`Risks
`Tjeerd Barf* and Allard Kaptein
`
`Perspective
`
`pubs.acs.org/jmc
`
`Drug Discovery Unit, Covalution Pharma BV, Ravenstein, The Netherlands
`
`■ INTRODUCTION
`
`In the relatively young but expanding field of irreversible kinase
`inhibitor drug discovery, there are two main developments that
`are of central importance. First, the patients have watched the
`first wave of low molecular weight protein kinase inhibitors
`becoming available to them over the past decade. These kinase
`inhibitors have been approved by the U.S. Food and Drug
`Administration (FDA) for
`therapeutic use in oncology
`indications and constitute an important addition to the arsenal
`of drugs to combat cancer (Table 1). Without exception, these
`marketed protein kinase inhibitors have been identified and
`developed using conventional approaches,
`i.e.,
`reversible
`inhibitors that (partly) occlude the ATP pocket in the catalytic
`domain of the kinase. Although protein kinases are regarded as
`an attractive drug target family, it took (and still takes) huge
`efforts to master the human kinome, which comprises more
`than 500 protein kinases. The search for clinically applicable
`kinase inhibitors that target the highly conserved ATP pocket
`has been thwarted by a couple of well-known hurdles.
`Selectivity, cellular potency, and an increasingly crowded
`intellectual property arena are major points of attention.
`The second important development is the renewed interest
`in covalent binding drugs (reviewed by Potashman and Duggan
`and by Singh et al).1,2 This recent revival results from a better
`understanding of the benefits of the covalent binding principle
`and the approval of effective and safe covalent drugs.
`Historically, drug discoverers have been taught to stay away
`from small molecular entities that harbor reactive electrophilic
`groups because these used to be equivalent to promiscuity.3
`Promiscuous hits that relied on reactive groups were tradition-
`ally hard to optimize toward leads, since these were more than
`often interfering with the biochemical assay rather than truly
`modifying the activity of the target of interest.4 Even if the
`target modulation was real,
`indiscriminant
`reactivity was
`believed to trigger
`insurmountable toxic events that may
`surface in late stage clinical
`trials when larger patient
`“suicide
`populations are involved. As a consequence,
`inhibitors”, “warheads”, and covalent
`irreversible inhibitors
`developed a negative flavor over time and became almost
`synonymous with toxicity in some organizations. The
`skepticism toward irreversible drugs may evaporate as more
`examples of
`irreversible drugs progress clinically that
`demonstrate good efficacy and safety margins.
`In a nutshell, the therapeutic applicability or the success of
`irreversible binding kinase inhibitors is dependent on whether
`or not the covalent bond can be confined solely to the protein
`kinase of interest. So this approach is in essence a story about
`two T’s: treatment and toxicity. When relying on the covalent
`binding principle,
`it
`is important
`to discern adduct-based
`toxicity and adduct-based treatment, since the adducting
`
`molecular entities in question obey overlapping fundamental
`rules in terms of reactivity. The only key difference is the nature
`and the function of the proteins that are covalently modified.
`Covalent kinase inhibitors with well-balanced recognition and
`reactivity should provide efficacy, selectivity, and ultimately the
`safety margins that are required for regulatory approval. If we
`strike the right balance, a third “T” will give enough comfort:
`therapeutic window.
`This Perspective aims to give a comprehensive account from
`a medicinal chemist's point of view on the progress of
`irreversible kinase inhibitor drug discovery. The “state of the
`art” is reviewed by means of reported irreversible kinase
`inhibitors profiles and their chemical structures. The potential
`upsides and pitfalls that are associated with this concept are
`highlighted to provide a general understanding of
`the
`differences with respect to conventional drug discovery, as
`well as the future potential of this approach.
`
`■ EFFECT OF RESIDENCE TIME ON THE
`THERAPEUTIC WINDOW
`Residence Time and Efficacy. A kinase inhibitor will only
`be efficacious when it
`is modulating the action of
`the
`physiological kinase and can only do so when the inhibitor is
`bound. This general paradigm was effectively captured by
`Copeland and co-workers, who described the potential
`advantages of
`long residence time in terms of duration of
`action and target selectivity.5 The equilibrium dissociation
`the drug−target binary complex as
`constant (Kd) for
`determined in a closed (in vitro) system is a more accurate
`measure than the IC50 or Ki. Yet the equilibrium setting is
`different in vivo, since the drug concentrations are no longer
`constant, and the efficacy is dependent on the on-rate (kon) and
`even more importantly the off-rate constant (koff). The
`importance of the related occupancy time or half-life for the
`protein−drug complex was corroborated in the context of
`pharmacological agents for whom longer
`residence times
`seemed to be beneficial
`in terms of biological or clinical
`efficacy.6,7 In this respect, kinase inhibitors that exert covalent
`and irreversible binding can achieve the ultimate physiological
`goal: Efficacy is maintained until the target kinase is physically
`restored by the body to physiologically relevant levels (Figure
`1).8 In other words, the pharmacodynamics will be a function
`of the de novo synthesis rate of the target protein rather than
`the trough level of the compound.
`Unlike for kinase inhibitors with a reversible binding mode,
`we anticipate that not all ADME parameters need to be as
`efficiently optimized for covalent
`inhibitors. The general
`
`Received: March 7, 2012
`Published: May 23, 2012
`
`© 2012 American Chemical Society
`
`6243
`
`dx.doi.org/10.1021/jm3003203 | J. Med. Chem. 2012, 55, 6243−6262
`
`Downloaded via Sharisse Geiger on November 21, 2022 at 14:09:42 (UTC).
`
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`
`SANDOZ INC.
`
`IPR2023-00478
`
`Ex. 1030, p. 1 of 20
`
`
`
`Journal of Medicinal Chemistry
`
`Perspective
`
`Table 1. FDA Approved Protein Kinase Inhibitors (as of March 2012)
`
`generic (brand) name
`imatinib (Gleevec)
`gefitinib (Iressa)
`erlotinib (Tarceva)
`sorafenib (Nexavar)
`sunitinib (Sutent)
`dasatinib (Sprycel)
`nilotinib (Tasigna)
`lapatinib (Tykerb)
`pazopanib (Votrient)
`vandetanib (Caprelsa)
`vemurafinib (Zelboraf)
`crizotinib (Xalkori)
`ruxolitinib (Jakafi)
`axitinib (Inlyta)
`
`year of approval
`2001
`2003
`2004
`2005
`2006
`2006
`2007
`2007
`2009
`2011
`2011
`2011
`2011
`2012
`
`company
`Novartis
`AstraZeneca
`Genetech, OSI
`Bayer, Onyx
`Pfizer
`Bristol-Myers Squibb
`Novartis
`GlaxoSmithKline
`GlaxoSmithKline
`AstraZeneca
`Roche, Plexxicon
`Pfizer
`Incyte
`Pfizer
`
`indication
`chronic myeloid leukemia (CML)
`non-small-cell lung carcinoma (NSCLC)
`NSCLC, pancreatic cancer
`hepatocellular carcinoma, renal cell carcinoma (RCC)
`gastrointestinal stromal tumor (GIST), RCC
`CML
`CML
`breast cancer
`RCC
`thyroid cancer
`CML
`NSCLC (ALK +ve)
`myelofibrosis
`RCC
`
`target kinase
`Abl, c-Kit, PDGFRα/β
`EGFR
`EGFR
`Raf, VEGFR2/3, c-Kit, PDGFRβ
`c-Kit, VEGFR, PDGFR, FLT3
`Abl, c-Kit, PDGFR, Src
`Abl, c-Kit, PDGFR, Src, ephrin
`EGFR, ErbB2
`VEGFR, PDGFRα/β, c-Kit
`VEGFR, EGFR, RET
`Abl, c-Kit, PDGFR, Src, ephrin
`ALK, MET
`JAK1/2
`VEGFR, PDGFRβ, c-Kit
`
`Figure 1. For conventional drugs, the pharmacodynamic effect is driven by the pharmacokinetics of the drug (left). For irreversible covalent drugs,
`the pharmacodynamic effect is driven by the turnover rate (de novo synthesis) of the protein target (right). The red hatched area is drug exposure
`not required for efficacy.
`
`requirements for absorption and distribution characteristics
`remain relatively unchanged with respect to the conventional
`approach. It is still advantageous to have good oral availability
`and rapid distribution, but in the case of an irreversible kinase
`inhibitor it would be favorable to clear the drug from the body
`once the target kinase is maximally occupied. Thus, the in vitro
`and in vivo criteria for optimization cycles with respect to
`metabolism and excretion are very much different and should
`accentuate a swift clearance of the drug.
`Residence Time and Toxicity. Drug-safety-related issues
`pose a serious problem to the pharmaceutical industry and have
`been a major contributing factor to attrition rates in the drug
`trajectories.9 As discussed above,
`since no
`development
`excessive circulating levels are required to maintain efficacy
`with irreversible binding drugs, rapid clearance should lead to a
`lower propensity for off-target related adverse effects. But this
`can only be achieved when off-target protein−drug interactions
`are short-lived because of the noncovalent nature of binding.
`On the other hand, toxic events can also be associated with
`covalent binding, and this can be either on-target or off-target
`related. If the toxicity is a direct consequence of on-target
`modulation, then irreversible inhibition might exacerbate the
`adverse effect potential. In this case, the clinical benefit should
`outweigh the adverse event. Alternatively, prolonged residence
`times can occur if collateral targets are being adducted by the
`potential drug. In that case, a high degree of in vivo selectivity
`can only be accomplished when the off-target turnover rate is
`(much) faster than the on-target resynthesis rate.
`For irreversible kinase inhibitors, the target selectivity thus
`should not only be discussed in terms of reversible binding to
`closely related kinases with respect
`to primary sequence
`homology. Especially kinases that (also) share the site of
`covalent modification are highly relevant to include in the
`selectivity panel. A good level of selectivity for those off-target
`
`kinases is suggestive of a high degree of overall selectivity
`within the human kinome, as well as for other protein classes.
`Also, potential adverse effects due to covalent inhibition of
`specific off-target kinases can be designed out early in the
`optimization cycles to yield a better therapeutic window.
`Much of the aversion to irreversible binders is predominantly
`the result of the negative experiences with reactive metabolites
`that could bind covalently to a variety of proteins.10,11 In such a
`case, the parent drugs are metabolically bioactivated to reactive
`species that potentially can bind covalently to all sorts of
`macromolecules. Next to the predictable intrinsic toxicity, this
`may lead to unpredictable idiosyncratic toxicity (IDT) in late
`stage clinical
`testing and beyond. In order to exclude or
`minimize the propensity of reactive metabolite formation,
`potential drugs are usually evaluated preclinically for (i)
`gluthathione (GSH) adduct formation and depletion when
`incubated with recombinant enzymes or human liver tissue
`preparations and (ii) magnitude of covalent protein binding in
`liver microsomal fractions and in rodents. It is obvious that
`these models are highly relevant
`for the optimization of
`irreversible kinase inhibitors as well. The advantage with
`irreversible kinase inhibitors in this respect is that the parent
`molecules are the reactive species, which should have a better
`predictive power for the reactivity assays and body retention
`models involved.
`Recently, the occurrence of IDT or hypersensitivity reaction
`has been linked by Nakayama et al. to the magnitude of
`covalent and irreversible binding to off-targets and the daily
`dose required in patients.12 A classification system to assess the
`risk of IDT was proposed based on covalent binding to human
`hepatocytes as the best predictor, in combination with the daily
`dose.
`that has to be considered for
`Another potential event
`covalent and irreversible kinase inhibitors is hapten formation,
`
`6244
`
`dx.doi.org/10.1021/jm3003203 | J. Med. Chem. 2012, 55, 6243−6262
`
`SANDOZ INC.
`
`IPR2023-00478
`
`Ex. 1030, p. 2 of 20
`
`
`
`Journal of Medicinal Chemistry
`
`i.e., triggering an immune response to the adducted protein.13
`Experiences with reactive compounds and reactive metabolites
`indicate that additional danger signals (e.g., mitochondrial
`damage) are required next to the hapten mechanism to induce
`an immune response (for a review, see Zhang et al).14 Except
`for the β-lactam antibiotics, absence of reports of
`immune
`responses for recently marketed irreversible binding drugs does
`not support this concern as a general problem for covalent
`drugs.
`Reactive Metabolite Considerations. Considerations
`that have been used historically within the pharmaceutical
`industry to assess the fate of conventional drugs with suspected
`reactive metabolites can provide guidance for
`irreversible
`binding kinase inhibitors as well.10 It
`is obvious that
`the
`“avoidance strategy”,
`i.e., complete elimination of reactive
`groups, cannot be applied on warhead-containing drugs, since
`the pharmacological effect is largely dependent on the reactive
`group. Perhaps it would be better to develop a reversible
`therapy for less severe disorders, especially if one can rationalize
`that reversible drugs should work equally well. Also,
`the
`anticipated human dose (for any drug) has to be considered, as
`the body burden has to be kept to a minimum (vide infra). In
`addition, a disease that needs acute dosing rather than chronic
`treatment might be more appropriate to engage with
`irreversible binding drugs. These and other questions need to
`be addressed before a covalent (kinase) inhibitor program can
`be progressed toward clinical development. Obviously, the key
`issue here is the benefit/risk ratio, and serious life threatening
`diseases may be particularly justified in this scenario.
`reactive
`One has to keep in mind though that most
`metabolites are by no means
`tuned, and they can be
`uncontrollable like “unguided missiles”. In contrast, deliberately
`designed irreversible kinase inhibitors can be subjected to safety
`assays during optimization cycles
`to ensure selection of
`compounds that meet stringent safety prediction criteria. In
`some other instances, bioactivation is even required to generate
`the active drug, such as in the case of proton pump inhibitors
`like omeprazole and P2Y12 receptor antagonist clopidogrel.1 No
`matter the type of covalent approach involved, a decision tree
`that incorporates several safety assessment parameters will be
`required to mitigate the risks.
`In addition,
`the level of
`acceptance that has been set for the magnitude of covalent
`binding of reactive metabolites in vitro and in vivo also applies
`for covalent binding parent drugs.
`
`■ THE CONCEPT APPLIED TO KINASES
`
`the members belonging to the
`The question is whether
`superfamily of protein kinases are suitable targets at all for
`irreversible inhibition. First of all, the replenishment of the
`target kinase has to be sluggish enough to allow drug dosing
`once or twice a day. This has to be checked on a kinase-by-
`kinase basis. Second, unlike for instance the enzyme families of
`(cysteine) proteases or phosphatases, kinases lack a catalytically
`active amino acid residue that is key to the function of a
`particular enzyme by virtue of its overt nucleophilicity. The
`relatively unreactive cysteines, serines, threonines, and lysines
`in kinases may have to be exposed to excessive reactivity to
`ensure covalent bond formation. There are plenty of examples
`of overly reactive electrophiles
`that despite the best of
`intentions, in retrospect never had a chance to make it into a
`safe drug. Nevertheless, these prototypical molecules definitely
`marked the beginning of this emerging platform.
`
`Perspective
`
`Non-ATP Competitiveness. There is one major upside
`that can specifically be exploited in the kinase field. Non-ATP
`competitive inhibitors offer distinct advantages to conventional
`ATP competitive binders, especially in terms of efficacy and
`selectivity.15,16 The highly conserved nature of the ATP binding
`pocket poses a challenge in the identification of selective kinase
`inhibitors, but more important is that high intracellular ATP
`concentrations affect the cellular potency of ATP competitive
`kinase inhibitors under physiological conditions.17 Besides
`allosteric binders of protein kinases, blockade of the ATP
`pocket with an irreversible inhibitor has emerged as an
`attractive and alternative strategy to achieve non-ATP
`competitive inhibition of kinase-mediated signaling.18 As a
`consequence, caution should be taken with the interpretation
`and/or comparison of the biochemical and functional IC50
`values, since irreversible inhibitors usually become more potent
`over time. The selection of the length of preincubation time
`will
`thus
`influence the observed value, unless maximal
`inhibition is achieved instantly. The inactivation of the kinase
`involves two steps: formation of a reversible complex based on
`affinity (Ki) followed by the covalent bond formation between
`the inhibitor and the kinase (kinact).19 The first step depends on
`the ATP level used in the various assays reported. The second
`step is primarily influenced by the inherent reactivity of the
`electrophilic group, as well as the distance to the nucleophilic
`trap. The better the juxtaposition of the reacting partners, the
`better the inactivation step. As the kinase is effectively removed
`from the equilibrium with ATP and substrate, a shift in IC50 is
`expected and, as demonstrated in Figure 2, the inhibitor shows
`an increasingly non-ATP competitive character as
`time
`progresses. Consequently, this would be beneficial for protein
`kinases with a low Km,ATP in particular.
`Many research groups have already embarked intentionally,
`or unintentionally, on the search for
`irreversible kinase
`inhibitors (partly reviewed by Garuti et al).20 In some instances,
`these inhibitors have become invaluable assets in the target
`
`Figure 2. IC50 determination depends on the ATP concentration and
`preincubation time for an irreversible kinase inhibitor (left), whereas
`an ATP competitive kinase inhibitor
`is only affected by ATP
`concentration (right). Data are from an unpublished kinase inhibitor
`study included for illustration purposes only.
`
`6245
`
`dx.doi.org/10.1021/jm3003203 | J. Med. Chem. 2012, 55, 6243−6262
`
`SANDOZ INC.
`
`IPR2023-00478
`
`Ex. 1030, p. 3 of 20
`
`
`
`Journal of Medicinal Chemistry
`
`Perspective
`
`Figure 3. Examples of lysine trapping kinase inhibitors.
`
`Figure 4. Schematic representation of available cysteines in or near the ATP binding pocket of kinases in the active conformation (based on the
`crystal structure of interleukin 2 tyrosine kinase, PDB code 1SM2). The spheres indicate the locations of cysteines that are in principle accessible for
`covalent modification: group 1, glycine-rich- or P-loop (dark blue); group 2, the roof of the ATP binding pocket (pink); group 3, hinge region and
`front pocket (yellow); group 4, DFG-motif neighboring cysteine (light blue). Group 5 (in dark red) denotes the cluster of cysteines that resides on
`the activation loop. The gatekeeper group is shown in orange. Adapted by permission from Macmillan Publishers Ltd.: Nature Reviews Cancer
`(http://www.nature.com/nrc/index.html), Copyright 2009.38
`
`validation stage. Others were deliberately designed to be the
`drugs of the future. The selection of examples that is reviewed
`in this Perspective covers both and in addition is primarily
`based on the diversity of cysteine locations within the known
`druggable binding regions of kinases. The electrophilic portions
`of these inhibitors (highlighted in blue in this Perspective) are
`either being trapped by the highly conserved lysine or cysteines
`at various positions of the kinases. To our knowledge, there are
`no reports to date that describe covalent kinase inhibitors that
`adduct other amino acid residues, such as serine or threonine.
`Lysine-Trapping Covalent Kinase Inhibitors. Because of
`the well-conserved nature and its abundant presence, the lysine
`involved in the phosphate transfer machinery is not the most
`ideal candidate amino acid for trapping electrophilic inhib-
`itors.21 This particular lysine resides in virtually every ATP-
`binding pocket of the known human kinases. Nevertheless, a
`couple of lysine trapping covalent inhibitors have appeared in
`the first reported is ATP analogue 5′-
`literature. One of
`fluorosulfonylbenzoyl adenosine (FSBA, 1), a tool frequently
`used as an affinity label for kinases like EGFR,22 p38γ,23 and
`brain PI4K (Figure 3).24 As per example, Fox and co-workers
`used 1 for analysis of kinetic mechanisms and ATP-binding site
`reactivity toward p38γ and found it to bind irreversibly to the
`unphosphorylated and active form.23 Digest maps showed that
`Lys56 of p38γ was selectively and covalently modified by 1.
`The use of 1 has been reported in more than 200 publications,
`underscoring the usefulness of this particular chemical probe.
`
`Bell and co-workers discovered a pyrrole-5-carboxaldehyde
`class of type 1 insulin-like growth factor receptor (IGF-1R)
`inhibitors with modest activity but good selectivity over a small
`panel of other kinases.25 These aldehyde-based inhibitors were
`shown to bind in the ATP-binding pocket of IGF-1R in a
`covalent but reversible manner via imine formation between
`Lys1003 and the aldehyde functionality. Reduction of the imine
`adduct with sodium borohydride irreversibly inactivated the
`kinase, enabling a tryptic digest and X-ray crystallography
`examination. A representative example is inhibitor 2, which
`demonstrated an IC50 of 0.49 μM for IGF-1R.
`A frequently studied lipid kinase inhibitor is represented by
`antiproliferative agent wortmannin (3, Figure 3). This PI3K
`inhibitor modifies the ATP pocket covalently and irreversibly
`via trapping of conserved Lys802 (p110α numbering).26
`Analogues of 3 inhibit PI3K (reported IC50 values in the low
`nanomolar range) by virtue of an activated furan ring as a fairly
`unusual electrophile, which opens upon nucleophilic attack of
`the side chain of Lys802.27 SAR studies on analogues of 3
`demonstrated that omission (or modification) of this furan ring
`results in significant reduction in inhibitory activity. An X-ray
`crystallographic structure revealed the exact nature of
`the
`binding mode of 3, and clear density was observed for the
`primary amine linking to the opened furane ring.28 Later on, 3
`was also shown to inhibit the protein kinase Plk1 (IC50 = 24
`nM), shedding some doubt on the pharmacology that was
`previously attributed to PI3K inhibition.29 Subsequently, it was
`
`6246
`
`dx.doi.org/10.1021/jm3003203 | J. Med. Chem. 2012, 55, 6243−6262
`
`SANDOZ INC.
`
`IPR2023-00478
`
`Ex. 1030, p. 4 of 20
`
`
`
`Journal of Medicinal Chemistry
`
`established that 3 also inhibits Plk3 by covalent modification of
`the conserved lysine in the ATP binding pocket.30 In general,
`deliberately designed covalent drugs that target the invariant
`lysine in kinases are not to be expected. However, uniquely
`positioned lysines elsewhere in or near the ATP binding pocket
`may well be suitable for a covalent inhibitor approach.
`Cysteine-Trapping Covalent Kinase Inhibitors. Unlike
`cysteines
`in kinases, cysteines
`involved in the catalytic
`for instance, protein tyrosine phosphatases31
`machinery in,
`and cysteine proteases32 are part of very specific environments
`made up of neighboring amino acids and ternary architecture of
`these particular proteins. Part of the active site architecture is
`the position of
`these cysteines, residing frequently at
`the
`“bottom” of one or more α-helixes. Backbone NH groups point
`downward generating a net positive charge that stabilizes the
`sulfhydryl anion and augments nucleophilicity.33 The pKa of a
`thiol in a typical cysteine residue is ∼8.5, whereas the presence
`of a polar or positively charged environment can decrease the
`pKa up to 5 log units. In many protein families, cysteines
`contribute to maintaining ternary protein structures via
`disulfide bond formation.
`In the ATP-binding pocket of
`kinases,
`reduced cysteines generally do not seem overly
`nucleophilic and the side chain functionality may have a
`rudimentary role at most. There are, however, several recent
`reports that suggest that reaction of a cysteine thiol can regulate
`and control even protein kinase function,
`for instance, by
`means of redox signaling34,35 or S-alkylation with endogenous
`electrophiles.36,37 For this very same reason, reducing agents
`such as dithiothreitol (DTT) or dithioerythritol (DTE) are
`indispensable ingredients in biochemical assays for kinases, in
`order to keep the enzyme in an active and sensitive state. This
`also supports the notion that these cysteines are sensitive
`enough for electrophilic modification by designer drugs.
`Recently, Zhang et al. have suggested a very useful
`classification system that originally binned the cysteines into
`four groups according to the relative positions in the ATP
`binding pocket (Figure 4).38 It is justified to add at least two
`additional groups to this initial classification,
`involving the
`cysteines located at the gatekeeper position and a fifth group.
`Group 5 covers cysteines residing at various locations in the
`activation loop. Some cysteines are more unique than others,
`and there are at least 200 kinases that seem tractable for a
`covalent approach. Given the known conformational flexibility
`of kinases, this number may be even higher. Structural analyses
`of the different conformations, including the active kinase state
`and inactive “DFG-out” and “C-helix-out” conformations,
`unveil additional cysteines that could be targeted.39 These
`cumulative analyses generate a fingerprint of the kinome active
`site toward cysteines, and the resulting “cysteinome” is of great
`value for the design of covalent kinase inhibitors. There are
`now several examples of covalent kinase inhibitors that target
`this region, and in the following sections the several groups are
`reviewed. The order of appearance is dictated by the group
`numbering as indicated in Figure 4.
`Group 1 Cysteines. So far, there is only one report that
`describes
`the identification of covalent kinase inhibitors
`targeting a cysteine on the lip of the glycine-rich loop (Figure
`4). An irreversible group 1B inhibitor
`is represented by
`fibroblast growth factor receptor (FGFR) inhibitor 4 (FIIN-1,
`Figure 5), which was identified via a structure-based approach
`using available reversible FGFR inhibitors.40 The FGFR family
`consists of four members, FGFR1−4 and plays an important
`role in tumor formation and progression as well as wound
`
`Perspective
`
`Figure 5. (Top) Chemical structure of FGFR inhibitor 4. (Bottom)
`The human kinome and the group 1B cysteines. Adapted illustration
`reproduced with permission from AAAS (Manning, G.; Whyte, D. B.;
`Martinez, R.; Hunter, T.; Sudarsanam, S. The protein kinase
`complement of the human genome. Science 2002, 298, 1912−1934)
`and courtesy of Cell Signaling Technology, Inc. (www.cellsignal.
`com).42
`
`healing.41 Compound 4 is a biochemically potent acrylamide-
`based inhibitor of FGFR1, FGFR2, and FGFR3 (IC50 values of
`9.2, 6.2, and 11.9 nM, respectively) and is somewhat less potent
`on FGFR4 (IC50 = 189 nM), with decent selectivity over the
`other group 1B cysteine kinases, such as Src, YES, and TNK1
`(Figure 5). The reversible propionyl analogue of 4 was 24- to
`100-fold less potent in FGFR1 and FGFR3 harboring cells,
`respectively. With the aid of a biotinylated version of 4 (Figure
`17), specific covalent wild type (WT) FGFR1 labeling was
`demonstrated in transfected HEK293 cells whereas the C486S
`mutant was barely modified. This is in line with the binding
`model that strongly suggests that 4 forms a covalent bond with
`FGFR1 via Cys486.
`Group 2 Cysteines. Within this cluster, subgroup 2B
`comprises only 11 kinase members and has been the sole
`subject of proven covalent kinase targeting within the group 2
`cysteines. Cohen et al. were the first to exploit this particular
`cysteine as a selectivity filter in addition to the threonine
`gatekeeper residue in p90 ribosomal protein S6 kinase (RSK).43
`The four closely related RSKs (RSK1−4) all have the P-loop
`cysteine, whereas RSK3 has
`the sterically demanding
`
`6247
`
`dx.doi.org/10.1021/jm3003203 | J. Med. Chem. 2012, 55, 6243−6262
`
`SANDOZ INC.
`
`IPR2023-00478
`
`Ex. 1030, p. 5 of 20
`
`
`
`Journal of Medicinal Chemistry
`
`Perspective
`
`Figure 6. (Left) Human kinome and the group 2B cysteines. Adapted illustration reproduced with permission from AAAS (Manning, G.; Whyte, D.
`B.; Martinez, R.; Hunter, T.; Sudarsanam, S. The protein kinase complement of the human genome. Science 2002, 298, 1912−1934) and courtesy of
`Cell Signaling Technology, Inc. (www.cellsignal.com).42 (Right) Chemical structures of group 2B inhibitor 5−8 and group 2C kinase inhibitor 9.
`
`methionine as a gatekeeper as opposed to threonine in the
`other three RSK kinase members. This was sufficient to develop
`a fluoromethylketone (FMK) derivative 5 as a highly potent
`and selective RSK1 and RSK2 inhibitor, with an IC50 value
`against RSK2 of 15 nM (Figure 6). Selective covalent labeling
`of WT (but not the C436V mutant) RSK2 was demonstrated
`by FMK-biotin in transfected HEK293 cells. Also, Michael
`acceptors such as an acrylate or acrylonitrile were appended
`from the same pyrrolopyrimidine scaffold, yielding irreversible
`RSK2 inhibitors with modest biochemical IC50 values.44 By
`contrast, the doubly activated electrophiles, as in cyanoacryla-
`mide 6, were 200-fold more potent (IC50 = 4 nM) in enzymatic
`assays than the monoactivated vinyl groups, yet in a fully
`reversible covalent binding mode. Owing to its greater
`reactivity, 6 demonstrated faster kinetics than 5 in cellular
`settings, but the duration of occupancy seemed to rely on
`protein resynthesis and was indistinguishable between the two
`inhibitors. Intriguingly, all RSK kinases also accommodate
`another cysteine which is located in the back of the ATP
`binding pocket, equivalent to Cys560 in RSK2 (group 4).45
`Potential trapping of derivatives of 5 by this cysteine was not
`investigated.
`NEK2 is a centrosomal kinase that interferes with the spindle
`assembly checkpoint, affecting mitosis. It harbors a cysteine
`(Cys22) in an identical position as the RSK family of kinases.46
`Henise and Taunton developed a reasonably potent (IC50 =
`770 nM) and cell permeable irreversible NEK2 inhibitor (7)
`based on an oxindole scaffold using a propynamide as the
`warhead. Selective covalent NEK2 inhibitors, such as 7, are
`important validation assets to investigate the consequences of
`intracellular inactivation of NEK2 on regulating the prolifer-
`ation of tumor cells overexpressing this kinase.
`The same specific cysteine on the P-loop of MEKK1 was
`previously identified as a glutathionylation site following
`
`oxidative stress.47 Bioactive nutrients harbor electrophilic
`isothiocyanates
`that
`intervene with stress-signaling kinase
`pathways. Reported activities include cell cycle effects and
`protection of experimental carcinogenesis in animal models.
`The SAPK/JNK signaling pathway was investigated by means
`of phenylethyl isothiocyanate (PEITC) 8, which was shown to
`inhibit WT MEKK1 (group 2B) but failed to intervene with the
`C1238V mutant in both biochemical and cellular assays.48 It
`was the second covalent kinase inhibitor reported for this
`subgroup, and although far from being druglike, it demonstrates
`the potential for covalent kinase inhibitors as research tools in
`the exploratory biology of kinase signaling pathways.
`One group 2C suspect has been reported, which is a p38α
`inhibitor with moderate biochemical potency (IC50 = 200 nM).
`Dialkynylimidazole derivative 9 did covalently modify p38α
`following prolonged incubation and as determined by ESI-
`MS.49 The authors propose trapping via an intricate cyclization
`and rearrangement mechanism involving both ethynyl groups
`of 9, but the exact mode of action and site of covalent binding
`were not confirmed. Cys39 could be adducted given its position
`in the roof of the ATP pocket of p38α and the known binding
`mode of the imidazole class of p38 inhibitors.50 However, the
`cysteine sulfhydryl group is directed away from the pocket
`itself, and as a consequence, this cysteine is unlikely to be
`involved in adduct formation.
`Group 3 Cysteines. The 3F cluster of the group 3 cysteine-
`containing kinases has received by far the most attention
`because of the clinical potential of kinase inhibitors belonging
`to this subgroup.18 In addition, the relatively high nucleophil-
`icity with respect to the other cysteines facilitates the reaction
`with all sorts of electrophilic agents. The pKa of the sulfhydryl
`group of this particular cysteine is most likely lower than
`average because of the position at the end of a C-lobe α-helix
`(Figure 4). The net positive charge through the backbone NH
`
`6248
`
`dx.doi.org/10.1021/jm3003203 | J. Med. Chem. 2012, 55, 6243−6262
`
`SANDOZ INC.
`
`IPR2023-00478
`
`Ex. 1030, p. 6 of 20
`
`
`
`Journal of Medicinal Chemistry
`
`thus making it more
`groups stabili