Fragments vs DHODH

Rapidly proliferating cancer cells require a steady supply of nucleic acids, and cutting that off is a potential therapy. The enzyme dihydroorotate dehydrogenase (DHODH), which is important for pyrimidine synthesis, is thus an interesting drug target. In a recent ACS Med. Chem. Lett. paper, Lindsey DeRatt, Scott Kuduk, and colleagues at Janssen describe their approach.

 

The researchers had previously used virtual screening and structure-based drug design to develop compound 1, which is potent in both biochemical and cell-based assays. However, the molecule is highly effluxed by P-glycoprotein, which can limit both oral bioavailability and brain penetration. Thus, they turned to fragments.

 

An SPR screen (about which sadly no details are provided) yielded compound 2, and crystallography revealed that the amide carbonyl makes a similar contact to tyrosine 356 (Y356) as does the carbonyl in the triazolone moiety of compound 1. Merging these led to compound 4, which was considerably more potent than compound 2 but much less so than compound 1. However, further optimization led eventually to compound 25. Although less potent in an enzymatic assay than compound 1, compound 25 is equally effective in cells. It also has excellent pharmacokinetics in mice and – importantly – a considerably lower efflux ratio.

 

Interestingly, when the researchers solved the crystal structure of a related molecule bound to DHODH, they found that the carbonyl no longer interacts with Y356 but is instead flipped 180º and interacts with a different residue. The researchers conclude by stating that they are designing new molecules to reengage Y356, which could further improve potency.

 

Several lessons emerge from this brief paper. First, the flipped urea moiety is another reminder that fragments do not always maintain their orientations, as also seen here, here, and here. Second, information from the fragment was used not to improve potency but rather to address other aspects of an existing lead series, as seen here and here. And finally, one could argue that the only critical feature of the fragment remaining in the final molecule is the NH of the urea. But the fragment did cause the researchers to examine their molecules from a different perspective, resulting in a better series. Perhaps you could call this an example of fragment-assisted drug discovery. As is so often the case, fragments can inspire new ideas that may otherwise be overlooked.

Fragments vs SHP2

One of the success stories we highlighted in last week’s summary of Fragments 2024 was the discovery of a potent inhibitor of SH2 domain-containing protein tyrosine phosphatase 2 (SHP2). James Day and colleagues at Astex and Taiho have just published the full account in J. Med. Chem.

 

Previous studies had shown that blocking SHP2 might be effective in certain cancers, particularly those dependent on mutant KRAS. As its name suggests, however, SHP2 is a phosphatase. This class of enzymes has highly charged active sites, which makes drug discovery notoriously difficult (see here for example). Indeed, a crystallographic fragment screen of the isolated phosphatase domain produced just one hit.

 

Simultaneously, the researchers performed NMR and crystallographic screens of the full-length protein, which contains two SH2 domains. This campaign was much more successful, with 88 crystallographically validated fragment hits. (Interestingly, a thermal shift assay of the same construct came up empty.) As Astex has previously reported, secondary binding sites on proteins are common, and SHP2 is no exception, with fragments binding to five sites. However, the vast majority – 83 of 88 – bound to what is called the tunnel region between the phosphatase domain and one of the SH2 domains.

 

The researchers note that “following completion of our Pyramid fragment screen, Novartis independently reported several SHP2 inhibitors” binding to the same site, which must have been both validating and irritating. Indeed, the Astex researchers did work on fragments binding to other sites, advancing one to a low micromolar inhibitor. But it’s hard to ignore a hot spot with dozens of bound fragments, and the tunnel region became their primary focus. One fragment was optimized to a low micromolar inhibitor. Another, fragment 3, had measurable affinity by ITC and respectable ligand-efficiency, and this was taken the furthest.

 

We’ve written previously about the importance of water in molecular interactions, and here the researchers performed solvent mapping molecular dynamics to identify water molecules that could be advantageously engaged. Scaffold hopping led to compound 15, and crystallography confirmed that the pyridine nitrogen forms a hydrogen bond to a water molecule. Increasing the lipophilicity around the phenyl ring and adding a basic amine to engage an electronegative region of the protein led to compound 18, with nanomolar biochemical activity and low micromolar activity in cells. Further structure-based design ultimately led to compound 28, with sub-micromolar cell activity. This compound has low efflux, low clearance and excellent oral bioavailability. When dosed orally in mouse xenograft models the molecule significantly inhibited tumor growth.

 

The exo-diastereomer of compound 28, in which the primary amine is facing down instead of up, shows interesting differences. It has a similar pKa as well as similar biochemical and cell-based activity but is plagued by high efflux and poor oral bioavailability. The researchers suggest that “steric shielding of the tropane bridge or pharmacophoric differences in efflux transporter recognition” may be responsible. There was considerable discussion at Fragments 2024 as to the precise source of the differences, but whatever the cause, this pair serves as a useful reminder that pharmacokinetics may vary dramatically even between nearly identical molecules.

 

Clinical development of SHP2 inhibitors has slowed due to a variety of reasons, including apparent on-target toxicity, but this is still a nice fragment-to-lead success story. Perhaps, as with capivasertib, it will just take time to find the right clinical strategy and patients who can benefit from these molecules.

Fragments 2024

Last week saw the first of four dedicated fragment meetings this year: Fragments 2024, the 9^th RSC-BMCS Fragment-based Drug Discovery Meeting, was held in historic Hinxton Hall, Cambridge, UK. I won’t attempt to cover the 17 talks, 40+ posters, and 20 exhibitors in detail but just try to hit on some broad themes.

 

One highlight was a talk by Chris Swain, whose Cambridge MedChem Consulting has come up several times at Practical Fragments. Chris has been systematically cataloging fragment hits reported in the literature, and his database now includes >2500 fragments from >300 papers that hit 265 targets. This has not been easy: as we’ve noted in our annual F2L reviews, papers don’t always mention fragments in the title or abstract; sometimes you need to dig deep into the experimental methods to find out the origin of the initial hits, and even then there are questions of interpretation. Chris noted that the the drug aprepitat originated from a fragment-like pharamacophore extracted from a more complex literature compound. That story was published in 1998, predating the term “fragment-based drug discovery,” but perhaps it would be considered FBDD today.

 

The fragments themselves are a diverse bunch, with an average Tanimoto similarity of just 0.09, but there are small clusters. Looking at them in more detail, the ten most common scaffolds are aromatic (benzene, indole), which is a departure from approved drugs. There is also a significant fraction of charged molecules, including 298 acids and 348 basic groups. About 10% of the fragments hit more than one target, exactly what you would expect from the theory of molecular complexity.

 

Chris’s talk was followed by a wide-ranging panel discussion that expanded on some of these themes. Solubility was recognized as important, though with different techniques being more persnickety: Justin Dietrich (AbbVie) noted that pre-screening is critical for SPR, but for protein-detected NMR the protein is present at high enough concentrations to act as a “phase transfer reagent.”

 

The topic of thermodynamics also came up, with Chris Murray noting that Astex collects lots of ITC data but uses it for assessing free energy (ΔG) values rather than enthalpic energy (ΔH) values. Helena Danielson (Uppsala University) noted that the early correlation between compound quality and enthalpy found with HIV protease inhibitors did not seem to apply to other targets despite significant investment in collecting data at multiple companies, as also noted by Chris Smith (Mirati) and Mike Hann (GSK). Rod Hubbard (Vernalis) puckishly suggested that the study of ΔH had produced “more heat than light.”

 

The topic of MiniFrags also came up during the panel discussion. Chris Murray noted that they had been tried on quite a few targets but, as Rod Hubbard confirmed, were more helpful in identifying binding sites than providing starting points. But Chris Smith pronounced himself a “complete convert” after a MiniFrag identified an induced pocket on a previously intractable target where fragments (and other techniques) had failed.

 

Covalent fragments also made several appearances, with Jonathan Pettinger describing a phenotypic screen at GSK looking for compounds that block the pro-inflammatory M1 polarization of macrophages. After screening some 2000 covalent fragments they used chemoproteomics to determine that one of the best compounds acted by modifying cysteine 817 of the kinase JAK1. Interestingly, this is the same cysteine identified independently by researchers from Vividion, which could speak to the centrality of this target, the reactivity of this particular cysteine, or both.

 

Pursuing residues other than cysteine is seen as difficult, with Mike Hann noting in the panel discussion that these may require more extensive non-covalent interactions and Chris Murray noting that the warheads themselves were less attractive. But these challenges have not dissuaded Peter Cossar (Eindoven University of Technology), who has introduced cysteine-reactive disulfide and lysine-reactive aldehyde moieties into the same fragment to crosslink a 14-3-3 protein to substrate ERRγ.

 

Another theme was screening crude reaction mixtures in a “direct to biology” approach. Vernalis was an early adopter with their off-rate screening, and a talk by Lucie Guetzoyan confirmed that they are continuing to invest here not just with SPR but also with affinity-selection mass spectrometry and X-ray crystallography. Lucie also described using flow chemistry to enable sensitive organometallic chemistries such as Grignard and Negishi couplings. John Spencer (University of Sussex) is also using crude reaction screening by crystallography and thought the approach can compress ten years worth of work into a few months.

 

As with most conferences these days there were plenty of success stories. Martina Schaefer (Nuvisan) described the discovery of the Bayer SOS1 inhibitor BAY-293, which we wrote about here. Anna Vulpetti (Novartis) described the discovery of IL-1β inhibitors, which we wrote about here. Nicola Wilsher (Astex) described the discovery of potent SHP2 inhibitors; I’ll write more about these later. And Matthew Calabrese described the discovery of allosteric activators selective for the γ3 subunit of AMPK, which could avoid the cardiotoxicity seen with less selective molecules. Three HTS screens had failed but fragments ultimately led to a potent tool molecule. Interestingly, some of the HTS compounds were later found to be hits but had been overlooked because they were so weak that they did not rise above the noise of the assay.

 

Finally, Justin Dietrich described several success stories, including against TNFα (which we wrote about here) as well as CD40 ligand. Justin noted that FBLD is used alongside HTS and DEL at AbbVie, and that the techniques can be complementary – a theme noted by several others.

 

Despite being so intimately integrated with other discovery approaches, FBLD continues to innovate and evolve and remain sufficiently quirky that stand-alone meetings are still valuable and rewarding. I’m looking forward to seeing what the next several meetings reveal.

The EU-OPENSCREEN fragment library

A well-curated fragment library is usually the starting point for fragment-based lead discovery, and not an insignificant investment. If you are just starting out you may want to use an existing library. One such option is described in an (open-access) paper in RSC Med. Chem. by Jordi Mestres and collaborators at IMIM Hospital del Mar Medical Research Institute and across Europe.

 

The EU-OPENSCREEN European Research Infrastructure Consortium (ERIC) allows researchers to access early lead discovery and chemistry resources. Among other components, it includes a set of more than 96,000 compounds for high-throughput screening, the European Chemical Biology Library, or ECBL. To complement this, the researchers have developed what they call the European Fragment Screening Library, or EFSL.

 

Recognizing that rapid follow-up is a critical next step in fragment-based lead discovery, the researchers designed EFSL based on ECBL. They did this by choosing fragments commercially available from Enamine that were sub-structures of ECBL members. Fragments were chosen to represent as much of the ECBL as possible, as well as for rule-of-three compliance. Fragments with multiple vectors for growing were also prioritized, similar to the “sociable fragments” concept we wrote about here. Finally, a set of 88 very small “minifrags” were also included.

 

Fragments were dissolved in deuterated DMSO at 100 mM (or 1000 mM for minifrags). Solubility and integrity were assessed at 1 mM (or 10 mM for minifrags) in PBS using ¹H-NMR using an internal standard; those with solubility < 0.2 mM were rejected, as were those with missing or extra peaks in the NMR spectra. Of 1056 compounds tested, 913 passed these QC criteria.

 

The EFSL is available for screening (via grant applications), and the paper summarizes the results of eight screens performed over two years using a range of detection technologies including crystallography, ligand-detected NMR, small-angle X-ray scattering, thermal shift, and BLI. Hit rates ranged form just 0.1% to 31.3%, though in the last case only a small subset of the library was tested.

 

After fragment screening and confirmation, four of the projects tested larger compounds from the ECBL in follow-up studies, and two were able to identify hits. One project targeting a bacterial beta-ketoacyl-ACP synthase 2 (FabF) used BLI to identify a fragment with a dissociation constant of 35 µM. Of the 147 compounds related compounds from the ECBL, two had slightly higher affinity, albeit at the expense of lower ligand efficiency. Perhaps exploring Enamine REAL Space as in this example would be more effective at finding significantly more potent molecules.

 

In summary, the EFSL seems to be a useful resource, particularly for academic labs. If you’ve got a target and no internal fragment-screening capabilities, it might be worth putting in an application.