Fragment linking: oil and water do mix

Fragment linking is one of the most seductive forms of fragment-based lead discovery: take two low-affinity binders, link them together, and get a huge boost in potency. But what’s appealing in theory is difficult in practice: the linked molecule rarely binds more tightly than the product of the fragment affinities, and sometimes there is not even an improvement over the starting fragments. In a recent paper in Molecular Informatics, Mark Whittaker and colleagues at Evotec suggest a strategy to maximize the chance of success.

The researchers start by briefly reviewing nine published examples of fragment linking where affinities for both fragments as well the linked molecule are provided (some of these have been discussed previously here, here, and here). Of these, only three examples showed clear superadditivity (in which the linked molecule has a significantly higher affinity than the product of the affinities of the individual fragments), and two of these examples are rigged systems in which a molecule already known for its potency (such as biotin) is dissected into fragments. The challenges of linking are succinctly summarized:

The keys to achieving superadditivity upon linking are to maintain the binding modes of the parent fragments, not introduce both entropy and solvation penalties while designing the linker, and also make any interactions with the intervening protein surface that need to be made.

Also, of course, the resulting molecule needs to be synthetically accessible. Having a certain amount of flexibility in the linker can be useful, as this will allow the fragments some room to shift around, but too much flexibility introduces an entropic cost that defeats the purpose of linking in the first place. Software tools such as those by BioSolveIT can help design the linker, but what if some fragments themselves are inherently better suited for linking?

All three of the examples that show superadditivity start with one fragment that is highly polar and makes hydrogen bonds or metal-mediated bonds with the protein. The researchers suggest that such fragments are likely to pay a heavy thermodynamic penalty when they are desolvated, and that this cost can be reduced by linking them to a hydrophobic fragment. Thus, to maximize your chances of successful linking, the authors suggest you should choose

a fragment pair that consists of one fragment that binds by strong H-bonds (or non-classical equivalents) and a second fragment that is more tolerant of changes in binding mode (hydrophobic or vdW binders).

This is an interesting proposal, though because there are so few examples it is hard to assess. Indeed, the only other case of clear superadditivity I found involves dimerizing a fragment that is reasonably hydrophobic (ClogP = 2.4), albeit negatively charged. Hopefully we’ll see more examples in the coming years, but in the meantime, linking a water-loving fragment to an oily one is worth a shot.

Hsp90 and fragment linking

There has been no shortage of fragment-based approaches directed toward the anti-cancer target Hsp90, most of which have relied on growing fragments (see here for some impressive recent examples). Researchers from Abbott published a report providing a couple examples of linking fragments against this target a few years back, but in those cases the ligand efficiencies of the linked molecules were dramatically lower than those of the initial fragments. In a recent paper in ChemMedChem, researchers from Evotec describe an example that maintains the ligand efficiency.

The research group had previously conducted a fragment screen against Hsp90, resulting in a number of hits. In the new paper, fragment hits 1 and 2 (see figure) were both found to have fairly low affinities, but were characterized crystallographically. Interestingly, fragment 2 could adopt at least two very different conformations, depending on whether it was co-crystallized in the presence of fragment 1. In the ternary structure, the two fragments come within about 3 Å of each other, and molecular modeling suggested that four atoms should be able to link them.

Gratifyingly, when such a compound was made and tested, it inhibited the enzyme several hundred-fold more tightly than either of the initial fragments. The crystal structure revealed that the compound binds similarly to the ternary structure of Hsp90 and fragments 1 and 2.

The authors note that “the binding free energy of the linked fragment 3c was found to be exactly the sum of those of the original two fragments.” Of course, this is still a long way from an ideal linking situation: as noted earlier this year a good linker should lead to super-additivity (an improvement of ligand efficiency), not just additivity (maintenance of ligand efficiency). Nonetheless, this example is still better than many attempts at linking, which often are less than additive.

ACS Spring Meeting 2010

The spring national meeting of the American Chemical Society has just concluded in (uncharacteristically sunny) San Francisco. The main fragment event was a full day session organized by Rachelle Bienstock of the NIH. The theme was “Fragment based drug discovery: success stories due to novel computational methods applications.” Rachelle is planning on getting some of the speakers to write chapters for a book, so I won’t do more than give a very brief overview here.

The session was very multinational, with speakers from France, Germany, Russia, and the UK, in addition to the US, and a good mix of companies and academics. On the computational corporate side John MacCuish from Mesa Analytics described the molecular shape fingerprints approach, Carsten Detering of BioSolveIT provided several examples of applying his company’s methods for fragment linking and scaffold hopping, and Francois Delfaud of MEDIT described mining the pdb for protein-fragment interactions and applying this to Eg5 inhibitors. On the computational academic side, Tobias Lippert of the Center for Bioinformatics in Hamburg discussed the Qsearch program, Vladimer Poroikov of the Institute of Biomedical Chemistry in Moscow discussed PASS, which relies on a large training set to predict actives and inactives, and Dima Kozakov of Boston University presented the FTMap approach for predicting fragment-binding pockets in protein-protein interactions.

Moving away from the purely computational, Yongjin Xu of Novartis described the application of virtual fragment screening to identify p38 and BRaf inhibitors, Vicki Nienaber of Zenobia described iterative fragment screening to identify potent and selective LRRK2 inhibitors, and I presented Carmot’s Chemotype Evolution approach. Finally, GPCRs appear to be increasingly amenable to FBLD; Richard Law of Evotec presented a number of applications of computational methods to various programs including histamine receptors, while Miles Congreve of Heptares presented their StaR Technology for generating stabilized GPCRs suitable for SPR, NMR, and crystallography and discussed applications to the adenosine A2A receptor and the beta-1 adrenoreceptor. In the later case, the researchers were able to obtain 9 co-crystal structures and found that agonists and antagonists bound somewhat differently.

There were also a few other relevant posters and talks throughout the conference. For example, I learned that Locus Pharmaceuticals has transformed itself into Ansaris; Fouzia Machrouhi presented a poster on developing nanomolar inhibitors of the protein kinase AMPK.

Finally, Andrew Woodhead presented an update on Astex’s CDK2 program. One of the earliest posts on Practical Fragments described Astex’s fragment-based discovery of AT7519, which is in clinical trials for cancer. However, with an oral bioavailability of less than 1%, this compound is administered intravenously. Extensive medicinal chemistry ultimately revealed that a relatively minor change – capping the secondary amine with a methyl sulfonamide – led to a molecule with dramatically improved oral bioavailabilty. This molecule, AT9311, also retains good activity in mouse xenograft models. This is a useful reminder that fragment-based methods are not a replacement for solid (and inevitably subsequent) medicinal chemistry.

Molecular Medicine Tri-Conference 2010

The first event on our 2010 calendar, the Molecular Medicine Tri-Conference 2010, was held in San Francisco earlier this week. There were fragment talks and a roundtable, as well as a number of vendors selling fragment libraries – we’ve recently noted how rapidly this area has expanded.

Michael Hennig presented a nice overview of the history and development of fragment-screening at F. Hoffmann-La Roche (Basel). Work done there back in the late 1990s relied on NMR and crystallographic screening of a library of 300 fragments, described in the seminal “needle-screening” publication in J. Med. Chem. Today that library has grown to 6000 compounds following a relaxed rule-of-3 (allowing in particular more hydrogen-bond acceptors and higher lipophilicity) and requiring at least one hydrogen bond donor or acceptor and at least one ring. Also, the primary screening technique is now surface-plasmon resonance (SPR), with crystallographic follow-up; the entire collection can be screened on a single Biacore instrument in four weeks.

Hennig shared two case studies, one on BACE-1, the other on chymase. In the second case, a dozen fragments were successfully co-crystallized with the enzyme, and all but one of these bound in the S1 pocket, revealing the importance of this site for binding. In response to a question about how widely FBDD is used at Roche, Hennig said that it is applied to all targets that are technically feasible.

In another talk, James Madden described fragment-based discovery at Evotec. An increasingly stringent series of assays (from high-throughput high-concentration functional assays, through SPR and/or ligand-detected NMR, and finally crystallography and/or protein-detected NMR) helps keep the number of compounds manageable at each step. Madden also presented two cases studies, BACE-1 (clearly a popular target for FBDD, perhaps due to its intractability to many other approaches) and PDE10a.

A fun talk with relevance beyond FBDD was “Examples of X-ray Bloopers”, by Edward Kesicki of the Infectious Disease Research Institute (IDRI) in Seattle, WA. He described several cautionary tales from his own experience. In one case, a chemist provided the structure of the wrong enantiomer to a crystallographer, who duly refined the data, resulting in weeks of confusion and time-consuming follow-up experiments. In two others, crystallographers inadvertently omitted methylene units in fitting electron density. We’ve previously commented on the dangers of taking crystallographic data at face value, and Kesicki also mentioned an effort by Stephen Warren of Gonzaga University to comb through and correct structures in the protein data bank. He has a lot of work to do: of the 1000 structures examined thus far, roughly 20% have problems with the ligands.

Finally, in a panel discussion on “medicinal chemistry drivers,” someone asked about the role of fragment-based drug discovery. Consistent with the idea that fragment approaches are becoming increasingly integrated with other lead-finding activities, Hing Sham of Elan said that he was neither pro-fragment nor anti-fragment – “it’s just another tool in the toolbox.”

Is FBDD a FADD?

Two reviews in the July issue of Drug Discovery Today provide an update on the state of FBDD.

The first, from researchers at the VU University, Amsterdam, and IOTA Pharmaceuticals, discusses 23 examples. Many of these have been reviewed elsewhere, but the paper also describes some studies that are unpublished or just reported at meetings. It’s a nice, thorough introduction to the field, and the organization of the review, by institution, gives a flavor of the diversity of approaches.

The second review, from researchers at Astex Therapeutics, provides a historical perspective and clinical focus. There are also useful tables of commercial suppliers of fragments as well as FBDD-derived compounds that have made it into clinical development.

In an accompanying editorial, Mark Whittaker of Evotec asks whether fragment-based drug discovery (FBDD) should really be called fragment-assisted drug discovery (FADD):

This is more than just a difference in semantics, but is, in fact, a broader question of when and how to apply fragment approaches to lead generation, either on their own or in concert with other hit finding techniques.

He goes on to explain that although fragment-based methods can be used by themselves to generate leads, they can also be complementary to other approaches to assess target druggability or focus later hit-finding. This conclusion is consistent with Practical Fragments’ latest poll, in which 85% of respondents reported that, far from being a fad, FBDD (or, if you like, FADD) is integrated in the hit finding stage at their company.

Ligand efficiency for antibiotics

Back in October of last year we highlighted a paper in Science that disclosed a new antibiotic targeting the bacterial protein FtsZ. The compound was derived through fragment-based techniques, though at the time no details were provided. A new paper in BMCL now provides some of the early medicinal chemistry, and also introduces an interesting new tool for evaluating antibiotics.

As mentioned in the Science paper, the researchers (led by Prolysis but with a number of contributors from Evotec and Key Organics) started with the fragment-like (MW = 151, 11 heavy atoms) 3-methoxybenzamide. An initial survey of “SAR by catalog” soon moved to the synthesis of analogs that could be assembled in up to four steps from commercially available compounds. This study found that the amide was essential, and only limited substitutions around the aromatic ring were tolerated. Turning to the alkoxy group, the authors took the classic “methyl, ethyl, butyl” approach, but kept going all the way to dodecyl. Intriguingly, a nonyloxy substituent proved to be optimal, better than either 8 or 10 carbon chains. Adding two fluorine atoms to the aromatic ring improved the potency further. Although the paper does not describe the final push to PC190723, the authors do describe the desire to replace the long alkyl chain and its likely attendant problems.



The paper also defines an interesting variation of ligand efficiency:

Antibacterial efficiency = -ln (MIC) / N, where
MIC = minimum inhibitory concentration (mg/ml) and
N = non-hydrogen atoms

Although the metric has a few quirks (for example, low-efficiency compounds can actually have negative numbers), “good” values correspond roughly to good LE values; clinically approved low molecular weight antibiotics have antibacterial efficiencies in the 0.26-0.32 mg/ml/atom range.

So for all you folks working on antibiotics, not only are fragments a viable starting point, you now have a new way to evaluate progress.