Dual drug targeting of mutant Bcr-Abl induces inactive conformation: New strategy for the treatment of chronic myeloid leukemia and overcoming monotherapy resistance
Abstract
Bcr-Abl is an oncogenic fusion protein which expression enhances tumorigenesis, and has been highly associated with chronic myeloid leukemia (CML). Acquired drug resistance in mutant Bcr-Abl has enhanced pathogenesis with the use of single therapy agents such as Nilotinib. Moreover, allosteric targeting has been identified to consequentially inhibit Bcr-Abl activity, which led to the recent development of ABL-001 (asciminib) that selectively binds the myristoyl pocket. Experimental studies have revealed that the combination of Nilotinib and ABL-001 induced a “bent” conformation in the C-terminal helix of Bcr-Abl; a benchmark of inhibition, thereby exhibiting a greater potency in the treatment of CML, surmounting the setbacks of drug resistance, disease regression and relapse. Therefore, we report the first account of the dynamics and conformational analysis of oncogenic T334I Bcr-Abl by dual targeting. Our findings revealed that unlike in the Bcr-Abl-Nilotinib complex, dual targeting by both inhibitors induced the bent conformation in the C-terminal helix that varied with time. This was coupled with significant alteration in Bcr-Abl stability, flexibility and compactness and an overall structural re-orientation inwards towards the hydrophobic core, which reduced the solvent-exposed residues indicative of protein folding. This study = will facilitate allosteric targeting and the design of more potent allosteric inhibitors for resistive target proteins in cancer.
Keywords: Bcr-Abl; oncogenic; catalytic; allosteric; myristoyl; C-terminal helix
1. Introduction
The success of targeted therapy with kinase inhibitors has been best shown in Bcr-Abl, wherein a drug such as imatinib selectively diminishes kinase activity with significant pharmacological effect in the treatment of chronic myelogenous leukemia (CML) (Hassan et al. 2010). However, the potency of kinase inhibitors in cancer therapeutics has been invariably impeded by the emergence of acquired resistance, thereby presenting one of the major challenges to the effective deployment of these agents in the clinic (Hassan et al. 2010). Prior to oncogenesis, activation of Abelson tyrosine kinase (ABL) occurs by the combination of the BCR (breakpoint cluster region) gene with the gene that codes for the intracellular non-receptor tyrosine kinase, which translates into the formation of an enormous protein of approximately 1150 residues. This possesses two distinct halves; the N-terminal half constitutes an N-terminal cap, SH3 and an SH2 domain sequentially while the C-terminal half is made up response elements for varying functionality, which includes domain for DNA and actin binding (Nagar et al. 2003). This is in close proximity to the phosphorylation site in the C-terminal half, and also includes proline-rich segments for nuclear localization and export signaling (Van Etten 1999). The N-terminal cap contains an approximate of 80 residues that play critical role in autoinhibition. This domain is followed by the SH3 domain, SH2 and ultimately a tyrosine kinase domain (Nagar et al. 2003). The interaction between the SH3 and the SH2 domains facilitates autoinhibition, which involves the binding of the SH3 domain to the linker sequence (polyproline) between the SH2 and kinase domains. This is followed by the interaction between the SH2 domain and the kinase domain’s C-terminal lobe to form a clamp structure (Hantschel et al. 2003; Nagar et al. 2006). The binding of the SH3 domains to peptides to form polyproline type II helices while on the other hand, SH2 domains interact with and bind to peptides that contain phosphotyrosine. In its normal state, the c-Abl is regulated with very low kinase activity, while in diseased states such as in chronic myeloid leukemia (CML), it is upregulated by a complex intramolecular interplay between the SH2, SH3, and the kinase domain (Colicelli 2010; Hantschel and Superti-Furga 2004). In this case, oncogenic activation is driven by the fusion of Bcr to c-Abl at its NH2 terminus (Hantschel et al. 2003; Panjarian et al. 2013). The linker region between SH2 and catalytic domains form a polyproline type II helix, which binds the SH3 domain, bringing it into close proximity with the catalytic domain (Hantschel and Superti-Furga 2004) (Figure 1).
Figure 1: Structural highlight of Bcr-Abl domains showing the SH2, SH3, kinase domains and the C-terminal helix.
Regulation of c-Abl occurs via an autoinhibitory mechanism, which has been elucidated to occur through the binding of myristoyl to the myristoyl pocket, an allosteric site situated on c-Abl. This results in its conformational reorientation to an inactive auto-inhibited kinase state evidenced by the bending of the C-terminal helix (Greuber et al. 2013). However, Bcr-Abl is not myristoylated due to the absence of the first Abl exon and the disruption of its regulatory mechanism by pertinent auto- phosphorylation results into uncontrolled oncogenic activity as seen in chronic myeloid leukemia (Greuber et al. 2013). Inactivation of c-Abl has been the basis of CML therapy and this has resulted into the design and use of imatinib, a target specific ATP-competitive inhibitor which functions by blocking the binding of ATP to the catalytic site, leading to the inactivity of Bcr-Abl and eventual amelioration of CML and improves overall survival of patients (Hochhaus et al. 2009). However, point mutations, which occur at this domain, has accounted for acquired resistance and reduced sensitivity to imatinib therapy. Advancement in therapeutic intervention led to the development of more potent kinase inhibitors such as nilotinib and dasatinib to treat CML associated with Imatinib resistance (Shah et al. 2004; Weisberg et al. 2005). Nilotinib (Figure 2A) shares more similarity with imatinib in its mode of binding and target specificity coupled with its physicochemical properties (Kantarjian et al. 2006; Vajpai et al. 2008). Although not modified by myristoylation, the presence of the pocket for myristate binding on Bcr-Abl presented additional target site for therapy, which led to the design of site-specific molecules to mimic the binding of myristate and facilitate auto-inhibitory regulation (Hantschel et al. 2003). These classes of inhibitors are ATP non-competitive but rather bind to the myristoyl pocket of the ABL kinase domain (Zhang et al. 2010), thereby decreasing Bcr- Abl aberrant kinase activity. Such inhibitors include GNF-2, GNF-5 and most recently, ABL-001 (asciminib), which possesses more potency and selectivity among this class of allosteric inhibitors. ABL-001 (Figure 2B), among others was able to bind specifically at the myristoyl pocket thereby inducing a ‘bend’ in the C-terminal helix, which structurally characterizes auto-inhibitory modulation (Nagar et al. 2003; Wylie et al. 2017).
Figure 2: 2D structure of c-Abl inhibitors: [A] Nilotinib [B] ABL-001
Combination therapy involving both the catalytic and allosteric (myristoyl pocket) inhibitors have been identified has a rationale for improving outcomes of CML treatment and surmounts setbacks associated with lone administration of either inhibitors such as resistance and intolerance (Greuber et al. 2013; Wylie et al. 2017). In other words, allosteric inhibition complements the therapeutic activity of a kinase inhibitor to effect the inactivation of Bcr-Abl oncogenic fusion protein as structurally evidenced by the bending of its C-terminal helix. Activated Bcr-Abl fusion oncoprotein has been described to exhibit flexibility in the linker chains characteristic of an increased kinase activity while binding of inhibitors to its active and allosteric induces rigidity, which is characteristic of inhibited Bcr-Abl (Gray and Fabbro, 2014; Hantschel, 2012) similar to a c-ABL autoinhibited state for reduced kinase activity (Nagar et al. 2006; Panjarian et al. 2013). This is schematically represented below (Figure 3).
Figure 3: Schematic representation of the structural effect of single-agent and dual inhibition on Bcr-Abl protein complex relative to its activation and inactivation conformation. [A] Active flexible and straight C- terminal helix [B] Active flexible and straight C-terminal helix [C] Inactive Rigid and bent C-terminal helix conformation.
There is therefore need to have an insight into the structural basis of dual inhibition in abrogating oncogenic Bcr-Abl kinase activity relative to induction of inactive C-terminal conformation. Therefore, in this study we present the first account of the dynamics and conformational basis of kinase inactivity effected by dual inhibitory mechanisms of ABL001 (allosteric) and Nilotinib (catalytic) (Figure 4).
Figure 4: Target site interactions between inhibitors and site residues showing bond types involved in stabilization at the respective sites. [A] Nilotinib and catalytic site residues [B] ABL-001 and myristoyl-pocket residues
This is to provide further insight into structural basis of inactivation induced by co-administration of allosteric and catalytic inhibitors. These findings would further validate the rationale of combination therapy in the treatment of CML and also aids the structure-based design of other potent allosteric inhibitors to enhance the efficacy of kinase inhibitors.
2. Computational Methodology
2.1 System Preparation
Crystal structure of c-Abl tyrosine kinase was retrieved from RSCB Protein Data Bank (PDBID: 1OPK) (Nagar et al. 2003) and was used to obtain the inactive C-terminal helix protein conformation while missing residues were added using Modeller (Eswar et al. 2007).Further preparation of the retrieved structure involved the removal of co-crystallized substrates that are irrelevant to study. Such include myristoyl acid present in the myristoyl binding pocket (allosteric site). This model was used because it constitute the kinase, SH2 and SH3 domains that characterizes Bcr-Abl oncogenic fusion protein unlike in other structures which were distinctively crystallized. T334I point mutation was carried out at the catalytic site (kinase domain) of the protein complex using UCSF chimera [18] to obtain the Nilotinib-resistant mutant. Both inhibitors; Nilotinib (ID: 644214) and ABL001 (ID: 72165228) were obtained from Pubchem (Kim et al. 2016). Protein- inhibitor docking was carried sequentially at the catalytic and myristic-binding pocket using Autodock Vina (Trott and Olson 2010) to obtain the Bcr-Abl-Nilotinib (BN) and ABL-001-Bcr-Abl- Nilotinib (BAN) complexes; both set up for molecular dynamics simulations. Molecular docking preceded the definition of grid boxes around the target sites (active and allosteric) of Bcr-Abl to which the inhibitors were docked respectively. System visualization and preparation was carried out using UCSF chimera (Pettersen et al. 2004).
2.2 Molecular dynamics (MD) simulations
This was carried out on the systems with the GPU version of AMBER 14 with an integrated SANDER module (Case et al. 2005). This followed a standard simulation protocol, which has been previously adopted in our previous studies (Machaba et al. 2016; Ramharack and Soliman 2017) and enumerated as follows. Parameterization of the inhibitors was carried out using the ANTECHAMBER module wherein atomic partial charges (Gasteiger – gaff, using the bcc charge scheme) were generated (Salomon-Ferrer et al. 2013) while the ff99SB AMBER force field was used to parameterize the proteins. Using the LEAP module, hydrogen atoms were added and system neutralized by the addition of counter ions, thereby generating ligand, protein and complex topologies and parameter files. The systems were explicitly solvated with water using the TIP3P orthorhombic box size of 10Å, which enclosed all atoms of the protein (Case et al. 2005). Prior to running the LEAP module, the pdb4amber script was executed to protonate the histidine residues at a constant pH (cpH). This was employed to automatically modify the protein system for use with tleap (Case et al. 2005). Both complexes were minimized initially for 2000 minimization steps applying a restraint potential of 500kcal/mol and then fully minimized for another 1000 steps of steepest descent without restrain.
This was followed by gradual heating the system with a temperature range of 0-300k for 50ps after which the system was equilibrated for 500ps and the temperature and pressure kept constant at 300k and 1bar respectively using the Berendsen barostat (Berendsen et al. 1984). Without restraints, 200ns MD simulation was initiated using a time step of 1fs and coordinates saved at 1ps interval followed by subsequent analysis of trajectories using the integrated PTRJ and CPPTRAJ module (Roe and Cheatham 2013). Visualization of the complexes and data plots were carried out using the graphical user interface of UCSF chimera (Pettersen et al. 2004) and Microcal Origin analytical software (Seifert 2014).
3 Result and Discussion
3.1 Protein stability
Stability of the protein systems was comparatively determined by measuring the root mean square deviation (RMSD) of the C-α atoms across the 200ns of the MD simulations. Moreover, deviation in protein structure as calculated by the RMSD could correspond to the stability of the protein wherein a high value implies a low stability and a low value could indicate an increase in protein stability. As seen in Figure 5, deviations in both systems varied with time wherein a highly unstable structure characterized mutant Bcr-Abl inhibited by Nilotinib alone.
Figure 5: Comparative C-α RMSD plot of single-agent (Nilotinib) and dual (Nilotinib+ABL-001) targeted Bcr- Abl complex
This could suggest one of the basis of mutation-induced acquired resistance to monotherapy with the kinase inhibitor; Nilotinib. However, dual targeting by Nilotinib and ABL-001 induced a reduction in C-α atoms deviation which could correlate with increase in stability. This could provide insight into the structural attributes of mutant Bcr-Abl targeted with ABL-001 and Nilotinib in the treatment of CML as revealed experimentally (Wylie et al. 2017). Average RMSD values include 3.03Å for BN while the allosteric inhibitor lowered the deviation to 2.33Å in the BAN complex.
3.2 Flexibility and activity interplay
Conformational studies have associated a compact and rigid structure with an inactive oncogenic Bcr- Abl fusion protein while flexibility characterizes its activity and resistance (Nagar et al. 2003; Reddy and Aggarwal 2012). Therefore, in order to structurally determine the effect of dual inhibition on the flexibility protein system, we measured the C-α root mean square fluctuation (RMSF), which is a metrics for estimating flexibility. The result revealed that both inhibitors significantly lowered the overall residual fluctuation indicative of rigidity, which characterizes the inactive form of the protein (Figure 6).
Figure 6: C- α RMSF plot showing the effect of single-agent and dual targeting on residual fluctuation and overall flexibility relative to inactivation. Inset are the SH3 (red), SH2 (gold) and the C-terminal helix (blue).
In other words, combination of Nilotinib and the allosteric inhibitor; ABL-001 reduces flexibility. However, with the catalytic inhibitor, Nilotinib alone, mutant Bcr-Abl maintained high flexibility, which characterizes the active form. This could indicate mutant Bcr-Abl Nilotinib-resistivity and serve as a basis for structural characterization. Reduction in flexibility was revealed as shown in the figure above across the respective domains that constitute Bcr-Abl fusion proteins. The complex with both inhibitors (BAN) has an average RMSF value of 7.68Å that was significantly lower than in BN: 12.08Å.
3.3 Exposed surface area and protein folding
The solvent accessible surface area (SASA) analysis is useful in understanding the folding-unfolding process of a protein in conformational studies as it measures the surface area of the protein exposed to solvent across the duration of the molecular dynamics simulation. Moreover, the process of folding decreases the protein surface exposed to the solvent while an increase in the exposed surface depicts an unfolding process (Lins et al. 2003). Conformational Bcr-Abl studies had also revealed that protein misfolding characterizes its activity while inactive in its folded state (Maru 2012; Panjarian et al. 2013). Our result agrees accordingly as shown in the SASA plot, which reveals that allosteric combined with catalytic targeting reduces the number of solvent exposed surfaces (Figure 7).
Figure 7: Comparative SASA plot showing single and dual inhibition-induced exposure of surfaces
This is indicative of protein unfolding which characterizes inactive Bcr-Abl state. However, there was an increase in the surfaces exposed to solvent in BN. In other words, dual targeting caused an inward re-orientation of residual side chains that were exposed to solvent towards the hydrophobic core of the protein. This could further suggest the potency of dual targeting in the treatment of Nilotinib resistant CML as compared to regimen with Nilotinib alone. Compactness of both systems (singly and dually inhibited) was measured by estimating the radius of gyration (RoG) where a high RoG value depicts a reduction or loss of structural compactness and vice versa (Lobanov, Bogatyreva, and Galzitskaya, 2008). The result is shown in Figure 8 and reveals that there was a loss of compactness in the structure of Bcr-Abl that was dually inhibited with Nilotinib and ABL001 as compared with that which was inhibited with Nilotinib alone. Moreover, the variations in structural compactness among the two systems could be relative to inhibition using single and dual inhibitors respectively as shown in other results.
Figure 8: RoG plot showing the effect of single (Nilotinib) and dual (Nilotinib and ABL001) inhibition on the compactness of Bcr-Abl structure.
3.4 Systemic induction of inactive C-terminal helix conformation
Conformational re-orientation of the C-terminal helix is crucial to the regulation of c-Abl activity and inactivity during its auto-inhibition, a regulatory mechanism in normal conditions. This occurs upon its myristoylation as a result of the binding of myristate to its binding pocket, resulting into a ‘bent’ C- terminal helix conformation, a highlight of an inactive or auto-inhibited c-Abl kinase (Nagar et al. 2003; Wylie et al. 2017). Although Bcr-Abl oncogenic fusion protein is not myristoylated due to the lack of the first Abl exon (Hantschel 2012), it still possess the myristate-binding pocket. This has been the subject of allosteric targeting in order to mimic auto-inhibitory type of regulation as it occurs in non-diseased conditions. Trajectory visualization revealed inhibition by nilotinib alone gradually assumed a straight C-terminal helix conformation (slightly bent at some trajectories as shown in Figure 9) at around 50ns and was steady until the end of the MD simulation. The distinct fluctuations in structure was seen in the RMSF plot. However, the bent C-terminal conformation was induced by dual inhibition with both Nilotinib and ABL-001. Other than inducing a ‘bend’ in the C-terminal helix, it appears that allosteric targeting by ABL-001 locked this conformation across the 200ns of MD simulation. This reveals the mechanistic effect of allosteric targeting in the potency of dual inhibition relative to Bcr-Abl inactivation and ultimately, CML treatment.
Figure 9: Superimposed single (gold) and dual (blue) targeted Bcr-Abl complex. Inset shows the trajectory visualization of the C-terminal helix conformations of the singly inhibited Bcr-Abl (Nilotinib alone – Magenta) and the dually inhibited Bcr-Abl (Nilotinib and ABL001 – Red) across the MD simulation time.
4. Conclusion
Advancement in the treatment of CML has been a subject of research due to the emergence of acquired resistance to kinase inhibitors due to mutations in the kinase domain of the target Bcr-Abl fusion oncoprotein. These inhibitors were designed as competitive inhibitors to prevent the binding of ATP to its kinase domain thereby preventing auto-phosphorylation, which accounts for its dysregulation in diseased conditions. Examples include Imatinib and Nilotinib. Setbacks in CML therapy with kinase inhibitors due to resistance and reduction in drug sensitivity led to allosteric targeting with second site inhibitors at the myristate-binding pocket in order to mimic the normal auto-inhibitory regulatory mechanism. An example of such allosteric inhibitor is ABL001 (Asciminib). Moreover, treatment of CML with the combination of both inhibitors has been shown to improve outcomes with no recurrence after treatment cessation compared to single agent treatment with kinase inhibitors. Since structural re-orientation characterize activation and inactivation of Bcr- Abl most importantly the straight and bent C-terminal helix conformation. In this study, we provided the structural basis of dual inhibitory mechanisms of Nilotinib (kinase inhibitor) and ABL001 (allosteric inhibitor) compared to single agent by Nilotinib alone, to support studies wherein combination therapies where used for improved treatment outcomes in CML. Our findings agreed with earlier experimental studies and revealed that single-agent inhibition with Nilotinib was unable to induce a ‘bend’ in the C-terminal helix but maintained its characteristic active straight conformation. Also, the protein was shown to exhibit high residual fluctuation, flexibility and a low stability that characterizes Bcr-Abl active conformation. However, both inhibitors induced an overall structural re-orientation in active Bcr-Abl, which increased its stability and conferred rigidity; structural attributes of inactive fusion Bcr-Abl oncoproteins. Also, dual inhibition reduced the residual side chains exposed to solvent as revealed by surface analysis indicative of protein folding, while single agent inhibition by Nilotinib had a higher exposed surface indicative of an unfolded or misfolded protein which has been experimentally shown to characterize its activity. Taken together, in agreement with previous experimental studies, therapies with allosteric combined with catalytic inhibitors possess a great propensity for inactivating Bcr-Abl and ultimately improved outcomes in the treatment of CML with acquired resistance to single-agent kinase inhibitors. This study will provide structural insight into the basis of dual inhibitory mechanisms and aid the design and development of more potent allosteric inhibitors to enhance inhibitory activities with kinase inhibitors.