carmot and Amgen on the discovery of covalent KRAS^G12C inhibitors.

You know a field is becoming popular when suppliers start selling reagents, and this is certainly the case for covalent fragments: Enamine and Life Chemicals both sell covalent fragment libraries. If you’ve had experience with them or others, please leave comments!

Natalia Kozlyuk (Vanderbilt) gave a nice presentation that emphasized some of the challenges that can arise in FBLD. Screening 14,000 fragments against the protein RAGE by NMR resulted in a number of hits, and crystallography revealed that a couple of these bind just 4 Å apart. Linking these together with variable-length linkers led to one molecule that bound as expected, while in another one of the linked fragments bound in a third site. Unfortunately, even the best dimeric molecules are quite weak, and they also cause the protein to precipitate. This appears to be a different mechanism from “classic” small molecule aggregation, in which the small molecules first form aggregates that block protein activity, though it is not unprecedented.

In a similar vein, Stijn Gremmen and Jan Schultz (ZoBio) described work done with Gotham Therapeutics to discover inhibitors of the methyltransferase complex METTL3/METTL14. HTS-derived compounds caused aggregation of protein as assessed by size exclusion chromatography and multiangle light scattering, but a fragment screen followed by optimization led to potent, well-characterized molecules.

Continuing the theme, Beth Knapp-Reed (GSK) described an HTS assay against the anti-cancer target LDHA in which 1.9 million compounds yielded 560 hits, almost all of which turned out to be false positives due to oxalic acid contamination. Fragment screening was more productive, yielding 16 crystallographically validated hits, and the researchers were able to improve the affinity more than 10,000-fold. And Puja Pathuri discussed successful efforts at Astex to discover ERK1/2 inhibitors (see here for more details).

Finally, last year we noted the increasing number of talks on PROTACs and targeted protein degradation. This year for the first time the conference held a full day and a half long program on the topic. PROTACs typically consist of two-component molecules in which one piece binds to a target of interest and the other piece binds to a protein called an E3 ligase which ultimately causes proteolytic degradation of the target. For example, Michael Plewe described how he and his colleagues at Cullgen have modified the fragment-derived vemurafenib to target an oncolytic mutant form of BRAF.

One of the interesting features of targeted protein degradation is that a high affinity ligand is not always necessary: Craig Crews (Yale) described how an 11 µM p38α ligand was sufficient to degrade 99% of the protein in cells. And fragments can help not just with target proteins, but in identifying new E3 ligase ligands as well. Carles Galdeano (University of Barcelona) described how fragment screening has yielded a couple sub-micromolar affinity ligands of an E3 ligase called Fbw7.

In the interest of time I’ll stop here, but if you were particularly struck by anything please mention it in the comments. And if you missed the meeting, be sure to mark September 15-18 on your calendar for next year!

Fragments find flexibility in fascin 1

Protein flexibility can be both an opportunity and a barrier – quite literally, when a solid wall of protein seems to block opportunities for fragment growing. But like secret doorways, protein domains can yawn open to expose tunnels and cavities. An example of this was published earlier this year in Bioorg. Med. Chem. Lett. by Stuart Francis and collaborators at the CRUK Beatson Institute.

The researchers were interested in fascin 1, which increases the invasiveness of multiple cancers by helping pack filamentous actin into bundles important for cell migration. The team began their search for an inhibitor by performing a surface plasmon resonance (SPR) screen of 1050 fragments, generating an impressive 53 hits. Although a number of these were reported to bind to multiple sites on the protein, only one is discussed.

Compound 1 binds between two domains of the protein in a pocket that does not exist in unbound fascin. However, the fact that the pocket completely envelopes the fragment “hampered attempts to develop the series.” Fortunately, the researchers were following the patent literature, and when they characterized compound 2 (not a fragment, and reported by a different group), they discovered that while it binds in the same pocket as compound 1, additional conformational changes occur to accommodate the larger molecule.

Next, the researchers looked for analogs of compound 2 and also performed a virtual screen against the enlarged pocket. Of 110 commercial compounds tested, three gave dissociation constants better than 100 µM, including compound 3. The researchers recognized that compound 3 lacks the halogens found on both the original fragment and compound 2, and by adding these they were able to improve the affinity more than ten-fold. Further optimization ultimately led to BDP-13176, with mid-nanomolar affinity by SPR and ITC as well as activity in a functional assay. Although the molecule has reasonable solubility and stability against liver microsomes, it has low permeability and high efflux.

This is a nice structure-based design story, and while the fragment did provide some information about the binding site, one could argue that the real breakthrough came with determining the binding mode of compound 2. Indeed, without this information, it would have been all too easy to assume that the pocket was not ligandable. This is an important reminder that crystal structures usually only reveal one form of a protein. The system is also a good test case for modelers who want to see how their algorithms perform against a dynamic protein. Breakthroughs are often unexpected, and it is always worth making a few compounds that don’t look like they’ll fit.

Fragments vs sepiapterin reductase, via 19F NMR

It has been two years since we’ve had a post devoted to fluorine NMR. Though I don't share Teddy’s “fetish” for ¹⁹F-based screening, I do think the technique can be quite powerful, as demonstrated in a recent J. Med. Chem. paper by Jo Alen, Markus Schade, and their colleagues at Grünenthal GmbH.

The researchers were interested in sepiapterin reductase, which is abbreviated as SPR but which I’ll spell out to avoid confusion with surface plasmon resonance. This enzyme performs the last step in the production of tetrahydrobiopterin, an essential cofactor for multiple enzymes, including some that synthesize neurotransmitters and produce nitric oxide. Sepiapterin reductase has been proposed as a target for non-opioid-based pain medications.

The primary assay involved displacement of a fluorine-containing inhibitor that binds in the substrate site of the enzyme; thus, the researchers could use ¹⁹F NMR without requiring fluorinated fragments. A total of 4750 fragments were screened at 250 µM, initially in pools of 12. The 26 hits were then tested in an enzymatic assay, and 21 showed activity better than 75 µM. The best, compound 3, was sub-micromolar.

Crystal structures were obtained for six compounds, including compound 3, and all bound in the substrate pocket as predicted from the original displacement assay. The phenolate of compound 3 makes hydrogen bonds to two critical catalytic residues. Not surprisingly, capping this moiety with a methyl group led to an inactive compound. The researchers made dozens of variants, but aside from compound 26, most of these were disappointingly less active. Compound 26 does show good solubility and permeability, though no cell data are provided, and the phenol will likely be glucuronidated in vivo.

This is a nice story that illustrates a not-infrequent frustration: after identifying the initial nanomolar hit from a small library, the researchers likely thought improving potency still further would be easy. Instead, it took more than 60 analogs just to gain another order of magnitude. That said, 57 nM is nothing to sneeze at. And this situation is certainly preferable to the more common alternative of starting with a weak fragment that remains weak no matter what you do to it!

Fragment vs hematopoietic prostaglandin D2 synthase: a chemical probe

Six years ago we highlighted work out of GlaxoSmithKline and Astex describing some of their efforts to find inhibitors of hematopoietic prostaglandin D2 synthase (H-PGDS), an enzyme implicated in asthma, lupus, and multiple other inflammatory diseases. A recent paper in Bioorg. Med. Chem. by David Deaton and collaborators describes another chemical series from that program.

As noted in the earlier publication, the researchers were graced with 76 crystallographic fragment hits, of which compound 1a was a weak but ligand-efficient member. Several other fragments that bound in the same region contained a methoxy group, and a quick survey of commercially available analogs led to compound 1b, with a nice bump in potency. Replacing the nitrile with an amide (compound 1d) improved activity further.

What do you do when you’ve got an amide? Make lots of them! This was effective, and the researchers show more than 60 analogs leading to low nanomolar inhibitors such as compound 1bg. One problem with amides is that they can be enzymatically cleaved in vivo, but this challenge was surmounted by tweaking the substituents.

The researchers also noted that the compounds are quite electron-rich, potentially leading to phototoxicity, and in fact some of the molecules degraded upon exposure to UV light. Also, methoxy substituents are prone to dealkylation in vivo. The researchers solved both problems by replacing the methyl with a difluoromethyl group, leading to GSK2894631A.

This molecule was put through a battery of tests and found to be orally bioavailable (at least in mice) with good pharmacokinetics. It is selective against related enzymes as well as a larger panel of receptors and transporters. Encouragingly, the compound showed potent activity in a mouse model of acute inflammation. In other words, this looks to be a useful chemical probe to explore the biology of prostaglandin signaling.

This is a nice story on several levels, and it also illustrates an important point that younger researchers and folks in academia sometimes overlook: it can take ages before work done in industry sees the light of day. Indeed, one of the authors on the paper left Astex more than five years ago, so the work described is likely several years older than that. Still, better late than never. Good science is always worth publishing, even if – like another paper we recently highlighted – it happened some time ago.