Identifying small molecule RNA binders using covalent tethering
September 21st, 2025 By John Widen
Some of the first small molecule therapies that were used to treat cancer covalently modify DNA (and proteins as well).
Mustards were originally developed as chemical warfare agents during the Great War (WWI). Mustards are historically referred to as
mustard 'gases', which is a misnomer because these substances are actually liquids.
Two types of mustards were developed using sulfur and nitrogen (Fig. 1A). Mustards decompose to produce HCl
when they are exposed to water.
These chemicals are dangerous and horrible because they are absorbed through the skin readily
and are reactive towards proteins and DNA. The symptoms that develop involve blistering, burns, and severe irritation to mucosul
membranes including the respritory system. Like many frontline cancer therapies, mustards paradoxically cause genetic mutations that
increase the risk of cancer. Newer versions of original mustards are cyclophosphamide (Fig. 1B) and platinum-based
therapies that operate by the same mechanism, which alkylate DNA (and proteins) to cause cell cycle arrest. The goal of these toxic
molecules is to kill quickly dividing cells (i.e. cancer) before healthy cells. These molecules are effective in killing cancer cells
but the side effects, as you can imagine, are not fun. However, DNA-linking chemotherapies have been around for over 50 years and are still
the frontline therapy for many cancers.
Mustards are reactive alkylating reagents based on one of my favorite physical organic chemistry principles, anchimeric assistance. This
is also (maybe more commonly) referred to as neighboring group participation. Normal alkyl halides are not that reactive. Most of the time
alkyl halides require forcing conditions to undergo SN2 displacement such as heat and strong base. In the case of mustards, a lone pair of
electrons from the sulfur, nitrogen, or oxygen two methylene groups away from a halide can drastically increase the
reactivity by displacing the halogen forming a 3-membered cation intermediate that can be ring opened by a nucleophile
(Fig. 1C). An ethyl halide substitution
has the fastest kinetics compared to linkers that would form a 4-membered ring. I don't have my textbook in front of me but I
believe linkers that can form 5 and 6-membered rings have reactivities somewhere between molecules that can form
3 and 4-membered ring intermediates.
This brings us to
this new publication
in JACS from the Disney Lab at the Unversity of Florida, Scripps Institute that describes a screening method using covalent fragments for the
identification of RNA binders. Nine nitrogen mustards were initially used to compared
electrophoretic mobility shift and MALDI-TOF mass spectrometry assays. Not surprisingly, MALDI-TOF is deemed more sensitive and detected three additional
molecules that were negative in the electrophoretic mobility shift assay. Also, not so surprising, was that several of the nitrogen mustards formed up to
seven adducts with the RNA oligo, likely explaining the larger EMSA shifts.
Only one mustard, Compound 9, formed a single adduct with the RNA oligo where all of the
other active mustards formed two or more (Fig. 2A).
Using the MALDI-TOF method a 2000 molecule libary was screened containing various warheads attached to heterocyclic scaffolds that are known in the
literature to bind RNA. The library only contains six nitrogen mustards and is focused on less reactive covalent moieties including traditional thiol
reactive electrophiles such as acrylamide and chloroacetamide. The majority of the library though focuses on activated ureas, thioureas, and amides. From a
perspective of covalent fragment screening this makes a lot of sense because highly reactive electrophiles such as mustards will give far too many non-specific
hits. Typically, a principle of covalent libraries is to reduce the non-specific activity so that a binding event has to occur prior
to covalent reaction with the target of interest; in this case RNA. I'll mention here that high-throughput covalent fragment screens typically use
a Rapid Fire MS by Agilent (or a similar system) instead of MALDI-TOF. Both give the same result but I'm not sure of the logistics of a high-throughput screen
with >10,000 small molecules is practical without a special setup using MALDI. In theory it would only require a plate stacker but prepping samples would
also be an issue.
The rest of the publication focuses on a new electrophile, 3-chloropivalamide, which made up the majority of the hits (24 of 34 primary hits).
Interestingly, 24 out of 49 molecules containing a 3-chloropivalamide within the screening library were
hits from the screen and all validated in subsequent assays. There is a proposed mechanism in the supporting information (SI) where the ketone of the amide group
displaces the chorline, therby forming a 4-membered cation intermediate, but unfortunately no mechanistic work was done to confirm this (Fig. 2B). Maybe this work is
being saved for a follow-up paper but it would be interesting to see if the reactivity can be tuned. Sounds like a great graduate student project to me!
Simplified 3-chloropivalamide derivatives were synthesized based on the predominant scaffolds identified from the screen. All of the simplified
molecules were reactive towards a guanidine mimic 4-(4-Nitrobenzyl)pyridine(NBP) and thiol 5,5-dithiobis(2-nitrobenzoic acid) (DTNB). These assays
demonstrate that this electophile has intrinsic reactivity towards nitrogen and thiol nucleophiles. The authors try to make the case that 3-chloropivalamide
is a guanosine-like electophile but the comparison was made to iodoacetamide, a very reactive electophile towards thiols. To me, this comparison is like
hitting your hand with a sludgehammer versus a regular hammer and saying that the regular hammer hurts far less. The bottom line is that 3-chloropivalamide
is still quite reactive towards thiols, but I think this electophile does provide an opportunity to identify weak binders to RNA scaffolds.
The authors demonstrate that some of the scaffolds from the screening hits have weak binding affinity
without the elecophile present using NMR (Fig. 2C). However, this result is bias because
the screening library uses well-known RNA binding scaffolds in the first place. The point still stands that the technique could identify weak binders that
are novel.
There are other non-mustard hits from the primary screen that I thought were interesting but not the focus of this paper. The structures of the
validated hits are in the supporting information. There are three chloroacetamides, one N-acylimidazole, one acrylamide, and several other heterocyclic
molecules that are worthy of medicinal chemists' nightmares (Fig. 3). The adduct formation ratio was less than one for the
acrylamide and chloroacetamides in the
validation studies but I think it is intriguing that these electrophiles reacted with RNA in a covalent manner at all. A
previous study demonstrates a
lack of reactivity towards RNA and DNA. There was no follow-up on these molecules but I wonder what functional group these electophiles react with
on RNA. Chloroacetamides are generally too reactive to be considered as a viable approach to develop
covalent small molecule therapies for humans but have certainly provided many tool compounds for biological reasearch.
This publication hints at a concept that I've thought about for quite some time but is not reported often in drug discovery. That concept is
using a covalent fragment or covalent small molecule library to identify ligands against challenging drug targets
with the goal to remove the electrophile and maintain all or some of the potency.
There are very limited examples of this occurring in drug discovery, but one prolific example
is KRAS. This target was considered undruggable until Kevan Shokat and Jim Wells' lab at UCSF
identified a cryptic pocket
adjacent to the GTP binding site using
a disulfide-fragment-based screening approach called tethering. That was in 2013. This discovery set off a race to identify drugs targeting
KRASG12C that is still ongoing today.
After a decade of research from many academic labs and biotech companies, KRAS scaffolds have been optimized to the point
where the electrophile can now
be removed and maintain low nM potency against multiple KRAS mutants.
The Disney lab publication hints at this idea of using covalency to identify novel RNA binders but doesn't explicity state it.
I am always on the look out for reports taking this approach. More often than not publications remove the electrophile
to demonstrate that there is a significant loss in potency. The concept of switching the approach to a non-covalent ligand
is usually not top of mind. But at some point in optimization, the binding component of the scaffold prior to reaction with the
nuclephilic residue should be high enough to measure. Losing 50 to 100-fold activity is tough to deal with but that loss in
potency could be gained back like it was for KRAS by forming other strong non-covalent interactions.
Next time you are optimizing the potency for a covalent scaffold, try removing the electrophile every once in a while
and see if the non-covalent binding affinity can be measured. Maybe you can transition to a non-covalent approach.
I'll stop there. Thanks for reading.
The site does not have a comments section yet! Hopefully, very soon! Until then please drop me a line at jwiden@chemjam.com. If you provide comments on my articles I reserve the right to post them on this website as additional commentary. My goal is to have an open discussion!
