Fragments versus CDK4 and CDK6

The cyclin-dependent kinases (CDKs) were some of the earliest protein kinases targeted for drug discovery. They are important for cell-cycle progression, and thus cancer. However, selectivity among the multiple CDK family members has been challenging. In the June issue of ACS Med. Chem. Lett., researchers from Astex and Novartis describe the optimization of a fragment to a selective inhibitor of CDK4 and CDK6.

Astex has been working on CDKs for some time; one of Practical Fragments’ first posts described AT7519, an inhibitor of CDK1, 2, 4, 5, and 9 that is in multiple phase 2 clinical trials. In the new paper, the researchers start with a fairly potent CDK6 hit (fragment A). Crystallography suggested that replacing the pyrrole with a pyridine would provide better vectors from which to grow the fragment, leading to Compound B, which was still active. Growing in two directions then led to Compound 1, and extensive structure-based design led ultimately to Compound 6, which is selective for CDK4 and CDK6 over CDK1 and CDK2. In a panel of 35 additional off-target kinases, the compound displayed IC₅₀ values of 5 micromolar or worse. Compound 6 also showed target modulation in mice and tumor xenograft activity, albeit at fairly high doses.

The authors note that selectivity was a key goal, and that in the course of optimization they were willing to sacrifice potency against their desired targets in order to avoid hitting CDK1 and CDK2. The success of this strategy illustrates again the importance of maximizing ligand efficiency at the outset, as drops in LE can then be used to “pay” for other desirable properties. (Note also that the drop in LLE_AT is not quite as severe as the drop in LE.)

Some enthusiasts have argued that fragments provide a more efficient path to the clinic, and this can certainly be the case, as illustrated by the rapid progress of vemurafenib from fragment to drug. However, the current paper illustrates that advancing fragments can still require considerable resources: with 36 authors on two continents, it is clear that this project was not a walk in the park. It is, however, another illustration of starting with a fragment to develop a useful molecule.

Are enthalpic binders more selective than entropic binders?

Thermodynamics is one of those abstract subjects that can have surprising real-world implications. The two components of free energy, enthalpy and entropy, are simplistically associated in drug discovery with polar interactions for the former and hydrophobic interactions for the later. Some researchers have suggested that enthalpically-driven binders are better starting points for optimization, and that best-in-class drugs rely more on enthalpy than entropy. In a recent paper in Drug Discovery Today, Yuko Kawasaki and Ernesto Freire at Johns Hopkins University suggest that enthalpic binders may also be more selective.

Medicinal chemists apply two general strategies to improve selectivity: increase the affinity of a compound for its target more than for off-targets, or decrease the affinity of a compound for off-targets. Kawasaki and Freire argue that the first is more likely to result from entropic interactions, while the second is more likely to result from enthalpic interactions. This is because nonpolar (entropic) interactions are often tolerant of mismatches; a hydrophobic substituent might improve the affinity of your ligand for its target, but, unless it causes a severe steric clash, it may also improve activity for off-targets – though hopefully less. Indeed, recent findings suggest that more lipophilic molecules tend to be more promiscuous than similarly-sized but less lipophlic molecules. On the other hand, due to the highly directional nature of polar interactions, a mismatched polar (enthalpic) interaction in an off-target is likely to be highly detrimental to binding.

The researchers consider two case studies involving HIV-1 protease inhibitors. In one example, adding two (non-polar) methyl groups improves the affinity of the inhibitor for its target as well as for two off-targets, though it improves the potency towards HIV-1 protease more, thus improving selectivity.

In the second case, a non-polar thioether is replaced with a polar sulfone. This slightly decreases the overall binding affinity for HIV-1 protease, but has a much larger negative effect on two off-targets, resulting in greater selectivity. In this case, the enthalpy of binding for HIV-1 protease is considerably improved, though the effect is compensated for by unfavorable changes in entropy. As the authors note, “even if a strong hydrogen bond does not contribute to affinity, it might contribute significantly to selectivity.”

Ideally you would want to use both strategies (improving affinity for your target and decreasing affinity for off-targets). However, since you probably don’t know all your off-targets, focusing on enthalpic binders may be the way to go, as mismatched polar interactions are likely to exclude lots of unknown off-targets.

Of course, two examples may not make a trend, but they do make a testable hypothesis. For example, there is a veritable plethora of kinase inhibitors with known specificity profiles: it would be interesting to correlate these with their thermodynamic profiles. But at any rate, this is yet another reason to hold down the hydrophobicity of your compounds.

Privilege or selectivity?

Fragment selectivity is something we’ve covered before (see here and here). Sarah Barelier and Isabelle Krimm at the Université de Lyon have published on this topic (see here), and in a recent issue of Current Opinion in Chemical Biology they review the subject and its implications.

The authors document what many investigators have independently observed: some fragments, as expected theoretically, are less selective than larger molecules, but other fragments are quite selective.

They also note that fragments that bind to any protein tend to be slightly more lipophilic than fragments that don’t bind to any target proteins, suggesting that hydrophobic interactions are important:

Hydrophobic interactions play a major role in protein–ligand interactions and are known to be non-directional, thus allowing binding to a multitude of pockets in different conformations. By contrast, hydrogen bonds were shown to confer specificity but do not always add much binding free energy. This is due to the cost of desolvating both the donor and acceptor of the hydrogen bond, which can nearly equal the benefit of the hydrogen bond formation. Therefore, if the hydrogen bond acceptors or donors are not satisfied in the complex, it is likely that more hydrophobic fragments will be preferred.

This observation—lipophilicity for binding energy, hydrogen bonds for specificity—is consistent with the recent publication from Mike Hann and Andrew Leach, which finds that promiscuity increases with increasing lipophilicity.

One figure in the Barelier and Krimm paper shows 30 fragment-like “privileged scaffolds” that should bind to multiple proteins. What struck me is these molecules’ overwhelmingly planar character: more than half are completely aromatic (such as quinoline and indole), and only one is completely aliphatic. Barelier and Krimm note that:

The low specificity of these molecules is probably owing to their rigid and aromatic structures, well-adapted to protein hydrophobic pockets where π-stacking with phenylalanine and tyrosine are commonly observed.

This reminds me of Tony Giannetti’s talk at the FBLD San Diego meeting earlier this month, where he also noted that fragment hits tend to be relatively flat. Of course, given the negative correlation between aromaticity and good pharmaceutical properties, just because aromatics are frequent hits doesn’t mean they are necessarily the best hits – they may be tricks rather than treats. All of which comes back to a key question for library design: do you focus on the flat “privileged” scaffolds that will likely have high hit rates in your assay but may have baggage, or on the more three-dimensional compounds that may have lower hit rates but may ultimately be more developable?

Fragment specificity

A frequent topic in fragment roundtable discussions concerns specificity: do fragments hit lots of targets, or just a few? Isabelle Krimm and colleagues at the Université de Lyon in France studied this question experimentally and report their results in a recent issue of J. Med. Chem. The paper provides data for the ongoing debate of whether and how much specificity a fragment should exhibit before being pursued for further lead development.

The researchers assembled a diverse set of 150 fragments and used NMR techniques to determine whether they bind to five different proteins. Three of the proteins, Bcl-xL, Bcl-w, and Mcl-1 are related members of the Bcl-2 family of antiapoptotic proteins, and at least the first of these has been successfully targeted using fragment-based methods. The fourth protein, PRDX5, has proven to be much less yielding to inhibitor discovery, while the fifth, human serum albumin (HSA), binds a wide variety of small molecules.

After applying 1D-NMR techniques (WaterLOGSY and STD) to all of their fragments against each of the five proteins, the researchers used more rigorous but less sensitive 2D-NMR (HSQC) to determine the binding sites of the hits. (This later study revealed, in agreement with previous results from the same lab, that the fragments all bind in the “hot spots” or active sites of the proteins.)

More than two-thirds of the fragments bound to at least one protein, a rather high hit rate. However, the hit rates for each protein varied considerably, with only 7 hits for PRDX5 and 72 for HSA (with a close second of 71 for Bcl-xL). Within the Bcl-2 family there was little specificity observed: Mcl-1, with 29 hits, shared all but one hit with either Bcl-xL or Bcl-2 or both; such non-specificity among related proteins has been discussed previously. In the case of HSA and Bcl-xL, although both proteins had similar numbers of hits, just over half of these were in common, demonstrating that fragment specificity is not difficult even with small-molecule sponges such as HSA. That said, many fragments were remarkably nonspecific, with 22 hitting four of the 5 proteins. Amazingly, all 7 of the hits against PRDX5 also hit all four other proteins.

The physicochemical properties of the fragments that hit one or more proteins were compared with those of the library as a whole, and although most of the parameters were similar, the ClogP values (a measure of hydrophobicity) were considerably higher for hits, and highest of all for the non-specific hits.

These findings are more evidence that, as predicted almost a decade ago, fragments can bind to more proteins than can larger, more complex molecules. The follow-up question, how much does this matter, is still up for debate. There are plenty of examples of developing specific inhibitors from non-specific starting points during the course of fragment optimization. But how non-specific is too non-specific? Would you feel comfortable pursuing any of the fragments that hit all of the proteins?