02 May 2016

A strong case for crystallography first

We noted last week that one theme of the recent CHI FBDD meeting was the increasing throughput of crystallography. Crystal structures can provide the clearest information on binding modes, and a key function of standard screening cascades is to whittle the number of fragments down to manageably small numbers for crystal soaking. Only a few groups have used crystallography as a primary screen. A team led by Gerhard Klebe at Philipps-Universität Marburg argues in ACS Chem. Biol. that crystallography should be brought to the forefront.

The researchers were interested in the model protein endothiapepsin. As discussed last year, they had previously screened this protein against a library of 361 compounds using six different methods, and the agreement among methods was – to put it charitably – poor. Nonetheless, many hits that did not confirm in orthogonal assays produced crystal structures when soaked into the protein. Thus emboldened, the researchers decided to soak all 361 fragments individually into crystals of endothiapepsin. This resulted in 71 structures, a hit rate of 20%, higher than any of the other methods (which ranged from 2-17%). Even more shocking, 31 of the fragments were not identified by any of the other methods, and another 21 were only identified by one other method. Thus, a cascade of any two assays would have found, at best, only a quarter of the crystallographically validated hits.

In agreement with other recent work, the fragments bound in multiple locations, including eight subsites within the binding cleft as well as three potentially allosteric sites. Not all of these sites were found using other methods.

But are these fragments so weak as to be uninteresting? To find out, the researchers performed isothermal titration calorimetry (ITC) to determine dissociation constants for 59 of the crystallographic hits. Three of the 21 most potent (submillimolar) binders were not detected by any of the other methods, and another seven were only found by one.

What factors led to this crystallographic bonanza? First, the researchers used the very high concentration of 90 mM for each fragment (in practice sometimes <90 mM because of precipitation). Not surprisingly, solubility was important: 97% of the hits had solubilities of at least 1 mM in aqueous buffer, and the soaking solution contained 10% DMSO as well as plenty of glycerol and PEG. Achieving such high concentrations is harder when multiple fragments are present, and the researchers argue from some of their historical data that the common use of cocktails lowers success rates.

How did different methods compare? Interestingly, functional assays such as high-concentration screening or reporter-displacement assays fared best, while electrospray ionization mass-spectrometry (ESI-MS) and microscale thermophoresis (MST) were close to random. This is in marked contrast to other reports for ESI-MS and MST, and the researchers are careful to note that “the choice and success of the individual biophysical screens likely depend on the target and expertise of the involved research groups.”

Primary crystallographic screening was an early strategy at Astex, and although this may not have been fully feasible 15 years ago, it seems they were on the right track. Of course, not all targets are amenable to crystallography, and not everyone has ready access to a synchrotron beam with lots of automation. But for those that are, it might be time to drop the pre-screens and step directly into the light.

6 comments:

Artem Evdokimov said...

Very insightful post. I have a few notes to corroborate Dr. Klebe's observations: about eight years ago while at Pfizer we did a fragment screen against an enzyme, and went directly to crystallographic screening. If I recall correctly, some of the observed fragments were very useful as starting points for ligand development and also not many of the biophysical methods agreed with these hits. Taking a real memory trip backwards, to the P&G Pharma days - we had a phosphatase fragment screen (of sorts, it wasn't all that large) and again fragments emerged that were unique and quite a lot of them weren't detectable by biophysical methods -- but were in good correlation with the enzymatic assay.

So in a nutshell, I agree: it is not as easy or cheap to go for structure-based screening but if you have the resources and the determination to do so, the results might pleasantly surprise you :)

Unknown said...

You should reference Vicki Nienaber's paper on this subject. We were playing around with this in 1998-2000 time frame at Abbott, following Hajduk and Fesik's SAR by NMR work.

Nienaber, VL, et al, Nature Biotechnology 18, 1105 - 1108 (2000)
doi:10.1038/80319

Peter Nollert said...

An interesting study indeed!
Looking at this new lead discovery space with a business perspective it is important to realize that historically, preference was given to technologies that identified good leads fast, at low cost and with a scalable technology (HTS, I'm looking at you). One way to use Gerhard Klebe's results is to be able to make a stronger case wherever the aim is to discover new allosteric sites, new modes of action and new chemical space.
In addition, it opens the door to re-working 'old' targets. This is a huge opportunity! For instance drug targets that are currently served with drugs that are due to fall off the patent cliff in a few years' time. These are highly validated targets, often situated in a well-understood biological pathway and, even better, with understood side effects. A good example for such a target is the M1 muscarinic acetylcholine receptor. Many pharma companies had developed HTS-based molecules that failed in late phase clinical trials due to M2&3 side effects. Now, with structures (and SBDD-ability) of M1, M2 and M3 receptors, highly selective M1 receptor agonists can be developed. Heptares/Sosei is planning on getting such a molecule into Phase II clinical trials this year.
Here's another thought: if endothiapepsin is not an outlier, it would be rather straight-forward to make an ROI case for many targets to be re-assessed by X-ray structure-FBDD. The remaining issue would be speed rather than cost. Instead of weighing the cost of HTS (and ESI-MS & MST to some extent) against crystallography-approaches the deciding factor may be compatibility with the 'narrow time window' forced onto drug discovery teams. So, in this subset you may find crystallographically established drug targets that have somewhat flawed small molecules.
Hardly revolutionary insight, but now supported with a systematic study (N=1).

Anonymous said...

I'm not really convinced that this paper makes a strong case to use crystallography first. I think what it does is to highlight the importance of the comments made by Dale Cameron in relation to the previous paper from the same group. This is another case where different approaches yield different hits with relatively little overlap. How then do we best assess which are the highest quality hits for progression? If the availability of a crystal structure of the complex is the used as the end-point in making the decision then perhaps the question is a loaded one. Crystallography in this case failed to identify hits from other techniques - although I found the detailed comparison somewhat difficult to follow as different fragments were screened in different assays (why?) and the readouts were different for the different techniques - e.g. if identifying allosteric sites is a goal of the screen it seems counterintuitive to include a competition step with an orthosteric ligand.

There has been considerable progress in improving the throughput of crystallography - highlighted by Radek Nowak and Jose Marquez in their talks at the recent CHI meeting. But it is not clear that these were employed in the Klebe work - so even with a small library of fragments it appears that this was a mammoth undertaking. Even if throughput and synchrotron access are not limiting I suspect there are many technical challenges that are not discussed in detail. The high hit rate with endothiapepsin may be due in part to the strategy of screening single compounds at very high concentrations - 90 mM in 10% DMSO. Is this strategy transferrable to other targets? I suspect not easily. We have one target which gives far better data with soaks at 10 mM than soaks at 20 mM. We have many other crystal systems that do not tolerate 10% DMSO. How many fragments are not soluble at 900 mM in DMSO?

Perhaps one thing that the article does provide is a system to try and better understand the reasons why some fragments fail in a particular method. For example, endothiapepsin is clearly amenable to ligand detected NMR analysis, the protein is able to transfer magnetisation to fragments that bind at the active site and yet a submiilimolar fragment identified by crystallography and confirmed by ITC failed to show up as a hit in the STD screen. Would an STD be observed if the NMR screen were conducted on single compounds? Higher concentrations of ligand?

Perhaps we should be screening by ITC??

Anonymous said...

Although "crystallography first" is an interesting apprach, its feasability remains to be evaluated by more studies.

After years of experience with biophsical screenings using a variety of techniques (including DSF, SPR, NMR and MST), we more often than not found a rather good overlap between the technologies. So from my point of view it is hard to tell how representative the presented results (and the low overlap/success rates of the used methods) are.

In this line, a recent study by Boehringer showed a very good correlation between different methods (http://pubs.acs.org/doi/abs/10.1021/acs.jmedchem.5b01865) with a number of exclusive, x-ray confirmed hits for each method (the most with MST, btw), resulting in the successful generation of BRD9 inhibitors. Additionally, from our own experience, especially MST can often identify potential false-positives which cause aggregation and/or over-stoichiometric binding, which is often missed by other technologies.

Andrey Grishin said...

Thank you Dan for sharing this article, very interesting. In your earlier post about HisRS, you say that “with a few exceptions, crystallography is rarely a primary screen”, but why?
From a synchrotron perspective, collecting data was never that easy:
• modern beamlines provide 10-20 structures/hour;
• automatic data analysis pipelines convert diffraction images into a ligand density (dimple);
• synchrotrons have industrial scientists to help with data collection;
• synchrotrons have a business development departments to negotiate contract terms. Pricing is usually set based on cost recovery;
• IP rights usually belong to the client;
Some synchrotrons went even further. For example, ESRF created a fully automated beamline Massif-1 for high throughput projects. Diamond has created XChem fragment screening laboratory, right by the beamline. Some synchrotrons adopted features of a business incubator (Ex. The Canadian Light Source spun off Canadian Isotope Innovations company).
Isn’t it time to engage even more with synchrotrons? Wouldn’t a company that partners with a synchrotron on a deeper level get an edge in drug design? I look forward to further thoughts on this.