Why HDX-MS is rare in FBLD – and practical tips to change this

Hydrogen/deuterium exchange mass spectrometry (HDX-MS) can help identify the binding site of a ligand, sometimes. Briefly, a protein-ligand complex is diluted into a solution of D₂O; exchangeable hydrogens on the protein will be replaced by deuterium, and those that interact with the ligand will be protected. A comparison with the protein alone will thus reveal which region interacts with the ligand. Practical Fragments discussed the technique back in 2012 and 2014, but since then it has been mentioned only a handful of times. In a new paper in J. Am. Soc. Mass Spectrom. Yoshitomo Hamuro and Stephen Coales (ExSAR) provide insights into why it is rarely used in FBLD, and offer solutions.

 

The researchers argue that the main problem with using HDX-MS in FBLD is that fragments often have low affinities and low solubilities. A theoretical analysis reveals that “the concentration of a ligand, not the molar excess of a ligand over a protein, is the key to drive the equilibrium to complex formation.” A series of calculations with hypothetical ligands having dissociation constants of either 100 µM or 1000 µM reveals that changing the concentration of protein is unlikely to have much effect on the outcome of the experiment, whereas increasing the concentration of ligand will give cleaner data. The problem is that many fragments may not be soluble at sufficiently high concentrations.

 

To solve this challenge, the researchers provide two solutions. First, they suggest spiking ligand into the D₂O exchange buffer; this will keep the ligand from being diluted.

 

A second fix is similar: rather than diluting a protein-ligand complex 1:9 into D₂O, the researchers suggest a 1:1 dilution, so the ligand concentration drops only by half rather than by 10-fold.

 

High concentrations of ligand can potentially interfere with the mass spectrometry measurements, so the researchers also suggest using smaller volumes with higher concentrations of protein.

 

These all seem like simple, practical measures to make HDX-MS more applicable to FBLD, but unfortunately the paper does not actually provide any experimental proof of concept data, so I’ll put the question to you, dear readers: have you found HDX-MS useful in FBLD? If so, under what conditions?

Nucleophilic fragments: the other kind of covalent inhibitors

Covalent fragment-based lead discovery is becoming increasingly popular, spurred on by the rapid discovery and approval of sotorasib. In general, covalent inhibitors contain cysteine-reactive electrophiles, though efforts are also targeting other amino acid residues such as serine and lysine. In all these cases though, the fragment contains an electrophile, while the protein contains the nucleophile. A new paper in J. Am. Chem. Soc. by Megan Matthews and collaborators at University of Pennsylvania and Oberlin College turns things around.

 

None of the twenty standard amino acids are electrophilic, but some proteins do use electrophilic cofactors, such as pyridoxal phosphate. Moreover, some proteins undergo post-translational modifications which introduce a pyruvoyl (Pyvl) or glyoxylyl (Glox) group onto the N-terminus; these contain, respectively, an electrophilic ketone or aldehyde. As we wrote about here, aldehydes and ketones can react covalently with hydrazines, and the new paper shows that the kinetics of this reaction vary – as expected – with the nucleophilicity of the hydrazine.

 

Next, the researchers assembled a library of 17 fragment probes containing both a nucleophile as well as an alkyne that could be used for click chemistry. These probes were screened against cells for 30 minutes at 37 °C, the cells were lysed, labeled proteins conjugated to a dye, and the whole gemish run on a denaturing gel; the results showed a wide range of reactivities for the different probes.

 

To assess which proteins were reacting with which probes, the researchers turned to isoTOP-ABPP, a chemoproteomic method we previously wrote about here in the context of electrophilic fragments. (Chemical biologists are fond of abbreviations, and they call this new approach with nucleophilic fragments “reverse-polarity activity-based protein profiling”, or RP-ABPP.) Three probes, P11, P12, and P13, were found to modify 98, 60, and 16 proteins, respectively. Remarkably, despite their small size and common hydrazine nucleophile, only a single protein was labeled by all three probes.

 

Two of the proteins labeled by P11 include secernin-2 and -3 (SCRN2 and SCRN3). The functions of these proteins are unknown, though genome-wide studies have associated SCRN3 with several diseases.

 

The requirement for the probes to contain both an alkyne handle and a nucleophile increases complexity, and the researchers recognized that they could use the probes in competition mode against fragments lacking the alkyne. They assembled a set of 45 nucleophile-containing fragments and treated cell lysates with these, followed by treatment with probe P11, click chemistry to introduce a fluorescent dye, and gel electrophoresis. Hydrazine-containing fragments that inhibited the binding of P11 were found for SCRN2, SCRN3, and the protein AMD1. Some of these fragments showed EC₅₀ values less than 1 µM and were up to 25-fold selective for SCRN3 over SCRN2 despite the 54% sequence identity shared between the two proteins.

 

An orthodox medicinal chemist might sniff at the hydrazine moiety in these molecules, but it is worth noting that P12, P13, and P17 are all derived from approved drugs (carbidopa, hydralazine, and phenelzine; substructures colored blue).

 

The functional roles of Pyvl and Glox modifications in proteins are poorly understood, and whether modulating them will prove useful in treating diseases remains uncertain. But the best way to answer this question will be by inventing suitable chemical probes. This paper suggests that nucleophilic fragments may prove useful.

Higher hit rates with heavier halogens

Halogen bonding is an esoteric type of molecular interaction. Any first-year chemistry student can tell you that halogens are electronegative. More advanced students learn that the electron density on a halogen attached to a carbon is not evenly distributed. Rather, an electron deficient region appears directly opposite the carbon bond on chlorine, bromine, and iodine atoms. This “σ-hole” can form attractive interactions with electron-rich moieties, such as backbone carbonyl atoms. These highly directional interactions can be useful alternatives to hydrogen bonds, especially since they allow a reduction in the number of hydrogen bond donors. But how to find them? This is the topic of a recent open-access paper in Frontiers in Chemistry by Frank Boeckler and collaborators at Eberhard Karls Universität Tübingen.

 

The researchers constructed a library of 191 commercially available halogen-enriched fragments (called HEFLibs), which we wrote about in 2019. Most fragments have a single halogen atom, though 15 have two of the same type (two chlorine atoms, for example). The initial publication had no screening data, but the new paper describes screening the library against four diverse proteins: the methyltransferase DOT1L, the oxygenase IDO1, and the kinases AAK1 and CAMK1G.

 

Ligand-detected STD NMR was used as the primary screen, with proteins present at 20 µM and fragments at 1 mM each in mixtures of two. Between 9 and 57 hits were found for each target, with unique hits for all the targets except DOT1L. Some fragments hit all four targets, including one similar to the "universal fragment" we highlighted here.

 

Interestingly, iodine-containing fragments gave higher hit rates than bromine-containing fragments, which in turn gave higher hit rates than chlorine-containing fragments. Specifically, 9 of 14 (64%) iodine-containing fragments hit at least one target, vs 51% and 35% for bromine- and chlorine-containing fragments.

 

To assess whether halogen bonding played a role, the researchers calculated maximum electrostatic potential (V_max) for each fragment; this is a measure of the size of the σ-hole. Fragment hits tended to have higher V_max values than non-hits.

 

One possible confounding influence is that aryl halides can react with cysteine residues in proteins, and indeed the researchers did find that some of their fragments are unstable in the presence of the cellular reducing agent glutathione.

 

To confirm the STD-NMR results with an orthogonal method, the researchers turned to isothermal titration calorimetry (ITC). Of 57 fragment-protein pairs tested, only ten gave K_D values less than 1 mM, and nine were against the kinases; there were even a couple single-digit micromolar binders for AAK1. ITC is less sensitive than NMR, so some of the other fragments may bind too weakly to fully characterize.

 

Unfortunately, crystallography has been unsuccessful so far, so it remains unclear whether any of the hits are actually making halogen bonding interactions with the proteins. Halogens are good at filling lipophilic pockets, so it is perhaps likely that less specific van der Waals interactions are the key affinity drivers. But the Boeckler group has been pursuing halogen bonding for more than a decade, so I look forward to seeing more on this topic.

 

And in the meantime, happy Pi Day!

Virtual screening succeeds against the SARS-CoV-2 main protease

Today marks exactly two years since Practical Fragments first mentioned SARS-CoV-2. Since then, COVID-19 has killed more than 6 million people worldwide. Multiple effective vaccines have been developed and approved, along with a couple small-molecule drugs, but the virus is here to stay, and more drugs will be needed. This brings us to an open-access paper published in J. Am. Chem. Soc. by Jens Carlsson (Uppsala University) and a large group of international collaborators.

 

The so-called main protease (M^pro, or 3CLp) has been an antiviral target since the earliest days of the pandemic; the work we highlighted two years ago focused on a crystallographic screen against this enzyme. The new paper describes two virtual screening approaches.

 

The first started with a library of 235 million virtual compounds, mostly from Enamine’s “readily available for synthesis” (REAL) collection. Each compound was docked in thousands of different orientations against the active site of M^pro using DOCK3.7. Despite the staggering numbers (more than 223 trillion complexes!), the screen took just a day on 3500 CPU cores. The top 300,000 compounds were clustered based on similarity, and 100 molecules were synthesized. Nineteen of these showed binding by SPR, and three also inhibited the enzyme. Crystal structures were obtained for two of these, and both bound similarly to the predicted binding modes.

 

Compounds 1 and 3 each contain a hydantoin moiety that makes multiple hydrogen bonds to the protein, and merging elements led to low micromolar compounds such as compound 15. Further optimization ultimately delivered compound 19.

 

Compound 19 was potent in SPR and biochemical assays. Though it binds noncovalently, it had comparable cellular activity to nirmatrelvir, the recently approved covalent inhibitor of M^pro. Compound 19 showed nanomolar cell potency against SARS-CoV-1 and MERS-CoV and good selectivity against ten human proteases. The in vitro stability and permeability of compound 19 are also promising.

 

In addition to this de novo virtual screen, the researchers performed a second screen starting from one of the fragments identified crystallographically at Diamond Light Source. Of 93 molecules purchased and experimentally tested, 21 showed binding by SPR and 5 of these also inhibited the enzyme, with the most potent compound showing low micromolar activity.

 

There are several lessons from this paper. First, despite searching hundreds of millions of compounds, the best hits had only modest activity. This is perhaps surprising given the high fragment hit rates observed against M^pro in crystallographic and NMR screens, though it is worth noting that those fragments were even weaker binders.

 

Second, the hit rate from the naïve virtual screen was similar to that from the experimentally derived fragment screen. The researchers suggest that perhaps docking “may be more proficient in ranking diverse chemotypes rather than differentiating between closely related elaborations of the same scaffold.” In other words, virtual screens seem better at evaluating diverse starting points rather many similar molecules.

 

Third, despite the fact that the de novo virtual screen was not explicitly fragment-based, compound 1 does actually adhere to the rule of three. From there, addition of just six atoms improved affinity by >600-fold while also improving ligand efficiency.

 

Finally, this work is a testament to the utility of combining massive virtual screening with readily synthesizable compounds: the researchers note that it took less than four months to progress from compound 1 to nanomolar inhibitors.

 

This work relied heavily on rapid chemical synthesis done in Ukraine. Indeed, the two most popular fragment suppliers are both largely based in that country. Over the years many of us have come to know Ukrainian scientists not just as trusted colleagues but also as friends. I wish them and their families safety, and strength.