Investigation of the Abl regulatory core dynamics by single molecule FRET and solution NMR

The physiological activity of Abelson tyrosine kinase (Abl) is linked to many cellular processes, such as control of cell growth and survival, oxidative stress response, DNA repair and F-actin-dependent mechanisms. Therefore, Abl requires a tight and fine-tuned regulation under healthy conditions, whereas deregulation leads to disease. In particular, the deregulated fusion protein Bcr Abl, which results from an abnormal reciprocal translocation between chromosomes 22 and 9, causes chronic myeloid leukemia (CML). The molecular details of Abl’s normal regulation and of deregulation in Bcr-Abl are not well understood.
Most knowledge on Abl regulation stems from a combination of structural methods and effects of mutations on kinase activity. It seems clear that in the inactive form, Abl is autoinhibited by interactions within its N-terminal half that consists sequentially of the myristoylated N-cap, the SH3 and SH2 domains, and the catalytic kinase domain (KD). In the autoinhibited state, this part of Abl adopts an assembled conformation where the SH3 and SH2 domains dock onto the KD. In contrast in the active conformation, the SH3 and SH2 domains are thought to disassemble from the KD. Very little is known about this active state, since under normal condition its population is very low and hard to detect by biophysical bulk methods. Furthermore, due to their large size and partially dynamic character, no successful structural or biophysical studies have been reported on full-length Abl or Bcr-Abl.
The central aim of this thesis was to obtain information on the conformational exchange between the inactive and active state of Abl by single-molecule FRET (Förster resonance energy transfer) experiments.
These experiments required Abl molecules labeled by FRET donor and acceptor fluorophores. After exploring many approaches, the labeling could be established by introducing the unnatural amino acid propargyl-lysine into Ablcore (consisting of the SH3, SH2 and KD domains) at two positions via amber codon reassignment and a subsequent copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC, ‘click’) reaction with Alexa Fluor 488 and 647 donor and acceptor fluorophores, respectively. Two different FRET pairs were introduced: Ablcore-196*-510* and Ablcore-225*-297*, which were positioned to maximally distinguish between different proposed structural models of the activated state.
The single-molecule FRET experiments carried out on these labeled Ablcore molecules revealed that the apo form is in dynamic equilibrium on the submillisecond time scale between the assembled conformation and a highly dynamic disassembled conformation of about 30 % population. This dynamic opening has not been observed before and presumably presents the first step in the activation mechanism. Adding the allosteric myristoyl-pocket inhibitors asciminib or GNF-5 to apo Ablcore almost completely abolishes this tendency to open, thereby explaining their inhibitory effects. In contrast, addition of type II ATP-site inhibitors disassembles Ablcore to a highly dynamic conformation on the submillisecond timescale, which is brought back to a mainly closed conformation by the further addition of the allosteric inhibitors. Such ternary complexes with type II ATP-site and allosteric myristoyl-pocket inhibitors, however, have still a higher tendency for opening than the apo form.
In addition to these main results, the following side projects provided insights into the effects of various factors on the Ablcore conformational equilibrium by solution NMR:
1) The KD C-terminal αI-helix adopts a bent conformation in the crystal structure of the assembled myristoyl-bound Ablcore, which was proposed to allow docking of the SH2 domain onto the KD. Without myristoyl, the αI-helix is flexible and was proposed to exert an entropic force toward the SH2 domain, thereby assisting Abl activation. We investigated this mechanism by stepwise truncation of the αI helix and subsequent quantification of the Ablcore disassembly and activity. The results show that the length of the αI-helix correlates with the degree of imatinib-induced core opening and the enzymatic activity. This firmly establishes that the αI-helix is an essential part of the disassembly and activation mechanism and that the type II inhibitor-induced opening occurs via an allosteric coupling from αI to the ATP site.
2) The influence of the Abl N-cap and its myristoylation on the core conformation had not been investigated in detail before in solution. Using isotope-labeled, N terminally myristoylated Ablcap comprising the N-cap, SH3, SH2, and KD (Abl residues 2-531) expressed in insect cells, we showed by NMR that the myristoylated N-cap additionally stabilizes the assembled conformation in solution. Intriguingly, the myristoyl pocket in the kinase domain is not permanently occupied by the myristoyl, but is in exchange with an empty state. Furthermore, we also investigated the non-myristoylated Ablcap G2A mutant, which also adopted mainly the assembled conformation.
3) We observed an interaction of the isolated C-terminal F-actin binding domain with the disassembled Ablcore complex. This pushes the equilibrium of the Ablcore to a more open state, which is also observed when the interaction between the SH3 domain and the SH2-KD linker is disturbed by point mutations. Thus, the interaction of the FABD with the Ablcore presumably involves this interface. This interaction, however, is not strong enough to open the assembled core.
4) We observed a disassembly of the apo form under high (~750 bar) pressure, which is probably due to empty cavities between the individual domains visible in the crystal structure of the assembled core. This pressure-induced conformational switch between the inactive assembled and active disassembled conformation shows that these interdomain cavities are important for Abl regulation.
In summary, we have established single-molecule FRET experiments on Ablcore by using unnatural amino acids and bioorthogonal labeling chemistry and further investigated the influence of several factors on Abl’s conformational equilibrium by solution NMR. The results give many new insights into the dynamic opening of the Abl regulatory core, which is the essential step in Abl’s activation.
The established fluorophore labeling technology will allow to study Abl’s dynamic conformational equilibria at the single-molecule level under many conditions. In particular, it should be possible to follow the structural transitions of not only Ablcore, but also of full length Abl and Bcr-Abl both in vitro and in living cells. Such experiments should provide a much deeper understanding of the molecular mechanisms causing CML and possibly lead to new therapeutic approaches.