Myristoyl’s dual role in allosterically regulating and localizing Abl kinase

c-Abl kinase, a key signaling hub in many biological processes ranging from cell development to proliferation, is tightly regulated by two inhibitory Src homology domains. An N-terminal myristoyl modification can bind to a hydrophobic pocket in the kinase C-lobe, which stabilizes the autoinhibitory assembly. Activation is triggered by myristoyl release. We used molecular dynamics simulations to show how both myristoyl and the Src homology domains are required to impose the full inhibitory effect on the kinase domain and reveal the allosteric transmission pathway at residue-level resolution. Importantly, we find myristoyl insertion into a membrane to thermodynamically compete with binding to c-Abl. Myristoyl thus not only localizes the protein to the cellular membrane, but membrane attachment at the same time enhances activation of c-Abl by stabilizing its preactivated state. Our data put forward a model in which lipidation tightly couples kinase localization and regulation, a scheme that currently appears to be unique for this non-receptor tyrosine kinase.


Introduction
which clamp the kinase in a more rigid state. We note that the differences between the tested To understand how Myr or the SH domains influence Abl dynamics and consequently activity we 120 used force distribution analysis (FDA) to decipher allosteric pathways (Fig. 2). In short, we compute 121 time-averaged forces between any pairs of residues as obtained from equilibrium MD simulations 122 and determine the difference between two conditions. Comparing simulations of the SH3-SH2-123 kinase complex with and without Myr shows high force differences at the active site ( Fig. 2A). They 124 are transmitted via the F helix. Residues G402 (DFG-motif), R381 and E305 ( C helix) are linked by a 125 90 pN cluster, which also includes additional residues from the A-loop, DFG-motif and C helix at 50 126 pN and 70 pN force difference between the Myr bound and unbound state. This likely contributes 127 to C helix outward rotation seen in inactive states of Abl by replacing the E305-K290 interaction Overview of simulated Abl models. Models referred to as 'whole' contain the kinase (KD, dark blue) and SH domains (magenta). The kinase was simulated with a kinked or straight I helix (light blue) conformation. All models were simulated in the absence or presence of Myr (yellow). (B) RMSF of the I helix. The thick and thin blue line at the bottom indicates the position of the folded or unfolded part of the kinked helix, respectively. Orange stars represent the hydrophobic residues at the helix the C-terminal part (C) First principal component. ** p-value vs respective model including Myr < 0.01 (D) RMSD of the kinase domain excluding the I helix. (E) Amplitude of N-C-lobe opening motion (difference between smallest and largest distance between COM of lobes). Centerlines of boxplots denote the mean, box edges the upper and lower quartile. Whiskers represent 1.5 × inter-quartile range. ×× p-value vs whole + Myr < 0.01 beyond locally controlling the conformation of the I helix. However, the SH domains, by locking the kinase domain, are required for the inhibitory effect of Myr to be properly transmitted to the 143 active site. 144 We then went on to focus on the allosteric impact of SH-domain binding to the kinase both in 145 the presence or absence of Myr. In simulations with Myr, medium force differences spread over the 146 whole kinase domain (Fig. 2C), while the highest differences are located at the hinge between the 147 N-and C-lobe and extend to the active site. They connect residues between the A-loop and C helix.

148
In detail, the cluster includes all three residues from the DFG motif (D400, F401, G402) at 70 pN 149 force difference and at 90 pN links R405 and R381 to E305 and E311 from the C helix to maintain 150 its outward rotation. In Abl's active conformation, R381 forms hydrogen bonds to R405 and pY412 151 to stabilize the extended conformation of the A-loop (Xie et al., 2020). By turning its sidechain 152 towards E305 this interaction is disrupted. Furthermore, the rotation of R381 might contribute 153 to the DFG-flip towards the inactivating out conformation and accompanying disruption of the R-154 spine, which consists of residues M309, L320, H380 and F401 (Azam et al., 2008). Levinson et al. 155 (2006) have proposed the DFG-flip to be facilitated by C helix outward movement since it creates 156 space for rotation of F401. Rotation of R381 additionally pulls the sidechain of the adjacent H380 157 away from F401, which might aid its rotation towards the active site. In the absence of Myr, we still 158 observe force differences spread over the kinase-SH-domain interface (Fig. 2D) It is commonly thought that Myr unbinding induces I helix straightening as the next step in the al-169 losteric activation pathway of Abl (Nagar et al., 2003). This is supported by the fact that the straight 170 helix is incompatible with kinase inhibition by SH2 domain binding to the C-lobe of the kinase as 171 observed in crystal structures of Abl (Fig. 3A). In contrast, as mentioned above, we did not observe 172 spontaneous straightening of the I helix in the absence of Myr or the SH-domains. However, the 173 folding process might happen on a timescale longer than accessible with conventional MD simu-174 lations. To accelerate the process and asking how I helix straightening impacts SH2 binding, we 175 turned to Metadynamics simulations. We enhanced the sampling of conformations with high he-176 licity (Pietrucci and Laio, 2009) of residues 515-521, which correspond to the region interrupting 177 the helical fold of the I helix in its kinked conformation. 178 We observed that straightening of the helix is possible in presence of the SH domains, with Abl kinase by Umbrella sampling for comparison is less straightforward due to the involved con-212 formational change of the I helix. In equilibrium MD simulations, we observed the kinked I helix 213 conformation to be destabilized if Myr is absent (Fig. 1B). X-ray structures with an empty Myr pocket 214 support this, and suggest helicity to prevail until residue ∼519, while more C-terminal residues are 215 largely disordered (Appendix Fig. 2). To determine the free energies for Myr unbinding from Abl 216 kinase that reflect this conformational propensity, we carried out Umbrella Sampling simulations 217 using an Abl model with a partially unfolded I helix obtained during Metadynamics simulations.

218
In addition, as extreme cases, a truncated version in which residues 518-531 are absent, and a 219 straightened helix conformation were used. For all of these models, the free energy of Myr unbind-220 ing from the protein is similar or smaller compared to unbinding from the membrane, supporting 221 the notion that Myr binding to the membrane is thermodynamically possible. This result is also 222 supported by previous measurements. The dissociation constant K D for binding of a myristoylated 223 peptide from Abl's kinase domain has been determined to be 2.3 M (Hantschel et al., 2003), cor-224 responding to a binding free energy of 32.4 kJ/mol (Table 1), in good agreement with the calculated 225 free energies using Abl models with a truncated or straight I helix conformation. In contrast, the 226 free energy for Myr unbinding from Abl was significantly higher when using an Abl model based 227 on the crystal structure with Myr bound (i.e. with a kinked I helix). This can be explained by the hy-228 drophobic residues lining the Myr binding pocket, which interact with Myr and are solvent exposed 229 after unbinding.

230
Overall, our Umbrella Sampling results indicate that indeed helix rearrangements have to be 231 taken into account to reflect the correct energy of unbinding. We note that these helix rearrange-232 ments of course contribute to the actual free energy of Myr unbinding but have not been covered 233 during the limited timescale of Umbrella Sampling. The comparison to the experimental value, 234 however, suggests that this approximation is feasible. We note that the free energy of Myr unbind-235 ing from a membrane obtained from simulations is higher than the value of 33.5 kJ/mol that has 236 previously been determined experimentally (Table 1)    and the R-spine (Fig. 2). This is in line with and explains at molecular detail that a kinase domain to be stored in a hydrophobic location or it will quickly rebind to its binding pocket and switch Abl 281 back into its inactive state. We here propose the membrane as an anchoring point for Myr, identi-282 fying a new role for Abl localization and regulation (Fig. 5), which is in line with the fact that protein 283 myristoylation is usually involved in membrane recruitment (Resh, 2016 consideration, membrane binding is energetically as favorable or potentially even more favorable 289 than protein binding, a tendency also found when comparing experimental measurements (Table   290 1) (Hantschel et al., 2003;Peitzsch and McLaughlin, 1993). Previous reports that Abl localizes to  (Nagar et al., 2003), which prevents unspecific binding. At the same time, it explains why 300 the membrane proximal fraction of Abl is not decreased by Myr deletion (Hantschel et al., 2003). 301 If Abl gets recruited to regions close to cellular membranes by binding to the cytoskeleton with 302 its C-terminal F-actin binding domain, the free energy difference that has to be overcome reduces 303 from ∼58 kJ/mol to ∼12 kJ/mol (ΔΔG between Myr unbinding from Abl I -kinked and membrane, 304 therefore be considered an extra regulatory layer, which can be carefully balanced by localization 308 cues or membrane composition. In fact, N-cap residues 15-60, which have been shown to be irrel-309 evant for Abl inhibition (Hantschel et al., 2003), encompass a number of basic residues (K24, K28, 310 K29, R33) and could interact with acidic membrane lipids found at focal contacts. Src kinase also 311 features basic residues near its myristoylation site which stabilize the interaction with PIP 2 -rich 312 membranes (Daday et al., 2022). We speculate that a membrane also enhances Myr unbinding 313 from Abl and Myr insertion by using these residues as a guide for Myr.

327
Since we aimed for a more physiological structure we replaced this inhibitor by ATP by doing the the Nosè-Hoover thermostat (Nosé, 1984;Hoover, 1985). We simulated 10 replicates for 500 ns for 353 each Abl model and thermostat. Since the sets of simulations with different thermostats show the 354 same trends (Appendix Fig. 5), we decided to combine their analyses. For integrating the equations 355 of motion, the leap-frog integrator with a time step of 2 fs was used. All bonds involving hydrogen 356 atoms were constrained using the LINCS algorithm (Hess et al., 1997).  were smoothly switched to zero between 1.0 and 1.2 nm using the force-switch method. Long   (Lemkul and Bevan, 2010). We created a triclinic box with at least 2 nm distance between the pro-392 tein and the box boundaries and elongated the box to 12 nm to provided space for pulling out Myr.

393
The dimension for the membrane system were approximately 5 × 5 nm along the membrane plane 394 and extended to 11 nm in the pulling direction vertical to the membrane plane. We used Gromacs Huang J, MacKerell J A D. CHARMM36 all-atom additive protein force field: validation based on comparison to Hub JS, de Groot BL, Van Der Spoel D.
ℎ -A free weighted histogram analysis implementation in-Narayan B, Fathizadeh A, Templeton C, He P, Arasteh S, Elber R, Buchete NV, Levy RM. The transition between active and inactive conformations of Abl kinase studied by rock climbing and Milestoning. Biochim Biophys Appendix 1-figure 2. Overview over I helix lengths in Abl crystal structures published on the protein data bank. The kinked I helix structure is shown for reference in dark red. The two thick line segments indicate the folded part, which is interrupted as shown by the thinner line segment. Light grey horizontal lines represent the position of the hydrophobic residues I521, L525 and V529. Light red lines depict resolved residues. Yellow lines represent residues that were present in the construct used for crystallization, but were not resolved. Vertical black lines indicate that the helices where not folded beyond this residue position, dark grey Xs that helix continuation was blocked by another copy of Abl in the crystal lattice. Blue labels illustrate that the straight helix conformation, especially the hydrophobic residues, where protected by another protein copy in the crystal lattice. It can be seen that the majority of helices that continue folded beyond the first hydrophobic residues are stabilized by another protein copy, while helices, which end much sooner are either blocked or the respective residues where not included in the construct used for crystallization.