Fragments of the future - part 2 (The intersection of chemical and biological space)

In a previous post about heterocycles that appear chemically feasible but have not been reported, we wondered whether these molecules would show biological activity. The structure of biologically relevant chemical space – that fraction of possible molecules that will exhibit some biological effect – is of great interest, but as yet unknown. Brian Shoichet and coworkers at UCSF have just published a thought-provoking analysis in Nature Chemical Biology that is also relevant to developing new fragment libraries. 

The researchers ask why it is that HTS collections of a million or so compounds, vanishingly small in comparison to the roughly 1,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 possible small drug-sized molecules, nonetheless so often succeed in identifying hits. A lovely paper by Tobias Fink and Jean-Louis Reymond had previously computationally enumerated all possible compounds with up to 11 C, N, O, and F atoms. Of these 26,429,328 molecules, 25,810 are commercially available. 

Shoichet and colleagues compared the structures of these compounds with the structures of metabolites and natural products (all of which have by definition been processed by at least one protein) and found that the commercially available compounds were much more similar to natural products and metabolites than were non-commercially available compounds. Indeed, the more similar a molecule is to a known natural product or metabolite, the more likely that it is available for purchase; 2918 of the commercially available compounds are in fact natural products or metabolites. 

The bias also increases exponentially with molecular size: a random 11-atom commercial compound is almost 1000-times more likely to resemble a natural product or metabolite than is a non-commercially available molecule, whereas the bias is only about 2-fold for 6-atom molecules. Similar results were observed with other libraries. 

The authors conclude that: A major reason why the screening of synthetic compounds ever finds notable hits is that our libraries are biased toward the sort of molecules that proteins have evolved to recognize.  

This resemblance is reasonable. After all, most commercially available compounds are ultimately derived from naturally occurring starting materials, so their similarity to natural products isn’t surprising. Moreover, historically much of chemistry was devoted to natural product synthesis, so many of the intermediates built up over the years resemble natural products. And of course, once you learn how to do chemistry on one moiety, you will tend to stick with it unless you have a good reason to do otherwise; each heterocycle behaves (often frustratingly) differently, so if a natural-product-like molecule does the job, why look for trouble? 

But does this “biogenic bias” mean that the rest of chemical space is a biological desert? Not necessarily. I can imagine at least two alternative models of chemical-biological diversity space. 

Let’s call one model “lamp posts in dark fields.” Consider a vast field of some crop that can only be harvested by night. There are lights scattered haphazardly throughout the field. One might expect that the crops immediately under the lamp posts would be harvested more intensively than crops in darker parts of the field, even if other areas are equally productive. In this scenario, the lamp posts reveal natural products and similar molecules, but much – or even most – of (unlit) chemical space may also be biologically active, it just hasn’t been sampled yet. 

Another possibility is the “oil-field model.” As we are all too aware, petroleum is distributed very unevenly across the globe. In some areas, such as Texas, oil was easy to find and easy to extract. In others, such as the deep ocean or the high arctic, oil is harder to find and more technically demanding to access. In this scenario, there are vast pockets of chemical space that are relevant to biology, they just haven’t been identified (let alone accessed) yet. 

These are fun, speculative questions, but the paper provides some practical data. Specifically, 83% of core ring scaffolds found in natural products are absent from commercial libraries. In fragment or lead-sized molecules of MW < 350 with less than three stereocenters, 1891 rings scaffolds found in natural products are not commercially available. These could be useful additions to fragment libraries, and the paper lists 18 examples. 

In fact, at least one company, deCODE, is explicitly enriching its fragment collection with molecules based on natural products and metabolites. This might be a good strategy. After all, even if the “lamp posts in dark fields” model is correct, there are plenty of brightly illuminated, unharvested chemotypes. At least for now, picking these may be more productive than venturing into the twilight-zone of uncharted chemical space.

Fragmenting natural products – sometimes PAINfully

Many drugs have their origins in natural products. But as any synthetic organic chemist will tell you, natural products often have complex architectures that can take years of effort and dozens of chemists to make in the lab. Thus, many of the compounds made in industry look quite different from natural products, particularly in the past few decades. High failure rates in drug discovery have led folks to return to natural products or similar compounds, such as those from diversity oriented synthesis (DOS). In a recent issue of Nature Chemistry, Herbert Waldmann and colleagues at the Max-Planck Institute in Dortmund examine whether natural products can serve as starting points for new fragments.

The researchers started by computationally deconstructing 183,769 natural products into 751,577 component fragments. After various filters (size, lipophilicity, reactivity, etc.) they arrived at 110,485 fragments sorted by similarity into 2000 clusters. The resulting fragments differ in their overall calculated properties from commercial fragments. This is all highly reminiscent of the Emerald (nee deCODE) “fragments of life”, though surprisingly that work is not referenced.

One challenge of designing new fragments is that you may not be able to buy them. In this case, nearly half of the clusters did have a compound that could be purchased – though perhaps this somewhat defeats the purpose of trying to explore novel chemical space. At any rate, 193 fragments were either bought or synthesized. These were tested in functional assays against p38a MAP kinase and several protein phosphatases. A number of hits were identified, and in the case of p38a, nine kinase-fragment co-crystal structures were solved. Some of these were similar to previously reported fragments, but others were more unusual. Together with the crystal structures, these fragments provide new ideas for a well-studied target.

Looking at the structures of some of the phosphatase inhibitors, however, I started to worry. One strong point of the paper is that it is very complete: the chemical structures of all 193 tested fragments are provided in the supplementary information. Unfortunately, the list contains some truly dreadful members; 17 of the worst are shown here, with the nasty bits shown in red. All of these are PAINS that will nonspecifically interfere with many different assays.

Compounds 15, 44, 49, 159, 166, 173, 174, and 175 are catechols; compounds 89 and 151 (yes, they are the same molecule – guess they really liked this one), 165, 166, 167, and 168 are quinones; compounds 55, 89/151, and 166 are hydroquinones; compound 20 is a Michael acceptor; compound 76 is an epoxide; and compound 184 is a redox cycler. In other words, these fragments are a depressing example of life imitating art (or at least satire).

To be blunt: none of these molecules should appear in a screening library today.

I don’t want to pick on these researchers; it is after all laudable that they fully disclosed the structures of their molecules.

However, I am concerned that other people may build libraries containing some of these fragments, or worse, that opportunistic vendors will start selling “natural-product derived fragments.” Indeed, most of these molecules are commercially available. It is disappointing that so many nuisance compounds would find their way into research published in a Nature family journal, and I think it is important to call it out. Only by publicizing the problems that can arise will people be made aware of the dangers.

Natural Product Derived Fragments against MMP-13

I have been lucky to work on a lot of systems that very much interest me.  I, in particular, love metallo-proteins.  I worked on rubredoxin as a post-doc and when I moved into industry I worked on a slew of metalloproteins.  So, I love it now when I see papers on targets I used to work on.  This paper does exactly that while also letting me riff (later) on Natural-Product-Derived Fragments (NPDF). 

NPDF has a long history in FBDD having been discussed here, here, here, and so on.  Many vendors and some companies have NPDF libraries (whether they call them that or not).  However, these libraries have yet to be proven to be an efficient route for "discovering clinical drug candidates".  Lanz and Riedl set out to do this against MMP-13 (how many of your just said, yeah I worked on that target?).  All MMP-13 clinical candidates with strong ZBG (Zinc-binding groups) have failed.  They are aiming to develop a MMP-13 without a strong ZBG.  Of course, we have seen a LOT of work towards this goal: here, here, and here for example.  The authors propose that the use of NPDF prevents the problem of using fragments with "debatable biological properties".  This seems to the be the argument used by the NPDF people, since these fragments are found in nature they have desirable properties.  I have never bought this line of reasoning for a variety of reasons.  

To their end, the authors selected uracil as their starting NPDF for these reasons: good synthetic starting points, cis amide bonds, and its found in a variety of natural products (nucleic acids).  They docked it in the S1' non-zinc binding site and found a strongly conserved binding site. [For me, and I would imagine a whole lot of people, this fits in the "things you already knew" category.]  The uracil interacted with the NH an CO of Met232 via its cis amide bonds and "addresses" Lys228.  Several compounds were made from the uracil starting point (Figure 1):

Figure 1.  2: 5 uM vs. MMP-13, < 50% Inhib against 1,2,3,7,8,9,12, and 14 at 20 uM. 3: 10 nM vs. MMP-13, < 50% Inhib against 1,2,3,7,8,9,12, and 14 at 20 uM2: 5 nM vs. MMP-13, < 50% Inhib against 1,2,3,7,8,9,12, and 14 at 10 uM

So, in the end, they have created a potent and selective compound.  They did use a NPDF as a starting point.  Making these compounds is not something that bowls me over either for a Technical Difficulty score or Artistic Merit.   However, I would not go so far as to say that they have validated the NPDF approach.  I think to show that a generic approach works you need more than one (relatively well known) target with more than one (relatively well known) fragment. 

Stabilizing protein-protein interactions

Despite the fact that the second FDA-approved fragment-derived drug targets a protein-protein interaction (PPI), these types of targets have a well-earned reputation for being difficult. Most researchers try to disrupt PPIs. An alternative is to stabilize PPIs. This is not as crazy as it sounds: rapamycin, tafamidis, and PROTACs all stabilize PPIs. In a paper just published in J. Am. Chem. Soc., Michelle Arkin, Christian Ottmann, and collaborators at UCSF, Eindhoven University of Technology, Novartis, and the University of Duisburg-Essen bring fragments to bear on the problem.

The researchers were interested in the protein 14-3-3δ, a “hub” protein that binds to more than 300 other proteins (not all at the same time). One of these is estrogen receptor α (ERα): binding prevents the transcription factor from dimerizing and binding to DNA. The natural product fusicoccin A (FC-A) binds at the interface of 14-3-3δ and ERα and stabilizes that interaction, thereby inhibiting the growth of breast cancer cells. Because FC-A is a structurally complex natural product, the researchers sought fragments that would have a similar effect. They used Tethering, in which reversible disulfide bond formation stabilizes a protein-ligand complex, allowing its identification (see here and here). Specifically, fragments that bind near a cysteine residue are resistant to reduction, and the extent of binding can be detected by mass spectrometry.

The 14-3-3δ protein conveniently contains a cysteine residue in the vicinity of the ERα binding groove; the researchers used this native protein and also created two additional mutant proteins in which the native C38 cysteine was removed and new cysteine residues were introduced nearby. These three proteins were then screened against a library of 1600 disulfide-containing fragments under mildly reducing conditions in the presence or absence of a phosphopeptide derived from ERα. Most of the hits against the native protein were weak, but several hits against the N42C mutant were both resistant to reduction and also bound preferentially to the 14-3-3δ/ERα peptide complex compared to 14-3-3δ alone. Thus, ERα could enhance the binding of fragments to 14-3-3δ.

Next, the researchers used a fluorescently labeled peptide derived from ERα to show that one fragment could improve the apparent dissociation constant for the peptide and 14-3-3δ about 40-fold, from 1.3 µM 32 nM. Crystallography revealed that the cooperative fragments bound at the PPI interface, as expected given the location of the cysteine residues. The cooperative fragments placed a phenyl group in close proximity to a valine residue from the ERα peptide.

The researchers then examined the selectivity of one of their stabilizing fragments for other 14-3-3δ client proteins. In the case of a phosphopeptide derived from TASK3, which has a similar sequence to that of the ERα peptide, the fragment also showed cooperative binding. However, two peptides from other client proteins competed with the fragment for binding, and crystal structures revealed that the binding modes would be incompatible.

This is a nice illustration of site-directed fragment discovery to identify fragments that can modulate protein function in a more sophisticated manner than simple inhibition. One of the nice features of Tethering is that – like crystallography – it is able to identify extraordinarily weak binders. Unfortunately, this sometimes makes the hits challenging to advance: NMR experiments do show binding between a non-disulfide-containing derivative of one of the fragments and the 14-3-3δ/ERα peptide complex, but at high concentrations. It will be interesting to see whether this can be built into a potent non-covalent binder, and/or whether other types of covalent modifiers will be able to produce useful chemical probes for this target.

Funky fragments

Natural products have led to many approved drugs, and there is an increasing appreciation that Nature often knows best. Indeed, several published fragment libraries incorporate natural products or natural product-like molecules (see for example here, here, and here). With all this attention, it was inevitable that commercial fragment suppliers would spot this market need.

SerpentesOleum, Inc. has just launched a library they call FUNK: Fragments Uncovered in Natural Kompounds. This set consists of several hundred natural products and derived fragments carefully selected to maximize hit rates. For example:

The company has screened their library against targets such as PTP1B and falcipain-1 and obtained remarkably high hit rates in functional assays. In fact, SerpentesOleum is so confident that they’re offering a money-back guarantee if you don’t obtain at least one active against your target, no matter what it is. Looking at the structures of their compounds, I have no reason to doubt their claim.

Fragment optimization without purification

Compound purification can be a major hassle: separating the desired product from starting materials, reagents, and byproducts often takes far longer than making the compound in the first place. As we’ve previously noted, this is especially true for small, polar fragments – which are particularly attractive for drugs. Two new papers address this challenge. (Shameless plug: my company Carmot Therapeutics also has a solution to this problem.)

In J. Med. Chem., Paul Brough and Vernalis colleagues describe their discovery of inhibitors of all four isoforms of pyruvate dehydrogenase kinase (PDHK), potential targets for diabetes and oncology. The ATP-binding site of these four enzymes is similar to that of oncology target HSP90, in which Vernalis has a long-standing interest.

A screen of 1063 fragments (each at 0.5 mM) against PDHK-2 yielded 78 hits that were positive in three different NMR-based assays and also ATP-competitive. These yielded a whopping 43 structures when soaked into crystals of the related isoform PDHK-3. Compound 6 was one, and the binding mode was very similar to that previously seen for the same fragment with HSP90. Fragment growing rapidly led to molecules such as compound 8, with low micromolar potency. This compound was almost equipotent against HSP90, but modeling suggested that it might be possible to further grow this molecule in a direction that would be accommodated in the PDHKs but not in HSP90.

The next step was to make a bunch of analogs, and here's where avoiding purification becomes advantageous. Specifically, the researchers turned to off-rate screening (ORS), which entails making compounds and then testing the impure mixtures using surface plasmon resonance (SPR) to look for those which dissociate more slowly. Since off-rate is not dependent on the concentration of ligand, a low yield shouldn’t change the results of the assay.

An initial library of 56 compounds led to the discovery of compound 18, and subsequent libraries and medicinal chemistry ultimately yielded VER-246608, which is a potent pan-PDHK inhibitor. As designed, it is also completely inactive against HSP90. The molecule is described more thoroughly in this Oncotarget paper, which reveals that despite activity against PDHKs in cells, VER-246608 is not particularly effective at slowing the proliferation of cancer cells. Still, it does appear to be a useful chemical probe for further exploring the biology of the PDHKs.

Shifting methods but staying with the theme of assaying impure compounds brings us to a paper in SLAS Discovery by Sten Ohlson, Brian Dymock, and colleagues at Nanyang Technological University and the National University of Singapore. The protein tested was HSP90, and the method used was weak affinity chromatography, or WAC (see here, here, and here).

Like SPR, WAC also uses an immobilized protein. However, whereas SPR provides the (kinetic) off-rate, WAC provides the (thermodynamic) dissociation constant, which is calculated from the change in retention time of the molecule as it passes through a column containing protein-bound resin. In this case the researchers synthesized a mixture of five different compounds which varied from 7-24% of the mixture. This crude sample was analyzed by WAC, and the resulting dissociation constants, ranging from 48-147 µM, were satisfactorily similar to the values obtained using pure compounds.

Both of these approaches should accelerate screening and facilitate the analysis of complicated mixtures, such as natural product extracts. It will be fun to watch for more examples.

Cussed curcumin

Teddy’s retirement from the blog has cut down on the number of PAINS-shaming posts, and truth be told there are so many candidate papers that they could easily swamp fragments, which I suspect would drive away most of the readership. That said, I did want to highlight an exhaustive Perspective about a particularly diabolical natural product just published today in J. Med. Chem. by Mike Walters and collaborators at the University of Minnesota, Brigham and Women’s Hospital, and the University of Illinois (and also covered in a news story in Nature.)

We’ve previously discussed some of the types of artifacts that can plague small molecule screens: aggregation, covalent adducts, redox cycling, fluorescence, photoreactivity, and more. Curcumin is a jack of all trades in that it is capable of all of the above. It’s also unstable even at neutral pH, and can decompose into other reactive species. It is the quintessential chemical con artist: if you have an assay, curcumin will probably be active in it.

The new paper is a thorough investigation (18 pages, with 164 references) of the chemistry and biology of curcumin, covering in gruesome detail all the many ways it can deceive. After discussing the history and physicochemical properties (and liabilities), several literature case studies where curcumin is proposed as having biological activity are explored and thoroughly demolished; one of these has been retracted but continues to be cited uncritically years later.

One might expect that something which hits so many assays would be toxic. This turns out not be the case: curcumin is present at 1-6% in tasty turmeric and only seems to show any adverse events at very high doses – several grams per day. The reason, the researchers show, is that curcumin’s pharmacokinetics are lousy, with oral bioavailability of less than 1%. This is a very literal example of the cliché “garbage in, garbage out.”

Sadly, these properties have not dampened interest in testing curcumin in people. The researchers identify 135 registered clinical trials, only eight of which have reported study results, with 49 either recruiting or not yet recruiting. The few examples where results have been reported are not particularly encouraging.

Typing curcumin into PubMed pulls up close to 10,000 papers, with more than 150 published in J. Med. Chem. alone. Will this devastating exposé help? For honest and diligent researchers, it should serve as a flashing warning to be extremely careful with any data gathered using curcumin. Unfortunately, some in the scientific community may not care as long as they are able to pump out papers. Indeed, according to Wikipedia, at least one prominent curcumin researcher had to retract several papers because of questionable “data integrity”. And there may be still darker motives: type curcumin into Google and the top results are ads touting the stuff. There’s money to be made, and even more if you slap on some scientific lipstick.

And despite specific J. Med. Chem. author guidelines to be cautious about “interference compounds” and “provide firm experimental evidence in at least two different assays that reported compounds with potential PAINS liability are specifically active and their apparent activity is not an artifact”, the journal recently published a paper fully devoted to the synthesis and SIR of rhodamine derivatives, with no consideration of mechanism nor mention that they can be problematic. (Indeed, the researchers do not even bother to include detergent in their enzymatic assay!)

All of which is to say that it’s easy to publish crap. But hopefully now, more people will recognize it as such.

Growing into closed pockets

Fragment-growing is a popular way to increase the activity of fragments, all the more so when there is an obvious place towards which to grow. In a paper published online in the J. Am. Chem. Soc., Iwan de Esch and colleagues at VU University Amsterdam describe the structural and thermodynamic consequences of one such effort, and conclude that binding in a certain normally closed pocket is enthalpically rather than entropically driven.

The researchers were interested in acetylcholine-binding protein (AChBP), a soluble and crystallizable homolog of an important class of ligand-gated ion channels. This is the same protein (and the same group) highlighted last year in the context of ligand efficiency hot spots. In the current work, a fairly potent fragment, compound 1, was co-crystallized with AChBP and the structure solved. This fragment binds in roughly the same position as the more potent natural product lobeline (compound 2), but lobeline contains a hydroxyphenethyl group that the fragment lacks (see figure). This moiety binds in a hydrophobic pocket that does not appear in the fragment complex due to the movement of a tyrosine residue. Recognizing the potential for the pocket to form, the researchers introduced this moiety into their own molecule, producing compound 3.

Gratifyingly, compound 3 binds about 50-fold more tightly than the initial fragment. This molecule was also co-crystallized with AChBP, and, as designed, the phenyl group binds in the “lobeline pocket". Moreover, compound 3 does not show a corresponding increase in affinity towards a version of AChBP from a different organism that does not have this pocket.

To correlate binding mode with thermodynamics, the researchers also characterized the binding of their compounds using surface plasmon resonance (SPR) and isothermal titration calorimetry (ITC). ITC measures enthalpy directly, and performing SPR analyses at different temperatures can also be used to dissect enthalpic and entropic terms. The two methods generally concurred, though the numbers did jump around a bit, and in a few cases SPR predicted negative (ie, unfavorable) entropy where ITC suggested positive (favorable) entropy for the same compound.

Both SPR and ITC indicated that the increase in potency for compound 3 over compound 1 is driven by enthalpy, not entropy. This is somewhat unexpected, as the contacts made by compound 3 are largely hydrophobic, and the simplistic view is that such contacts are usually entropy-dominated. (The added hydroxyl in compound 3 doesn’t appear to be doing anything useful, and in fact removing it increases potency roughly three-fold). The researchers suggest that, for poorly solvated hidden pockets such as this, enthalpy may dominate. Perhaps also the protein rearrangement necessary to open the pocket is entropically costly.

There is much more data in the paper than can be summarized here, and the notion that ligands that induce conformational changes in proteins could be enthalpic rather than entropic binders is an intriguing hypothesis. However, as a vigorous debate last year demonstrates, it is still unclear whether knowing the answer – as scientifically interesting as it may be – will have practical implications for drug discovery.

Fragment-based Drug Discovery Down Under (FBDD-DU) 2024

The end of June brought me to Brisbane for the fifth FBDD-DU Conference, which was meeting for the first time outside Melbourne. This was also my first FBDD-DU conference since 2019, and it was nice to see a wide range of talks from around Australia and beyond. As always, I won’t attempt to be comprehensive, so if you attended, please feel free to add your observations.

 

Techniques

Experimental techniques received considerable attention. Félix Torres (NexMR) described using an inexpensive benchtop NMR that doesn’t require liquid helium. Fragments were screened using photochemically induced dynamic nuclear hyperpolarization (photo-CIDNP). The method is so rapid that it is limited more by sample handling than data collection, and the Torres team is speeding things up using flow technology. Right now photo-CIDNP is still very much DIY, but rumor has it that Bruker may soon launch a photochemical module for their benchtop instrument.

 

We’ve written about high-throughput crystallographic screening at the Diamond Light Source, and synchrotrons around the world are building similar platforms. Kate Smith described integrated systems at the Swiss Light Source which automate crystallization, fragment screening, data collection, and data processing. She also described increasing automation of fragment screening using the free-electron laser (FEL), which we wrote about here. Current throughput is around 40 compounds per day and requires large amounts of protein, but these are still early days.

 

Australia is building their own high-throughput crystallography platform, and various components were described by Roxanne Smith (University of Melbourne), Gautham Balaji (Monash Univesrity), and Yogesh Khandokar (ANSTO-Australian Synchrotron). Watch this space!

 

Speaking of Australia, Nyssa Drinkwater described Compounds Australia, a national repository of more than 2.5 million molecules, including several fragment collections. Members, who can be from outside Australia, can store their own libraries within the facility to ease collaborations with other groups, and they can also access public libraries of compounds, including unusual Antipodean natural product extracts. I was fortunate to be able to visit the facility at Griffith University and can attest that it is easily the equal of those in large pharma.

 

Turning to mass spectrometry, Sally-Ann Poulsen (Griffith University) described covalent library screening against PRMT5, a target we’ve written about here. Sally-Ann is also a pioneer of (conventionally non-covalent) native mass spectrometry, and she described applying this methodology to screen small molecules against RNA.

 

But the star of the conference was SPR, appearing in multiple talks. Long-time readers may recall an instrument made by SensiQ, with its gradient injection capability to accelerate data collection. This is now marketed by Sartorius, and Lauren Hartley-Tassell (Griffith University) described using it to screen a glycoprotein. The larger plumbing in the instrument is less prone to clogging, and Lauren said it can even accommodate screening of whole cells.

 

Anything to accelerate the (sometimes painful) process of advancing fragments is always welcome. As Jason Pun (Monash University) noted, eight of nine targets screened in Martin Scanlon’s group started with fragments having affinities worse than 100 µM. Off-rate screening, an SPR technique we wrote about here, can rapidly identify more potent molecules from crude reaction mixtures, but data processing can be tedious. Jason described new software tools to automate this process, and hopefully he will publish the methodology and code. (An aside: over coffee Yun Shi of Griffith University noted that off-rate screening, or ORS, should really be called off-rate constant screening, which would give the more amusing acronym ORCS.)

 

Targets

Turning to targets, Ben Davis (Vernalis) described a collaboration with Servier to advance oncology target USP7 inhibitors from a literature fragment to a preclinical candidate. Crude reaction mixture screening was used extensively, not just by SPR but even in microsome stability studies. Unfortunately the project ended when on-target toxicology effects emerged, which were perversely more severe in higher animal species than they were in mice.

 

Yun Shi described finding tiny heterocyclic fragments that react with the NAD⁺ cofactor of neurodegenerative target SARM1 in situ to generate a potent inhibitor, as we wrote about here. Yun is using ¹⁹F NMR to follow the base-exchange reaction to identify inhibitors to other glycohydrolases too.

 

Deaths due to E. coli are – somewhat surprisingly – more common than those caused by any other pathogen, and Christina Spry described her work at the National Australian University to discover inhibitors of the essential dephosphocoenzyme A kinase (GPCK) enzyme, which catalyzes the final step in the synthesis of Coenzyme A (CoA). Fragment screening by DSF and NMR identified a weak (K_D=380 µM) binder, and fragment growing has led to a low nanomolar inhibitor that is selective against the human form of the enzyme.

 

Continuing the E. coli theme, several talks discussed efforts against the challenging bacterial virulence target DsbA, a twenty-year campaign in Martin Scanlon’s group at Monash as noted by Yildiz Tasdan. The enzyme has a shallow, hydrophobic active site, but the discovery of fragments binding to a cryptic site and crude-reaction screening by ORS (ORCS?) and affinity-selected mass spectrometry (ASMS) has finally led to molecules with dissociation constants around 1 µM.

 

Finally, in his closing keynote address Alvin Hung, who recently founded NeuroVanda, described a wide range of fragment success stories, many of them covered on Practical Fragments, against targets including pantothenate synthetase, GSK3β, PKC-ι, and MNK1/2. Although structural enablement helped in many cases, Alvin was not rigid about the need for atomic-level details: in response to the question whether he would advance a fragment in the absence of structure, he answered simply, “of course.” Perhaps it's time to redo my poll on this subject.

 

I’ll wrap up here, but if you missed this or earlier events this year there are still a couple more conferences in Boston, and 2025 is already starting to take shape.

LELP fragments reach their potential

In last month’s issue of Nature Reviews Drug Discovery, György Keserü of Gedeon Richter and Gergely Makara of Merck published a thought-provoking analysis of recent trends in lead discovery. Their results illustrate the potential of fragment-based methods, but also point out the still sizable gap from current practice.

The authors assembled a database of 335 hit-lead pairs derived from high-throughput screening (HTS) that were published between 2000 and 2007. They also assembled a database of 84 non-HTS hit-lead pairs published between 2000 and February 2008, consisting of fragment-based, virtual screening, natural product, and miscellaneous examples. They then compared properties – such as potency, molecular mass, logP, logS (a calculated measure of solubility), and ligand efficiency – of the initial hits with the resulting leads.

The results for HTS hits are not pretty: the lipophilicity as assessed by logP was considerably higher on average for HTS hits than for hits from any other methods, and this only increased as the hits were progressed to leads. The same goes for (in)solubility (as measured by logS). Even more alarming, the average properties of even the hits are worse than those of a collection of 541 approved drugs.

Fragment hits start out with the lowest lipophilicity and highest predicted solubility, but during hit-to-lead optimization these properties deteriorate to the point where they are similar on average to leads derived from HTS. Also surprisingly, the ligand efficiency of fragment hits and leads are, if anything, lower than their HTS counterparts, contrary to expectations. Even the average molecular weight of fragment-derived leads is not much lower than HTS-derived leads.

So what’s going wrong? The authors point out that, at most larger companies, fragment-based approaches are often only attempted after HTS has failed, suggesting that the targets tackled by fragment-based methods may be inherently more difficult. But they also suggest that in the early stages of hit-to-lead optimization the primary measure of success is how many compounds are delivered to lead optimization, which could encourage rapid hit expansion with simple chemistries to rapidly boost potency by adding grease, leading to more hydrophobic, less soluble molecules that will ultimately struggle in the clinic.

The authors suggest a new metric, ligand-efficiency-dependent lipophilicity, or LELP, to help avoid this trap:

LELP = (log P / LE)

Since a desirable logP range is between 0 and 3, and a desirable ligand efficiency is above 0.4, one should strive for LELP values between 0 and 7.5. There are already lots of metrics out there for evaluating molecules: see, for example, discussions of %LE, antibacterial efficiency, and fit quality (also here and here). Is a new one really necessary? Perhaps, if it gets people to focus on non-lipophilic means of increasing potency.

The authors end on a positive note for fragments:

Bearing in mind the sampling of chemical space, hit properties and synthetic accessibility, we consider that fragment hits are the optimum starting points for lead discovery and optimization.

There is, however, a burden on the team transforming a fragment hit into a viable lead: it is important to focus not merely on improving potency, but on maintaining as many of the fragment-like properties that make fragments attractive starting points in the first place. Although this goes without saying, analyses like this one suggest that it still needs to be said – and heard.

Tool Discovery Done Right

I have been on a bit of a jag lately pummeling academic "Drug discovery".  Dan recently hopped on, in his much more circumspect manner.  My big problem is calling something drug discovery that is not.  I may come off as harshly anti-academic; I am not.  I am pro-good science and even more for the right things in the proper place.  I think this one is one of those.  In this paper, the authors describe their use of fragments to find inhibitors of Mycobacteria.  Thiolactomycin is a natural product of beta-ketoacyl acpM synthase (KAS) from M. tuberculosis, but exhibits broad spectrum anti-biotic activity. While it has rather low potency against M. tuberculosis KAS (200uM, 0.26 LEAN), it has favorable physicochemical properties, low cytotoxicity, high bioavailability, and activity in animal infection models.  This makes it an excellent target for optimization, and considering the activity and lack of selectivity, lots of it!

The authors decided to use inter-ligand NOE to guide their efforts.  This is an approach that has been used previously, and in a case of cosmic togetherness, against a different M. tuberculosis target.  Amzingly, that work is NOT cited in this paper.   As pointed out, iLOE has some issues, aggregation and so on.  The authors do not seem worried by this and instead are interested in utilizing selective iLOE.  One thing the authors do NOT do is include any detergent in their samples.  They synthesized PK940 and used this for the NMR studies.  

They were unable to detect iLOEs with mixing times shorter than 500 ms (that's long!) using the standard 2D NOESY, so they used a selective experiment.  This allowed them to shorten the mixing time, increase sensitivity, and solve some other NMR based problems.  [N.B. They had to use 900 MHz to get good enough separation for some of the methyl resonances.  I guess you have to justify that humongous tin can somehow.]  This was able to generate the needed NOE data to put into a model to give them an idea on how to make compounds, suggesting elaboration of the thiolactone at the 3 or 4 position (A below).

They started to explore these hypotheses, and found that activities were all within 2x fold of the original.  So, unlike others, they decided this was flat SAR. Although they do make some conjectures that would seem to be easily testable (FOLLOW UP PAPER?). None of the compounds were significantly better than the lead and many in fact lost the "slow onset" binding that is considered important for good PD.  There is nothing really ground breaking here, but what I like about this paper is where it was published Journal of Biological Chemistry, not Journal of Medicinal Chemistry.  This is where these types of papers belong.  As I keep on saying, your computation is only as good as your follow up.  In this case, the computation is to support the NMR and not the other way around.

Deconstructing an HTS hit for GyrB inhibitors

COVID-19 is deservedly engaging most of our collective mindspace when it comes to infectious diseases. Unfortunately, plenty other threats are out there, including antibiotic-resistant bacteria. A paper recently published in ACS Omega by Fumihito Ushiyama and colleagues at Taisho reports progress in this area.

The researchers were specifically interested in the protein DNA Gyrase B (GyrB), which is essential for bacterial replication (see here for previous work on the same target). A high-throughput screen against the E. coli protein led to a few dozen hits that were validated using a variety of biophysical methods including SPR, isothermal titration calorimetry (ITC), and crystallography. Compound 1 binds in the ATP-binding site, which is also where the natural product inhibitor novobiocin binds. The latter molecule makes an interaction with an arginine residue in the protein, but introducing a carboxylic acid moiety onto compound 1 to make a similar interaction was not successful (compound 8e).

Taking a step back, the researchers stripped compound 1 down to the core fragment 2a, which makes both polar and hydrophobic interactions with GyrB. Unfortunately, this fragment was too weak to show any affinity by ITC, as were 120 related fragments.

Looking closer at the structure of compound 1 bound to the protein revealed a small unfilled hydrophobic pocket near the 2-quinolinone fragment. Making appropriately substituted fragments was “relatively complicated,” and most of them were inactive. However, compound 2d showed binding by ITC as well as excellent ligand efficiency. Growing from this fragment ultimately led to compound 13e, with low nanomolar affinity. In addition to binding, compound 13e is a potent inhibitor of GyrB and is selective against a panel of 96 human kinases. Unfortunately though, it displays only modest antibacterial activity, likely due to efflux.

Nonetheless, this is a nice example of thoughtful structure-based design. In particular, the dramatic boost in potency gained by filling a small pocket (nearly 400-fold from compound 8e to 13e) validates the willingness to explore difficult chemistry rather than sticking with available analogs. The paper ends by noting that optimization is continuing, and I wish them well. By my count only a single fragment-derived antibacterial agent has entered clinical development, and that program is no longer active. We could use more.

Fragments in the Clinic: AT13387

We recently discussed BACE, a target that has been tackled by FBDD due to its intractability to other methods. The subject of this post is quite the opposite: the anticancer target Hsp90 has proven very amenable to a variety of approaches, including fragment methods (see here and here); close to a dozen compounds targeting Hsp90 are in the clinic. Now Astex has detailed their work in this area with two back-to-back papers in a recent issue of J. Med. Chem. describing the discovery of AT13387.

The first paper, by Christopher Murray and colleagues, actually presents the discovery of two separate series of inhibitors. The researchers started with a library of about 1600 fragments and used NMR techniques (water LOGSY) to identify hits against Hsp90. Competition with ADP allowed them to identify molecules that bind to the nucleotide binding site. In all, 125 fragments were taken into crystallography, using both co-crystallography and soaking, resulting in 26 co-crystal structures. Four of these structures are described in some detail, with two leading to potent inhibitors. Throughout the process, isothermal titration calorimetry was used to measure dissociation constants.

In the first series, compound 1 was identified as a weak hit (see Figure 1). Virtual screening led to the purchase of a few variants, including compound 5, with roughly 100-fold improved affinity. Interestingly, the crystal structure of compound 1 bound to Hsp90 showed that the molecule was twisted around the bond connecting the two aromatic rings, despite this not being energetically optimal for the unbound molecule. By substituting the phenyl ring of compound 5 to stabilize this twisted conformation the researchers were able to improve the potency another 20-fold (compound 9), along with a boost in ligand efficiency. Further structural work suggested adding another chlorine to fill a lipophilic site as well as adding a solubilizing group, ultimately leading to compound 14, with low nanomolar binding affinity and low micromolar cell activity.

Figure 1

In the second series, compound 3 (which is actually itself a drug, ethamivan) had only modest ligand efficiency, but crystallography suggested that replacing the methoxy group with something slightly larger and more lipophilic would improve the interactions, a hypothesis borne out by the increased activity of compound 17 (see Figure 2). Increasing the lipophilicity of the amide side chain to take advantage of protein flexibility led to a further two orders of magnitude increase in potency (compound 28). Finally, the researchers were able to use the known binding mode of a natural product to add an additional hydroxyl group, leading to compound 31, with sub-nanomolar affinity (more than a million-fold more potent than the initial fragment!) and mid-nanomolar cell activity.

Figure 2

An impressive feature of both these examples is that, through the use of elegant medicinal chemistry, the researchers were able to improve ligand efficiency throughout the course of affinity improvement. Of course, it helps that they were working on a crystallographically friendly target for which several other groups had published extensive SAR, but these are nonetheless beautiful case studies. As the researchers point out, “in terms of the efficiency of the added groups, the two fragment to lead campaigns… are among the most efficient ever reported.”

But the story doesn’t end there. The second paper, by Andrew Woodhead and colleagues, describes the further optimization of compound 31 to the clinical candidate AT13387. Despite its impressive biochemical and cell potency, compound 31 had only modest activity in a mouse xenograft model, as well as a short plasma half-life. Not surprisingly the hydroxyl groups were found to be points of metabolism, but initial efforts at capping these or changing their electronics either proved detrimental to activity or did not improve the pharmacokinetics. This led to a medicinal chemistry focus on the isoindoline portion of the molecule: a number of positively charged moieties were added at various positions to try to change the overall properties of the molecule. Several substituents were tolerated, and seven related molecules were taken into preclinical candidate selection to look for optimal in vivo properties, solubility, and selectivity against P450 and hERG. AT13387 (see Figure 2) was chosen as the molecule having the best overall profile and entered human clinical trials for solid tumors.

This second paper is a valuable companion to the first: it is particularly notable that, on the simple measures of biochemical and cell potency, AT13387 is no better than compound 31. This emphasizes yet again that affinity is only the first step in drug discovery – it’s a long road from a good lead to the clinic, and an even longer road from there to a marketed drug.

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.