Irreversible and reversible inhibitors

Another option in drug design is the addition of a chemical group enabling a covalent, irreversible interaction with the target. The use of irreversible inhibitors has always been evaluated carefully, as specific and unspecific irreversible reactions might lead to unpredictable side effects. However, the cellular potency of ATP-competitive inhibitors is affected by intracellular ATP levels, so even if kinase binding can be observed in vitro, the inhibitor might not be able to compete with ATP inside the cell²³⁷. Here, irreversible inhibitors offer a solution, as covalent, irreversible inhibition increases residence time at the target protein. Irreversible inhibitors have gained considerable attention as three inhibitors have been approved recently (Afatinib and Osimertinib for EGFR inhibition in

48

NSCLC, Ibrutinib for treatment of mantle cell lymphoma and chronic lymphatic leukemia). In our set, nine inhibitors target the EGF-receptor and two are directed against BTK (Bruton tyrosine kinase). They can bind covalently to a cysteine in the active center of these kinases (C797 in EGFR and C481 in BTK) and are considered more selective than their reversible counterparts. These inhibitors might also bind reversibly to other kinases and are expected to be irreversible for all proteins having a cysteine at this position in the active site²³⁸. Druggable cysteines in kinases have been separated into different groups, depending on their position within the kinase domain^{33, 239}. Kdapp values for irreversible inhibitors lack the influence of kinetic behavior of this binding mode⁶³, therefore their binding affinities need to be evaluated with care. For comparison to reversible inhibitors, Kdapp values were used as the relative ranking of targets did not change, whereas EC50

values were considered for characterization of Pelitinib binding kinetics.

Figure 24 depicts the distribution of targets and their pKdapps of reversible and irreversible EGFR inhibitors after 45 min pre-incubation prior the Kinobeads enrichment. The only two designated BTK-inhibitors in clinical trials are irreversible; therefore, they are also included in this plot. Ibrutinib profiling revealed 23 additional targets, whereas the structurally similar compound Ono-4059 analogue is very selective with only two targets. Since EGFR/HER2 inhibitors represent the majority of this binding type in the panel, 24 EGFR inhibitors will be discussed further. At first sight, there was no obvious difference in proteins binding to the inhibitors. Both binding types can have many or few targets. Notably, the reversible inhibitors AC-480, Lapatinib and TAK-285 showed no additional targets in the profiling using the four cell line mix. 24 EGFR-inhibitors were also profiled in BT-474 lysate, a breast cancer cell line driven by HER2 expression and with high EGFR levels²⁴⁰. In this more disease relevant setting, the inhibitors showed comparable selectivity to the profiles obtained in the four cell line mix. Also, the Kdapps were comparable. However, HER2 could not be identified as target, suggesting a different binding mechanism to the Kinobeads. Here, MAP2K5 and STK17A were identified for AC-480 and TAK-285, respectively, albeit with lower potency. MAP2K5 was identified as low confidence target of both inhibitors in the four cell line mix and thus not considered in the analysis. EGFR was always the most potent target, except for Neratinib and Rociletinib in the cell mix. In the BT-474 cell line, the determined EC50 was slightly better than that of the off targets. This is likely due to higher EGFR expression levels that led to better EGFR quantification and thus better curve fitting. The multi-kinase inhibitors that bind EGFR reversibly showed no trend towards higher Kdapps for EGFR in a systematic manner. The difference in Kdapp

between EGFR and other targets is more than two log10-units for the reversible Sapitinib (and AC-480 in BT-474) and for the irreversible Afatinib, AV-412, Canertinib, Dacomitinib and Poziotinib.

These inhibitors can thus be very selective for EGFR if choosing the right dosage window. Icotinib and Vandetanib have very poor affinity towards EGFR, which could be due to the unique compound structure of Icotinib, whereas Vandetanib is a designated VEGFR-inhibitor with known less potency towards EGFR²⁴¹. GAK (Cyclin G associated kinase, grey dot) was identified as a common off-target of EGFR inhibitors. It was hit by reversible and irreversible inhibitors alike with comparable potency to EGFR for reversible inhibitors. GAK is a negative regulator of EGFR signaling and its down regulation might promote tumorigenesis²⁴². Moreover, GAK inhibition might be one molecular reason for the side effect of pulmonary alveolar dysfunction observed in Gefitinib therapy²⁴³. Therefore, GAK inhibition should be carefully evaluated in drug design and therapy, as this might lead to undesired consequences.

49 Figure 24: Reversible and irreversible EGFR inhibitors. Scatter dot plots for targets of designated reversible and irreversible inhibitors against EGFR or BTK. EGFR inhibitors were also profiled in BT-474 cell lines (bottom panel). EGFR is marked in pink, BTK in blue and TEC in light blue; they have a cysteine in the same position as EGFR. GAK (grey) seems to be a common off-target of EGFR inhibitors. Irreversible and reversible inhibitors do not necessarily differ in their selectivity. Some reversible inhibitors can be quite selective (e.g. AC-480, Lapatinib, TAK-285). The main target of irreversible inhibitors often shows the most potent inhibition and spans a wide affinity range to the next off-target (e.g. Afatinib).

Remarkably, BTK is commonly targeted by all irreversible inhibitors, but not by Afatinib and Dacomitinib. Except for Poziotinib, the designated EGFR inhibitors were more potent towards EGFR.

For designated BTK inhibitors the expected preference for BTK was observed. Interestingly, irreversible inhibitors do not bind to BTK in BT-474 cells. BT-474 cells express a different transcript variant of BTK with an alternative promoter and an additional sequence at the N-terminus²⁴⁴. Tesevatinib, which is the only reversible inhibitor targeting BTK as well, also binds to the kinase in BT-474 lysate. One might speculate that the reactive cysteine in the alternative transcript cannot be reached by irreversible inhibitors anymore, whereas Tesevatinib interacts with different residues in the kinase domain. EGFR, BTK and TEC all have a highly nucleophilic cysteine in the hinge binding region, especially facilitating electrophilic reactions^{33, 239}. For that reason not only BTK, but also TEC-kinase is targeted by this set of irreversible inhibitors and is often amongst the top three hits of an inhibitor.

Afatinib for example has quite a few targets but spans a wide range of Kdapps. The Kdapp for EGFR is around 8 nM (after 45 min preincubation), whereas the next target is only targeted with a potency of around 1 µM. At the respective dosage, Afatinib can thus also be considered a selective inhibitor.

50

Selectivity for the main target is not only dependent on affinity but also on the residence time of the drug at the target. This often results in the application of lower dosing concentrations, with which these drugs can be applied.

The irreversible inhibitor Pelitinib was investigated further concerning irreversibly and reversibly bound off-targets. Therefore, incubation times of the drug with the lysate were extended to 6 h, 14 h and 20 h prior to the 30 min Kinobeads incubation. Figure 25 shows the inhibition curves for selected targets of Pelitinib and their time dependence. EGFR as main target shows an EC50 of 22 nM after 45 min, which improves to 6 nM and 5 nM for the 6-14 h and 20 h time points. The same behavior can be observed for BTK, with an EC50 of 7551 nM after 45 min and 212 nM after 20 h.

GAK shows the opposite behavior, with 57 nM at 45 min increasing to an EC50 of 562 nM after 20 h incubation. Therefore, the obtained EC50 values in the screen should be compared carefully, as irreversible inhibitors show better potency over time confirming the irreversible binding behavior²³⁹. This was especially severe for BTK, where no real curve fit and EC50 determination was possible for the earliest time point.

Figure 25: Time dependence of obtained EC50 values for the irreversible EGFR inhibitor Pelitinib. Irreversible binding mechanisms, e.g. EGFR and BTK, improve in their EC50 upon longer incubation, whereas the EC50 of targets bound reversibly (GAK, WEE1, MAP4K5, MAP3K1) is not affected or decreases over time.

The other targets of Pelitinib do not have a cysteine at the respective position of EGFR²³⁸. They are therefore likely binding reversibly and this binding should therefore not be influenced by incubation time. The kinase WEE1 shows similar binding curves for 45 min, 6 h and 20 h (EC50 between 174 nM to 470 nM). MAP4K5 and MAP3K1 have similar EC50s for the two earlier time points (~40 nM and

51

~140 nM, respectively), which increase for the two latter time points (~130 nM and 1091/475 nM).

Members of the SRC-family of kinases as well as the non-kinase protein FECH (see 3.5), showed binding at higher doses for the 45 min time point but were not competed after 20 h. Therefore, targets with no cysteine in the active center (MAP4K5 and MAP3K1) or at a distinct position from the EGFR/BTK cysteine group (e.g. WEE1, GAK) showed no difference in EC50 or decreased binding affinities over time²³⁸. The shift towards less potent EC50s for reversible targets at longer time points might be an artefact of the experiment itself. One potential explanation is given by potential side reactions of the irreversible inhibitor with DTT, diminishing the effective concentration of free inhibitor present in the lysate. Experiments with lower DTT concentrations, however, did not show severe differences (data not shown). Moreover, with longer incubation times, more drug will be bound irreversibly by the main target. Unspecific binding to cysteines in other proteins might also occur. This might affect the pool of free inhibitor as well.

In conclusion, selective EGFR inhibitors can be designed that inhibit only their designated target or show increased potency because of longer residence time in the irreversible binding mode.

Irreversible inhibitors can be profiled with Kinobeads, but the kinetic aspect of their mode of action and therefore, influence on the target profile, has to be considered.