Fragments in Texas

The meeting Development of Novel Therapies through Fragment Based Drug Discovery was held last week in Houston, Texas, organized by the Gulf Coast Consortium for Quantitative Biomedical Sciences. Although it was only a single day, it was packed, with thirteen speakers, a couple vendor lunch talks, and some two dozen posters. Below is just a flavor – please add your own impressions if you were there.

I kicked off the first session by giving an overview of FBDD, highlighting both pitfalls and successes. Beth Knapp-Reed (GlaxoSmithKline) then discussed efforts against LDHA, a target previously tackled successfully using fragment linking (see here and here). In this case, an HTS screen of 1.9 million compounds produced only a single hit that resulted in a crystal structure, while fragment screens yielded 16 structures at three binding sites. An NMR-based functional screen (using ¹³C-labeled substrate) was key to obtaining robust SAR, and using information from both the HTS hit and the fragments ultimately led to nanomolar inhibitors. Next, Tom Davies provided an overview of Astex’s discovery platform, focusing on a success with KEAP1. We recently highlighted research suggesting that crystallography should be used as a primary screen, which Astex does for some targets. Tom noted that doing so currently takes about a month, though only after spending somewhere between 3 and 12 months establishing a robust protein construct as well as crystallization and soaking conditions.

Jane Withka opened the next session by discussing the continuing evolution and use of the Pfizer fragment library. Out of 32 targets screened, only one produced no hits, and this was a particularly flexible protein. Interestingly, despite being relatively balanced among basic, acidic, and neutral members, hits were strong enriched in neutral compounds and strongly depleted in basic fragments. Neutral fragments were exactly what was sought by Daniel Cheney (Bristol-Meyers Squibb), who discussed successful efforts to replace a basic amidine moiety in Factor VIIa inhibitors. And Brad Jordan (Amgen) discussed the successful application of ¹⁹F NMR screening to find fragments that could be linked to previously discovered inhibitors to obtain selective picomolar inhibitors of BACE1.

Alex Waterson (Vanderbilt) started the first afternoon session by discussing how fragments had been successfully applied to RPA, RAS, and MCL-1. In the last case, the best compounds now have dissociation constants in the picomolar range, are active in cells, and show activity in xenograft models. IND-enabling studies are slated to begin as early as this year, with the hope of developing a cousin of venetoclax. Inna Krieger (Texas A&M) described how fragments could be used to understand the mechanism of M. tuberculosis malate synthase, while Dawn George (AbbVie) described selective (but inexplicably toxic) PKCθ inhibitors, which are now being made available to researchers to probe the biology. Finally, Damian Young (Baylor) gave an update on his sp³-carbon enriched fragments. Jane had mentioned that following up on hits with multiple stereocenters was not always easy, but Damian’s DOS approach efficiently and systematically yields each possibility. Whether these will meet the Safran-Zunft challenge remains to be seen.

The last session was focused on success stories. Michael Mesleh (Broad Institute) discussed Cubist’s bacterial DNA gyrase inhibitors. Marion Lanier (Takeda) described how fragment screening and careful medicinal chemistry led to a low nanomolar, selective inhibitor of BTK. With a molecular weight of just 318, the molecule is scarcely larger than a Texas fragment, and has good pharmacokinetics and activity in a rat arthritis model. And Yi Liu (Kura) discussed the optimization of covalent KRAS inhibitors originally discovered using Tethering.

This was my first visit to Houston, and I was struck by the number of researchers who had relocated from around the world, particularly from (previously) large(r) pharma companies. Whenever scientists meet the talk often turns to funding shortages, but not here: everyone seemed to have plenty of money and resources, and one of the organizers announced that he was trying to fill several positions. This was the first major fragment meeting in Texas but likely not the last – there is talk of turning it into a recurring event. And there are still several good upcoming events this year; early registration for FBLD 2016 closes in just a few weeks.

Calculating hotspots in detail

In the eight years since Practical Fragments first started, Moore’s law has held strong and computational power has increased accordingly. Last year we described how tools such as FTMap can be used to identify hot spots – regions on proteins where fragments are most likely to bind. Although FTMap is quite successful at identifying these, it is less able to point to specific interactions (such as hydrogen bond donors or acceptors) that are likely to drive binding. In other words, computational chemists have become adept at identifying where fragments might bind but lag in predicting how. A new paper in J. Med. Chem. by Chris Radoux at the Cambridge Crystallographic Data Centre and collaborators at UCB and the University of Cambridge addresses this challenge.

The approach starts with a set of three simple molecular probes: toluene, to look for hydrophobic interactions; aniline, to look for hydrogen bond acceptors; and cyclohexa-2,5-dien-1-one, to look for hydrogen bond donors. These probes are larger than those (such as ethanol) used in many other programs, the idea being that too-small molecules might find hot spots so small as to be useless. Indeed, with 7 non-hydrogen atoms, these probes are near the low end of the consensus size for fragments.

Calculations are performed on protein structures – either with no ligand bound or with a bound ligand computationally removed – to determine whether each surface atom of the protein is a hydrogen bond donor, acceptor, or hydrophobic, as well as how exposed the particular atom is. The three probes are then mapped onto the proteins to look for favorable interactions. Regions where multiple probes can bind are scored higher, with hotspots defined as those regions of the protein having the highest scores. The type of probe with the highest score also describes what type of interactions are likely to be favorable at various regions within a given hot spot. Although the researchers note that multiple software packages could be used for these calculations, they used a program called SuperStar, and calculations took just a few minutes on an ordinary laptop.

To validate the approach, the researchers used a previously published data set (discussed here) of 21 fragment-to-lead pairs against a variety of proteins for which crystal structures and binding affinities were available. In general, the method was able to identify the fragment binding site quite effectively; the one outright failure was on the fragment with the lowest affinity, which also had poorly resolved electron density in the crystal structure. Importantly, the fragments tended to have the highest scores, with added portions of the leads scoring lower. This data set was used to calibrate the scoring system for identifying hot spots, as well as specific molecular interactions within each hot spot.

Having thus validated the approach, the researchers took a more detailed look at two published fragment-to-lead programs for protein kinase B and pantothenate synthetase. In both these cases, group efficiency analyses had previously been performed to establish which portions of the ligands contributed most significantly to binding. Gratifyingly, the computations correctly predicted these.

Overall this approach appears promising. At a minimum, it is another tool for assessing the ligandability of potential targets. More significantly, by highlighting the hottest bits of hot spots, it could be useful for medicinal chemists trying to optimize and grow fragments and leads. Unfortunately, as currently described, the process will require a skilled modeler. It would be nice if the authors built a simple web-based interface for people to upload pdb files for analysis, as is the case for FTMap. Also, all the data presented are retrospective – a prospective example would be the true test. Does anyone have experience to share?

Fragments vs JAK – but phototoxicity

The four members of the JAK family of kinases have received plenty of attention due to their role in inflammation. Two drugs that inhibit these targets, tofacitinib and ruxolitinib, have been approved for rheumatoid arthritis and myelofibrosis, respectively. Psoriasis is another possible indication, but for this disease a topical drug might be useful, particularly one with low systemic bioavailability. The search for such molecules is the subject of a paper recently published in ACS Med. Chem. Lett. by Andrea Ritźen and colleagues at LEO Pharma.

The researchers screened 500 fragments at 100 µM each using surface plasmon resonance (SPR) against JAK2. Hits were then tested in a biochemical assay against JAK1; in general, there was good correlation, suggesting a (desirable) lack of selectivity between the two family members. One of the more attractive hits was compound 1, which was characterized crystallographically bound to JAK2.

Compound 1 is an indazole, which is often seen in kinase inhibitors binding to the so-called hinge binding site where the adenine of ATP binds. Other indazoles have previously been reported as JAK2 inhibitors. Nonetheless, the sulfonamide moiety of compound 1 provides an interesting new vector. Moreover, a search of published structures suggested that adding a phenol moiety could make additional contacts to the proteins. This led to the design and synthesis of compound 2, which showed a dramatic boost in potency as well as measurable cell activity. Further optimization led ultimately to compound 34, with low nanomolar biochemical activity and mid-nanomolar cell activity. Compound 34 was also reasonably selective in a panel of 20 kinases.

A topical drug needs to be stable in sunlight, but unfortunately compound 34 showed phototoxicity. This led to the testing of a few other compounds, revealing that even the initial fragment 1 is unstable over a period of a few hours in simulated outdoor light. Indazole itself seems reasonably stable, suggesting that perhaps adding different substituents could fix the problem.

This is a brief but satisfying example of using published information as well as medicinal chemistry to advance a fragment hit. Although the program does not appear to have led to a drug lead, it is laudable that the researchers describe the photoinstability of the indazoles. With these moieties appearing so frequently in campaigns against kinases, this could be a valuable cautionary tale to others pursuing similar scaffolds.

Fragments vs KEAP1 – crystallographically

Last week we discussed how soaking crystals in high concentrations of fragments could identify useful molecules. Indeed, last month’s Drug Discovery Chemistry conference featured a talk by Tom Davies (Astex) illustrating the power of this approach. In a recent paper in J. Med. Chem., Tom, Jeffrey Kerns (GlaxoSmithKline), and their collaborators provide the full story.

The researchers were interested in the protein KEAP1, which binds to and blocks the activity of the transcription factor NRF2. Small electrophilic molecules covalently react with KEAP1, causing NRF2 to dissociate and upregulate various cytoprotective genes, which could be useful for a variety of diseases. Indeed, this is at least partly how the approved drug dimethyl fumarate seems to work. However, dimethyl fumarate is quite reactive; a more specific molecule could have a better therapeutic profile, and would certainly be useful for probing the complex biology. With this in mind, the researchers sought a non-covalent inhibitor.

NRF2 interacts with a bowl-shaped “Kelch domain” of KEAP1 largely through electrostatic interactions. Thus, not only is this a challenging protein-protein interaction, coming up with a cell-permeable molecule is all the more difficult. The researchers soaked crystals of the Kelch domain against 330 fragments overnight at concentrations of 5-50 mM. Fragments were observed binding in three adjacent hot-spots, and although no functional activity could be detected, the binding modes suggested a path forward.

Growing from fragment 1 toward a hot-spot occupied by an aromatic fragment (magenta below) led to compound 4, with detectable activity in a fluorescence-polarization assay as well as clear binding in an isothermal titration calorimetry (ITC) experiment. Growing compound 4 toward another hot-spot occupied by a sulfonamide-containing fragment (green below) led to the sub-micromolar compound 6, and further optimization resulted in the low nanomolar compound 7. As is often (but not always) the case, the fragment portion of the final molecule binds in virtually the same position as the initial fragment 1 (blue).

Comparison with the structure of the NRF2 peptide reveals that the carboxylic acid in compound 7 binds in a very similar fashion to a glutamate residue in the peptide, and some of the other peptide contacts are also mirrored, but with very different moieties. Importantly, compound 7 has only a single negative charge, balanced lipophilicity, and fills the binding pocket more effectively than the peptide. These properties translate to good biological activity in multiple different cell-based assays, where the compound causes NRF2 translocation to the nucleus, upregulates appropriate gene expression, and prevents glutathione depletion when cells are treated with an organic peroxide. Although the pharmacokinetics have yet to be optimized, it also shows encouraging activity in a rat model of ozone exposure. Finally, it is reasonably selective in a panel of 49 undesirable off-target proteins. All these properties make this at least an excellent chemical probe.

There are several important lessons in this paper. First, as we’ve noted previously, it is possible to replace a highly polar peptide with a much more drug-like molecule. Second, fragments don’t need measurable affinity to be useful starting points. Crystallography is ideally suited for finding such fragments, though Astex researchers reported a similar success story with NMR last year. Finally, progressing these fragments won’t necessarily be easy, as reflected in the 25 authors on the paper. That said, such intensive efforts can pay off, as illustrated last month by the approval of venetoclax against another difficult protein-protein interaction.

A strong case for crystallography first

We noted last week that one theme of the recent CHI FBDD meeting was the increasing throughput of crystallography. Crystal structures can provide the clearest information on binding modes, and a key function of standard screening cascades is to whittle the number of fragments down to manageably small numbers for crystal soaking. Only a few groups have used crystallography as a primary screen. A team led by Gerhard Klebe at Philipps-Universität Marburg argues in ACS Chem. Biol. that crystallography should be brought to the forefront.

The researchers were interested in the model protein endothiapepsin. As discussed last year, they had previously screened this protein against a library of 361 compounds using six different methods, and the agreement among methods was – to put it charitably – poor. Nonetheless, many hits that did not confirm in orthogonal assays produced crystal structures when soaked into the protein. Thus emboldened, the researchers decided to soak all 361 fragments individually into crystals of endothiapepsin. This resulted in 71 structures, a hit rate of 20%, higher than any of the other methods (which ranged from 2-17%). Even more shocking, 31 of the fragments were not identified by any of the other methods, and another 21 were only identified by one other method. Thus, a cascade of any two assays would have found, at best, only a quarter of the crystallographically validated hits.

In agreement with other recent work, the fragments bound in multiple locations, including eight subsites within the binding cleft as well as three potentially allosteric sites. Not all of these sites were found using other methods.

But are these fragments so weak as to be uninteresting? To find out, the researchers performed isothermal titration calorimetry (ITC) to determine dissociation constants for 59 of the crystallographic hits. Three of the 21 most potent (submillimolar) binders were not detected by any of the other methods, and another seven were only found by one.

What factors led to this crystallographic bonanza? First, the researchers used the very high concentration of 90 mM for each fragment (in practice sometimes <90 mM because of precipitation). Not surprisingly, solubility was important: 97% of the hits had solubilities of at least 1 mM in aqueous buffer, and the soaking solution contained 10% DMSO as well as plenty of glycerol and PEG. Achieving such high concentrations is harder when multiple fragments are present, and the researchers argue from some of their historical data that the common use of cocktails lowers success rates.

How did different methods compare? Interestingly, functional assays such as high-concentration screening or reporter-displacement assays fared best, while electrospray ionization mass-spectrometry (ESI-MS) and microscale thermophoresis (MST) were close to random. This is in marked contrast to other reports for ESI-MS and MST, and the researchers are careful to note that “the choice and success of the individual biophysical screens likely depend on the target and expertise of the involved research groups.”

Primary crystallographic screening was an early strategy at Astex, and although this may not have been fully feasible 15 years ago, it seems they were on the right track. Of course, not all targets are amenable to crystallography, and not everyone has ready access to a synchrotron beam with lots of automation. But for those that are, it might be time to drop the pre-screens and step directly into the light.