Fragments vs RNA revisited: the power of two

RNA can assume complex three-dimensional structures just like proteins, and given the many roles it plays it is perhaps surprising that there are so few drugs that target this class of biomolecules. One problem is that ribonucleic acids are less diverse than amino acids, so there is less scope for developing small molecules that bind to specific regions of RNA. Nonetheless, a few brave souls have tried, some using fragment-based approaches. The latest such effort appears in J. Mol. Biol.

The researchers, led by Gabriele Varani of the University of Washington, Seattle, took a two-step approach to find two fragments that could simultaneously bind to the TAR element of HIV-1, a short stem-loop element essential for viral replication. The protein that normally binds to TAR contains a critical arginine residue, so the researchers started by purchasing a set of 16 arginine mimetics and using NMR to determine if any of them bound to TAR RNA. Several did, and one guanidine-containing molecule (MV2003) gave a strong NMR signal and also contained a hydrophobic element. The researchers decided to use this to hunt for a second fragment.

To find the second binder, the team screened 250 generic (ie, not targeted to RNA) fragments from Maybridge in pools of 5-8 in the presence of the first fragment. Remarkably, saturation transfer difference (STD) experiments, which detect changes in ligand NMR signals upon binding to macromolecules, suggested that more than 100 of these generic fragments appeared to bind to TAR RNA. However, more careful study of 20 representative fragments from 13 different scaffolds rapidly winnowed the set: 5 didn’t repeat when tested outside the pool, 6 gave signals in the absence of RNA, and 3 were not dependent on the presence of MV2003, suggesting that they bind nonspecifically. However, the remaining 6 only produced signals in the presence of both TAR RNA and MV2003, indicating a specific ternary complex. Although two of these fragments contain a (positively charged) primary amine, the rest are likely either neutral or only partially protonated at physiological pH. Interestingly, one of these is closely related to an RNA-binding fragment identified in previous work by a different group.

Next, the researchers constructed a model of how MV2003 bound to RNA. They used NMR data (nuclear Overhauser effects, or NOEs) to determine which atoms of MV2003 were close to which atoms of TAR RNA. Unfortunately no intermolecular NOEs were observed between any of the six fragments and the RNA, but it was possible to observe interligand NOEs (ILOEs) between the fragments and MV2003, and this enabled additional modeling suggesting that the fragments bind in a small pocket that only forms when MV2003 binds to RNA.

The paper ends with a cliff-hanger:

The formation of a new binding pocket allows binding of other fragments and suggests that more powerful ligands can be generated by linking the fragments together.

Although fragment linking is easier said than done, the hydrophobic moiety in MV2003 may improve the odds here, as described in the previous post. Practical Fragments hopes they will give it a shot!

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.

Who's doing FBLD in 2011?

It’s been almost two years since our last attempt at cataloging companies doing fragment-based lead discovery. That list contained 19 entries, and 5 others were mentioned in the comments section. A paper just published online by Phil Hajduk and colleagues at Abbott in J. Comput. Aided Mol. Des. provides another list of 19 companies, which has inspired Practical Fragments to combine the two to provide what we hope is the most comprehensive list of companies working in FBLD. (The paper itself is worth reading too for insights into how fragment screening has evolved. For example, Abbott pioneered the use of NMR for finding and characterizing fragments, but now functional screening, modeling, and X-ray crystallography are dominant.)

Companies doing FBLD:

Abbott Laboratories
Ansaris (previously Locus)
AstraZeneca
Astex
Beactica
BioFocus (Galapagos)
BioLeap
Biosensor Tools
BioSolveIT
Boehringer Ingelheim
Carmot Therapeutics
Crelux
Crown Biosciences
Crystax Pharmaceuticals (web site seems down - are they still around?)
Eli Lilly
Emerald BioStructures (from deCODE)
Evotec
Genentech (Roche)
Genzyme (Sanfi-Aventis)
GlaxoSmithKline
Graffinity Pharmaceuticals (NovAliX)
iNovacia
IOTA Pharmaceuticals
Johnson & Johnson
Kinetic Discovery
MEDIT
Merck
Nerviano Medical Sciences
NovAliX
Novartis
Pfizer
Pharma Diagnostics
Plexxikon
Polyphor
Proteros
Pyxis Discovery
Roche 
Schrodinger
Selcia
Sprint Bioscience
Structure Based Design
Vernalis
Zenobia Therapeutics
ZoBio

Unlike the previous list, this one also includes large pharmaceutical companies known to be active in FBLD. Companies that have been acquired or merged are listed separately if they maintain separate web sites. The list excludes companies solely focused on selling fragment libraries, as these are covered separately.

The current list includes some 44 companies, which illustrates how widespread FBLD has become. The continuity is also encouraging: despite the challenging economic environment of the last few years, aside from a few acquisitions, a name change, and a spin-off, all the companies from the 2009 list are still around (with the possible exception of Crystax).

I’m sure the list is still incomplete, so if you know of someone else please add them to the comment section.

Biolayer interferometry (BLI)

Surface plasmon resonance (SPR) has become a primary tool for finding fragments. One of its attractions is that, in addition to requiring only small amounts of protein, it can provide dissociation constants (Kd values) and, for tighter binders, on-rates and off-rates. However, SPR is not the only biosensor-based technology out there. Biolayer interferometry is a related technique, and, as judged by the discussion following the FBLD 2010 meeting, is clearly of interest to many people. A paper published online by Charles Wartchow and colleagues in J. Comput. Aided Mol. Des. provides a description of the technology and comparison with other methods.

Like SPR, BLI requires immobilization of the protein target to a surface; the current paper uses biotin-labeled proteins and streptavidin coated biosensors from ForteBio. Unlike SPR, the technology does not rely on samples flowing through tiny capillaries, and up to 16 protein-labeled sensors can be simultaneously dipped directly into different solutions of small molecules arrayed in a 384-well plate. BLI relies on changes in the interference pattern of light between the sensor and the solution caused when a small molecule binds to a protein on the surface of the sensor.

In the current study, the authors studied three proteins: Bcl-2, JNK1, and eIF4E. Initially a library of 140 fragments was screened in triplicate at 200 micromolar concentration against each of the three targets. Both JNK1 and Bcl-2 gave very high hit rates (24 and 21%, respectively), but eIF4E gave a much more “fragment typical” hit rate of 3.5%. This protein was subsequently screened against 6500 compounds, a task which required 1 mg of protein, 10 days, and 700 sensors (which needed to be periodically replaced throughout the campaign).

After curating the eIF4E hits to remove compounds that gave anomalously high signals or slow off-rates, the remaining molecules were then retested in a second screen, which confirmed 50% of the remaining hits, for an overall hit rate of 1.3%. However, many of these still looked suspicious when they were tested in 8-point titration curves; it seems that, like SPR, BLI is also prone to false-positive problems.

The researchers also ran biochemical and SPR screens on some of the targets. For eIF4E, the overlap between hits coming from BLI and those from biochemical screens was 52%, though many of these are derivatives of a single scaffold. Another subset of the common hits gave non-ideal behavior, calling into question their mechanism of action. It remains unclear whether the BLI hits that were not active in biochemical assays are real, and if so, relevant.

In the end, the authors conclude that:

These fragment screening studies demonstrate that BLI is suitable for small molecule characterization and fragment screening.

But they continue:

Hit assessment… with BLI and SPR is non-trivial, however, and although numerous hits from the BLI, SPR, and biochemical assays were characterized, most of the BLI and SPR data obtained from the examination of a concentration series in the micromolar range showed linear relationships with respect to concentration, unreasonably high signals, or slow off-rates.

Clearly, like all techniques, one should not rely on BLI alone. What remains to be seen is whether BLI has advantages over related techniques such as SPR, whether in terms of speed, sensitivity, resistance to artifacts, or cost. Several of the authors of the paper are from Roche, but the paper does not make clear whether BLI is becoming integrated into the workflow there. Is anyone else out there using BLI? If so, what has been your experience?