Reactive and undesirable functionality

Some functional groups are reactive enough to cause trouble: they alkylate proteins indiscriminately, they interfere with assays, or they simply do not survive long enough to be tested. This module teaches you to spot them — and then complicates the picture, because the same chemistry that makes a screening artefact makes a modern covalent drug.

An electrophile is an electron-poor centre that a nucleophile will attack. Proteins are full of nucleophiles — cysteine thiols above all, then lysine amines, serine and threonine hydroxyls, histidine nitrogens. A sufficiently reactive electrophile will find them, and it will not be fussy about which protein it finds them on.

Electrophiles come in two flavours. Hard electrophiles are charge-controlled and prefer hard nucleophiles such as amines and alkoxides. Soft electrophiles are orbital-controlled and prefer soft nucleophiles — above all the cysteine thiol, which is by some distance the most reactive nucleophile in a typical protein. Almost every covalent drug and almost every alkylating artefact reacts with cysteine.

The single most important reactive motif to recognise is the Michael acceptor: a carbon-carbon double bond conjugated to an electron-withdrawing group, usually a carbonyl. The alpha,beta-unsaturated carbonyl.

The conjugation makes the far carbon of the double bond electrophilic, and a thiol adds across it. Acrylamides, acrylates, vinyl sulfones, enones and quinones all behave this way. Learn the shape: C=C next to C=O.

Strained rings and good leaving groups do the same job by a different mechanism. Epoxides and aziridines are three-membered rings under enough strain that a nucleophile opens them. Alkyl halides on sp³ carbon — particularly benzylic or allylic ones — are straightforward alkylating agents. Aryl halides are not: an aryl C-Cl is stable and appears in dozens of safe drugs.

Some groups cause trouble not by reacting directly but by being turned into something reactive.

• Quinones cycle between oxidised and reduced forms, generating reactive oxygen species each turn. They are also Michael acceptors, so they are doubly flagged.
• Catechols oxidise readily to quinones, so they inherit the problem.
• Nitroaromatics are reduced to nitroso and hydroxylamine species, which are electrophilic and are associated with mutagenicity.
• Anilines are oxidised to hydroxylamines by the same route. An aniline is not a reason to abandon a series — there are many marketed anilines — but it is a reason to look at the metabolic data early.

A separate category: not toxic, just fragile. If a compound decomposes in buffer, the assay result is about something other than the compound you think you tested.

• Imines (Schiff bases) hydrolyse back to the amine and carbonyl in water, often within minutes.
• Hemiketals and hemiacetals are in equilibrium with their open forms.
• Esters in the wrong place are hydrolysed by plasma esterases, which is the intent in a prodrug and a nuisance everywhere else.
• Anhydrides and acyl halides are simply too reactive to be drugs.

Some compounds appear as hits in assay after assay against unrelated targets. They are almost never promiscuous binders. The mechanisms are mundane and worth knowing by name, because they determine what you should do next.

• Colloidal aggregation — the compound forms particles that sequester enzyme non-specifically. Adding detergent abolishes it, which is the standard diagnostic.
• Redox cycling — the compound generates hydrogen peroxide, which inactivates the enzyme. Adding catalase abolishes it.
• Fluorescence interference — the compound absorbs or emits at the assay wavelength. Changing the readout abolishes it.
• Non-specific covalent reactivity — the compound modifies the protein. A mass spectrometry experiment shows it.

Notice that each mechanism has a specific experiment attached. That is the difference between using an alert well and using it badly, and module 21 is entirely about that difference.

Here is the complication. The acrylamide warhead of a modern kinase inhibitor is the same chemistry as a Michael acceptor you would reject from a screening hit list. Ibrutinib, afatinib and osimertinib all carry one, and all are approved drugs.

So what separates a warhead from an artefact? Three things, and none of them is the substructure.

• Targeting. The molecule binds reversibly and specifically first. The warhead only reacts once the reversible complex has held it next to the right cysteine, for long enough. Reactivity without affinity is just a poison.
• Tuned reactivity. The warhead is deliberately made as unreactive as it can be while still working. A hot acrylamide alkylates glutathione before it reaches its target.
• Intent and evidence. Someone measured the selectivity, found the adduct by mass spectrometry, and showed the effect depends on that specific cysteine.

The substructure is not the problem. The context is. That sentence is the whole module, and it is also the reason unit 5.2 exists.