15 October 2018

FBLD 2018

Ten years ago, Vicki Nienaber (Zenobia) enlisted a small group of fellow enthusiasts to help her organize an independent fragment-based lead discovery conference in San Diego. That event was so successful that it was repeated in York in 2009, Philadelphia in 2010, San Francisco in 2012, Basel in 2014, and Cambridge (USA) in 2016. Last week, to celebrate its first decade, Derek Cole (Takeda), Rod Hubbard (University of York) and Chris Smith (COI) brought FBLD 2018 back to San Diego, along with some 200 fragment fans. With around 30 talks, more than 40 posters, and nearly 20 exhibitors, I won’t attempt to present a comprehensive overview, but just focus on broad themes.

Success Stories
I estimate that, in 2008, 14 fragment-based programs had entered the clinic, none of which had advanced beyond phase 2. That list has now grown to more than 40, so naturally success stories were a focus.

Andy Bell (Exscientia) discussed NMT inhibitors for malaria and the common cold (see here); the AI-driven approach took < 500 molecules to get to molecules with animal efficacy. Steve Woodhead (Takeda) revealed potent inhibitors of TBK1, a kinase involved in the innate immune response. It took just three months to go from a fragment hit to an animal-active lead, though unfortunately that molecule also showed apparent on-target toxicity. And Rosa María Rodríguez Sarmiento (Roche) described the discovery of COMT inhibitors (see here).

Mary Harner (BMS) described the discovery of sub-micromolar KAT II inhibitors in just a few months, enabled by parallel chemistry and the synthesis of 833 compounds. Several series turned out to be aggregators, and BMS has instituted a routine β-lactamase screen (an enzyme particularly sensitive to aggregators) to catch these early.

Keith McDaniel (AbbVie) described the discovery of the BET-family bromodomain inhibitor ABBV-075. This program also made rapid progress: just six months from the initial fragment hit, although the team did spend another year trying to find better molecules. This effort eventually paid off, as the same fragment has now led to a BD2-selective molecule, ABBV-744, that has recently entered the clinic.

And Paul Sprengeler (eFFECTOR) described the discovery of eFT508. This too was a rapid success: just 1 year and 170 compounds, enabled by 30 co-crystal structures, and in the end a dozen molecules competing for candidacy.

Notice that many of these projects moved quickly. Feel free to send this summary to anyone who worries that fragment programs move too slowly to be practical.

Technologies have always had a starring role in FBLD conferences, and this one was no exception. Ben Cravatt (Scripps) discussed his fragment-based target discovery methods (see here and here). As I speculated recently, he is now using these approaches to discover new protein degraders. And his "fully functionalized fragments" are being adopted by others, as described in a poster by Emma Grant and collaborators at GlaxoSmithKline and University of Strathclyde.

Surface plasmon resonance (SPR) was used routinely by many of the speakers, but there is plenty of room for innovation. John Quinn (Genentech) described how to extend kinetic measurements to the very fast and the very slow. John also noted that gathering kinetic data earlier to deprioritize series with slow on-rates may be wise. And for those who wonder about the limits of detection for SPR, John measured the affinity of imidazole for NTA: just 13.6 mM!

Miles Congreve (Sosei Heptares) described multiple methods applied to GPCR targets along with a number of success stories. He also noted that, in the PAR2 program we mentioned recently, fragments were able to identify a buried pocket that could not be found using DNA-encoded libraries of several billion members, presumably because the pocket would not be accessible to a DNA-bound ligand. Interestingly, this pocket could be detected computationally using FTMap, as shown in a poster presented by Amanda Wakefield (Boston University).

Pedro Serrano (Takeda) described a variety of biophysical methods applied to GPCRs, the most stunning of which is an SPR microscope capable of performing kinetic binding assays on whole cells. He has tested this Biosensing Instrument on four different GPCRs, and although there are technical challenges, the data seem usable.

But the light shone most brightly on crystallography, illuminated by Stephen Burley (Protein Data Bank) among others. In order to justify continued public funding and free access (yes, there were suggestions to put the PDB behind a paywall), the PDB was asked to demonstrate its usefulness to society. Their analysis found that of the 210 new molecular entities (NMEs) approved by the FDA from 2010 through 2016, 184 had PDB entries for the target and/or the NME – for a total of 5914 structures, 95% of which were crystallographic. Most of these structures had been deposited at least 10 years before the drug was approved, so in many cases they probably played an important role.

John Barker described how Evotec has jumped into high-throughput screening by crystallography in a collaboration with the Diamond Light Source, which is now capable of doing 700 soaks per day. They have run 10 screens over the past 18 months with a small library of 320 fragments, with hit rates typically around 8%.

We have written about how high concentrations can improve success in crystal soaking experiments, and both Chris Murray and Dominic Tisi of Astex described how they’ve taken this to an extreme: 1 M soaks, with the fragment dissolved directly in the soaking solutions. Obviously this requires highly soluble fragments, so they’ve built a library of 81 “MiniFrags” having on average just 6.4 non-hydrogen atoms. They have tested these against five targets that diffract to high resolution and have found impressively high hit rates of 20-60%, compared to the 2-20% in the original 100 mM soaks for the same targets. Some of the sites are exploited by previously reported inhibitors or substrates, while others are new. And while the “universal fragment” 4-bromopyrazole did well, 1,2,3-triazole did even better – binding to all five targets in a total of 22 sites.

Crystallographers should not become complacent. Gabe Lander (Scripps) gave an update on cryo-EM, which we’ve written about here. The number of cryo-EM structures deposited in the PDB eclipsed those from NMR in 2016, and resolution continues to improve, with the current (as of late September) record at 1.56 Å. Still, the technique is not nearly as fast as crystallography: best case is 8 hours from data collection to refinement, although Gabe did think that 10 structures per day would be possible within the next few years. And Chris Murray noted that, if present trends continue, “we’ll all be doing cryo-EM in five years’ time.” Backing this up, he showed what I suspect may be the first clear density map of a fragment bound to a test protein.

This was the last major fragment event of the year, but next year’s calendar is already shaping up nicely. And mark your calendar for September 2020, when FBLD 2020 will move to the original Cambridge (UK).

06 October 2018

Fragments in the clinic: 2018 edition

To celebrate FBLD 2018, we're updating the list of FBLD-derived drugs. The current list contains 40 molecules - 25% more than the last compilation two years ago. As always, this table includes compounds whether or not they are still in development (indeed, some of the companies no longer even exist). Drugs reported as still active in clinicaltrials.gov, company websites, or other sources are in bold, and those that have been discussed on Practical Fragments are hyperlinked to the most relevant post.


VenetoclaxAbbVie/GenentechSelective Bcl-2
Phase 3

PLX3397PlexxikonCSF1R, KIT
Phase 2

AT9283 AstexAurora, JAK2
IndeglitazarPlexxikonpan-PPAR agonist
Navitoclax (ABT-263)AbbottBcl-2/Bcl-xL
Phase 1

ABT-518AbbottMMP-2 & 9
AT13148AstexAKT, p70S6K, ROCK
AZD5099AstraZenecaBacterial topoisomerase II
BI 691751Boehringer IngelheimLTA4H
MAK683NovartisPRC2 EED

I have no doubt that this list is incomplete, particularly in Phase 1. If you know of any others (and can mention them) please leave a comment.

01 October 2018

Sixteenth Annual Discovery on Target

CHI’s Discovery on Target took place in Boston last week. With >1300 attendees from over two dozen countries, this is the older, larger cousin of the San Diego DDC meeting; at some points ten tracks were running simultaneously. Although more heavily focused on biology, there were still plenty of talks of interest to fragment folks.

Michael Shultz (Novartis) provocatively asked “do we need to change the definition of drug-like properties?” Long-time readers will recall that his earlier papers on ligand efficiency led to considerable debate, which seems to have been settled to everyone’s satisfaction with the exception of Dr. Saysno.

His new study, which has just published in J. Med. Chem., analyzes the molecular properties of all 750 oral drugs approved in the US between 1900 and 2017. Contrary to what strict rule of five advocates might expect, the molecular weight has increased over the past couple decades, as has the number of hydrogen bond acceptors. In contrast, the number of hydrogen bond donors (#HBD) has remained constant, suggesting that this may be more important for oral bioavailability. (Indeed, #HBD is the only Lipinski rule not broken by venetoclax.) Although Shultz did not examine “three dimensionality,” he laudably includes all the raw data – including SMILES – in the supporting information. This will be a useful resource for data-driven debates.

Molecular properties are carefully considered by Ashley Adams, who discussed the four fragment libraries used at AbbVie. The first is a 4000-member “rule of three” compliant library. For tougher targets, a 9000-member Ro3.5 library is available, as is a specialized fluorine library for 19F NMR (2000 members) and a 1000-member “biophysics” library, in which all compounds are less than 200 Da. Fragment optimization is often challenging, and since the C-H bond is most common but perhaps least explored, the AbbVie database is annotated with references on C-H bond activation relevant to each fragment.

Anil Padyana spoke about the metabolic enzymes being targeted at Agios. As we mentioned recently, these are very difficult targets, so the researchers often use parallel (as opposed to nested) screening using different techniques to minimize false negatives. Anil also described an interesting SPR assay in which fragments were introduced to the protein after the addition of an activating substrate.

High-quality protein constructs are essential for any fragment screen, and Jan Schultz described ZoBio’s technology for generating these. The company’s “protein domain trapping” approach entails high-throughput generation and screening of tens or hundreds of thousands of truncations of a given protein and rapidly selecting stable, high-expressing, and active variants.

Trevor Perrior mentioned that Domainex has a similar technology, which has been able to produce soluble protein domains in 90% of its attempts. Trevor also described a separate project in which a 656-fragment compound library was screened using SPR against the enzyme RAS. They found fragments that bind in a previously discovered site but, unlike the earlier work, the Domainex researchers were able to optimize these to nanomolar inhibitors.

Another success story was presented by Dean Brown (AstraZeneca), who described a collaboration with Heptares to discover inhibitors of protease-activated receptor 2 (PAR2). As the name suggests, this GPCR is activated when a protease cleaves the N-terminus, allowing the remaining N-terminal residues to fold back and activate the GPCR. The researchers used a stabilized form of PAR2 in an SPR screen of 4000 fragments and obtained >100 binders in multiple series. This led to AZ8838, which blocks signaling by binding in an allosteric pocket. It also has a slow off-rate, which is often an attractive feature – particularly in the context of intramolecular activation.

A number of talks were focused on protein degraders such as PROTACs (PROteolysis-TArgeting Chimeras). These are generally two-part molecules connected by a linker: one part binds to a target of interest, while the other engages the cellular degradation machinery to destroy the target. As Shanique Alabi, a graduate student in Craig Crews's lab at Yale demonstrated, the molecules are catalytic – a single PROTAC molecule can cause the destruction of multiple copies of a target protein. This “event-driven” pharmacology is thus different from most historical drugs, which are “occupancy-driven.” Is there a role for fragments?

One of the strengths of FBLD is that if a ligandable site exists, it can be found. As Astex demonstrated, the majority of proteins seem to have secondary sites, away from the active site. Although some of these may be allosteric, others probably have no functional activity, particularly in the case of protein-protein interactions where secondary sites may be located some distance from the interface. The power of degraders is that non-functional sites can be made functional. The power of FBLD is that it can find small-molecule binding sites, which could then be used as anchoring sites for one side of a degrader. Watch this space!

24 September 2018

Fragments in the clinic: Asciminib

Imatinib is the early poster child of personalized medicine. The drug famously works by binding to the mutant kinase BCR-ABL1, and its approval by the FDA for chronic myelogenous leukemia in 2001 arguably launched hundreds of programs targeting kinases. Although imatininb is remarkably effective, resistance sometimes develops, usually caused by mutations that lead to loss of affinity for the drug. Imatinib and other approved drugs that target BCR-ABL1 all bind in the active (ATP-binding) site of the kinase, and they all have various off-targets that can lead to toxicity. To sidestep these issues, researchers at Novartis have developed an allosteric inhibitor, as described by Andreas Marzinzik and colleagues in a new paper in J. Med. Chem.

The ABL1 kinase is naturally autoinhibited by the binding of a myristoyl group to an allosteric pocket. Although the pocket exists in BCR-ABL1, the site that is normally myristoylated is lost. The researchers wanted to create a molecule that would mimic the function of the myristoyl group and exert its inhibitory effect within the allosteric pocket.

An NMR-based screen of 500 fragments yielded 30 hits – perhaps not a surprisingly high hit rate given the lipophilicity of the pocket. Compound 2, with low micromolar affinity, had a high ligand efficiency. Unfortunately, it and similarly high-affinity fragments showed no cell-based activity. A crystal structure of compound 2 bound to the protein revealed that, although the fragment binds in the myristate pocket, its binding mode would actually prevent the conformational change necessary for allosteric inhibition. Tweaking and growing the fragment led to molecules such as compound 4, which were still inactive.

To determine why, the researchers developed a clever NMR assay based on a specific valine residue located in a disordered region of the protein that becomes helical in the allosterically inhibited state of the protein. This assay allowed them to distinguish which protein conformation molecules bound and revealed that, contrary to design, compound 4 did not in fact bind to the inhibited form of the protein. Other researchers had found a different series of molecules that also bind in the myristate pocket, and these all contained a trifluoromethoxy group. When this moiety was grafted onto compound 4, the resulting compound 5 showed cell-based activity.

Now the medicinal chemistry began in earnest. Crystallography revealed a lipophilic cleft in the allosterically inhibited form of the protein which could be filled with a pyrimidine, and the cationic solubilizing group in compound 5 was replaced by the neutral moiety in compound 7. This compound showed some hERG channel inhibition, which could be fixed by replacing the pyrimidine with a pyrazole. Also, crystallography revealed that there was a little extra space near one of the fluorine atoms, which could be replaced with a chlorine in the clinical compound asciminib (ABL001). A crystal structure of this molecule shows it binding to the inactive conformation of the protein (the helix that forms is in the upper right).

Asciminib effectively inhibits proliferation of cells containing either wild-type or T315I BCR-ABL1, the latter being one of the more pernicious resistance mutations. The compound is also highly selective against > 60 other kinases, and is only active against CML cell lines in a panel of 546 cancer cell lines, suggesting that it should be well tolerated. Mouse xenograft models were also impressive, and the compound is currently in a phase 3 clinical trial.

This is a thorough, clearly written account combining biophysics, modeling, chemistry, and biology to discover a first-in-class drug. It is also a useful reminder that binding alone may not be sufficient to cause desired effects. As with all the clinical-stage programs, Practical Fragments wishes everyone involved the best of luck!

17 September 2018

Fragments in the clinic: ASTX660

Three years ago we highlighted a paper from Astex describing the discovery of an extraordinarily weak fragment and its advancement to a dual inhibitor of the anti-cancer targets cIAP1 and XIAP. We ended that post by writing, “whether or not this leads to a drug, it does look like another candidate for a useful chemical probe.” As three papers now make clear, the program has indeed led to an experimental drug.

The first paper, by Emiliano Tamanini and colleagues, was published in J. Med. Chem. last year and describes the optimization of Compound 21, one of the best compounds from the 2015 report. The researchers noticed that compounds in the series were chemically unstable: the amide bond was subject to hydrolysis. Fortunately this was readily fixed by repositioning the pyridyl nitrogen.

Optimization of the benzyl group was complicated by the fact that it binds in the P4 pocket, which differs between cIAP1 and XIAP. In the end, adding a fluorine gave a slight potency improvement against both proteins. The bulk of the work was focused on elaborating the methoxy group of compound 21. Detailed modeling experiments were used to choose moieties that would fold back on the core of the molecules in solution, thus pre-orienting them for binding as well as shielding a critical hydrogen bond. These efforts led to AT-IAP, with low nanomolar cell activity against both proteins as well as activity in mouse xenograft models.

Although AT-IAP is orally bioavailable in mice and rats, the bioavailability is much lower in monkeys, and it also inhibits the hERG channel, which can lead to cardiac toxicity. Fixing these problems is the focus of a paper published last month in J. Med. Chem. by Christopher Johnson and colleagues.

Metabolite identification studies revealed that the morpholine ring of AT-IAP is cleaved by CYP enzymes, so this was one area the researchers tried to modify. Although somewhat successful, hERG was still a problem, and this correlated with lipophilicity. Knowing how the molecules bound allowed the researchers to introduce small hydrophilic substituents without disrupting critical interactions, ultimately leading to ASTX660. Not only did the added hydroxymethyl group decrease hERG binding, it also improved bioavailability – a reminder that decreasing lipophilicity can have useful effects even on distant parts of the molecule.

More characterization of ASTX660 is provided in a paper by George Ward and colleagues in Mol. Canc. Ther. This reports the crystal structure of the molecule bound to XIAP. As Johnson et al. note, the polar interactions made by the molecule are conserved from the original fragment – the additional protein interactions that improve affinity by more than a million-fold are all hydrophobic.

Ward et al. also provide more detailed mechanistic cell biology, pharmacokinetics, and xenograft data. In particular, ASTX660 is a much more potent antagonist of XIAP activity in vivo than other clinical-stage compounds, which will hopefully translate to better efficacy. The compound is currently in a phase 1-2 study.

Collectively these papers provide a valuable lesson in structure- and property-based drug design and illustrate just how much effort can be required to go from fragment to clinical compound. I’ll end this post with an echo of the original: whether or not this leads to an approved drug, it is a lovely story of perseverance combined with creative chemistry and biology. Practical Fragments wishes everyone involved the best of luck.

10 September 2018

Fragment flipping during optimization

Last month we highlighted a study that asked how often the binding mode of a fragment changed during optimization. A new paper in J. Med. Chem., by Swen Hoelder and collaborators at Institute of Cancer Research, University of Oxford, and Universitat de Barcelona provides an interesting case study.

The researchers were interested in the kinase ALK2, which is implicated in an aggressive and universally fatal childhood cancer called diffuse intrinsic pontine glioma. They started by screening a library of fragments designed to target kinases, which yielded compound 1. This compound actually contains two moieties that are known kinase hinge binders, a quinazolinone and a pyrazole. Unfortunately the researchers could not obtain a crystal structure of the fragment bound to ALK2, but SAR suggested that the pyrazole was not essential, and indeed replacing this with a quinoline led to compound 7, with sub-micromolar activity.
Next the researchers introduced methyl groups at various positions around the quinazolinone and found that these neither significantly improved nor decreased binding. Modeling based on similar reported molecules led them to grow the molecule towards solvent, ultimately leading to the mid-nanomolar compound 16 (blue in figure below), which they characterized crystallographically bound to ALK2. The molecule bound as expected, with the unsubstituted nitrogen of the quinazolinone forming a hydrogen bond to the hinge region of the kinase.

So far so good, but the researchers were still curious about some of their earlier SAR. In particular, the methyl groups added to some of the molecules should have been incompatible with the observed binding mode of compound 16, suggesting an alternative binding mode for these molecules. This insight proved correct, and in fact adding two methyl groups to compound 7 led to compound 21, which is more potent than compound 7 and binds such that the amide of the quinazolinone core forms hydrogen bonds with the hinge region, as confirmed by crystallography (green in figure).
Compounds 16 and 21 have similar affinities yet different binding modes, so what about selectivity? Testing them in a panel of ~110 kinases revealed both to be quite selective for ALK family kinases, though they had different off-targets. The selectivity of compound 21 is particularly impressive given its small size – it teeters on the edge of being rule of three compliant. A related molecule also showed activity in a cell-based assay.

An interesting unanswered question is the binding mode of the initial fragment. Perhaps it binds in multiple orientations, which could explain why crystallography was unsuccessful. Regardless, this is a nice study that illustrates how close attention to confusing SAR can lead to attractive new series.

03 September 2018

From generic fragment to selective BET-family BD2 inhibitor

Fragments have been a rich source of leads against bromodomain-containing proteins, epigenetic readers that recognize acetylated lysine residues and are implicated in a variety of diseases. The four members of the BET family in particular have been heavily explored. Each of these proteins actually contains two separate bromodomains, called BD1 and BD2, and most reported inhibitors hit both of them more or less equally. To follow up on some intriguing biological hints that this may not be necessary, Robert Law and collaborators at GlaxoSmithKline and University of Strathclyde pursued selective BD2 inhibitors, which they describe in J. Med. Chem.

The researchers started with a fragment they first reported six years ago, and which has been used by Forma as the starting point for one of their own programs. Although fragment 10 is equipotent against BD1 and BD2 of BRD4, growing led to compound 12, with an encouraging 60-fold selectivity for BD2 (all values shown below are for BRD4 BD2). A crystal structure of a close analog suggested several opportunities for further growth to improve potency.

Changing the methyl group to a cyclopropyl group improved selectivity, and introducing a hydroxymethyl substituent off the phenyl ring (compound 44a) improved potency for BD2. This molecule was fairly lipophilic, so the researchers explored adding a variety of polar substituents to improve solubility, ultimately resulting in GSK340.

GSK340 was profiled against 35 bromodomains and found to be at least 40-fold selective for the BD2 domain compared to the BD1 domain of the four BET family members. It showed the highest affinity for BRD4 but also bound tightly to the BD2 domains of BRD2, BRD3, and BRDT and was selective against non-BET family bromodomains. The compound was cell permeable and inhibited the release of the inflammatory cytokine MCP-1, supporting the notion that BD2 domain inhibition alone could have useful anti-inflammatory effects. Unfortunately GSK340 shows sufficiently high clearance in rat and human hepatocytes that the researchers suggest its utility will be limited to in vitro assays. Still, this paper provides another illustration that – with the help of creative medicinal chemistry – a generic, non-specific fragment can lead to a novel and selective chemical probe.

23 August 2018

256th American Chemical Society National Meeting

This week thousands of chemists converged on the venerable city of Boston for the Fall National ACS meeting. One brief but rich symposium was entitled “Best practices in fragment-based drug design,” organized by Amy Hart, David Marcoux, and Heidi Perez (BMS).

After an introductory presentation by me, Anil Padyana described FBDD at Agios. The company is focused on metabolic enzymes, many of which have dynamic, shallow, and polar active sites – challenging even for fragments! Indeed, a summary of 11 targets screened using a variety of methods revealed generally low hit rates, usually < 3%. The company’s first approved drug came out of an HTS screen against IDH2 of 80,000 compounds that yielded just 24 hits. The molecule that ultimately led to enasidenib was essentially a (large) fragment, with 21 atoms. Agios’ current fragment library is just over 5000 about 10,000 molecules, though they are in the process of expanding this to 20,000 – perhaps part of a general trend. Given the history of enasidenib, they are including molecules beyond the rule of 3, with an upper molecular weight limit of 350 Da, far higher than most respondents in our recent poll.

Cullen Cavallaro presented an early, though still unpublished, FBDD story from BMS: KAT II, a brain enzyme implicated in schizophrenia. Screening 3700 fragments using NMR, SPR, and TSA yielded 236 hits, only 6 of which were common to all methods. All 236 hits were soaked into crystals of KAT II, resulting in 43 structures, 13 of which bound in the active site. Strikingly, 12 of these contained carboxylic acids, which generally don’t cross the blood brain barrier. The lucky thirteenth fragment showed no activity in an enzymatic assay, no thermal stabilization, and only a marginal STD NMR signal. However, through a combination of library synthesis and structure-based design the researchers were able to obtain nanomolar inhibitors. Unfortunately the project was stopped when BMS exited neuroscience.

Anna Vulpetti provided an overview of work done by her and her Novartis colleagues to discover inhibitors of Factor D. A high-throughput screen of the serine protease didn’t yield anything useful, but a combination of fragment screening and structure-based design led to multiple series of inhibitors. Anna is a proponent of fluorine NMR, and Novartis has recently expanded its fluorinated fragment library to 4000 members. Like Agios, they have chosen to include some larger fragments, up to 350 Da.

Finally, David Norton described the initial work done at Astex to discover an orally available ERK1/2 inhibitor, which entered phase 1/2 clinical trials in May of this year. We highlighted some of this work a couple months ago so I won’t cover it in detail, but among other lessons David emphasized the importance of initial fragment optimization before starting to grow.

There were plenty of fragment talks outside the symposium too. Last year we highlighted Ben Cravatt’s strategy for performing fragment screening in cells. Chris Parker, the first author on that paper, has just launched his independent academic career at Scripps Florida, and provided an update. Chris noted that, for phenotypic screening, the approach is essentially a target-finding method, and indeed more than 4000 proteins have been identified, varying over five orders of magnitude in abundance. Having proper controls is critical, and recent efforts include screening pairs of enantiomeric fragments and looking for differences.

Taekyu Lee provided an update of the Vanderbilt MCL-1 program, most recently described in this paper. Some of the molecules shown have low picomolar affinity, mid nanomolar cell activity, and are more than 10,000-fold selective for MCL-1 over BCL-2 and BCL-xL. The program was partnered with Boehringer Ingelheim earlier this year, and they’ve got competition: four other molecules have entered the clinic.

Covalent fragments were also a theme. Peter Wipf (University of Pittsburgh) described the construction of a 300 compound “mercaptophilic” library. In contrast to other academic reactive fragment libraries we’ve covered (see here, here, and here), this one contains a wide variety of different warheads with varying reactivities.

Finally, Jeff Neitz (UCSF) described efforts against Taspase-1, which is involved in cancer cell proliferation. A high-throughput screen of 242,000 molecules yielded seven chemical series – all of which ultimately proved to be artifacts. The enzyme is a threonine protease but contains a cysteine residue near the active site, so the researchers conducted a Tethering screen with 1280 disulfide-containing molecules, which led to 64 hits in five classes. Converting the disulfide to more drug-like warheads ultimately led to nanomolar molecules with cell-based activity, and the researchers even had some success removing the warhead entirely.

If you missed the meeting, you still have time to catch what should be an epic conference: FBLD 2018 returns to San Diego where it originated ten years ago. People still talk fondly about that meeting, so don’t miss this one!

20 August 2018

Poll results: the modern fragment library

Our most recent poll has just closed, and the results provide a snapshot of fragment libraries in 2018.

The first two questions asked about the smallest and largest fragments (in terms of non-hydrogen or heavy atoms) you would include in your library. As shown below, the median lower bound is 7-8 heavy atoms, while the median upper bound is 17-18 heavy atoms. This is comparable to what we saw when we last asked these questions five or six years ago.

(Methods note: of the 98 respondents to the question “What is the largest number of heavy atoms you would allow in a fragment,” 10 answered ≤ 10 atoms, but no one answered 11-12. I suspect that the 10 answers were erroneous and that they actually meant to answer the second question, “What is the smallest number of heavy atoms you would allow in a fragment,” which was answered by 89 respondents. Therefore I have excluded these answers from the figure. Please leave a comment if I’ve mischaracterized your vote!)

The third question asked how many fragments people have in their libraries, and more than 40% of the 99 respondents answered 1001-2000.

The distribution is similar to the results from four years ago when we last asked this question. Notably though, the number of respondents with very large libraries has more than doubled, admittedly from a small base.

Overall then this poll could probably be summarized as, “plus ça change.” Or, as the growing number of clinical success stories attests, if it ain’t broke, don’t fix it!

13 August 2018

Fragment growing and merging: an inverse agonist of RORγt2

The nuclear receptor transcription factor RORγt2 is involved in the differentiation of Th17 cells, and is thus a target for inflammatory diseases. The protein contains a large, hydrophobic ligand binding site, and as a result most known inverse agonists have less than ideal physicochemical properties. In a paper recently published in J. Med. Chem., Samuel Hintermann and colleagues at Novartis have taken a fragment-based approach.

The researchers screened a library of 1408 fragments using a “differential static light scattering (DSLS)” assay, which is a type of thermal shift assay that measures denaturation and aggregation of the protein. A few dozen molecules that stabilized RORγt2 were tested in dose-response curves prior to crystallization trials, ultimately yielding 13 structures. Compound 1 was particularly interesting because it binds in the center of the cavity, providing growth vectors in two directions. It also makes a couple hydrogen bonds with the protein, as opposed to purely hydrophobic interactions.

Growing from the ethoxy position quickly led to improvements in affinity. To avoid the possibility of toxic iminoquinone metabolites, the researchers replaced the central phenyl ring with a pyridine, resulting in the low micromolar inverse agonist compound 8.
To further improve affinity, the researchers merged an element from a previously reported GlaxoSmithKline molecule (compound 2) onto compound 8, resulting in the potent compound 9, which was characterized in a battery of assays.

The crystal structure of compound 9 bound to the protein revealed that the core fragment moiety binds in the same manner as the original compound 1, though the added benzyl ether binds in a subpocket that had not previously been observed to bind ligands.

Kinetic studies using a reporter displacement assay revealed that compound 9 has both a slow on-rate as well as a slow off-rate, consistent with the fact that it is fully enclosed by the protein. The researchers performed molecular dynamics simulations to try to determine how the ligand could enter or leave, which suggested large conformational changes in a flexible region of the protein. Isothermal titration calorimetry (ITC) showed that the binding of compound 9 is enthalpically driven, with an unfavorable entropy. Although interpreting thermodynamics is fraught, this result makes intuitive sense given the hydrogen bonds formed and the fact that the molecule seems to rigidify the protein.

Biophysics is interesting, but of course biology is what was driving the program. Compound 9 is potent in a variety of cell assays and is also selective for RORγt2 over other nuclear hormone receptors. However, it is also mostly insoluble, and although it did show efficacy in a rodent inflammation model, plasma concentrations of compound 9 were highly variable between individual rats, which the authors attribute to poor physicochemical properties.

This is a nice application of fragment growing and merging that demonstrates how difficult it is to find useful leads for lipophilic sites: even with favorable biochemistry and biophysics, the pharmacokinetics are a slog. That said, others have made progress against similarly hydrophobic targets, so it will be fun to watch this story progress.

06 August 2018

Conservation of fragment binding modes revisited

A common assumption when growing or linking fragments is that the binding mode will remain the same. This is often the case, but exceptions occur frequently enough to keep life interesting. Last year we highlighted a study that tried to answer the question of when ligands changed their binding mode by analyzing the protein data bank (PDB). In a new J. Med. Chem. paper, Esther Kellenberger and collaborators at Université de Strasbourg and Eli Lilly have conducted an even more exhaustive study.

The researchers considered all protein structures deposited in the PDB between 2000 and mid-2016 solved to at least 3 Å resolution. This yielded 1079 different fragments (MW < 300 Da) and 1832 larger (“drug-like”) ligands, as well as 126 crystallization additives such as buffers and detergents. In comparing the same protein with different ligands, care was taken to remove mutant proteins that could cause a change in binding mode.

This dataset was used to address several questions.

First, how often does the same fragment bind to the same pocket in the same manner? Often a crystal structure will have several different copies of the same protein in the asymmetric unit. In nearly three-quarters of cases, the fragments bound in a similar manner to the different copies. The exceptions often involved protein conformational changes, in some cases due to different crystal contacts.

Second, how often does a fragment maintain its binding mode when incorporated into a larger molecule? The data set included 359 pairs of ligands on 51 proteins. Again, about three-quarters of fragments had similar binding modes as their larger counterparts. When binding modes changed, protein flexibility often played a role. Polar contacts such as hydrogen bonds were much more highly conserved than hydrophobic contacts. As the earlier study also found, binding modes of very small fragments (MW < 110) were most likely to change, while fragments with MW > 150 almost always retained their binding modes.

Third, do fragments and larger ligands make similar interactions? The data included 235 proteins in which at least one structure contained a fragment and another structure contained a larger ligand.  (The larger ligand didn’t necessarily contain the fragment.) Obviously larger ligands are able to make more interactions than smaller ligands, but, as Stephen Roughley and Rod Hubbard observed back in 2011, enough fragments should allow you to map out the important interactions. After systematically exploring the data, the current researchers suggest that fully mapping a pocket requires nine or more different fragments, a high bar satisfied by just 11 proteins.

Finally, do crystallization additives behave as fragments? The researchers looked at all additives with MW < 300, and separately considered those bound to otherwise free (apo) proteins and those bound to proteins containing other ligands. In general additives showed more variation in their binding modes, though those binding to apo proteins often made similar contacts as made by fragments and larger molecules. Intriguingly, small polar molecules such as DMSO and glycerol often made hydrophobic interactions with proteins.

There is plenty more in the paper than can be summarized here. Laudably, the researchers have provided all of their data in a convenient web portal that even supports chemical substructure searches. Overall the results reassuringly suggest that the binding mode of a fragment usually remains the same as it is optimized. But of course these types of analyses are subject to survivor bias: fragments that change binding mode unexpectedly may be more difficult to optimize, and thus less likely to lead to larger ligands.

The odds may be ever in your favor, but look out for the exceptions.

30 July 2018

Dimerization: elegant but not essential

A special case of fragment linking is dimerization, in which two copies of the same fragment bind to adjacent sites in a protein and are subsequently linked together (see for example here, here, and here). A recent example was published in J. Med. Chem. by Bernard Pirotte, Julien Hanson (University of Liège), Lionel Pochet (University of Namur), Jette Kastrup (University of Copenhagen) and their collaborators.

The researchers have for some time been interested in AMPA receptors, critical components in neuronal synaptic transmission. Increasing their activity could be useful for treating diseases such as depression and schizophrenia, but increasing activity indiscriminately is known to be toxic. One approach has been to develop positive allosteric modulators (PAMs), which increase the activity only in the presence of the natural ligand glutamic acid, thus amplifying the normal biological signal.

AMPA receptors themselves are dimers of dimers. Many different PAMs have been reported for AMPA receptors, and some of these are in fact dimeric molecules that span two adjacent binding sites across the dimer interface. A crystal structure of a molecule closely related to compound 35 revealed that each molecule binds to two adjacent protein subunits, so the researchers designed compound 22, which pairs the molecules through a simple ethylene moiety. The strategy paid off with a low nanomolar activator, which crystallography confirmed binds as expected.

Interestingly, conceptually cleaving the bond connecting the two fragments generates a compound (33) which is slightly less active than the initial fragment 35; it is possible the methyl groups are too close to one another when two copies of compound 33 are bound.

As the researchers point out, compound 22 is one of the most potent AMPA receptor PAMs reported. However, it is also quite large, particularly since it needs to cross the blood-brain barrier. No animal data are reported, but a simple metric called the CNS MPO desirability score is reasonably predictive. This score is based on the molecular weight, lipophilicity, total polar surface area, number of hydrogen bond donors, and basicity; higher scores are better. By this measure, compound 22 is predicted not to have high brain penetration, though of course any metric needs to be taken with caution.

However, a separate J. Med. Chem. paper by many of the same researchers revealed that dimerizing the molecules is not essential: simply growing compound 35 could also generate a low nanomolar AMPA receptor PAM (compound 8). Crystallography revealed that the added phenyl group binds where the second molecule of compound 35 would normally bind. Moreover, compound 8 has a higher ligand efficiency as well as a higher CNS MPO desirability score than the dimeric compound 22, suggesting that it is more likely to be able to cross the blood-brain barrier.

In the absence of pharmacological or pharmacokinetic data, if forced to choose I would probably focus on compound 8 rather than compound 22. All of which is to say that although there is a certain elegance to dimerizing molecules, you might be able to replace one of them with a smaller, simpler moiety.

23 July 2018

Fragments score a win against WDR5-WIN

Protein-protein interactions (PPIs) can be difficult targets for multiple reasons. First, the contacts often cover large, flattish areas with few “ligandable” pockets. Second, they can involve multiple proteins; imagine trying to disrupt a huge multicomponent machine with a little widget. The protein WDR5 falls into the second category. It serves as a scaffold around which other proteins assemble to regulate epigenetics. One of these proteins, MLL1, is implicated in certain leukemias and binds to WDR5 through the WDR5 INteraction (WIN) motif, making this protein-protein interaction an intriguing anti-cancer target. In a recent paper in J. Med. Chem., Stephen Fesik and colleagues at Vanderbilt University describe their efforts towards this target.

Unlike some PPIs, the WIN motif does contain a nice little pocket which normally recognizes arginine residues. However, since the highly basic guanidine moiety of arginine is undesirable in drugs, the researchers conducted a fragment screen to find new WIN-site binders. A two-dimensional (1H-15N HMQC) NMR screen of a large fragment library (>13,800 fragments, more than the majority of respondents in the poll to the right) identified 47 hits that produced similar spectral changes as a peptide that binds in the WIN site. Compound F-1 was the most potent.

A crystal structure of compound F-1 bound to WDR5 revealed that the imidazole moiety binds in the same deep pocket normally occupied by the arginine side chain, with the phenyl ring pointing up out of the pocket. Initial growing off the phenyl ring into nearby hydrophobic pockets produced more potent compounds, but at best these were still micromolar binders. The researchers had more success by targeting a slightly more distant pocket with compounds such as 4a and subsequently compound 4i. A crystal structure of compound 4a bound to WDR5 suggested that the biologically active conformation might not be the lowest energy conformation of the free molecule. Introducing a ring to restrict the conformation led to more potent molecules such as 6e, with sub-nanomolar affinity.

Unfortunately, though potent in biochemical assays, compound 6e and related molecules were about 2800-fold less potent in cell-based assays. The compound is cell permeable and not effluxed, so the disconnect must be due to something else – perhaps the multiple other proteins in the cellular environment. Anyone who has spent much time doing medicinal chemistry will have encountered frustrating situations like this. Perhaps a new chemotype is needed, or perhaps the compounds need to be made even more potent. Indeed, several years ago the Fesik group reported nanomolar binders of MCL-1, but it was not until they improved affinity to picomolar that they saw good cell potency. Stay tuned!

16 July 2018

Rise of the machines for fragment optimization

Our latest poll (please vote on the right-hand side of the page!) is about fragment libraries. Once you have your library, you can screen it using a variety of approaches. But what do you do once you get hits? Computational methods are increasingly being adopted; just this year we’ve discussed two approaches: growing via merging and AutoCouple. A new paper in J. Med. Chem. by Philippe Roche, Xavier Morelli, and collaborators at Aix-Marseille University and several other institutions describes a method that combines virtual screening with automated real-world synthesis in a platform called diversity-oriented target-focused synthesis (DOTS).

The process is best described with an example, and the test case presented is the first bromodomain of BRD4, BRD4(BD1). The researchers, who had previously identified a xanthine-containing series of inhibitors, pared this back to fragment-sized compound F1. Crystallography revealed a nearby pocket, which the researchers attempted to target with DOTS.

The researchers built a virtual library of 576 sulfonamides extending off the para position of the phenyl ring of compound F1. These were then virtually screened against BRD4(BD1) using the S4MPLE molecular modeling tool in which the F1 portion was constrained in the crystallographically observed conformation while the variable bits were allowed to move. The 100 top-scoring molecules were examined more closely, and 17 representatives were chosen to be synthesized on an automated robotic platform. This was actually a fairly modest set, as the Chemspeed system they used can run up to 96 parallel reactions. The crude products were then tested in a fluorescence assay, and all of them showed improved activities compared to the initial fragment. The majority, such as compound 17, showed high nanomolar inhibition.

The 13 submicromolar compounds were then resynthesized, purified, and validated in thermal shift and isothermal titration calorimetry (ITC) assays; these orthogonal methods confirmed their activities. The crystal structure of compound 17 bound to BRD4(BD1) was also solved, and this revealed that – as designed – the initial fragment retained its binding mode while the added portion makes new interactions with the protein.

The fact that 14 of the 17 molecules synthesized were at least an order of magnitude more potent than the initial fragment is satisfying, though it is worth noting that bromodomains are not the most difficult targets. Also, all of the new molecules have lower ligand efficiencies than the initial fragment. Still, advances and combinations of computational and robotic approaches will certainly increase the throughput of synthesis and testing, and I expect to see more of these examples.

09 July 2018

Practical Fragments turns ten, and celebrates with a poll on the modern fragment library

Ten years ago today Teddy launched Practical Fragments with a simple question about screening methodologies. More than 660 posts later we've returned to that topic several times, most recently in 2016. But before you can start screening you need a fragment library, which is the subject of our new poll.

Back in 2012 we asked readers the maximum size (in terms of "heavy", or non-hydrogen atoms) they would consider for fragments in their library. The results were mostly consistent with the Rule of 3, so beloved by Teddy that he compared it to a powerful wizard.

There has since been a trend toward smaller fragments, driven in part by empirical findings that smaller fragments have better hit rates, in agreement with molecular complexity theory.

At some point, though, ever smaller fragments will mean lower hit rates: fragments that are too small will bind so weakly they will be difficult to detect. And practical issues arise: organic molecules with just a few non-hydrogen atoms are often volatile.

Therefore, we’re revisiting this question: What is the smallest fragment you would put in your library?

As long as we're on the subject of libraries, how many fragments do you have in your primary screening library, or how many do you screen on a regular basis?

Please vote on the right-hand side of the page. If you have multiple fragment libraries (for example one for crystallographic screening and one for biochemical screening) you can respond for each library; you will need to press "vote" after each answer. Please feel free to leave comments too.

Thanks to all of you for making Practical Fragments a success and for your comments over the years – looking forward to the next decade!

02 July 2018

Fragment events in 2018 and 2019

Hard to believe we're already halfway through the year, but there are still some exciting events ahead, and 2019 is already starting to take shape.

August 19-23: The 256th National Meeting of the American Chemical Society, which will be held in Boston, includes a session on "Best practices in fragment-based drug design" on August 20.

September 25-28: CHI's Discovery on Target will also be held in Boston, and there will be lots of presentations of interest to readers of this blog, particularly in the Lead Generation Strategies track. Mary Harner and I will be presenting a FBDD short course over dinner on September 27.

October 7-10: Finally, FBLD 2018 returns to San Diego, where it was born a decade ago. This will mark the seventh in an illustrious series of conferences organized by scientists for scientists. You can read impressions of FBLD 2016FBLD 2014,  FBLD 2012FBLD 2010, and FBLD 2009.

March 20-22: Although not exclusively fragment-focused, the Sixth NovAliX Conference on Biophysics in Drug Discovery will have lots of relevant talks, and will be held in lovely Nice. You can read my impressions of the 2018 event here, last year's Strasbourg event here, and Teddy's impressions of the 2013 event herehere, and here.

March 24-26: The Royal Society of Chemistry's Fragments 2019 will be held in the original Cambridge. This is the seventh in an esteemed conference series that alternates years with the FBLD meetings. You can read impressions of Fragments 2013 and Fragments 2009.

April 8-12: CHI’s Fourteenth Annual Fragment-Based Drug Discovery, the longest-running fragment event, will be held in San Diego. You can read impressions of the 2018 meeting here, last year's meeting here, the 2016 meeting here; the 2015 meeting herehere, and here; the 2014 meeting here and here; the 2013 meeting here and here; the 2012 meeting here; the 2011 meeting here; and 2010 here.

November 20-22: If you can't make it to Nice, NovAliX will also be holding a biophysics meeting for the first time in the lovely city of Kyoto.

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