Fragments and nanodiscs: beware nonspecific binding

Membrane proteins make up roughly a quarter of human proteins, including many important drug targets. Biophysical methods for fragment screening typically require pure, isolated proteins, and removing membrane proteins from their native environment is not always possible. One solution has been to create nanodiscs which, as we described previously, are isolated little membranes containing the protein of interest. These nanodiscs can be immobilized to the sensor chips used for surface plasmon resonance (SPR), one of the most popular fragment finding methods. But in a recent open-access Chem. Biol. Drug Des. paper, Marcellus Ubbink and collaborators at Leiden University and ZoBio show that the precise composition of the nanodiscs can have a profound effect on the results.

 

The researchers chose cytochrome P450 3A4 (CYP3A4) as a model membrane protein. This enzyme metabolizes a large fraction of drugs and has a capacious active site able to bind a wide variety of substrates. Four different lipids were chosen for the nanodisc, all of which contained phosphatidylcholine headgroups and differing hydrophobic tails: POPC, DPPC, DMPC, and DPhPC.

 

Nanodiscs were prepared either with or without CYP3A4 and immobilized to SPR chips. Unlike some membrane proteins, it is possible to isolate and immobilize CYP3A4 in the absence of membranes, though the protein forms physiologically less relevant oligomers.

 

Next, the researchers examined 13 known (non-fragment) CYP3A4 ligands. Unfortunately, most of these bound to the empty nanodiscs, and in some cases more than ten ligands bound to a single empty nanodisc. This nonspecific binding correlated with lipophilicity, with only the three least lipophilic molecules showing no binding to empty nanodiscs. One of these was the antifungal drug fluconazole, with a clogP = 0.4. Happily though SPR studies using either free or nanodisc-bound CYP3A4 yielded dissociation constants of 10-20 µM, consistent with published values.

 

Thus encouraged, the researchers screened a diverse set of 140 fragments at 250 or 500 µM against empty and CYP3A4-loaded nanodiscs using SPR. Just as with the larger molecules, there was a good correlation between cLogP and nonspecific binding to empty nanodiscs. Fragments that bound to one type of nanodisc (POPC, for example) also tended to bind nonspecifically to other types of nanodiscs (DPPC, DMPC, and DPhPC). Fewer fragments bound nonspecifically to DMPC nanodiscs than to the others, suggesting this may be the best lipid to use.

 

Fragment hits were defined as those binding to CYP3A4-containing nanodiscs more than they bound empty nanodiscs (or, for isolated CYP3A4, the unmodified SPR chip). Hit rates varied dramatically, from 9 of 140 fragments tested against CYP3A4 in POPC nanodiscs to 33 of 140 tested against CYP3A4 in DMPC nanodiscs. There were also 33 hits against free CYP3A4, 11 of which were unique. However, all 11 of these are somewhat lipophilic (average cLogP ~2.3) and most also bound significantly to empty nanodiscs. The researchers suggest that these bind “aspecifically” to CYP3A4.

 

A Venn diagram of all the hits shows only two that bind to free CYP3A4 as well as all nanodiscs containing CYP3A4, and the researchers highlight these two as the most promising. Unfortunately these are not further characterized.

 

Near the beginning of the paper, the researchers note that very few fragment screens have been conducted against membrane proteins incorporated into nanodiscs. This analysis suggests why this is so. If you use nanodiscs, make sure to consider different types of ligands. And look carefully for nonspecific binding.

Better properties from fragments: c-Abl kinase activators

Last year we described the discovery of asciminib, an allosteric inhibitor of the kinase BCR-Abl that binds in the enzyme’s myristoyl-binding pocket. As we also highlighted nearly a decade ago, molecules that bind in this pocket can either inhibit or activate the enzyme. Although inhibitors have the most obvious therapeutic potential as anti-cancer agents, activators of the ubiquitously expressed c-Abl protein could potentially treat chemotherapy-induced neutropenia. In a recent J. Med. Chem. paper, Sophie Bertrand and coworkers at GlaxoSmithKline describe their efforts in this area.

The researchers started with a high-throughput screen of 1.3 million compounds. Among the hits was fragment-sized compound 2, which showed good binding and activation in biochemical assays but only modest activity in cells. Building off the left side of the molecule improved biochemical potency, but cell activity still lagged. SAR studies on the dichlorophenyl moiety suggested that this hydrophobic group was probably optimal, and a crystal structure of an analog bound to the enzyme confirmed this. Replacing the central thiazole with other aromatic rings also did little to improve cell activity.

The researchers acknowledge “that the chemistry strategy was largely pursuing compounds with rather poor physical properties,” notably low solubility, high lipophilicity, and high aromatic character. As co-author Robert Young has noted previously, physical properties matter. Happily, a fragment screen identified compound 28.

Adding the acetyl group from the HTS hit generated compound 29, with improved activity compared to the fragment. Moreover, this molecule had better solubility and permeability compared to the more lipohilic, thiazole-containing compound 2. Compound 29 also showed significantly improved activation of c-Abl in a cellular assay. Crystallography revealed that it bound in a similar fashion as compound 2, but with a twisted, more “three-dimensional” shape.

Further optimization, in part informed by previous work done on the thiazole series, ultimately led to compound 52, the most active compound synthesized. Another molecule in the pyrazoline series showed good pharmacokinetic properties in mice. Unfortunately, in vivo efficacy studies had to be halted early due to unexpected (and not clearly understood) toxicity.

This paper nicely illustrates several points. First, the power of fragment-assisted drug discovery, in which information from both HTS and FBLD is combined for lead optimization. Second, the inherently fuzzy line between FBLD and other discovery approaches: had compound 28 been tested in the HTS collection, it likely would have been a hit. Third, the importance of physicochemical properties. And finally, the inadequacy of potency and physicochemical properties alone to produce a developable compound. You can optimize your molecule to the best of your ability but still be sideswiped by nasty surprises such as toxicity. It is helpful to be clever in drug discovery, but you need to be lucky too.

Fragments in the clinic: ASTX660

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

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

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

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

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

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

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

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

Fragments vs histone KDM4 lysine demethylases: Celgene’s story

Last year we highlighted a paper from academia in which modeling was used to discover potent inhibitors of the lysine demethylase KDM4C, a potential anti-cancer target. In a recent paper in ACS Med. Chem. Lett., Michael Wallace and collaborators at Celgene, the European Institute of Oncology, and the University of Chicago report a chemical probe for KDM4 family members.

The researchers started with a literature screen of fragments known to bind to KDM4, leading them to compound 1, which previous work had shown binds to the catalytic iron through the pyridine ring nitrogen. Researchers from GlaxoSmithKline had also reported that growing off compound 1 could lead to more potent compounds, a strategy that proved successful here in the case of chiral compound 2a, which improved affinity by more than two orders of magnitude. Interestingly, the enantiomer had dramatically lower activity.

Armed with this information but no crystal structure, modeling suggested that further growing off the tetrahydronaphthalene would be productive, which turned out to be the case for compound 3, with similar affinity but improved activity in an antiproliferative cell assay. Further experiments showed that the compound increased levels of trimethyl-lysine on lysines 9 and 36 of histone 3, known substrates of KDM4 family members.

A crystal structure of compound 3 bound to KDM4A, which is closely related to KDM4C, suggested further room to grow. Compound 3 contains a carboxylic acid and has a low cLogD, traits that tend to reduce cell permeability. The researchers thus focused on increasing the lipophilicity of the molecules, leading to QC6352. Despite the fact that this molecule is less potent in the enzymatic assay, it has significantly improved cellular potency. It also has reasonable pharmacokinetics and oral bioavailability, and showed activity in a mouse xenograft model. QC6352 hits KDM4A, 4B, 4C, and 4D, but is quite selective against most other KDMs.

This paper illustrates three important points. First, as discussed previously, you don’t need a novel fragment to get to novel leads – you just need creative scientists. Indeed, the increasing number of fragment hits reported for various targets provides a wealth of starting points even for organizations that don’t do in-house fragment screening. Second, you don’t necessarily need a crystal structure as long as you have good modelers. And finally, while excess lipophilicity is rightly avoided, it is important to remember that compounds can also be too polar. As Oscar Wilde noted, “everything in moderation, including moderation.”

Rule of five versus rule of three

Metrics (such as ligand efficiency) and rules (such as the rule of three) seem to be some of the more controversial topics around here. If you aren’t experiencing metric-fatigue, it’s worth checking out a recent (and free!) “Ask the Experts” feature at Future Med. Chem., in which four prominent scientists weigh in on the utility of the rules of five and three.

Monash University’s Jonathan Baell (of PAINS fame) notes that, as of early 2013, the original 1997 Lipinski et. al. rule of five paper (and the 2001 reprint) had been cited more than 4600 times! Baell holds that, of the properties covered by the rules – molecular weight, lipophilicity, number of hydrogen-bond donors (HBD), and number of hydrogen-bond acceptors (HBA) – the property lipophilicity is probably the most important. Although he agrees that rules can be too strictly applied, he also asks:

What sum value is represented by the dead-end investment that the world never saw because of application of a Ro5 mentality?

I think this is a good, often-overlooked point. It is easy to find examples of drugs that violate the rule of five or programs that were killed by rule-bound managers with limited vision, but, as GlaxoSmithKline’s Paul Leeson says, “there is massive unexplored chemical space within the Ro5, which is available to innovative chemists.” Why not put much of the focus here?

Of course, readers of Practical Fragments are probably thinking as much about the rule of three as the rule of five, and one of the main criticisms of that rule, particularly by Pete Kenny, has been the fact that it is not clear how to define hydrogen-bond acceptors: do you count all nitrogen and oxygen atoms, including for example an amide –NH? I think the common-sense answer would be no, and Miles Congreve, the first author on the original rule of three paper, seems to agree. He also notes that the number of hydrogen bond acceptors seems to be less important in general than the number of hydrogen bond donors, which is negatively correlated with solubility, permeability, and bioavailability.

Given last year’s poll on the maximum size of fragments people allow in their libraries, it looks like most people are already capping molecular weight well below 300 Da, which skews the other parameters toward rule of three space. That said, Congreve does warn that commercial fragment libraries “contain too many compounds that are close to 300 Da, rather than containing a distribution of compounds in the range of 100 – 300 Da,” a statement borne out by by Chris Swain’s analyses. Of course, the larger you get, the more possibilities there are, and the optimal property distribution of a fragment library is still a matter of debate.

Ultimately I think many people will agree with Leeson, who says that “there are probably sufficient metrics in the literature today,” and with Celerino Abad-Zapatero, who notes that “additional rules will not be the answer in the long run.” On this note I promise no more posts on metrics or rules – for at least a month!

Purifying hydrophilic fragments

Lipophilicity is a topic that comes up periodically. Lipophilic molecules are increasingly viewed as problematic from a drug development standpoint. Even if the correlation studies indicting lipophilicity are not as strong as they appear, at the end of the day we would prefer most of our drugs to be nicely water soluble.

That said, many of the molecules we make are on the greasy side. GDB-17, Jean-Louis Reymond’s recent computational enumeration of small molecules with 17 or fewer heavy atoms, reveals that most potential molecules tend to be much more polar than similarly sized compounds that have actually been made. One likely reason for this is that purifying highly water-soluble molecules is difficult; it’s hard to wash away inorganic reagents, and they often stick to the normal silica gel that chemists use to purify conventional molecules. Reverse-phase HPLC is useful, but can be tedious and low throughput.

In a recent issue of Drug Discovery Today, Andrew Hobbs and Robert Young of GlaxoSmithKline provide practical tips on using reverse-phase flash chromatography as an alternative to HPLC. They report working at scales from milligrams to tens of grams and are able to separate some very polar molecules. There’s a lot of good stuff in this paper on choosing columns, solvents, and loading techniques. A lot of these details get pretty nuanced, so it’s nice to have them in one place. If you’re trying to isolate hydrophilic molecules, definitely check it out.

LLE vs LELP

Besides ligand efficiency (LE), a slew of other metrics has been proposed to help evaluate what compounds to take forward in drug discovery. As seen in our poll and a recent round-table discussion, lipophilic ligand efficiency (LLE) is quite popular:

LLE = pIC₅₀ (or pK_i) – ClogP (or logD)

However, because this metric is not size-adjusted, it is not particularly useful for evaluating fragments, which often have low potency. In contrast, the metric LELP accounts for size:

LELP = logP / LE (where LE = ligand efficiency)

In a recent issue of J. Med. Chem., György Keserű and colleagues evaluate how these two metrics compare in a variety of settings.

The authors examine eight different compound sets: fragment hits and derived leads, HTS hits and derived leads, leads that subsequently became drugs (ie, “successful leads”), development candidates, compounds that entered phase II trials, and drugs on the market. Not surprisingly, drugs and phase II compounds had better LLE and LELP scores than other molecules. Also not surprisingly, fragments scored misleadingly poorly on the basis of LLE but well on the basis of LELP. What is perhaps unexpected, though, is that LELP was better at identifying successful leads than was LLE. Moreover, when compounds were evaluated for pharmacokinetic and safety parameters, LELP was more effective at predicting problems than was LLE. The authors state:

In summary, evaluation of pharmacokinetic and safety parameters revealed that LELP has benefits over LLE, as compounds with acceptable in vitro ADMET profiles are discriminated from compounds with significant liabilities.

Despite these potential advantages, LELP doesn’t seem to be widely used, perhaps because it is less intuitive than some of the other metrics. Indeed, it would be interesting to see these studies repeated using LLE_AT, which also takes lipophilicity into account but has the same scale as LE.

Pushing the Rule of 3

The Rule of 3 (Ro3) is commonly used to design fragment libraries. First published as a brief 450-word (shorter than this post!) “update” in the discussion forum of Drug Discovery Today in 2003 by researchers at Astex, it has become the fragment equivalent of Chris Lipinski’s famous Rule of 5. Like that rule, it has its critics, notably our friends at FBDD and Molecular Design. A key point of contention is whether the Ro3 is too restrictive. A new paper in J. Med. Chem. from Gerhard Klebe’s group at Philipps University Marburg addresses this question.

The definition of the Rule of 3 provided by Astex is as follows:

The study indicated that such hits seem to obey, on average, a ‘Rule of Three’, in which molecular weight is <300, the number of hydrogen bond donors is ≤3, the number of hydrogen bond acceptors is ≤3 and ClogP is ≤3. In addition, the results suggested NROT (≤3) and PSA (≤60) might also be useful criteria for fragment selection.

One of the criticisms leveled at the Ro3 is that it is vague in terms of what constitutes a hydrogen bond acceptor. For example, does the nitrogen in an amide count? What about the nitrogen in an indolizine? Presumably for simplicity Lipinski assumed that any nitrogen or oxygen atom would count as a hydrogen bond acceptor. At the risk of engaging in exegesis, I propose that only oxygen or nitrogen atoms most medicinal chemists would consider as acceptors should be counted as acceptors, and that the limits on the number of rotatable bonds (NROT) and polar surface area (PSA) are optional.

In the recent paper, which is also discussed at FBDD and Molecular Design, Klebe and colleagues assembled a library of 364 fragments in which the average properties of the fragments were within Ro3 guidelines (with the exception of “Lipinski acceptors,” which would include the nitrogen of a tertiary amide), but there were some outliers. They then performed a fluorescence-based competition screen against the model protein endothiapepsin, resulting in 55 fragments that inhibited at least 40% at 0.5 or 1 mM concentration. These fragments were taken into crystallography trials, resulting in 11 structures. The paper presents lots of nice analysis of how these fragments bind to the protein. It also notes that:

Only 4 of the 11 fragments are consistent with the rule of 3. Restriction to this rule would have limited the fragment hits to a strongly reduced variety of chemotypes.

This may be an overstatement. Looking at the fragment hits more closely, all of them have molecular weights less than 300, and only one has ClogP > 3. Personally, given the problems of molecular obesity and the dangers of lipophilicity, I’d say that these aspects of the Ro3 are the most important, and find it notable that the hits were so compliant given that the library did contain larger, more lipophlic members.

All 11 of the crystallographically characterized fragments also have 3 or fewer hydrogen bond donors and TPSA < 60 Ų. Only two of the fragments have more than 3 rotatable bonds, but where the majority of the fragments fail to pass Ro3 is in the number of “Lipinski acceptors,” where 6 of the 11 have > 3. However, if you count hydrogen bond acceptors more judiciously (ie, compound 291 would have 3 acceptors rather than 4, since the aniline nitrogen would not be counted), only 1 of the 11 fragments has more than 3 acceptors.

Like most rules, the Rule of 3 should never be treated as a strait-jacket. That said, given the number of possible small fragment-sized molecules, and the necessarily limited size of any fragment collection, there seems to be plenty of room within the Rule of 3 for attractive chemical diversity.

A decade of molecular complexity

Molecular complexity is one of the key reasons why fragment-based lead discovery should work. As described in 2001 by Mike Hann and colleagues at GlaxoSmithKline, the idea is that very small, simple molecules are likely to be able to bind to many different sites on many different proteins; think of the water molecule as being an extreme example of this. As molecules become larger and more complex, they are less likely to bind to any given site on a protein, though if they are complementary to a site the potency will be greater. Similarly, more complex molecules are more likely to have a single binding mode than smaller, less-decorated molecules, which could assume multiple orientations at a single site. These intuitive ideas were supported by a simple computational model that suggested that there is an optimum complexity where molecules would be simple enough that they would bind to several different targets (and thus be useful in a screening collection) while still being complex enough to bind in single, defined orientations with sufficient potency to permit detection. Mike Hann and Andrew Leach now have a new paper in Current Opinion in Chemical Biology that analyzes how this idea has weathered the past decade.

A central tenet of the molecular complexity model is that more complex molecules should be less promiscuous (bind to fewer protein targets) than less complex molecules. Although defining complexity is itself complex, the authors summarize a number of studies that examine promiscuity as a function of various molecular properties that could be used as proxies for complexity. Interestingly, many of these studies find that as molecular weight or – especially – lipophilicity increases, promiscuity actually increases, an apparent contradiction of the complexity model. Indeed, Hann and Leach present internal data showing that, for a given molecular weight, promiscuity increases with increasing lipophilicity.

The authors consider several explanations for this, such as the notion that larger, lipophilic molecules may not need to be perfectly complementary to a protein: one portion could bind, while the rest of the molecule remains unbound. One explanation that the authors don’t address but that could account for much of the discrepancy is the validity of the measurements from the studies surveyed. Practical Fragments has previously discussed the issue of aggregation artifacts, which can occur even at nanomolar concentration – well below the 10 micromolar cutoff used in many of the cited studies. Indeed, Brian Shoichet has commented that the majority of hits from HTS screens could be artifacts, and an alarming proportion of "active molecules" in published work are also bogus. Thus, the apparent promiscuity of more lipophilic compounds may reflect merely assay artifacts, not true binding.

In other words, I propose at least two kinds of promiscuity. “Legitimately promiscuous” compounds actually bind to multiple proteins in a one-to-one defined fashion. Perhaps these are rare, in line with the complexity model. “Apparently promiscuous” compounds simply interfere with the assay, whether through aggregation, fluorescence artifacts, or other PAINful mechanisms. Given how many discovery programs get side-tracked by these phenomena, these compounds are likely to vastly outnumber legitimately promiscuous molecules, thus distorting the results of data-mining exercises.

There is plenty more in the paper than can be summarized here, and if this piques your interest Mike Hann will be discussing both molecular complexity as well as molecular obesity at the SLAS webinar series starting next month.