21 September 2020

Eighteenth Annual Discovery on Target Meeting

Last week CHI held Discovery on Target – virtually of course. There were 20 tracks over three days and more than 650 attendees, down from 1100+ last year. Because the event was more fragmented (pun intended) than the recent DDC, which had at most four parallel tracks, the Q&As and discussions seemed smaller, though that could have just been the ones I attended. On the other hand, one of the huge advantages of the format is being able to watch concurrently scheduled talks later. More thoughts on virtual conferences are here, and if you have not already done so please take our poll on the right side of the page.
This conference has always been more biology-focused than DDC, with tracks on antibodies, immunology, NASH, gene therapy, disease modeling, and fibrosis, among others. But there were also plenty of talks on targets and methodology, which is where I’ll focus most of this post. Please add your highlights and thoughts in the comments.
Julien Orts (ETH Zurich) presented an update on his NMR2 method, which uses information from intermolecular NOEs to computationally determine protein-ligand structures without requiring full NMR assignment of the protein. We wrote about this technique in 2017 and at the time we questioned how applicable it would be to fragments due to low affinities, multiple binding modes, and fewer contacts. As it turns out, very: Julien described successes with proteins including HDM2, DsbA, bromodomains, and Pin1. Even with as few as 10-12 intermolecular NOEs he has been able to get good agreement with crystal structures. Currently he is applying this approach to SARS-CoV-2 proteins as part of the COVID-19-NMR project.
William Pomerantz (University of Minnesota Twin Cities) also presented NMR techniques. He is particularly known for his protein-observed 19F (PrOF) NMR screening, in which fluorinated tyrosine and tryptophan residues are introduced into proteins. Ligand binding changes the chemical shifts of the fluorine atoms, and by varying the concentration of the fragment, accurate dissociation constants can be determined. In early work, a screen of 930 fragments in pools of 5 against BRD4 took 11 hours (and another 11 hours for deconvolution) and provided multiple hits. We’ve covered some of his more recent work using shapely fragments here, and in unpublished work he has been screening the dual-domain construct of BRD4 and finding fragments that are ten-fold selective for one over the other bromodomain. He is further improving throughput by screening two proteins simultaneously.
Rounding out NMR, Andrew Petros (AbbVie) presented a beautiful fragment-to-lead success story on TNFα, a trimeric cytokine that has been the subject of numerous (often unsuccessful) lead-discovery efforts. A 2-dimensional NMR screen of 18,000 fragments gave just 11 hits. Crystallography of one showed two copies binding in close proximity, and linking these ultimately led to a low nanomolar binder. The series showed high clearance and no oral bioavailability, so they performed additional screens to identify different fragments that were ultimately advanced to potent compounds with animal efficacy. I look forward to reading the paper when it is published.
Finding fragments is important, but so is avoiding artifacts, the subject of a talk by Samantha Allen (Janssen). Around 2% of screening compounds can form small molecule aggregates that can interfere with assays, and if these aren’t weeded out they can quickly overwhelm an assay. Samantha described the use of resonance waveguide grating (RWG) technology, as used in the Corning Epic BT. This label-free technology is similar to SPR, but RWG can be run in 384 or 1536 well plates. Samantha showed that RWG compares favorably to dynamic light scattering for detecting aggregates. It is also 4-5 times faster and less prone to false-positives.
Covalent fragments were a theme of last month’s DDC, and they were prominent here as well. Four years ago we highlighted work out of Ben Cravatt’s lab doing covalent fragment screening in cells, but this was a rather time-consuming process. Steve Gygi (Harvard) has streamlined activity-based protein profiling and was able to screen 288 fragments in just 7 days and identify more than 1500 modified cysteine residues.
Dan Nomura (UC Berkeley) continued the theme with a wide-ranging presentation using chemoproteomics to discover covalent ligands for a variety of targets, including new E3 ligases, which can be used for developing targeted protein degraders. (Shameless plug/disclosure: Dan Nomura is a founder of my company, Frontier Medicines, and we are actively hiring across multiple positions and levels.)
Targeted protein degraders such as PROTACs were the subject of one track at last year’s DoT meeting, and this year two sequential tracks were devoted to the topic. As I suggested in 2018, fragments could be ideal starting points given that high affinity is not always necessary. This year, Stewart Fisher confirmed that he and his colleagues at C4 Therapeutics often “detune” chemical matter, lowering the binding affinity to get efficient degraders. That doing so can improve physicochemical properties is a nice bonus.
Finally, although not directly fragment-related, William Kaelin (Dana Farber Cancer Institute) gave an inspiring talk on the discovery and development of MK-6482, an allosteric HIF2α inhibitor in late-stage clinical trials for cancer linked to Von Hippel-Lindau disease; data released just last week shows durable responses in patients with kidney cancer. The science itself was lovely, but he reminded us of the ultimate stakes: “It’s not about what journal your paper is published in or whether you can fool reviewer 3, it’s about whether you publish things that are true and robust and can be built upon by others.”
Words to live by.

13 September 2020

Poll: Conferences in the age of COVID-19

The last fragment-relevant conference of 2020 is happening this week – virtually of course. I attended my first virtual conference last month and wrote up my thoughts here. As I noted, I look forward to the resumption of in person events. However, despite rapid progress, we do not know when a safe, effective COVID-19 vaccine will be widely available.
In the meantime, conference organizers need to plan – usually months ahead. If you would like to help, please answer the brief (six question) survey on the right-side of the page. Also, please leave comments below – anonymously if you’d prefer – on what virtual conferences you’ve attended, what worked, and what didn’t. We'll post results early next month.
Hope to see you next year – ideally in-person!

07 September 2020

From noncovalent to covalent fragment for NSD1

As evident at the CHI Drug Discovery Chemistry meeting a couple weeks ago, covalent fragment-based lead discovery is becoming increasingly popular. Normally this entails screening a library of electrophile-containing fragments. However, it is also possible to start with a noncovalent fragment and add the “warhead” later. This is the approach taken by Jolanta Grembecka, Tomasz Cierpicki, and collaborators at University of Michigan, Memorial Sloan Kettering Cancer Center, and Columbia University in a paper just published in Nat. Chem. Biol.
The researchers were interested in NSD1, one of three related histone methyltransferases whose potential role in cancer is uncertain – in part due to the lack of good chemical probes. They started with a two-dimensional (1H-15N HSQC) NMR screen of 1600 fragments in pools of 20, with each fragment present at 250 µM. BT1 was one of the hits, and synthesis of several analogs led to BT2, which showed low micromolar affinity as assessed by ITC as well as an IC50 of 66 µM in a functional assay. Crystallography was unsuccessful, but NMR experiments suggested considerable changes in an autoinhibitory loop that blocks the substrate-binding region of the so-called SET domain.

For whatever reason, the researchers tested thiocyanate analog BT3 and found that this binds covalently as a disulfide to a cysteine in the autoinhibitory loop. They were able to get a crystal structure of BT3 bound to the protein, which revealed significant rearrangements that allow BT3 to bind deep in the SET domain, where it makes multiple polar and hydrophobic contacts. Prudently they chose to replace the unstable thiocyanate warhead, and while acrylamide derivatives were inactive, aziridine BT5 modified the protein as assessed by mass spectrometry.
BT5 inhibited NSD1 with an IC50 of 5.8 µM after a four hour incubation and was somewhat selective against the related proteins NSD2 and NSD3 as assessed both by mass spectrometry and activity assays. It also showed good selectivity at 50 µM against 20 other epigenetic enzymes and 291 kinases. Interestingly though, at this concentration BTK was inhibited by 41% and EGFR was inhibited by 49%; both these kinases are targeted by approved covalent drugs.
Next, the researchers conducted multiple cell assays. A cellular thermal shift assay (CETSA) revealed that BT5 stabilized NSD1 but not NSD2 or NSD3. Growth of an NSD1-dependent cancer cell line was inhibited with GC50 = 1.3 µM after 3 days, NSD1-mediated histone methylation was suppressed, and several target genes showed reduced expression. BT5 also inhibited growth of non-NSD1-dependent cell lines, though at 6-8 higher concentrations, and did not alter methylation or target gene expression. Finally, the compound impaired colony formation of cells from a primary patient-derived sample with an NUP98-NSD1 translocation.
This is a nice, carefully conducted study. Refreshingly, the researchers do not attempt to oversell their results, and acknowledge that “further optimization is needed to develop NSD1 chemical probes.” But they’re off to a good start, and it will be fun to see what they – and others – will come up with.

31 August 2020

Fifteenth Annual Fragment-Based Drug Discovery Meeting

What a strange year for conferences! Everything fragment-related in the first half of the year was canceled or rescheduled. And when it turned out that postponing Drug Discovery Chemistry (DDC) from April to August was overly optimistic, Cambridge Healthtech Institute went virtual for the first time, with the meeting running last week from August 25-28. Although I was skeptical, in the end the meeting was quite successful, with more than 600 attendees – roughly three-quarters last year’s in-person attendance. 
To accommodate multiple time zones each of the parallel tracks was shorter. Days typically began around 10:00 AM EDT, which was 7:00 for me in San Francisco but midnight for folks in Melbourne. Despite this inconvenience, 40% of participants this year came from outside the US, up from 30% last year.

Most of the talks were pre-recorded, which meant that speakers could answer questions in real-time in a chat box. More importantly, the talks are available to attendees for a year, which means you can watch relevant talks from other tracks, thus relieving the FOMO inevitably experienced in meetings of this sort. It also means you can replay particularly relevant talks to your colleagues, though I wonder if this also makes speakers wary about disclosing the freshest information. The on-demand access to posters is especially useful, as it is too easy to overlook these in the usual melee. The organizers also did a nice job of trying to foster the feel of an in-person meeting, with multiple live Q&A panels, breakout sessions, and other interactive events.

There are drawbacks, chief among them the lack of spontaneous and serendipitous meetings that are one of the main benefits of in-person conferences. And of course there were technical snafus. Most talk slots were only 20 minutes, but some of the pre-recorded talks went as long as 28 minutes, forcing attendees to choose between missing part of one talk or another (or a live Q&A). Although you can fast-forward, it would be nice if you could also speed up playback speeds. And on at least one occasion the wrong talk was played, though I was able to go back and watch the correct one later.

But enough about process, what about the event itself? With dozens of talks over four days I can’t be exhaustive, so please add your thoughts to the comments.

Faster, cheaper, better.
This describes one consistent theme. Frank von Delft (Diamond Light Source) gave an update on their high-throughput crystallographic screening, which has led to > 3000 fragment hits on > 150 targets since 2016. We’ve written previously about their efforts against COVID-19, which have led to three separate sub-micromolar lead series against the viral protease MPro. This rapid progress has been enabled by crowdsourcing across more than 30 separate groups, but Frank is also moving toward crystallographic screening of crude reaction mixtures, similar to the REFiL approach described by Beatrice Chiew (Monash University, see here for a longer description). Frank hopes to make good compounds a commodity, with a 5-year vision of achieving “biologically relevant potency routinely, cheaply (£10k) and quickly (weeks).” It’s an audacious goal, and we’ll check back in 2025 to see how far the field has come.

Key to the success of these types of approaches is what plenary speaker Phil Baran (Scripps) calls “boring chemistry” that consistently works in multiple contexts. To Phil, inventing chemistry that becomes boring is a great compliment, and he showed examples of running interesting transformations in tea, beer, and wine. As the name of this blog suggests, I have a soft spot for this sort of thing, and wholeheartedly agree with his statement that “you can’t give a Nature paper to a cancer patient.”

Covalent fragments (or not)
Covalent drug discovery was also a major theme, with John McCarter (Amgen) and Matt Marx (Mirati) describing discovery of two covalent clinical compounds against an oncogenic mutant form of KRAS (see here). Dom Esposito (Frederick National Laboratory) is also pursuing KRAS using covalent Tethering. They have prepared 91 cysteine mutants and screened 18 of them against nearly 1200 disulfide-containing fragments, yielding a plethora of hits.

Alexander Statsyuk (University of Houston), Maurizio Pellecchia (UC Riverside), and Nir London (Weizmann Institute) all also discussed covalent lead discovery. Maurizio has been targeting lysine residues using sulfonyl fluorides and fluorosulfates; for the latter warhead he has been able to show reasonable pharmacokinetics in rodents. We’ve previously discussed some of Nir’s work, but here he described a nice case study against the challenging anti-cancer target Pin1 which led to potent and surprisingly selective chloroacetamides with activity in mice.

Interestingly, while chloroacetamides were the main class of MPro fragment hits, the three most advanced lead series Frank mentioned are all non-covalent. There were plenty of other nice non-covalent fragment-based success stories too, including potent selective inhibitors of the lipid kinase Vps34 (Jenny Viklund, Sprint Biosciences) and selective inhibitors of one kringle domain of apolipoprotein(a) (Jenny Sandmark, AstraZeneca).

As always, there were many talks on methods, especially during the Fifth Annual Biophysics Friday track. John Quinn (Genentech) provided a state-of-the-art update on SPR, and mentioned that they are able to screen 3000 fragments in a day or two using the Biacore 8K. This allows them to assess ligandability for new targets, and the ligandability score is predictive for how well the target performs in HTS or DEL screening.

Amit Gupta described NanoTemper’s new Dianthus instrument, which relies on the temperature-related intensity change (TRIC) of a fluorophore bound to a protein. This is similar to their MST approach though it appears to be higher-throughput, and a paper benchmarking these techniques against DSF and SPR should be coming out later this year.

Also on the subject of thermal shifts, Justin Hall (Pfizer) gave a provocative presentation on using these to determine ligand affinities. He noticed a correlation in his own research, but the prevailing wisdom held that irreversible thermal denaturation (as seen for most proteins) would not provide thermodynamic parameters. Nonetheless, perhaps because proteins are fundamentally similar (consisting as they do of chains of 20-odd amino acid residues), the temperature-dependent Arrhenius functions and activation energies of unfolding are also similar, and thus for heating rates of 4°C/minute and 100 µM ligand one can extract dissociation constants. However, he did mention that this approach is restricted to reasonably tight ligands (KD < 20 µM). Also, if a ligand binds to the unfolded state of the protein, or to multiple sites, all bets are off.

In the interest of time I’ll stop here, but if you’d like to experience a virtual conference yourself, there will be a number of good FBLD talks at Discovery on Target next month, and I hope to “see” you there. But I especially hope that in-person conferences will resume next year once our industry – and competent governments – get COVID-19 under control.

24 August 2020

SAR by MicroED?

Of the dozen-plus methods to find and characterize fragments, only two have historically been able to provide detailed binding information: protein-detected NMR and X-ray crystallography. Earlier this year we described how researchers at Astex are using cryo-EM for FBLD. In a new open-access Communications Biology paper, Hongyi Xu and collaborators at Stockholm University, Lund University, and SciLifeLab describe another variation of cryo-EM.

Microcrystal electron diffraction (MicroED) is something of a cross between standard crystallography and cryo-EM. Like more “conventional” cryo-EM, an electron microscope is used to collect data on flash-frozen samples. But rather than painstakingly reconstructing thousands of images of individual protein molecules, MicroED uses diffraction of electrons from crystals that are far too small for standard X-ray crystallography. Even though electrons rather than photons are being scattered, diffraction is diffraction, so well-established X-ray crystallography methods can be used for processing MicroED data.

The researchers focused on human carbonic anhydrase II (HCA II), a popular model protein that has also been used to showcase X-ray crystallographic, native MS, and SPR methods. The microcrystals were less than 500 nm thick, smaller than most bacteria and at least 100-times smaller than typically used for X-ray crystallography. MicroED data were collected on microcrystals of native HCA II as well as microcrystals that had been soaked for 20 minutes with the known ligand acetazolamide.  At just 13 heavy atoms, this approved drug is still comfortably a fragment.

Data were collected to 2.5 Å resolution, which is modest especially compared with the 1.1 Å resolution of a published crystal structure. Nonetheless, the ligand density was clearly visible, and the refined model was very similar to the published crystal structure as well as another published structure of the complex determined by neutron diffraction. The researchers note that the observed features are similar to those that could be expected of a crystal structure solved at the same resolution.

The researchers note several potential advantages of MicroEM over X-ray crystallography. It can sometimes be difficult to obtain sufficiently large crystals for crystallography, particularly when protein is limited. Smaller crystals may allow faster diffusion of ligands into the crystals. And at higher resolution, individual hydrogen atoms are more easily resolved using electron diffraction than X-ray diffraction.

Will these advantages be enough to make MicroED a truly practical method for FBLD? It is not clear from the paper how long the data collection and processing took, HCA II is a friendly protein to work with, and acetazolamide is a high affinity ligand. That said, MicroEM was first described only in 2013, and this paper demonstrates that ligand binding modes can be determined. It will be fun to watch this technique develop.

17 August 2020

Merge and grow: Fragment-based activators of SOS1

The RAS family of proteins is implicated in roughly one third of cancers, and as such has been a long-standing target for drug discovery. Earlier this year we highlighted how covalent fragment-based approaches were instrumental in discovery of direct KRAS inhibitors. A recent paper in J. Med. Chem. by Stephen Fesik and colleagues at Vanderbilt University takes a more unusual approach.

RAS proteins are activated when guanine exchange factors (GEFs) such as Son of Sevenless 1 (SOS1) exchange GDP for GTP. Clinical compounds bind to a mutant form of KRAS and block this process. Previous high-throughput screening in Fesik’s group had found molecules that bind to and activate SOS1-mediated nucleotide exchange. While it might seem counterintuitive to activate a known oncogene, these molecules can actually block downstream RAS signaling by inducing a feedback mechanism. Here, the researchers used fragment screening to look for a new series.

The catalytic core of SOS1 is ~65 kD, relatively large for the protein-detected NMR methods beloved of the Fesik group, so they produced proteins in which the methyl groups of Ile, Val, Leu, and Met were 13C-labeled. Selective Ile to Ala mutations allowed them to assign the various methyl groups. An 1H-13C HMQC screen of nearly 14,000 fragments yielded 59 hits (~0.1%), all quite weak: only five had dissociation constants better than 1 mM. Crystal structures were obtained for 16, revealing that all of them bind in the same site previously identified (see also here for similar work from a different group).

Fragments F-4 and F-7 bound in similar positions as each other and also as the HTS-derived compounds, so the researchers merged them to yield molecules such as compound 1b, with improved affinity and ligand efficiency.

Crystallography suggested that a nearby aspartic acid residue could be engaged through fragment growing, leading to molecules such as compound 2d. In addition to low micromolar affinity, this molecule also activated SOS1-mediated nucleotide exchange. In a cell-based assay, the compound caused enhanced phosphorylation of downstream target ERK at low concentrations and decreased phosphorylation at high concentrations, similar to what had been seen for the earlier series of molecules. Presumably, the biphasic response is due to a negative feedback loop that ultimately clamps down RAS signaling.

This is a nice example of structurally enabled fragment-merging and growing, assisted by knowledge of other ligands. While the compounds are probably not sufficiently potent to serve as chemical probes, they could be useful starting points. Activating the RAS pathway may or may not be a good approach for treating cancer, and we need suitable chemical tools to answer this question.

10 August 2020

A fragment library designed for merging: application to PKCζ

Advancing fragments in the absence of structural information has a reputation for being so challenging that some people do not even attempt it. Modeling can help, but what if you could improve your odds by designing your library strategically? This approach has been demonstrated in a recent J. Med. Chem. paper by Masakazu Atobe and colleagues at Asahi Kasei Pharma.

To facilitate fragment merging, the researchers synthesized a library of 5000 substituted isoquinoline fragments. As illustrated by the drug fasudil, isoquinoline is a privileged pharmacophore for binding to the hinge region of kinases. Importantly, isoquinoline has 7 different positions from which to grow: screening monosubstituted versions would potentially allow rapid merging of hits. This approach is conceptually similar to that used to discover vemurafenib.

The target of interest was protein kinase C ζ (PKCζ – that’s a zeta, by the way), one of the 11 members of the PKC family that has been implicated in diseases ranging from diabetes to cancer. Previously reported inhibitors are insufficiently potent or selective, in part because no crystal structure of the kinase has been reported. The researchers were interested in developing a chemical probe to better understand the biology.

A biochemical screen of the 5000-member isoquinoline library at 100 µM yielded just a dozen hits, with IC50 values ranging from low to mid-micromolar. Importantly, substituents were found at four different positions, thus facilitating fragment merging. The researchers first merged fragment 6 with fragment 8, resulting in mid nanomolar inhibitor 10. Further optimization yielded compound 21, which is highly selective for PKCζ in a panel of 216 kinases and also has good pharmacokinetic properties in mice. However, cell potency is relatively modest.

Next, the researchers merged fragment 7 with fragment 9 to generate sub-nanomolar compound 26. This molecule also inhibited protein kinase A, but further optimization led to compound 37, which showed excellent selectivity in a panel of 381 kinases as well as good mouse pharmacokinetic properties and mid-nanomolar activity in a cellular assay. Encouragingly, the compound also showed good activity in a collagen-induced arthritis mouse model. The aniline – which adds about ten-fold to the affinity – may ultimately need to be removed, but clearly this molecule is well-suited for further optimization. 

This paper provides two lovely examples of fragment merging by design, but how general is the approach? One of the key advantages of fragment-based screening is the ability to survey huge swaths of chemical space. Building an entire library around a single fragment obviously constricts this. The fact that the hit rate (0.24%) was so low perhaps illustrates this point; it would be interesting to know how ligandable PKCζ is, or whether a library built around a different privileged pharmacophore would yield a higher hit rate. Lower expected hit rates necessitate larger libraries; 5000 fragments is already more than average according to our poll. And of course, if you are going to build a library of thousands of similar fragments, you had better be certain you choose one that has good pharmaceutical properties, further limiting your choices. Despite all these cavaeats, clearly the investment paid off for PKCζ. It will be fun to see what else comes out of this effort.

03 August 2020

In situ click chemistry on RNA

In templated or in situ reactions, bonds form between two fragments that are brought together in the context of a larger molecule such as a protein. We have written previously about dynamic combinatorial chemistry (DCC), which depends on reversible bond formation where the larger molecule shifts the equilibrium toward the linked fragments. For irreversible bond formation, the larger molecule effectively catalyzes the formation of an inhibitor (or at least a binder). In a recent Angew. Chem. Int. Ed. paper Jyotirmayee Dash and colleagues at the Indian Association for the Cultivation of Science describe an application of the latter that uses RNA as the template.

The target of interest was TAR RNA, a short region of viral RNA essential for HIV replication. We have previously highlighted a few examples of fragment screening against RNA (here, here, and here), including TAR, but most of the hits were weak.

The researchers used azide-alkyne cycloaddition, the quintessential click chemistry reaction. They built a small library of four alkynes (only one of which was fragment-sized) and 11 azides. All of these were incubated together (4 µM of each alkyne and 1 µM of each azide) in the presence of 5 µM biotin-labeled TAR RNA for 72 hours. (The reaction is typically slow at room temperatures unless catalyzed by metal ions.) Magnetic streptavidin-coated beads were then used to capture the RNA and any bound ligands, which were identified by HPLC-MS. Control experiments were run with TAR DNA or a mutant form of TAR RNA lacking an essential bulge. The result was one fairly potent compound (below) that was specific for TAR RNA, as well as a couple other molecules that were both weaker and less specific.

The affinity of compound 3ba for TAR RNA was measured by isothermal titration calorimetry (ITC) as well as by a fluorescence assay, which were in good agreement. Importantly, the ITC data suggested 1:1 binding, which is particularly important given that the ligand contains two 2-aminothiazoles, a moiety that has been called a PrAT for its promiscuous behavior. Finally, the ligand could displace the Tat peptide at low micromolar concentrations, suggesting that it is binding at the biologically relevant site of TAR.

I do have a few quibbles. It would have been interesting if the researchers had reported the affinity of the azide and alkyne themselves to see how much of a boost they got by linking. And since the most potent molecule is not always selected from target-guided synthesis, it would have been interesting to make and test other possible cycloaddition products to see if they missed anything useful.

Still, it is nice to see a submicromolar RNA binder come out of an in situ screen. Targeting RNA with small molecules has recently become trendy, and it will be fun to see how far approaches like these can go.

27 July 2020

Flatland: a nice place to be

The ideal shape of compounds used for biological screens is a subject of vigorous debate, with some arguing that shapely molecules may be superior in various ways to the “flatter” aromatic compounds that tend to dominate libraries. This view was expressed more than a decade ago in the paper, “Escape from Flatland: Increasing Saturation as an Approach to Improving Clinical Success.” However, those conclusions have been challenged. Since many of us are trying to discover drugs, it is worth asking what actual drugs look like. This is the subject of a new ACS Med. Chem. Lett. paper by Seth Cohen and colleagues at University of California, San Diego.

Assessing shapeliness is itself contentious. Here the researchers chose the intuitive metric, principal moment of inertia (PMI), which uses a simple triangle plot to assess whether a molecule is more rod-like, disk-like, or sphere-like. The degree of shapeliness (3D Score) can be calculated by summing the x- and y-coordinates to give values between 1 (rod- or disk-like) and 2 (sphere-like).

The researchers first extracted more than 8500 drugs and nutraceuticals from DrugBank, all of which had associated three-dimensional structures and MW >100. PMI calculations revealed that nearly 80% were linear or planar, with 3D Scores < 1.2. Another 17.5% had 3D Scores up to 1.4, while only 0.5% were greater than 1.6. Interestingly, this distribution is similar to that of the ZINC database of small molecules. You might expect a correlation between size and shapeliness, with larger molecules being more three-dimensional, but this was not the case. Perhaps related, a separate analysis found no correlation between shapeliness of fragments and resulting leads.

The 3D structures of compounds in DrugBank are calculated for energy-minimized conformations, which are not necessarily the biologically relevant conformations. So the researchers next went to the protein data bank (PDB) and its crystal structures of 502 unique DrugBank molecules bound to various proteins. Some molecules were represented multiple times (1036 structures of sapropterin!), and for these the PMIs were averaged. The results of this analysis were similar, with 83.5% of molecules having a 3D Score < 1.2 and just three molecules with a 3D Score > 1.6. As with the DrugBank data, there was no correlation between 3D Score and molecular weight.

Further analyses of compounds with multiple crystallographic structures was interesting. For diclofenac, with 51 PDB entries, 3D Scores ranged from 1.03 to 1.52, with the minimized score being 1.22. However, some of these structures are likely low affinity with questionable biological relevance. In contrast, for five approved HIV drugs, the PMIs remained very similar for molecules bound in the active sites.

Getting out of flatland is surprisingly difficult: the researchers examined the PMIs for several fragments from libraries designed to have shapely members and found that none had 3D Scores > 1.4. They suggest clever ways of increasing three dimensionality, such as building organometallic molecules. While this is likely to increase novelty and patentability, it also introduces unknown biological risks. One analysis that would be interesting is whether natural-product-derived drugs are significantly shapelier than their purely synthetic counterparts.

The researchers conclude:

The true need for topological diversity in feedstocks and final drug molecules remains unclear given the overwhelming number of linear and planar drugs. The question remains as to whether more 3D compounds represent attractive and untapped therapeutic space, or if more linear/planar molecules are indeed the best topologies for bioactive molecules.

This is indeed an interesting question, and I hope that chemists – particularly those in academia – continue to make and test ever more exotic molecules. But since the first word of this blog is “Practical,” I would not discount the more planar molecules that make up most of our pharmacopoeia.