reactivity than the neutral thiol form. Notably, cysteine has the strongest intrinsic
nucleophilicity among the proteinogenic amino acids except the rare selenocysteine.
While catalytic cysteines, that occur for example in cysteine proteases or phosphates,
are pK a -depressed to favor the more nucleophilic deprotonated form, the neutral thiol
is often dominant for non-catalytic cysteines at physiological pH making them more
difficult to address.
Recent computational studies have suggested that most cysteine residues in
kinases’ active sites have even higher pK a ’s than the aforementioned reference
value rendering those moieties comparably weak nucleophiles [14]. Nevertheless,
even cysteines with a very high predicted pK a of >20 (e.g., Cys814 in PDGFRα or
Cys788 in c-KIT [14]) have been amenable to covalent targeting [15]. It should also
be taken into account that the kinase conformation but also the presence of the ligand
itself can influence a cysteine’s pK a and thereby its reactivity [16]. Moreover, not
only the distance but also the orientation between the electrophilic warhead and the
nucleophilic amino acid as well as the flexibility of the latter two are important
determinants for the efficiency of covalent bond formation and may require further
consideration in the design process [17, 18].
The prototypical warhead type for targeting non-catalytic cysteine residues are
acrylamides and analogous attenuated Michael acceptors (vide infra). Usually,
rational TCI design starts from an appropriate non-reactive ligand which is equipped
with a warhead to target a proximal nucleophilic amino acid residue [19]. The design
process is normally guided by structural information from X-ray crystallography
allowing for the rational selection of suitable linkers and attachment points to install
the reactive group. On the other hand, the off-target profiles of known covalent
ligands can be used to identify starting points for re-design and optimization
[6]. Alternative TCI discovery strategies that have recently been pursued include
fragment-based approaches employing electrophiles of low structural complexity
[20–22], which can then be optimized to become potent and specific TCIs. Moreover, DNA-encoded libraries featuring reactive compounds may be screened
[23, 24]. A schematic overview of these strategies is depicted in Fig. 2.
In order to be useful for TCI design, warheads must fulfill several criteria. Ideally,
reactivity is just sufficient to ensure rapid, proximity-driven covalent bond formation
with the targeted residue while being too low for promiscuous bonding to other
physiological nucleophiles. Since the requirements vary between different targets,
the reactivity of the employed functional groups should be tunable over a wide
range. An appropriate balance between target engagement and promiscuity needs to
be found for each individual application. Moreover, warheads (and their metabolites)
should be non-toxic and sufficiently stable against metabolic degradation. In some
cases, however, rapid warhead depletion may be beneficial to make use of kinetic
selectivity while minimizing off-target modification, especially if more reactive
electrophiles are employed. As mentioned, α,β-unsaturated amides have been most
frequently used as cysteine-targeted warheads since they feature a relatively low
intrinsic reactivity, which can be adjusted by the addition of steric bulk or by tuning
the electronic properties of the amide N-substituents [25]. Such moieties rapidly
react with cysteine thiol(ate)s (Scheme 1a), and less frequently with other
Covalent Kinase Inhibitors: An Overview
47
nucleophilicity among the proteinogenic amino acids except the rare selenocysteine.
While catalytic cysteines, that occur for example in cysteine proteases or phosphates,
are pK a -depressed to favor the more nucleophilic deprotonated form, the neutral thiol
is often dominant for non-catalytic cysteines at physiological pH making them more
difficult to address.
Recent computational studies have suggested that most cysteine residues in
kinases’ active sites have even higher pK a ’s than the aforementioned reference
value rendering those moieties comparably weak nucleophiles [14]. Nevertheless,
even cysteines with a very high predicted pK a of >20 (e.g., Cys814 in PDGFRα or
Cys788 in c-KIT [14]) have been amenable to covalent targeting [15]. It should also
be taken into account that the kinase conformation but also the presence of the ligand
itself can influence a cysteine’s pK a and thereby its reactivity [16]. Moreover, not
only the distance but also the orientation between the electrophilic warhead and the
nucleophilic amino acid as well as the flexibility of the latter two are important
determinants for the efficiency of covalent bond formation and may require further
consideration in the design process [17, 18].
The prototypical warhead type for targeting non-catalytic cysteine residues are
acrylamides and analogous attenuated Michael acceptors (vide infra). Usually,
rational TCI design starts from an appropriate non-reactive ligand which is equipped
with a warhead to target a proximal nucleophilic amino acid residue [19]. The design
process is normally guided by structural information from X-ray crystallography
allowing for the rational selection of suitable linkers and attachment points to install
the reactive group. On the other hand, the off-target profiles of known covalent
ligands can be used to identify starting points for re-design and optimization
[6]. Alternative TCI discovery strategies that have recently been pursued include
fragment-based approaches employing electrophiles of low structural complexity
[20–22], which can then be optimized to become potent and specific TCIs. Moreover, DNA-encoded libraries featuring reactive compounds may be screened
[23, 24]. A schematic overview of these strategies is depicted in Fig. 2.
In order to be useful for TCI design, warheads must fulfill several criteria. Ideally,
reactivity is just sufficient to ensure rapid, proximity-driven covalent bond formation
with the targeted residue while being too low for promiscuous bonding to other
physiological nucleophiles. Since the requirements vary between different targets,
the reactivity of the employed functional groups should be tunable over a wide
range. An appropriate balance between target engagement and promiscuity needs to
be found for each individual application. Moreover, warheads (and their metabolites)
should be non-toxic and sufficiently stable against metabolic degradation. In some
cases, however, rapid warhead depletion may be beneficial to make use of kinetic
selectivity while minimizing off-target modification, especially if more reactive
electrophiles are employed. As mentioned, α,β-unsaturated amides have been most
frequently used as cysteine-targeted warheads since they feature a relatively low
intrinsic reactivity, which can be adjusted by the addition of steric bulk or by tuning
the electronic properties of the amide N-substituents [25]. Such moieties rapidly
react with cysteine thiol(ate)s (Scheme 1a), and less frequently with other
Covalent Kinase Inhibitors: An Overview
47